lable at ScienceDirect

Electrochimica Acta 296 (2019) 1055e1063

Contents lists avai

Electrochimica Acta

journal homepage: www.elsevier .com/locate/electacta

Sintered powder-base cathode over vacuum-deposited thin-filmelectrolyte of low-temperature solid oxide fuel cell: Performance andstability

Jung Hoon Park a, b, Seung Min Han b, d, Byung-Kook Kim a, Jong-Ho Lee a, c,Kyung Joong Yoon a, Hyoungchul Kim a, Ho-Il Ji a, Ji-Won Son a, c, *

a High-temperature Energy Materials Research Center, Korea Institute of Science and Technology (KIST), Seoul, 02792, South Koreab Dept. of Materials Science and Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, 34141, South Koreac Nanomaterials Science & Engineering, KIST School, Korea University of Science and Technology (UST), Seoul, 02792, South Koread Graduate School of EEWS, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, 34141, South Korea

a r t i c l e i n f o

Article history:Received 10 August 2018Received in revised form27 October 2018Accepted 4 November 2018Available online 13 November 2018

Keywords:Low-temperature solid oxide fuel cellSintered cathodeVacuum-deposited thin electrolytePerformanceStability

* Corresponding author. High-temperature EnergyKorea Institute of Science and Technology (KIST), Seo

E-mail address: jwson@kist.re.kr (J.-W. Son).

https://doi.org/10.1016/j.electacta.2018.11.0180013-4686/© 2018 Elsevier Ltd. All rights reserved.

a b s t r a c t

To expand the processing options for low-temperature-operating solid oxide fuel cells (LT-SOFCs), thehybridization of powder processing and vacuum deposition is attempted. Nanostructured nickel-yttria-stabilized zirconia (Ni-YSZ) anode functional layer (AFL) and YSZ/gadolinia-doped ceria (GDC) bi-layerelectrolyte are fabricated over a sintered anode support by pulsed laser deposition (PLD), a physicalvapor deposition technology. The most common powder-processed (screen-printed and sintered)La0.6Sr0.4Co0.2Fe0.8O3-d-Gd0.1Ce0.9O1.95 (LSCF-GDC) composite cathode is applied over vacuum-depositedthin-film components. When LSCF-GDC is sintered at a general sintering temperature of 1050 �C thenthe continuity of the GDC buffer is lost and excessive interdiffusion between the cathode and theelectrolyte has occurred at the interface. On the other hand, if the sintering temperature is lowered to950 �C, peak power density more than 1.7W cm�2 at 650 �C is obtained. Moreover, the operation stabilityof the hybrid SOFC (degradation rate ~8%/100 h) is superior to that of the SOFC with a vacuum-processednanostructure cathode (degradation rate ~21%/100 h) when exposed to 0.7 A cm�2 at 650 �C, which is asignificantly harsh degradation test condition for LT-SOFCs.

© 2018 Elsevier Ltd. All rights reserved.

1. Introduction

A solid oxide fuel cell (SOFC) is a fuel cell made of solid oxides,i.e., ceramic materials, and conventionally operates in the highesttemperature regime (generally over 700 �C). This allows significantmerits such as fuel flexibility, high efficiency, and high powerdensity/specific power [1,2]. In spite of its numerous advantages,substantial research efforts were expended to lower the operatingtemperature of SOFC because high-temperature operation exertssignificant burdens in terms of cost-effectiveness, system integra-tion, and life-time of the SOFC, which delays commercialization ofthe technology [2e4]. One approach for the successful proof ofconcept of low-temperature-operating SOFCs (LT-SOFCs) is micro-

Materials Research Center,ul, 02792, South Korea.

SOFCs employing ultrathin electrolytes and nanostructure elec-trodes, and fabricated by microfabrication based on vacuumdeposition such as pulsed laser deposition (PLD), atomic layerdeposition (ALD), or sputtering [5e9]. Cell performance compara-ble to that which can be obtained from a conventional powder-processed SOFC at temperatures 200e300 �C higher was achievedbecause the thin electrolytes and nanostructured electrodes effec-tively reduced the ohmic and polarization resistances at lowertemperatures. However, the structural stabilities of micro-SOFCsusing thin film components are extremely poor due to in-stabilities of the thin and nanostructured components at elevatedtemperature [10e12]. The majority of reported micro-SOFCs failedwithin an hour [7].

