Performance Degradation of Lanthanum Strontium Cobaltite after Surface Modiﬁcation

In the present study, the surface of La 0.6 Sr 0.4 CoO 3 (LSC) was modiﬁed via atomic layer deposition (ALD) using CoO x as the modiﬁcation material. The effect of the ALD CoO x treatment on the LSC-based anode-supported solid oxide fuel cell was analyzed and the ALD CoO x -treated cell was found to exhibit lower power density than that of the bare cell. Based on the electrochemical impedance spectroscopy measurements, it was concluded that this degradation stems mainly from the increased polarization loss, which results from the deterioration of the oxygen surface exchange property. Similarly, examining the bode plot revealed that the impedance at frequencies lower than 1 kHz increased mainly after the CoO x treatment; this increased impedance is believed to be associated with the limitation of O 2 adsorption and dissociation. © The

Conventional solid oxide fuel cells (SOFCs) are operated at high temperatures (≥800 • C) in order to ensure rapid kinetics and fast ion transport. 1 However, lower operating temperatures are required to decrease the thermal budget and guarantee long time stability. 2 For this reason, thin film electrolyte SOFC systems such as μ-SOFCs, were proposed and reported to have high power density owing to low ohmic loss. [3][4][5][6][7][8] The performance of SOFCs at low temperatures (<650 • C) is hindered by the sluggish kinetics of the cathode. 9,10 Therefore, developing a cathode which exhibits low polarization loss is essential to further improving this performance. Various mixed ionic electronic conducting (MIEC) ceramics such as La 1-x Sr x CoO 3-δ (LSC), La 1-x Sr x FeO 3-δ (LSF), and La 1-x Sr x Co 1-y Fe y O 3-δ (LSCF) have been proposed as cathode materials for low-temperature operation of the SOFC. [11][12][13] The charge transport at the MIEC cathode consists of five steps: (1) oxygen adsorption and dissociation; (2) oxygen ionization; (3) oxygen ion incorporation; (4) bulk diffusion; (5) oxygen ion transfer at the cathode/electrolyte interface. Among these processes, the surface kinetics, which includes steps (1) to (3), is typically considered as the rate-determining step rather than a stage involving bulk oxygen ion conduction. [14][15][16] Composite cathodes, which are fabricated by mechanically mixing different materials, can lead to improved cathode performance. Although electrolyte materials such as yttria-stabilized zirconia (YSZ) have been widely used as additives in forming the composite cathodes, transition metal oxides such as cobalt oxide can also be added. For example, Chen et al. reported that a composite of Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3-δ and 10 wt% of Co 3 O 4 exhibited better total conductivity and fuel cell performance compared to that of pure Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3-δ . 17 A surface modification technique has recently been introduced as an alternative method of increasing the catalytic activity of SOFC cathodes. In contrast to the composite cathode strategy, surface modification ensures adhesion between the cathode and electrolyte without the use of high sintering temperatures. 18 Surface modification of the backbone cathode with a nanoparticle or thin film has been suggested as one means of enhancing the surface kinetics. Furthermore, the addition of ceramic nanoparticles or a thin film to the SOFC cathode, via infiltration, can decrease the polarization loss and enhance the stability. 19 Both mixed conducting oxides, and even binary oxides with low ionic conductivity and negligible catalytic activity, have been reported to increase the performance of cathodes. 20 24 In addition, although strontium oxide (SrO) is considered to be detrimental to the SOFC cathode owing to Sr segregation phenomena, 25 modification with SrO was reported to lead to a decrease in the polarization resistance. 26 The aforementioned discrepancies can be attributed to infiltration-induced changes in the cathode morphology and inhomogeneous infiltration layers owing to the complex structure of the cathode. 14,23,24 Therefore, it is difficult to determine whether the enhanced cathode performance truly results from the additive.
