Solid State Ionics 183 (2011) 40–47
Contents lists available at ScienceDirect
Solid State Ionics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s s i
Oxido-reduction properties of La0.7Sr0.3Co0.8Fe0.2O3−δ perovskite oxide catalyst E. Siebert a,⁎, C. Roux a, A. Boréave b, F. Gaillard b, P. Vernoux b a Laboratoire d'Electrochimie et de Physico-chimie des Matériaux et des Interfaces, LEPMI (UMR5631 CNRS-groupe Grenoble INP), Ecole nationale supérieure de physique, électronique et matériaux (Grenoble INP-Phelma), Domaine universitaire – 1130 rue de la piscine – BP 75, 38402 Saint Martin d'Hères cedex, France b Université de Lyon, Lyon, F-69003, France; Université Lyon 1, France; CNRS, UMR 5256, Institut de Recherches sur la Catalyse et l'Environnement de Lyon (IRCELYON), Villeurbanne, F-69626, France
a r t i c l e
i n f o
Article history: Received 18 May 2010 Received in revised form 13 October 2010 Accepted 8 November 2010 Available online 14 January 2011 Keywords: LSCF perovskite oxide Cyclic voltammetry Oxygen thermodesorption Hydrogen Temperature programmed reduction
a b s t r a c t The oxido-reduction of La0.7Sr0.3Co0.8Fe0.2O3−δ (LSCF7382) perovskite oxide catalyst was studied using O2 thermodesorption and H2 temperature-programmed reduction techniques as well as cyclic voltammetry performed on LSCF7382 thin film deposited on gadolinia doped ceria solid electrolyte. The O2-TPD profile of the attrited powder used for thin film elaboration evidenced two types of desorbed oxygen; i.e. weakly adsorbed oxygen species and lattice oxygen. The annealing treatment decreased the amount of weakly adsorbed oxygen. H2-TPR and cyclic voltammetry evidenced two reduction processes that retained the perovskite structure in LSCF7382. They were tentatively assigned to the successive reduction of Co4+ to Co3+ and Co2+. The anodic peak observed on the cyclic voltammograms at positive overpotentials could be related to the phenomenom of oxygen back-spillover from the triple phase boundary to the LSCF7382/gas interface. © 2010 Elsevier B.V. All rights reserved.
1. Introduction LaCoO3 substituted with Sr on the La site and Fe on the Co site have been widely investigated for use as cathode material in solid oxide fuel cell operating at low temperature [1] and oxygen permeable membrane [2]. They are also good candidates for catalytic applications, for example in methane, propane [3] and Volatile Organic Compounds (VOCs) [4] combustion. Their good performances are related to their high oxygen ionic conductivity in addition to a high electronic conductivity. This results in an enlargement of the available area for oxygen reduction in electrochemical systems and a high oxygen surface exchange coefficient which play a key role in oxidation catalysis. The paper of Gellings and Bouwmeester [5] gives a survey of the catalytic properties of solid oxides which display oxygen ion or mixed conductivity. Recently, iron- and cobalt- containing perovskites (LSCF) were also proposed as catalyst in electrochemical promotion (EPOC) of propene combustion in air excess [6]. The LSCF catalyst was interfaced with yttria stabilized zirconia (YSZ). The reaction was promoted when a cathodic potential was applied to the catalyst. The promotion was non-Faradaic according to the definition given by Vayenas et al. [28]. Most of these applications involve operation under reducing conditions due either to the nature of the gas phase or to the applied potential. Therefore, it is interesting to characterize the redox properties of those catalyts in order to better understand the relationship between the catalytic properties and the nature of the catalytic active species.
⁎ Corresponding author. Tel.: +33 4 76 82 65 66; fax: +33 4 76 82 67 77. E-mail address:
[email protected] (E. Siebert). 0167-2738/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.ssi.2010.11.012
In the present work, single phase perovskite La0.7Sr0.3Co0.8Fe0.2O3−δ (LSCF7382) which displays large surface area was successfully prepared. This composition was selected because it exhibited a high specific surface area and was interesting for catalytic combustion of toluene in air [13]. The reducibility was first characterized by the conventional oxygen thermodesorption (O2-TPD) and hydrogen temperature programmed reduction (H2-TPR) techniques. Recently, the reducibility of a Co rich LSCF7382 catalyst was characterized by the TPR technique [7]. The perovskite oxide was reduced in two zones: the first in the temperature range 300–400 °C with a rather sharp peak and the second one in the temperature range 550–850 °C with a broad band. Then cyclic voltammetry (CV) was used. This technique has already been reported to be a strong tool to investigate catalytic reaction in heterogeneous catalysis [8]. It has been currently used to characterize the redox properties of perovskite oxide catalysts [8–10]. In those studies, CV was performed at room temperature in aqueous electrolyte. In the present work, CV was carried out under the operating conditions of the catalyst, i.e. in the temperature range 300–500 °C and under controlled oxygen partial pressure. Thin films of LSCF7382 were deposited on Gd doped CeO2 (CGO) solid electrolyte. This electrolyte was preferred to the standard YSZ electrolyte due mainly to the reactivity of LSCF7382 at high temperature with YSZ which results in the formation of insulating compounds at the interface [11]. CV of Fe rich LSCF interfaced to a double layer CGO/YSZ electrolyte has been reported by Bebelis et al. [12] in the temperature range 600–850 °C for investigation of the electrochemical reduction of oxygen in SOFC research. The observed peaks were related to electrochemical redox reactions of the Fe and/or Co ions. In the present work, the aim was to identify the electrochemical processes taking place on LSCF7382 oxide catalyst under cathodic and
E. Siebert et al. / Solid State Ionics 183 (2011) 40–47
41
anodic polarization in order to better understand the catalyst behavior under EPOC and SOFC operating conditions. The second objective was to correlate the CV data with O2-TPD and H2-TPR data in order to clarify the role of the oxygen vacancy, the oxidation state of the Co or Fe ion and the oxygen spillover species. Such correlation was not discussed so far in the literature.
attrition diminished the size of the agglomerates that became more homogeneous in size. A zoom of the SEM picture (not shown in the paper) evidenced that the agglomerates were highly porous and constituted of small particles with an average size of 50 nm.