To avoid the instability of LT-SOFC, we developed a thin-film-based SOFC called multiscale-architectured SOFC in the pastdecade [13e17]. With this platform, the thin film compo-nentsdincluding a nanostructured anode functional layer (AFL), amultilayer thin electrolyte, and a nanostructured cathodedwere

J.H. Park et al. / Electrochimica Acta 296 (2019) 1055e10631056

deposited using PLD over a powder-processed fully sintered anodesupport. The cell platform is called a multiscale-architecture sincethe structural scales of the unit cell range multiple scales likecentimeter (lateral dimension of the cell), micrometer (micro-structural scale of the powder-processed anode), and nanometer(microstructural scale of the deposited cell components, i.e. anode-electrolyte-cathode). Significantly high cell performances at tem-peratures lower than 650 �C and thermo-mechanical stability sur-viving 50 thermal cycles were achieved [15e17]. The platform wasexpanded to the successful realization of thin-film-based protonicceramic fuel cells (PCFC) [18,19] and to the modification of thecatalyst at the AFL [20]. Moreover, we demonstrated thatsputteringda commercial vacuum deposition technologydcanreplace PLD to fabricate a large-area multiscale-architectured SOFCup to the yttria-stabilized zirconia (YSZ) and gadolinia-doped ceria(GDC) bi-layer electrolyte in a previous study [21]. A unit cell witharea 5 cm by 5 cmwas successfully fabricated with a 2-inch (~5 cm)sputtering system, which indicates much larger cells can be fabri-cated by using a bigger sputtering system.

Despite these achievements, there have been concerns becausethe method used to realize the multiscale-architecture SOFC wasPLD, which possesses known limitations for commercialization.One is a common concern of all vacuum deposition technologies:manufacturing cost issues stemming from expensive apparatus.Recently, however, there was a cost analysis showing that the highperformance of SOFC employing thin-film deposition, along withreproducible production leading to high yield, can decrease theproduction cost [22], which may reduce the economic concerns atthe mass production stage. The other limitation is a problem spe-cific to PLD, which is the lack of scale-up capability due to theconfined deposition area. Although we could fabricate up to thethin-film bi-layer electrolyte of a larger area cell by using sputteringinstead of PLD [21], PLD had to be used to deposit a lanthanumstrontium cobaltite (La0.6Sr0.3CoO3-d, LSC) cathode layer because itis not a simple task to deposit optimized complex oxides withvolatile components like SOFC cathode materials by sputtering[23,24].

Generally, vacuum deposition (like PLD and sputtering) isappropriate for depositing thin and dense layers, which are thedesired properties for thin electrolytes in SOFCs. On the other hand,the cathode layer should have a porous structure and be relativelythick to secure the number of electrode reaction sites [25]; thesestructural characteristics are not easily realized by general vacuumdeposition technologies. PLD is capable of realizing complex cath-ode materials with proper structural and electrochemical proper-ties [15,25e27], but its use in practical fabrication is limited, asmentioned above.

It is of a significant interest to employ a non-vacuum fabricationmethod proper for the cathode over vacuum-deposited thin-filmelectrolyte. The biggest concern for hybridizing a vacuum-processed thin electrolyte and a non-vacuum-processed cathodeis the processing temperature discrepancy. If the thin electrolyte isfabricated by non-vacuum processing with high temperature sin-tering (�1000 �C), the high-temperature processing of the cathodefollowing electrolyte formation would not damage the thin elec-trolyte and the SOFCs with thin-film electrolyte would functionproperly [28e32]. However, the deposition temperatures ofvacuum-processed thin electrolytes are much lower than 1000 �C(�700 �C in general due to the limited heat transfer in vacuum) andthe application of cathodes fabricated at higher temperatures is notconsidered because they are thought to destroy the vacuum-deposited thin electrolytes.