Atomic layer deposition (ALD) can be used to study the effect of the surface modification of the SOFC cathode on the performance. ALD is a thin film deposition tool, which is based on chemical vapor deposition. In the ALD process, a vaporized precursor adsorbs on the substrate and the unreacted precursor is purged by an inert gas. A supplied oxidant then reacts with the adsorbed precursor and the byproduct is purged by an inert gas. The precursor/purge/oxidant/purge step is defined as one cycle of ALD; this cycle-by-cycle process enables precise control of the film thickness. ALD can be performed at lower deposition temperatures than that of other techniques, and can thus prevent structural change of the backbone electrode during modification. 27 In addition, the self-limiting nature of the ALD ensures that the modification material is uniformly formed on the porous substrate. 28 Yu et al. concluded that ALD-added CeO x and SrO had a detrimental effect on the performance of various cathodes. 27 In other studies, ALD Al 2 O 3 was intentionally applied to cathodes in order to examine the relationship between geometric blocking and the performance. 29,30 Another study suggested that conformal coatings of ALD ZrO 2 film could significantly enhance the stability of LSC cathodes. 31 Although studies of CoO x -treated LSM-and LSCF-based cathodes have been well documented, there is no consensus regarding the performance of LSC cathodes surface treated with CoO x . [21][22][23]32,33 As such, in this work, we studied the effect of ALD cobalt oxide (CoO x ) treatment on an LSC cathode based on an anode-supported SOFC cell. The low deposition temperature used in this study (200 • C) results in minimal 3-dimensional change in the cathode structure during modification. We compared the power performance and electrochemical impedance of the bare and CoO x -treated cells at various temperatures. One of the main findings of this study is that the ALD CoO x treatment results in a decrease in the cell performance owing to the degradation of the cathode surface; this degradation is related to the surface exchange, especially to the limitation of O 2 adsorption and dissociation.

Experimental
A 1 cm × 1 cm anode-supported cell was used for this study. The NiO/stabilized yttria-zirconia (NiO-YSZ) substrate was fabricated by a conventional powder process. The 8 μm-thick NiO-YSZ anode functional layer and 6.5 μm-thick YSZ electrolyte were fabricated by screen-printing. In addition, pulsed layer deposition (PLD) was used to insert a 300-nm-thick gadolinia-doped ceria (GDC) buffer layer between the YSZ and LSC in order to prevent undesirable cathode/electrolyte reactions. 34 The 2.5-μm-thick LSC (La 0.6 Sr 0.4 CoO 3-δ ) thin film cathode was deposited by PLD, as described in a previous study. 35 Figure 1 shows the microstructure of the cell used in this study.
The ALD treatment with CoO x , was performed using bis(1,4-diiso-propyl-1,4-diazabutadiene)cobalt [C 16 H 32 N 4 Co, Co(dpdab) 2 ] (UP Chemical Co.) as a precursor and ozone as an oxidant. Nitrogen gas (purity 99.99%) flowing at a rate of 2.5 sccm was used to purge the unreacted precursor and by-product. Therefore, one cycle of ALD consists of N 2 purge/Co(dpdab) 2 pulse/N 2 purge/O 3 pulse and Co(dpdab) 2 pulse for durations of 30 seconds and 4 seconds, respectively. The ALD process was performed with a thermal ALD system (ICOT Inc.) that is composed of a stainless steel tube type chamber whose vacuum is maintained by using a rotary pump; details of the configuration of the system are provided elsewhere. 36,37 The pressure during deposition was maintained at 0.1 Torr and deposition was performed at 200 • C. In addition, Co(dpdab) 2 was heated to 95 • C and the gas line for ozone supply was maintained at room temperature. Bare and 30cycle CoO x -treated anode-supported SOFC cells were prepared for a cell performance test. To minimize the deposition of CoO x on the electrolyte and anode during the ALD process, the entire cell, except for the cathode, was mechanically masked with a punched polyamide (Kapton) tape. ALD CoO x was also deposited on a single crystalline Si(100) wafer in order to determine the film growth rate and to perform photoelectron spectroscopy (XPS) and X-ray diffraction (XRD) analyses. Prior to deposition, the silicon wafers were cleaned via sonication with acetone, ethanol, and methanol followed by rinsing with deionized water.
The thickness of the ALD film and microstructure of the cathode were analyzed by using a scanning electron microscope (SEM; Hitachi S-4300). Moreover, XPS (ULVAC-PHI X-TOOL) and XRD (Rigaku Model D), respectively, were used to determine the valence state and the crystallinity of the ALD CoO x . The obtained XPS spectrum was calibrated to the carbon line (C1s, 284.6 eV). The corresponding electrochemical impedance spectroscopy (EIS) and currentvoltage (I-V) characteristics were analyzed using an electrochemical analyzer (Gamry Reference 3000 Potentiostat/Galvanostat/ZRA). A customized button cell characterization set-up was used in order to examine the electrochemical cell performance. Au and Ni meshes were used as respective current collectors for the cathode and anode and a high-temperature ceramic paste was painted to the edge of the unit cell to for proper sealing. The cell test was started at a temperature of 600 • C, which was subsequently reduced to 550 and 500 • C. For each temperature, EIS data were obtained at frequencies ranging from 10 5 to 0.2 Hz with ac amplitude of 10 mV and the corresponding current was measured by a voltage sweep from the open circuit voltage (OCV) to 0.2 V. In addition, air (P O 2 = 0.2 atm) and humidified hydrogen were supplied at a rate of 100 sccm to the cathode and anode, respectively.