2. Experimental
Oxygen Temperature Desorption (O2-TPD) and Hydrogen Temperature Programmed Reduction (H2-TPR) experiments were performed on attrited powders without annealing or after calcination at 1000 °C during 1 h. This annealing treatment was performed because it was similar to that used for the thin film deposition procedure. O2-TPD experiments were carried out at atmospheric pressure with 0.1 g of the catalyst in a continuously flowing fixed-bed quartz reactor. They were performed under He (1.8 l h− 1) at a heating rate of 20 °C min− 1. TPR experiments were also carried out in a fixed-bed reactor at atmospheric pressure with 0.05 g of catalyst powder. They were performed under 5% H2/He flow (1.8 l h− 1) at a heating rate of 10 °C min− 1 from room temperature to 800 °C for both TPD and TPR measurements. Preliminary experiments with various masses of solid and gas flows were conducted in order to check that no mass transfer limitation occurred, as reported in the literature for TPD from a packed bed [29]. Gaseous oxygen was previously adsorbed at 600 °C for 1 h under pure oxygen and during the cooling down to room temperature. Gases were analyzed by an INFICON Transpector 2 CIS quadrupole mass spectrometer. Signals at m/e = 2, 18, 28, 32, and + + + + + 44 amu corresponding to H+ 2 , H2O , N2 or CO , O2 , and CO2 ions, respectively, were recorded during the heating ramp. The QMS calibration was performed by using calibrated gas mixtures (Air Liquide ALPHAGAZ 2) and by-passing the reactor. The gas mixtures for TPR experiments were 1% H2/He, 5% H2/He and the zero baseline was checked with pure He (30 cc/min.). QMS calibration for oxygen was performed using the 1% O2/He mixture generally used for TPO and TPD experiments. The instrument response was checked for linearity in our working concentration domain for the molecules we usually analyse (no mass discrimination at the capillary inlet has been observed).
2.1. Catalyst powder preparation LSCF7382 was prepared by a combined citrate-EDTA complexing method as described elsewhere [13]. The X-ray diffraction (XRD) pattern of the as-prepared powder showed that the catalyst had the pure perovskite structure (rhombehedral R-3c symmetry) without any impurity phases. According to the Scherrer equation, the powder crystallite size was of the order of 20 nm. The specific surface area measured by nitrogen adsorption (BET measurement) was equal to 22 m2 g− 1. The as-prepared powder was ground by attrition before use for thin film preparation. A scanning electronic microscope (SEM) picture of the catalyst corresponding either to the as prepared powder and the as prepared powder after attrition is shown in Fig. 1. Those samples were denominated as-prepared and attrited powder, respectively. The as-prepared catalyst powder consisted of agglomerates with sizes in the range of 0.1–100 μm. As expected, grinding by
2.2. TPD and TPR measurements
2.3. Thin film deposition LSCF7382 thin films were deposited on a gadolinia doped ceria (CGO, 10 mol% Gd2O3 in CeO2) pellet of 1.9 mm thick and 16 mm diameter. Deposition was made by spray from a slurry containing: 40% LSCF7382 powder, 30% ethylic alcohol, 27% terpineol, 2% polyvinil-butyral and 1% polyvinyl-pirolidone in weight. The slurry was sprayed on the dense CGO pellet in a circular shape of 14 mm diameter and annealed at 1000 °C, in air, during 1 h. This temperature was chosen to obtain a suitable adherence of the thin film and to minimize the grain growth. Therefore, it corresponded to the middle of the sintering curve measured on the attrited LSCF7382 powder. Indeed, the densification of the attrited powder was found to begin at 800 °C and was almost complete at 1350 °C. The resulting deposit is illustrated on the SEM micrograph in Fig. 2. The cross section picture of the deposit shows that it was homogeneous and uniform over the whole section. The porous layer was approximately 10 μm thick with particle size in the range of 100–150 nm. A grain growth was observed in the film due to the thermal treatment of the deposition procedure. 2.4. Electrochemical cell and measurements
Fig. 1. SEM pictures of the (a) as prepared powder (b) attrited LSCF7382 powder.
The three-electrode electrochemical cell was symmetrical. The working and counter electrodes were made of LSCF7382. They were located face-to-face on the opposite sides of the CGO pellet. The dimensions of the CGO pellet were: 16 mm in diameter and 1.9 mm in thickness. The reference electrode which consisted of an Ag wire was applied on the periphery of the CGO cylinder in the middle of the cylinder height. This geometry ensured a rather uniform current
42
E. Siebert et al. / Solid State Ionics 183 (2011) 40–47 5.0 4.5
m/e = 32 amu / ua
4.0
(a)
3.5 3.0 2.5 2.0
(b)
1.5 1.0 0.5 0.0 100
300
500
700
900
Temperature /°C Fig. 3. O2-TPD profiles recorded for the (a) LSCF7382 attrited powder calcinated at 1000 °C for 1 h in air (b) LSCF7382 attrited powder. Fig. 2. SEM picture of the cross section of the LSCF7382 film deposited on CGO.