To expand processing options for the cathode excludingvacuum-deposition (like PLD), the feasibility of applying aconventionally processed cathode over a thin-film-based SOFC

should be investigated. Therefore, in this study, the most commonconventional powder-processed (i.e. screen-printed and sintered)La0.6Sr0.4Co0.2Fe0.8O3-d-Gd0.1Ce0.9O1.95 (LSCF-GDC) composite cath-ode [33e35] was applied to multiscale-architectured thin-filmSOFC and the possibility of hybridization of the processingmethodsto realize LT-SOFC was studied. For this, LSCF-GDC screen-printedcathodes were applied to the thin-film bi-layer electrolytes of themultiscale-architectured LT-SOFC as shown in Fig. 1. First, applica-bility of the routine LSCF-GDC cathode processing temperature(~1050 �C) was studied by analyzing the microstructures andinterfacial chemical reaction; and then lowering the cathode sin-tering temperature was attempted to realize a functioning cell. Theresulting properties like the effects of the powder-processedcathode on the performance and stability of the multiscale-architectured LT-SOFC were investigated.

2. Experimental

For a multiscale-architectured SOFC up to the electrolyte, a2 cm� 2 cm, 1mm-thick NiOeYSZ composite (NiO/YSZ¼ 56:44wt%) anode support was used as a substrate for PLD deposition. The150 mm-thick NiOeYSZ tapes with poly(methyl methacrylate)(PMMA) pore-forming agent and 30 mm-thick NiOeYSZ tapeswithout PMMAwere fabricated by tape casting. A 30 mm-thick tapewithout PMMAwas laminated on top of seven layers of the 150 mm-thick tape with PMMA, under uniaxial pressure of 15MPa at 75 �C.The laminated substrate was sintered at 1300 �C for 4 h to fabricatea completely rigid substrate.

For PLD, a KrF excimer laser (l¼ 248 nm, COMPEX Pro 201F,Coherent) was used as laser source. The target-to-substrate dis-tance was 5 cm and the laser energy density was ~2.5 J cm�2 at thetarget surface. A 2 mm-thick NiOeYSZ anode functional layer (AFL)was deposited onto the tape-cast anode support by PLD at 700 �Cand ambient oxygen pressure (Pamb) of 6.67 Pa. The support withAFL was post-annealed at 1200 �C in air for 1 h to suppress Niagglomeration [36]. Then, 1 mm-thick YSZ electrolyte and 200 nm-thick GDC buffer layers were deposited using PLD at a substratetemperature of 700 �C and Pamb of 6.67 Pa. This fabrication sche-medexcept for the cathodedis identical to previous studies[17,20].

For the fabrication of powder-processed LSCF-GDC cathode,pastes of an LSCF-GDC cathode functional layer(LSCF:GDC¼ 50:50 vol%) and an LSCF current collecting layer wereprepared by mixing ceramic powders with a dispersant (HypermerKD-1), binder (polyvinyl butyral, PVB), and plasticizer (dibutylphthalate, DBP) in a-terpineol using a planetary mill. These pasteswere sequentially screen-printed over a 1 cm� 1 cm area to formthe cathode. Then the entire cell was subjected to cathode sinteringat two different temperaturesd1050 and 950 �Cdin air for 2 h.Cathode formation and sintering around 1050 �C is the routineprocess for preparing the LSCF-GDC cathode in conventionalpowder-processed SOFCs [28,33,35]. The resultant thickness of thecathode functional layer and current collecting layer was approxi-mately 12e16 mm, making the entire cathode ~25 mm thick. Aschematic showing the cell configuration is presented in Fig. 1.