Results and Discussion
To investigate the valence state of ALD CoO x , we prepared asdeposited and annealed ALD CoO x films and analyzed the XPS spectra of the Co core levels. In the case of the as-deposited film (Figure 2), the Co2p 3/2 and Co2p 1/2 photoelectron peaks occur at 780.6 and 796.1 eV, respectively, with two satellite peaks occurring at slightly higher (i.e., by ∼5-6 eV) binding energies. These satellite peaks indicate that Co 2+ ions are present in the ALD CoO x film. 38,39 Furthermore, annealing at 600 • C for 4 hours in air resulted in respective shifts of the Co2p 3/2 and Co2p 1/2 photoelectron peaks to 779.8 and 795.3 eV and simultaneous weakening of the satellite peaks compared to those in the as-deposited film; this weakening is consistent with the general band profile of Co 3 O 4 . 40,41 The cobalt monoxide (200) peak occurs at 2θ = 42.1 • in the XRD pattern ( Figure 3) of the as-deposited ALD CoO x film and other orientations are either relatively weak or absent. In order to determine the phase of the film at the SOFC cell operating temperature (600 • C), the ALD CoO x film was annealed at 600 • C for 4 hours in air. CoO oxidizes easily in air at temperatures higher than 250 • C and is consequently   A growth rate of 0.6 Å cycle −1 was calculated for the ALD CoO x , which was deposited at 200 • C on the silicon substrate, by dividing the film thickness by the number of ALD cycles. Therefore, 30 cycles of ALD CoO x treatment applied to the LSC cathode should yield ∼1.8 nm of CoO x film. Figure 4 shows the surface microstructure of the LSC cathodes before and after the ALD CoO x treatment; the microstructure remained unchanged with the ALD treatment. Moreover, owing to its low deposition temperature (200 • C), CoO x will not be incorporated into LSC; therefore, the properties of the cathode surface are influenced mainly by the treatment via ALD. 29 I-V measurements of the bare and the ALD CoO x -treated sample were performed at various temperatures and the corresponding power density (mW/cm 2 ) was calculated as the product of the current density and voltage ( Figure 5). The bare (and the ALD CoO x -treated) samples exhibited max power densities of 492 (422), 246 (208), and 107 (91.9) mW/cm 2 at 600, 550, and 500 • C, respectively. As in the case of ALD Al 2 O 3 -treated materials, 30 the lower values of the latter compared to those of the former indicate that modification using ALD CoO x is detrimental to the power performance of the cell. However, almost identical values of 1.13 (1.12), 1.14 (1.13), and 1.15 (1.14) V at the aforementioned respective temperatures, reveal that ALD CoO x has only a slight influence on the OCV.