ð1Þ
This was done in order to keep the same potential scale. 3. Results 3.1. Catalyst powder characterization Fig. 3 shows the TPD profiles obtained for the attrited LSCF7382 powders as prepared and fired at 1000 °C for 1 h. These TPD profiles for the attrited powders are in full agreement with those reported in the
9
7 6
(a)
m/e = 2 amu
(b)
5
8 7 6
4
5
3
4 3
2
(b) 1
m/e = 18 amu
0 100
1
(a) 200
2
m/e = 18 amu /ua
RT EðV = 1atmÞ = EðV = xatmÞ + Lnx: 4F
literature for substituted perovskites powders [3,30]. A low temperature desorption peak, usually attributed to the desorption of oxygen species weakly adsorbed on the LSCF7382 surface and denoted as α oxygen, was observed at 250 °C and 320 °C for the as-prepared and calcinated powder, respectively. The amounts of the desorbed oxygen relative to this peak were 26 μmol g− 1 and 12 μmol g− 1, respectively. The oxygen vacancies created by the partial substitution of La3+ cations by Sr2+ were suggested to be the adsorption sites [5]. This is in agreement with the shift to higher temperature and the decrease of the oxygen amount observed on the powder following the thermal treatment. Indeed, the annealing treatment decreased the surface area of the catalyst powder. The large width and asymmetry of TPD α-peaks shown in Fig. 3 suggest that they involve several adsorbed oxygen states, as previously reported in the literature [3,31–33]. A recent ITPD study of LSCF7382 [34] evidenced two kinds of oxygen adsorbed states and reported the corresponding apparent activation energy of desorption (Eapp). The first one, for lower desorption temperatures (below 290 °C) exhibited an Eapp distribution from 105 to 125 kJ mol− 1 and the second one, for higher desorption temperatures (between 290 and 400 °C) corresponded to Eapp = 139 ± 5 kJ mol− 1. For higher temperatures in the TPD, a plateaulike profile can be observed up to 800 °C, as already reported by Zhang et al. [26]. It is generally attributed to the desorption of the lattice oxygen and denoted as β oxygen. In agreement with this interpretation, it was found to be independent on the thermal treatment. It is noteworthy that the TPD profile of attrited powder (Fig. 1-b) is similar to the profile observed for LSCF7382 before attrition [35] and therefore it can be concluded that the mechanical treatment does not modify significantly the oxygen desorption properties of the solid. Fig. 4 shows the TPR profiles for both types of LSCF7382 powder. As seen from this figure, the annealing treatment shifts the reduction
m/e = 2 amu /ua
density through the cell thus solving the problem encountered in solid-state electrochemistry with the position of the reference electrode [14]. Gold meshes and wires were used as current collectors. The diameter of the gold mesh was smaller than 16 mm. The gold surface area was equal to 1.13 cm2. Electrical contact between the mesh and the LSCF7382 layer was established by mechanical pressure. The reactor for electrochemical measurements was of singlechamber type where all electrodes were exposed to the same atmosphere. The temperature was measured with a thermocouple placed near the CGO pellet. It was varied between 300 and 500 °C. The gas composition consisted of O2-He mixtures that were adjusted by mass flow controllers (Brooks). The overall gas flow was 50 ml min− 1. For the electrochemical measurements, a potentiostat (SI 1287 Solartron) coupled to a frequency response analyzer (SI 1250 Solartron) was used. Impedance measurements were carried out under open-circuit voltage (OCV) conditions in the frequency range 65 kHz–0.01 Hz with an amplitude of 20 mVrms. The Zview software non-linear fitting program was used to fit the impedance data. The voltammetry measurements were performed with scan rates, v, in the range of 0.5–20 mV s− 1. They were carried out starting either from OCV or from a pre-treatment consisting of a polarization at a constant potential (holding potential). On the voltammograms, the potential, E, corresponded to the applied potential not corrected from the ohmic drop. Unless otherwise specified, this potential was given with respect to the Ag/CGO reference electrode exposed to the surrounding atmosphere containing oxygen. As a consequence, E was expressed in V/x atm O2 where x was the value of the actual oxygen partial pressure. The peak potentials, Ep, were corrected from the ohmic drop, due to the low conductivity of CGO in the temperature range of our experiments, by using the electrolyte resistance deduced from the impedance diagram. When the oxygen partial pressure was varied, Ep was calculated with respect to 1 atm of oxygen according to the Nernst equation expressed as:
300
400
500
600
700
800
0 900
Temperature /°°C Fig. 4. H2-TPR profiles recorded for the (a) LSCF7382 attrited powder calcinated at 1000 °C for 1 h in air (b) LSCF7382 attrited powder.