The cell performance was measured by full-cell operation [17].During the fuel cell test, air and humidified H2 (3% H2O) were usedas the oxidant and fuel, respectively; the flow rate was 200 sccm oneither side. The detailed cell test configuration is described in ourprevious study [37]. First, reduction of the anode was executed bysequentially changing the fuel gas composition from 10% to 100%wet H2 in N2 at 600 �C. The cell operating temperature was variedfrom 650 to 450 �C in intervals of 50 �C and the electrochemicalimpedance spectra (EIS) and currentevoltageepower (IeVeP)curves were obtained at each temperature. Each EIS was observed

Fig. 1. A schematic showing the configuration of the multiscale-architecture SOFC fabricated by processing hybridization.

J.H. Park et al. / Electrochimica Acta 296 (2019) 1055e1063 1057

over the frequency range 105 Hz to 10�1 Hz. The AC amplitude ofthe impedance measurements was 50mV. An Iviumstat electro-chemical analyzer (Iviumstat, Ivium Technologies) was used toobtain EIS and IeVeP curves. To investigate stability, the cell wassubjected to a constant load of 0.7 A cm�2 at 650 �C for 114 h. EIS at0.75 V cell voltage and IeVeP curves were measured every 1 hduring the stability test. After the cell tests, the microstructures ofthe cell were observed using scanning electron microscopy (SEM).For the case which requires chemical reaction analyses at thecathode/electrolyte interface, transmission electron microscopy(TEM) and its energy dispersive spectroscopy (EDS) wereemployed.

3. Results and discussion

3.1. Multiscale-architecture SOFC with a 1050 �C sintered cathode:‘Cell-1050’

In Fig. 2, surface microstructures of the cathode and the elec-trolyte of Cell-1050 are displayed. Fig. 2(a) and (b) are a low and ahigh magnification SEM pictures of the cathode surface, i.e. theLSCF current collector, respectively. There was no significant

Fig. 2. Surface morphologies of Cell-1050: (a) and (b) surface of the LSCF cathode (currentInset in (c) is the surface of the electrolyte in the multiscale-architecture thin-film SOFC w

abnormality in terms of the morphology. An appropriate porousstructure is fabricated and the particles are well connected, as otherpowder-processed LSCF cathode processed at the identical condi-tion [33]. Fig. 2(c) is the surface morphology of the GDC/YSZ bi-layer observed at the cell surface which is not covered by thecathode. The inset in Fig. 2(c) is thin-film electrolyte surface of themultiscale-architectured SOFC with thin-film-processed cathode[15,17], which did not experience higher temperature than 700 �C.The main and inset images compared in Fig. 2(c) are at the identicalmagnification and it appears that no substantial grain growthoccurred due to cathode sintering at 1050 �C.

Low and high magnification cross-sectional SEM pictures ofCell-1050 are displayed in Fig. 3. In a lowmagnification SEM pictureshowing the overall structure (Fig. 3(a)), the cell componentsappear to be well integrated. However, Fig. 3(b), the highermagnification back-scattered electron (BSE) SEM picture which canpresent material differences in gray contrast, clearly indicates colorchanges in the GDC buffer layer. It means that there is an obviousinterfacial chemical reaction across the GDC buffer layer. Theidentification of the reaction phase was attempted by using X-raydiffraction (XRD) on Cell-1050, however, it was not successfulbecause there are too many diffraction peaks from the multilayer

collecting layer); (c) surface of the electrolyte layer where the cathode does not cover.ith thin-film-processed cathode.

Fig. 3. Cross-sectional microstructures of Cell-1050: (a) low-magnification picture presenting all cell components; (b) high-magnification picture showing cathode/electrolyteinterface.

J.H. Park et al. / Electrochimica Acta 296 (2019) 1055e10631058

cell components and the interface of the interest is embeddedunder the thick cathode, as shown in Fig. 3(a).