EIS of the bare and ALD CoO x -treated cell was conducted at 600, 550, and 500 • C at OCV in order to determine the basis for performance degradation after ALD treatment. The corresponding area specific resistance of ohmic contribution (ASR ohm ) of each cell can be inferred from the high-frequency intercept with the real part (x-axis) of the Nyquist plots ( Figure 6). The ALD CoO x treatment resulted in negligible change in the ASR ohm , irrespective of the operating temperature. In addition, the area specific polarization resistance (ASR pol ) can be inferred from the distance between the high-and low-frequency intercept and includes contributions from both the cathode and anode. The ASR pol of the 30-cycle ALD CoO x -treated sample is higher than that of its bare counterpart. This suggests, as was reflected in the I-V results, that the increased polarization loss stems mainly from the degradation of the electrode with the ALD treatment. Specifically, the increased ASR pol owing to the ALD treatment, is believed to result from the degradation of the cathode that has more sluggish reaction kinetics than the anode. The Nyquist plots in Figure 6a, 6b, and 6c were fitted with an equivalent circuit, which consists of one resistance and three parallel RQ circuits in series, as shown in Figure 6d.  Figure 7 shows an Arrhenius plot of the fitted data of the polarization resistance of both samples. By using the temperature dependence of the polarization resistance, we can calculate the activation energy in terms of the ASR pol . As the figure shows, the ALD CoO x -treated cell has a higher (1.21 eV vs. 0.92 eV) activation energy of ASR pol than that of the bare sample. This implies that the former is less suitable for low-temperature operation than the latter. In addition, the difference in activation energy results mainly from the differing properties of the bare and CoO x -treated cathodes and as such, the rate determining step of the cathodes can differ from each other. 35 Bode plots (Figure 8) were used in order to identify the rate determining step for both samples. We confirmed that the surface chemistry is affected by the ALD CoO x modification, although the microstructure remains unchanged. Therefore, the impedance-modified frequency regime shown in the bode plot should be related to changes in the surface property of the cathode. This surface property is associated with the non-charge transfer process, which includes gas transport, adsorption-desorption of oxygen, and surface diffusion of adsorbed oxygen species. [44][45][46] In contrast, the transport of oxygen ions across the electrode/electrolyte interface is considered a charge transfer process. 46 Figure 8 shows that the bare and CoO x -treated samples have the same impedance in the high frequency region (f ≥10 kHz); therefore, it can be concluded that this frequency domain is not associated with the oxygen surface exchange. This is plausible since the high-frequency domain is associated with oxygen ion transport at the cathode/electrolyte interface; this ion transport is referred to as the charge transfer process. 47,48 In this study, the CoO x treatment results in changes in the impedance at frequencies lower than 1 kHz; the modification should therefore, degrade the surface exchange property of the cathode including the non-charge transfer process. 46 The higher activation energy of the ASR pol of the CoO x -treated cell, compared to that of the bare cell (Figure 7), can be attributed to the degradation of surface exchange characteristics owing to modification. Mechanisms of cathode degradation owing to ALD treatment have been proposed in previous studies. 27,29 Küngas et al., concluded that an ALD Al 2 O 3 layer can block the oxygen vacancy site on LSF and the consequent decreased vacancy concentration results in limited O 2 dissociative adsorption. 29 Yu et al. reported a similar mechanism in the case of ALD-deposited SrO, which deactivated the cathode surface by forming blocking layers; however, CeO x prevented the O 2 dissociation process. 27 CoO x treatment via ALD is believed to limit the O 2 adsorption and/or dissociation process. This limitation of adsorption and dissociation is associated with the low-frequency regime (f ≤100 Hz). 27,29 In this study, the CoO x -treated sample exhibits a higher impedance in the low-frequency domain than that of its bare counterpart; this implies that the CoO x -treated cell exhibits limitation of adsorption and/or dissociation.
The activation energy of the imaginary impedance at various frequencies was analyzed (Figure 9) in order to further understand this limitation. As previously stated, the high-and low-frequency regimes are associated with the charge transfer and non-charge transfer processes, respectively. 44 For frequencies higher than 100 Hz, the activation energy remained above 0.52 eV in both cases indicating that the bare and the CoO x -treated samples exhibit thermally activated charge transfer processes. 49 In addition, for frequencies ranging from 100-0.2 Hz, the activation energy of the latter remains relatively high (with a maximum value of 1.61 eV) while that of the former decreases gradually. Since the general activation energy of the limiting process including both adsorption and dissociation is high (1.5-1.6 eV), 50,51 then with a maximum activation energy of 1.61 eV, it can be concluded that the increased impedance owing to the CoO x treatment results from both the adsorption and dissociation limiting step.

Conclusions
Atomic layer deposition was used to determine the effect of CoO x treatment on LSC. XPS and XRD analyses revealed that as-deposited ALD CoO x has a valence a state of Co 2+ and a CoO crystalline phase. Annealing at 600 • C converted the CoO to Co 3 O 4 , which implies that ALD CoO x treated on an SOFC cell can exist as Co 3 O 4 . Moreover, the I-V characteristics and electrochemical impedance of the bare and the ALD CoO x -treated cells were distinct; i.e., the latter exhibited lower power density than that of the former. This lower power density resulted from the degradation of the cathode performance stemming from the detrimental effect of CoO x on the oxygen surface exchange. By observing the increase of impedance at low frequency, we speculate that the ALD CoO x results in limited oxygen adsorption and dissociation. Despite the detrimental effect on the performance, however, the aforementioned results elucidate the effect of cathode modification through the use of CoO x .