E. Siebert et al. / Solid State Ionics 183 (2011) 40–47
peaks to higher temperatures. This shift on TPR profiles can be plausibly explained by substantial increase in size of oxide particles following the thermal treatment as well as the decrease of the specific surface area and oxygen vacancies concentration. The TPR profile of both types of catalyst powder shows an initial and non-symmetrical broad peak spanning the temperature range 300–450 °C followed by a narrow peak at 650 °C for the attrited powder and 685 °C for the attrited calcinated powder. This result is in agreement with the TPR profile, reported by Yuan et al. [7] also showing that LSCF6482 was reduced in two zones. We have performed XRD analysis on the attrited LSCF7382 powder after a TPR run in hydrogen (5% H2/He) up to 500 °C or 800 °C. The first TPR run was stopped at 500 °C and only evidences the first complex reduction peak at 400 °C while the second one stopped at 800 °C contains the two reduction peaks centred at 400 and 700 °C. Once cooled down to room temperature under air atmosphere, the two XRD patterns were completely different showing that after the TPR up to 800 °C, the perovskite structure La0.7Sr0.3Co0.8Fe0.2O3 ([01 - 089 – 1265] JCPDS file) was decomposed (Fig. 5). Different phases, not very well crystallised, appeared in its place SrLaFeO4 ([00 - 029 – 1305] JCPDS file); Co ([00 - 015 – 0806] JCPDS file); La(OH)3 ([00 - 036 – 1481] JCPDS file); La2O3 ([00 - 005 – 0602] JCPDS file). In addition, the cobalt seemed to be in a metallic form. On the contrary, after the TPR stopped at 500 °C, the perovskite La0.7Sr0.3Co0.8Fe0.2O3 structure was maintained and no additional phase was evidenced. So far, we may assume that the first reduction step (broad peak at 400 °C) is reversible contrary to the second reduction step (700 °C) which leads to the decomposition of the perovskite structure. Moreover, it can be suggested that the low temperature broad band may be related to at least two reduction processes in LSCF7382. The amount of consumed H2 relative to the low temperature broad band was slightly dependent on the thermal treatment. It was equal to 2.1 μmol g− 1 and 2.4 mmol g− 1 before and after thermal treatment, respectively. This could indicate that the low temperature broad peak may be related to a bulk process. During the TPR experiments, the transition metal ion B reduction from BxB to B′B (one-electron reduction) occurred according to the following reaction: ∘∘
2BB + H2 + Oo →H2 O + VO + 2BB′ x
X
ð2Þ
where BxB and B′B denote different oxidation states of the Co and Fe ions. The amount of consumed H2 (nH2 = 0.12 mmol) was compared to the amount of Co (nCo = 0.18 mmol) and Fe (nFe = 0.045 mmol) ions in the LSCF7382 sample. This value is higher than that required for the oneelectron reduction of Co3+ to Co2+ and Fe4+ to Fe3+ ions in LSCF7382. Huang et al. [15] have performed quantitative TPR and TPD analysis on La0.7Sr0.3Co0.8Fe 0.2O3 SrLaFeO4 Co La(OH)3 La2O3
43
LaCoO3. The first reduction peak of complex shape observed at a maximum of 410 °C was attributed to the one-step reduction process of Co3+ to Co0 atomically located in the perovskite lattice provided that the perovskite structure was retained after reduction. The values of nH2, nCo and nFe previously reported in this study could be in agreement with this interpretation but XRD results did not favor this assumption. Furthermore, J. Niu et al. [16] have proposed another interpretation and attributed the band below 500 °C to the reduction of Co3+ to Co2+ in Sr doped LaCoO3 [16]. In addition, the existence of Co4+ was evidenced by Berny et al. [17] in La 0.5Sr 0.5CoO3. After heating at 520 °C in the reducing H2/N2 atmosphere, this recent study found that all of the Co4+ cations were successively reduced into Co3+ and metallic Co with the concomitant formation of strontium and lanthanum oxides. In the present work, the values of nH2, nCo and nFe as well as the existence of at least two processes in order to explain the low temperature broad band on the TPR profiles are also in favor of the existence of Co4+. The successive reduction of Co4+ to Co3+ and Co2+ could be responsible for the low temperature broad band on the TPR profiles and could explain the preservation of the perovskite structure after a TPR experiment until 550 °C. 3.2. Thin film characterisation To check the reproducibility of the thin film deposition procedure, the impedance was measured on the two identical LSCF7382 films deposited on both sides of CGO. The study was carried out as a function of the temperature for PO2 = 0.1 atm. Fig. 6 shows typical impedance spectra recorded at 408 °C. The low-frequency straight line reflects the impedance of the LSCF7382/CGO interface. A good agreement was observed between the results obtained on each film, denoted as film 1 and film 2, respectively. This was verified in all the temperature range. This indicates that the film deposition procedure was reproducible. The high-frequency axis intercept, RHF, is likely to reflect primarily the resistance associated with the bulk ionic conductivity of CGO electrolyte. This was checked by plotting the Arrhenius diagram of the conductivity calculated from the following equation: σ=
1 l RHF; 1 + RHF; 2 S
ð3Þ
where RHF,1 and RHF,2 were the high frequency intercepts deduced from the impedance spectra of the films 1 and 2, respectively, l was the thickness of the CGO pellet and S was the surface area of the Au grid. The result is given in Fig. 7. For comparison, the theoretical values of the CGO conductivity calculated from the equation of the conductivity of CGO reported by Steele [18] are also incorporated in the figure. The plot in Fig. 6 indicates that for temperatures higher than 400 °C, the values of the conductivity of CGO deduced from 600 0.05 Hz
Film1 Film 2
500
-Im Z /
400
(a)
300 200 65 kHz
100
(b)
0 0 5
15
25
35
45
55
65
100
200
300
400
500
600
700
800
Re Z /
degre 2 Fig. 5. XRD patterns of LSCF7382 powder following TPR treatment at 550 °C (a) and 800 °C (b).
Fig. 6. Typical impedance spectra of two identical LSCF7382 electrodes deposited on both sides of a CGO pellet, measured at 408 °C and PO2 = 0.1 atm (frequency range 65,000–0.05 Hz).