Therefore, to analyze this chemical reaction at the cathode/electrolyte interface, cross-sectional focused ion beam (FIB) sam-pling near the interface and TEMwith areal EDS were used. A high-angle annular dark field (HAADF) image and correspondingelemental distribution images of Sr, Ce, and Zr are displayed inFig. 4. Dashed lines on each image indicate the interface betweenthe GDC buffer and the YSZ electrolyte layers. It can be clearly seenthat the GDC buffer layer (bright parts in HAADF) has lost conti-nuity. The darker parts at the discontinuity of the GDC layer (someare indicated with arrows) are composed of Sr-Zr intermixing. It isobvious that Sr diffused down to the GDC/YSZ interface and Zrdiffused up, crossing the GDC layer to produce the Sr-Zr intermixingphase over substantial area of the interface. It demonstrates thatthe routine LSCF-base cathode sintering temperature of 1050 �Ccannot be employed for the hybridization of the powder-processedcathode and the thin-film-deposited electrolyte.

A question can be brought up is whether this result indicates theintrinsic instability of the GDC buffer layer which is vacuum-deposited at a lower temperature (700 �C in this case). Contem-plating over our numerous experiences with the vacuum-depositeddense GDC buffer layer [25,38e41], it is highly possible that themalfunctioning of the GDC buffer layer of the multiscale-architectured SOFC is severer, especially when compared withother vacuum-deposited GDC buffer layers with much bigger grainsizes. The grain size of the GDC buffer layer can be formed differ-ently in spite that the identical PLD deposition condition isemployed. In our previous studies using TEM [39,40], we havedemonstrated that the PLD deposited GDC buffer layers grow as aperfect epitaxial layer over each YSZ grain, thus the grain size ofGDC is identical to that of the YSZ electrolyte. In multiscale-architectured thin-film SOFC (TF-SOFC) like the present study[39], the diameter of the YSZ columnar grain is around a fewhundred nm and thus is the GDC grain size. In powder-processedSOFC [40] where the grain size of the YSZ electrolyte is severalmicrons, the grain size of the deposited GDC buffer layer is the sameto that of YSZ as well.

In an exemplary case of the latter where the grain size of theGDC buffer layer and the YSZ electrolyte is several microns [41],PLD-deposited LSC-GDC composite cathode was applied over theGDC buffer layer and post-annealed up to 1000 �C. There was nochemical reaction crossing the GDC buffer in spite that 1000 �C is ahigher temperature than general sintering temperatures of thepowder-processed LSC-GDC cathode, � ~950 �C [34,35,42]. It is

expected that the PLD-deposited LSC-GDC is much more reactivethan the powder-processed LSCF-GDC cathode due to the morereactive nature of LSC to LSCF and the nanoscale grain size.Furthermore, in a work by Han et al. [28], a thin (1-2 mm-thick) YSZelectrolyte fabricated by sintering, a vacuum (PVD)-deposited GDCbuffer layer could be combined with a 1040 �C sintered LSCFcathode, without any indication of chemical reaction and/or cellperformance deterioration. The most significant difference of Hanet al.'s work from the present study is that the grain size of the YSZ.Despite that the thickness of the GDC/YSZ bi-layer electrolytes aresimilar, the grain size of YSZ in Han et al.'s work was ranging1e4 mm because the thin electrolyte is sintered at a high temper-ature of 1400 �C [28], while that of the PLD deposited YSZ is aboutseveral hundred nm. Considering these results, it is postulated thatthe smaller grain size and resulting more grain boundaries aggra-vate chemical reaction across the GDC buffer layer. This may limitthe upper boundary of the sintering temperature of the powder-processed cathode over the vacuum-processed buffer layer andthin-film electrolyte.

This massive interfacial chemical reaction at Cell-1050 wouldlead to poor cell performance. Cell-1050 was subjected to the celltesting and the cell voltage rose when the reduction started, but iteventually decreased and never reached the desired open cellvoltage (OCV) over 1 V (Fig. S1(a)). Even if the GDC buffer layer doesnot function properly, there is no reason for poor OCV unless thereare massive breakage and/or complete consumption of thin YSZelectrolyte by the chemical reaction, which are not observed inboth SEM and TEM analyses. To check if this result was an experi-mental error or not, another Cell-1050 was fabricated and testedagain, but a similar trend was observed in terms of both the cellvoltage and the chemical reaction, as shown in Fig. S1(b). Althoughthe exact reason for lowOCV is not clear, it is obvious that 1050 �C ishigh enough to damage the vacuum-deposited thin film compo-nents by interfacial chemical reaction, eventually resulting in cellmalfunction.