44
E. Siebert et al. / Solid State Ionics 183 (2011) 40–47 1,00E-02
-0,2
307°°C, PO2 = 0.1 atm
-0,25
EPA1 / V
/ S.cm-1
This work Steele
1,00E-03
-0,3
-0,35
-0,4
1,00E-04 1,2
-0,45 1,3
1,4
1,5
1,6
1,7
1,8
-5
-4
1000 / (T(K)) Fig. 7. Arrhenius plot of the conductivity of CGO deduced from the impedance measurements according to Eq. (2). Comparison with the conductivity of CGO calculated from the equation reported by Steele [18].
impedance measurements are very close to those calculated from the equation proposed by Steele [18]. But as the temperature decreases, the conductivities deduced from the high frequency intercept on the impedance diagram are lower than expected theoretically. This observation is consistent with a non-negligible electrical resistance of the LSCF7382 thin film at low temperature. As expected, the lower the temperature, the higher the contribution of the LSCF7382 in plane resistance is. Fig. 8 shows the typical voltammetric response obtained with a scan rate of 1 mV s− 1, at 307 °C under PO2 =0.1 atm. The curve corresponds to the cathodic scan starting from the open circuit voltage. It was not corrected from the ohmic drop. It exhibits two distinct reduction peaks denoted PC1 and PC2. The scan rate, v, was varied from 20 to 0.1 mV s− 1. During the first cathodic scan, both cathodic peaks PC1 and PC2 were observed for scan rates lower and equal to 1 mV s− 1. For scan rates higher than 1 mV s− 1, only one cathodic peak, corresponding to PC1, was observed. This indicates that two reduction processes occurred in LSCF7382 with different reduction rates. During the reverse anodic scan, one main peak, noted PA1, was observed whatever the scan rate. It mainly corresponded to the reoxidation of the species formed during the PC1 process. This was checked by plotting the cyclic voltammograms for various reverse cathodic potentials varying from −0.5 V to −1 V. A shoulder on the left side of the anodic peak was observed for v≤1 mV s− 1. This was attributed to the re-oxidation of the species corresponding to PC2. From the effect of the scan rate on the existence of the redox peaks, it
306°°C, PO2 = 0.1 atm PA1
is believed that the redox process responsible for the PC1/PA1 peaks is more rapid than that occurring in PC2/PA2. The effect of the scan rate on the PC1/PA1 system was studied. The electrode was pre-treated at +0.1 V for 1 min. Then, the cyclic voltammogram was recorded from +0.1 V down to the potential corresponding to the end of PC1 and back to +0.1 V. Increasing scan rate caused a shift toward more cathodic potentials of the peak potential EPC1 and to more anodic potentials of EPA1. Fig. 9 gives the peak potential values, EPA1, corrected for the ohmic drop as a function of v. The transition from an irreversible behavior for which the peak potential is a function of the scan rate to a Nernstian (reversible) system for which the peak potential is scan rate independent is clearly evidenced [22]. At high scan rates a linear correlation, EP vs log(v), is obtained while at low scan rates (v b 0.2 mV s− 1) EP tends to a constant value. From the slope of the linear part, the transfer coefficient α can be determined [19,20]. In case of an irreversible one-step one-electron electrochemical reaction, the slope is given by 1.15 RT/αF. The value of α, calculated from the slope in Fig. 9, was found to be equal to 0.45 which was consistent with a oneelectron transfer step. The peak current, IPA1, was also studied as a function of v. In the case of a one-electron Nernstian or reversible system, governed by semi-infinite linear diffusion [19,20], the peak current is given by: Ip = 0446FCS
1 = 2 F Dv RT
1=2
I/A
Ip = 0:496FCSðαÞ
-1,0E-04
-2,0E-04
-3,0E-04 -1,5
-1
ð4Þ
1 = 2 F Dv : RT
ð5Þ
Plateau
0,0E+00
PC2
-1
Fig. 9. Effect of the scan rate on the anodic peak potential, PA1 (cell 1), electrode pretreatment: 0.1 V for 1 min. The potential was corrected from the ohmic drop using 387 Ω as the electrolyte resistance.
2,0E-04
Shoulder
-2
where C is the concentration and D the diffusion coefficient of the electroactive species. S is the electrode surface area. For an irreversible system [19,20], the peak current is given by:
3,0E-04
1,0E-04
-3
log (v in V/s)
PC1
-0,5
0
0,5
E/V Fig. 8. Cyclic voltammogram of LSCF7382/CGO electrode (cell 2), v = 1 mV s− 1. Cathodic scan starting from OCV.
As a consequence, for a one electron transfer step governed by semiinfinite linear diffusion, the current is expected to be proportional to v1/2 with a decrease in the slope from a Nernstian to an irreversible system. The linear v1/2 dependency of Ip is evidenced in Fig. 10. The residual current at v = 0, on Fig. 10, may be related to the oxygen reduction reaction. Fig. 10 also reports IPA1 values corrected from the residual current, denoted IPA1*. The plot of log(IPA1*/v1/2) vs log(v) shows the transition from a Nernstian to an irreversible behavior as expected from theoretical expressions (4) and (5) when v increases. The oxygen partial pressure was varied from 0.1 to 10− 4 atm. As expected the baseline current on the cathodic scan was found to decrease when PO2 was decreased indicating that the baseline current
E. Siebert et al. / Solid State Ionics 183 (2011) 40–47 0,9
307°°C, PO2 = 0.1 atm
0,5 log (IPA1 / v^1/2)
IPA1 / mA
0,7
0,3
0,1
-0,64
-0,68
-0,72 -5
-4
-3
-2
-1
log (v in V/s)
-0,1 0
1
2
3
4
5
Square root of v / (mV/s)^1/2 Fig. 10. Effect of the scan rate on the anodic peak current, PA1 (cell 1), electrode pretreatment: 0.1 V for 1 min. The label gives the log (IPA1⁎/v1/2) vs log(v) plot where IPA1⁎ = IPA1 − IPA1(v = o). It shows the transition from a reversible to an irreversible electrochemical system.