3.2. Multiscale-architecture SOFC with a 950 �C sintered cathode:‘Cell-950’

As can be seen in section 3.2, the sintering temperature of theLSCF-GDC cathode should be lowered to avoid the unwantedchemical reaction at the cathode and the thin-film GDC/YSZ bi-layer interface. However, since the LSCF-GDC cathode is powder-based, we cannot reduce the sintering temperature far below900 �C because the interfacial adhesion would not be developed.

Fig. 5. (a) I-V-P curves and (b) EIS of Cell-950 at various temperatures. In (b), the EIS at 650 and 600 �C are magnified in the lower graph.

Fig. 4. TEM HAADF image of Cell-1050 and corresponding elemental distribution mapping of Sr, Ce, and Zr. Dashed lines indicate the original interface between GDC and YSZ.

J.H. Park et al. / Electrochimica Acta 296 (2019) 1055e1063 1059

Table 1Peak power density comparison between Cell-950 and TF-SOFCs [17].

Cathode Anode

Peak power density (Power density at 0.7 V) [mWcm�2]

650 �C 600 �C 550 �C 500 �C

Cell-950 950 �Csinteredscreen printedLSCF-GDC

Nanostructure AFL 1753 (1439) 1316 (907) 794 (461) 367 (202)

TF-SOFC-1 [17] 650 �C post-annealedPLD LSC

Microstructure AFL 1650 (1396) 1149 (811) 645 (357) 274 (140)

TF-SOFC-2 [17] 650 �C post-annealedPLD LSC

Nanostructure PLD AFL 1775 (1707) 1443 (1237) 958 (674) 462 (302)

J.H. Park et al. / Electrochimica Acta 296 (2019) 1055e10631060

Therefore, a lower sintering temperature chosen was 950 �C. Thiscell will hereafter be denoted ‘Cell-950’. When was operated, Cell-950 exhibited fair cell performance as shown in the I-V-P curvesand EIS in Fig. 5. Open cell voltage (OCV) near 1.1 V was obtainedand the ohmic resistance values were 31.5 and 44.5mUcm2 at 650and 600 �C, respectively, which are in the similar range to that ofTF-SOFCs having the same bi-layer electrolyte [15,17]. These resultsindicate that significant damage was not induced in the thin-filmcomponents below the powder-processed cathode that was sin-tered at 950 �C. Since many highly performing cathode materialslike lanthanum strontium cobaltite (LSC) are sintered at tempera-tures around 900e950 �C [34,35,42], we believe that hybridizationof the powder-processed cathodes and the vacuum-deposition inthe multiscale-architecture SOFC platform is plausible.

In Table 1, a comparison of the cell power densities of Cell-950and a previously reported TF-SOFCs [17] with the same PLD

Fig. 6. (a) Cell voltage change while load of 0.7 A cm�2 is applied at 650 �C to Cell-950; (b) I-0.75 V every 1 h (Nyquist and Bode plots). Data measured every 10 hrs are selectively disp

deposited 200 nm-thick GDC/1 mm-thick YSZ bi-layer electrolyte ispresented. Cell-950 exhibited better cell performances at all tem-peratures in comparison with TF-SOFC-1. This is due to the signif-icant contribution of the nanoscale particle size of the AFL to thelow temperature performance [17]. It is impressive that Cell-950produced a similar peak power density with TF-SOFC-2 at 650 �C.For other temperatures Cell-950 exhibited lower peak powerdensities, but still, Cell-950 produces quite high cell performancesat low temperatures, and especially when Cell-950 vs. TF-SOFC-2and TF-SOFC-1 vs. TF-SOFC-2 are compared, the power drop byreplacing nanostructure electrode with powder-processed micro-scale electrodes is less significant at the cathode than that at theanode. The peak power density of the Cell-950 at 500 �C is 367mW/cm2. Moreover, since the active area is 1 cm2, the total poweroutput is identical, i.e. 367mW. In general, micro-fabricated free-standing membrane based thin-film SOFCs yield very high areal

V-P curves measured every 1 h during the degradation test, and (c) EIS at cell voltage oflayed for clearly representing the change.