Table 1 Peak potentials calculated with respect to 1 atm of oxygen for two different PO2 and two different cells. Experimental conditions
PC1 Ep (V/1 atm)
PC2 Ep (V/1 atm)
PA1 Ep (V/1 atm)
10− 4 atm, 307 °C, cell 1 0.1 atm, 307 °C, cell 1 0.1 atm, 306 °C, cell 2
− 0.55 − 0.51 − 0.505
− 0.81 − 0.78 − 0.85
− 0.45 − 0.41 − 0.405
could be related to the oxygen reduction reaction. The peaks PC1, PC2 and PA1 were observed whatever PO2. The peak potentials measured for PC1, PC2 and PA1 at 307 °C and 1 mV s− 1 were calculated with respect to 1 atm O2. The results are assembled in Table 1. They correspond to two different PO2 and two different cells. A good reproducibility was obtained. Two reduction processes occur at 307 °C in LSCF7382 for potentials equal to − 0.5 V/1 atm and −0.8 V/1 atm, respectively. The corresponding re-oxidation processes take place at potentials close to each other near − 0.4 V/1 atm. The charges corresponding to the reduction peaks PC1 and PC2 and the oxidation peak PA1 on the cyclic voltammogram were calculated by integration of the current taking into account the baseline correction. The results are reported in Table 2. The oxidation charge QPA1 exceeded the reduction charge QPC1 meaning that the reduced species of the redox system is still present in LSCF7382. Moreover, anodic pre-treatment increased QPC1. This indicates that the first oxido-reduction peaks may be related to the transition ion B4+/B3+ redox couple according to the electrochemical reaction: ∘ x ∘∘ x ∘∘ x 2BB + 2e′ + OO;LSCF + VO;CGO ⇔2BB + VOLSCF + OO;CGO :
ð6Þ
After anodic pre-treatment, the charge corresponding to the two reduction processes PC1 and PC2 was equal to that corresponding to the complex re-oxidation peak. This suggests the successive reduction of B4+ into B3+ and B2+. Taking into account that the CV experiments are reproducible from one cycle to the other, one may assume that the redox processes do
45
not altered the LSCF7382 structure. Therefore they are related to the first TPR reduction step, and not to the second one. As a consequence the apparent asymmetry of the TPR peak could be related to PC1 and PC2 and may confirm two kinds of oxygen sites. We suggest that those peaks may be related to the successive reduction of Co4+ to Co3+ and Co2+. This assignment is consistent with the variation of the charge in the cyclic voltammogram reported in Table 2 as well as the H2 consumption corresponding to the low temperature broad band on the TPR profile. This is also in agreement with the one-electron transfer step evidenced by cyclic voltammetry. The results reported by Bebelis et al. [12] on LSCF8228 and LSF82 electrodes interfaced to CGO/YSZ also confirm this hypothesis. They observed one broad cathodic peak near −0.64 V/1 atm at 600 °C and 20 mV s− 1 on the Co-containing perovskite (LSCF8228) with the corresponding reoxidation process located at −0.54 V/1 atm. Those peaks are very similar to those observed in the present work. On the contrary, no similar redox peaks were obtained with the iron-containing perovskite LS2F confirming the assumption of Co cation reduction. The differences in the redox potentials between Co and Fe reported for ions in aqueous solution [24] as well as those calculated from thermodynamics data [25] for Co and Fe in binary oxide compounds also favor this interpretation. In both cases, Co is more easily reduced than Fe since its redox potential is higher. Typically, the thermodynamic potentials of B8/3+/B2+ corresponding to the B3O4/BO equilibrium were equal to −1.23 V/1 atm for Fe and −0.42 V/1 atm for Co at 600 K, respectively. Furthermore, the reproducible and reversible redox behavior observed with cyclic voltammetry is in good agreement with the retaining of the perovskite structure after the reduction/reoxidation process. This was confirmed by XRD performed on the LSCF7382 thin film after operation. As expected, the pattern was similar to that given in Fig. 5 for the LSCF7382 perovskite structure. This result is also in agreement with the XRD measurements performed when stopping TPR experiments after the low temperature broad H2 consumption peak, i.e. at 550 °C (Fig. 5). Then, we may conclude that applying a cathodic polarization induces the oxygen reduction reaction in parallel with the creation of oxygen vacancies in LSCF7382 due to the successive reduction of Co4+ to Co3+ and Co2+, retaining the perovskite structure. Those oxygen vacancies may be at the origin of the good properties of this material as cathode of SOFC and its behavior under EPOC conditions. As seen from Fig. 8, a constant current (Iplateau) was obtained in the potential range from −0.2 V to +0.2 V on the cyclic voltammogram. The effect of the scan rate on Iplateau is illustrated in Fig. 11. Iplateau was found to vary linearly with the scan rate. This linear dependency corresponds to a capacitive behavior. The capacitance was evaluated from the slope of the straight line for two different oxygen partial pressures equal to 0.1 and 10− 4 atm, respectively. A value in the range of 0.02–0.03 μF cm− 2, at 306 °C was found whatever PO2. This value is several orders of magnitude higher than the electrostatic double layer capacity. It can be expected by storage of charge under the form of chemical species, called pseudocapacity [21]. The PO2 independency of the capacity as well as the position of the first peak on the O2-TPD profile suggest that weakly adsorbed oxygen species coming from the gas phase cannot be at the origin of this capacity. Oxygen ion spillover species could be responsible for this capacitive behavior as proposed by Roche et al. [22] on the LSM/ YSZ and Jaccoud et al. [23] on the Pt/YSZ systems.
Table 2 Charges corresponding to the reduction peaks and the complex re-oxidation peak, at T = 307 °C and v = 1 mV s− 1. Experimental conditions
Cathodic reverse potential (V/1 atm)
QPC1 (Cb)
QPA1 (Cb)
No-pre-treatment 0.1 atm Anodic pre-treatment, 0.1 atm No-pre-treatment 10− 4 atm Anodic pre-treatment, 10− 4 atm
− 0.78 − 0.78 − 1.11 − 1.11
− 0.018 − 0.025
+ 0.042 + 0.040
QPC1 + QPC2 (Cb)
QPA1 + shoulder (Cb)
− 0.07 − 0.098
+ 0.09 + 0.1
46
E. Siebert et al. / Solid State Ionics 183 (2011) 40–47 0,4
At high anodic potential, the chemisorbed oxygen, O–s, desorbs to the gas phase as molecular oxygen O2 due to oxygen evolution. As a consequence, the size of the reverse cathodic peak, PC3, was found to be smallest as shown in Fig. 12. The absence of PC3 on the voltammogram recorded under He indicates that O–s was no longer present on the electrode surface. This is in agreement with the O2-TPD profile showing that desorption of surface oxygen begins at 260 °C on the attrited LSCF7382 powder calcinated at 1000 °C for 1 h in air. In order to verify this point, the cyclic voltammogram was recorded at lower temperature (i.e. 260 °C). As expected, PC3 was present. Therefore, we may conclude that under anodic polarization, oxygen adsorbed species are created on the LSCF7382 surface. Those species desorbs due to the oxygen evolution reaction. This could explain that applying a high anodic polarization had no effect on the propene combustion in excess air as previously evidenced [6].
307°C, PO2 = 0.1 atm
Iplateau / mA
0,3
0,2
0,1
0 0
5
10
15
20
v / mV/s Fig. 11. Effect of the scan rate on Iplateau, (cell 1), electrode pre-treatment: 0.1 V for 1 min.
Fig. 12 shows an anodic scan of cyclic voltammogram, not corrected from the ohmic drop, obtained at 307 °C, under PO2 = 0.1 atm and v = 1 mV s− 1. The low conductivity of CGO was responsible for the straight line during the reverse scan at anodic potential. An anodic peak, noted PA3, was evidenced. The peak potential corrected from the ohmic drop was found to be equal to 0.24 V/1 atm. The corresponding reduction peak of smallest amplitude was located near −0.3 V/1 atm. Changing the oxygen atmosphere from 0.1 atm to He had no effect on the anodic peak. In the same time, PC3 disappeared. This allows us to suggest that the PA3/PC3 coupled redox peaks may be related to oxygen back-spillover from CGO to the electrode surface according to: −
x
OO; CGO ⇔O−s + 2e
∘∘
+ VO; CGO :
ð7Þ
During this process, an O2−ion reaches the triple phase boundary region (TPB) and then releases two electrons to the electrode forming a chemisorbed oxygen atom on the electrode surface. This process may also be related to the phenomenom of oxygen back-spillover from the TPB which modifies the work function of the electrode. The charge corresponding to PA3, QPA3, was calculated by integration of the current taking into account the baseline correction. QPA3 was found to be in the range of 0.02–0.03 Cb. A rough estimation of the quantity of adsorbed oxygen species, O–s, was made assuming that the specific surface area of LSCF7382 thin film was of the order of 4 m2 g− 1. A value of 10 μmol m− 2 was obtained. This amount is larger than a monolayer of adsorbed oxygen since the typical value for the LSCF perovskite material is reported to be equal to 4 μmol m− 2 [26,27]. 6,0E-04
4,0E-04
2,0E-04
I /A
PA1
0,0E+00
PC3
-2,0E-04 PC1
-4,0E-04 -1,5
-1
-0,5
0
0,5
O2-TPD, H2-TPR experiments performed on LSCF7382 powder as well as cyclic voltammetry carried out on LSCF7382 thin film deposited on CGO were used to study the oxido-reduction behavior of LSCF7382 oxide catalyst. The desorption of two types of oxygen (α and β) was observed as expected from the literature data [5]. The α oxygen was associated with surface oxygen accommodated in the oxygen vacancies of LSCF7382. The β (or lattice) oxygen was connected with the reduction of the B cations to a lower valency. LSCF7382 powder was reduced in two zones: the first in the temperature range 300–500 °C with a broad band composed of at least two processes and the second near 700 °C with a rather sharp peak. The high temperature sharp peak corresponded to the decomposition of LSCF7382. The amount of H2 consumed relative to the low temperature broad band suggested the successive reduction of Co4+ to Co3+ and Co2+. Two oxido-reduction processes were also evidenced by cyclic voltammetry under cathodic scan. They were related to the low temperature broad band observed on the H2-TPR profiles. Indeed, the variation of the peak current and potential as a function of the scan rate evidenced a one-electron transfer step. Moreover, the charge under the peak was also in agreement with the successive reduction of Co4+ to Co3+ and Co2+. The anodic peak evidenced on the cyclic voltammogram at anodic overpotential was related to oxygen back-spillover from the TPB toward the LSCF7382/gas surface, corresponding to more than a monolayer of adsorbed oxygen. Acknowledgements The authors gratefully acknowledge financial support by the French government (ACI Energie, Conception durable 2004). They also thank Dr Ning Li of IRCELYON for the catalyst powder preparation and Marc Henault of LEPMI for making the thin film deposition. References
PA3
306°C, PO2=0.1 atm
4. Conclusion
1
1,5
E/V Fig. 12. Cyclic voltammogram of LSCF7382/CGO electrode (cell 2), v = 1 mV s− 1. Anodic scan starting from OCV.