J.H. Park et al. / Electrochimica Acta 296 (2019) 1055e1063 1061

power densities, e.g. over 1W/cm2 at 500 �C, but due to theextremely poor scalability, the total power outputs per cell aremuch less than 1mW [7]. Considering this, the cell architectureused for the Cell-950 appears to be a potential approach to acquirea high-performance LT-SOFC.

More importantly, it is expected that application of the powder-processed cathode to the TF-SOFC may slow down the degradationbecause agglomeration of the nanostructured electrode is consid-ered the most significant origin of degradation in TF-SOFCs [43]. Toassess degradation with acceleration, we applied Cell-950 withconstant current load of 0.7 A cm�2 at 650 �C. To the best of ourknowledge, this is the harshest degradation test condition appliedto SOFCs with vacuum-deposited electrolyte components [7]. Theresults are presented in Fig. 6. During 114 h of constant load, the cellvoltage degraded by ~8%/100 h (Fig. 6(a)). The I-V-P curves and EISmeasured intermittently every 1 h are shown in Fig. 6(b) and (c). EISpresented in Fig. 6(c) is at the cell voltage 0.75 V. As can be seen inthe Nyquist plots, the resistance increase was mainly due to po-larization resistance. A small change in the ohmic resistance in-dicates that the electrolyte-electrode interface quality did notdegrade much. In the Bode plots, the low frequency componentappears to increase more significantly, which indicates that thesurface reaction of the electrode degraded. This may be attributedto finer particle sizes of the LSCF-GDC cathode due to the lowersintering temperature which will be discussed below with the

Fig. 7. SEM micrographs of Cell-950 after cell operation: (a) cross-section of the active cell cocathode; and (d) surface of the electrolyte.

microstructure shown in Fig. 7. As mentioned above, degradation isexpected to be suppressed relative to that of the cell with a thin-film-process nanostructured cathode. Therefore, to assess the sta-bility level of the powder-processed cathode quickly, a TF-SOFCwith PLD LSC was subjected to the same degradation test condi-tion. The degradation of the cell voltage is shown in Fig. S2 and thedegradation rate was ~21%/100 h, which is significantly higher thanthat of Cell-950. Since the particle size of the cathode of Cell-950 isin submicron level (a few hundred nm to micron, Fig. 7(c)) and thatof the particle size of the thin-film cathode is in 10 s nm level [39],the surface area of the latter is extremely larger than that of theformer. Therefore, it is thought that the driving force for theagglomeration and surface degradation is much less in the former.This comparison implies that hybridization of the powder-processed cathode can substantially improve the stability of themultiscale-architectured SOFC.

The post-mortem microstructure of Cell-950 is displayed inFig. 7. (a) and (b) are cross-sectional micrographs at differentmagnifications. As can be seen from Fig. 7(a), all active cell com-ponents maintained their structures. No sign of chemical reactioncan be seen in Fig. 7(b), unlike in Cell-1050 (Fig. 3(b)). Fig. 7(c) is thesurface of the LSCF cathode and (d) is the surface of the electrolyteof Cell-950 (again the inset is the thin-film electrolyte surface of theTF-SOFC). The grain size of the sintered cathode is much smallerand the connection between the grains appears to be weaker than

mponents; (b) magnified image of the electrolyte/electrode interface; (c) surface of the

Fig. 8. TEM HAADF image of Cell-950 and corresponding elemental distribution mapping of Sr, Ce, and Zr. Dashed lines indicate the original interface between GDC and YSZ.