[1] A. Esquirol, N.P. Brandon, J.A. Kilner, M. Mogensen, J. Electrochem. Soc. 15 (2004) A1847. [2] O. Büchler, J.M. Serra, W.A. Meulenberg, D. Sebold, H.P. Buchkremer, Solid State Ionics 178 (2007) 91. [3] M. Alifanti, J. Kirchnerova, B. Delmon, D. Klvana, Appl. Catal. A 262 (2004) 167. [4] H. Huang, Y. Liu, W. Tang, Y. Chen, Catal. Commun. 9 (2008) 55. [5] P.J. Gellings, H.J.M. Bouwmeester, Catal. Today 12 (1992) 1. [6] F. Gaillard, X. Li, M. Uray, P. Vernoux, Catal. Lett. 96 (2004) 177. [7] W. Yuan, X. Hu, L. Li, J. Nat. Gas Chem. 15 (2006) 58. [8] J. Zhu, Z. Zhao, D. Xiao, J. li, X. Yang, Y. Wu, J. Mol. Catal. Chem. 238 (2005) 35. [9] J. Zhu, Z. Zhao, D. Xiao, J. li, X. Yang, Y. Wu, Electrochem. Commun. 7 (2005) 58. [10] P. Wang, L. Yao, M. Wang, W. Wu, J. Alloy. Comp. 311 (2000) 53. [11] H.Y. Tu, Y. Takeda, N. Imanishi, O. Yamamoto, Solid State Ionics 100 (1997) 283. [12] S. Bebelis, V. Kourmoutis, A. Mai, F. Tietz, Solid State Ionics 179 (2008) 1080. [13] N. Li, A. Boréave, J.P. Deloume, F. Gaillard, Solid State Ionics 179 (2009) 1396. [14] J. Tutman, I. Riess, Solid State Ionics 179 (2008) 108. [15] L. Huang, M. Bassir, S. Kahaguine, Appl. Surf. Sci. 243 (2005) 360. [16] J. Niu, J. Deng, W. Liu, L. Zhang, G. Wang, H. Dai, H. He, X.Z., Catal. Today 126 (2007) 420. [17] F.J. Berny, J.F. Marco, X. Ren, J. Solid State Chem. 178 (2005) 96. [18] B.C.H. Steele, Solid State Ionics 129 (2000) 95.
E. Siebert et al. / Solid State Ionics 183 (2011) 40–47 [19] A.J. Bard, L.R. Faulkner, Electrochimie, Principes, méthodes et applications, Masson, Paris, 1983. [20] J.P. Diard, B. Le Gorrec, C. Montella, Cinétique électrochimique, Hermann, éditeurs des sciences et des arts, Paris, 1996. [21] E. Schouler, M. Kleitz, J. Electroanal. Chem. 64 (1975) 135. [22] V. Roche, J.P. Deloume, T. Pagnier, R. Revel, C. Roux, E. Siebert, P. Vernoux, Catal. Today 146 (2009) 266. [23] A. Jaccoud, C. Falgairette, G. Foti, C. Comninellis, Electrochim. Acta 52 (2007) 7927. [24] Atlas d'équilibres électrochimiques à 25°C, M. Pourbaix, Gauthier-Villars et cie ed, Paris, 1963. [25] J. Eliott, M. Gleiser, Thermochemistry for steelmaking, Addison-Wesley Pub company, inc, Massachusetts, USA, 1960. [26] H.M. Zhang, Y. Shimizu, Y. Teraoka, N. Miura, N. Yamazoe, J. Catal. 121 (1990) 432.
47
[27] S. Royer, F. Bérubé, S. Kaliaguine, Appl. Catal. Gen. 282 (2005) 273. [28] C. Vayenas, S. Bebelis, C. Pliangos, S. Bosda, D. Tsiplakides, Electrochemical activation of catalysis, Kluwer Academic Plenum Publ, New York, 2001. [29] R.A. Demmin, R.J. Gorte, J. Catal. 90 (1984) 32. [30] A. Mai, F. Tietz, D. Stöver, Solid State Ionics 173 (2004) 35. [31] J.P. Joly, D. Klvana, J. Kirchnerova, React. Kinet. Catal. Lett. 68 (1999) 249. [32] S. Royer, F. Bérubé, S. Kaliaguine, Appl. Catal. A Gen. 282 (2005) 273. [33] M. Zawadzki, H. Grabowska, J. Trawczyński, Solid State Ionics 181 (2010) 1131. [34] F. Gaillard, J.-P. Joly, Ning Li, A. Boréave, J.-P. Deloume, Solid State Ionics 179 (2008) 941. [35] F. Gaillard, J.P. Joly, A. Boréave, P. Vernoux, J.-P. Deloume, Appl. Surf. Sci. 253 (2007) 5876.