J.H. Park et al. / Electrochimica Acta 296 (2019) 1055e10631062

those of Cell-1050 shown in Fig. 2(b). The smaller grain size canincrease the reaction site density, however on the other hand,smaller grain size and weaker interconnectivity may accelerate thedegradation of the electrode surface reaction. Nevertheless, thesurface morphology deterioration was not as significant as in thethin-film cathode (shown in Fig. S3(a)) which was subjected to theidentical degradation test. The electrolyte microstructure does notchange much from that of the TF-SOFC. More detailed interfacialanalyses were performed by using TEM and the results are dis-played in Fig. 8. Unlike the interface of Cell-1050 shown in Fig. 4,there is no sign of interdiffusion and the structural integrity of theGDC buffer layer is maintained. Considering that this cell hasexperienced more than 100 h of 650 �C degradation test, the sta-bility of the thin-film components appears to be structurally sound.These microstructural features confirm that 950 �C cathode sin-tering did not damage the thin film components, as can be foreseenfrom the electrochemical test results.

The results on Cell-950 demonstrate the feasibility of hybridi-zation of powder processing and vacuum-deposition for LT-SOFCs.It appears possible to apply a powder-processed cathode over avacuum-deposited electrolyte if the thin-film SOFC platform is asstructurally sound as the multiscale-architecture SOFC. Thin-film

deposition of the cathode is not a trivial task due to the compli-cated composition of the material; hence, the present result opensthe opportunity for versatile processing combination options forrealizing LT-SOFCs.

4. Conclusions

The combination of powder-processed LSCF-GDC cathode andvacuum-deposited multiscale-architectured thin-film half-cell wasinvestigated and it was demonstrated that hybridization is plau-sible when the sintering temperature is not too high. When thesintering temperature was 1050 �C, undesirable chemical reactionsoccurred massively at the cathode and thin-film GDC/YSZ bi-layerelectrolyte. On the other hand, when the cathode was sintered at950 �C, cell performance reached peak power density exceeding1.7W cm2 with superior operating stability under high current loadin comparison with the cell with vacuum-processed cathode. Itappears possible to apply a powder-processed cathode over avacuum-deposited electrolyte if the thin-film SOFC platform isstructurally sound. We believe that the results and outlook pre-sented in this article open opportunities for versatile processingcombination options for realizing LT-SOFCs.

J.H. Park et al. / Electrochimica Acta 296 (2019) 1055e1063 1063

Acknowledgements

The authors are grateful to the Global Frontier R&D Program ofthe Center for Multiscale Energy Systems (Grant No. NRF-2015M3A6A7065442) of the National Research Foundation (NRF) ofKorea, funded by the Ministry of Science & ICT and to the Institu-tional Program of Korea Institute of Science and Technology (KIST)for financial support.

Appendix A. Supplementary data

Supplementary data to this article can be found online athttps://doi.org/10.1016/j.electacta.2018.11.018.

References

[1] B.C.H. Steele, A. Heinzel, Nature 414 (2001) 345e352.[2] E.D. Wachsman, K.T. Lee, Science 334 (2011) 935e939.[3] Z. Gao, L.V. Mogni, E.C. Miller, J.G. Railsback, S.A. Barnett, Energy Environ. Sci. 9

(2016) 1602e1644.[4] J.-W. Son, Metal oxide thin film-based low-temperature-operating solid oxide

fuel cell by interface structure control, in: N. Pryds, V. Esposito (Eds.), MetalOxide-based Thin Film Structures: Formation, Characterization and Applica-tion of Interface-based Phenomena, Elsevier, 2017.

[5] J. An, Y.-B. Kim, J. Park, T.M. Gür, F.B. Prinz, Nano Lett. 13 (2013) 4551e4555.[6] M. Tsuchiya, B.-K. Lai, S. Ramanathan, Nat. Nanotechnol. 6 (2011) 282e286.[7] I. Garbayo, D. Pla, A. Morata, L. Fonseca, N. Sabate, A. Tarancon, Energy Envi-

ron. Sci. 7 (2014) 3617e3629.[8] S. Oh, S. Hong, H.J. Kim, Y.-B. Kim, J. An, Ceram. Int. 43 (2017) 5781e5788.[9] J.D. Baek, K.-Y. Liu, P.-C. Su, J. Mater. Chem. A 5 (2017) 18414e18419.

[10] R.P. Reolon, S. Sanna, Y. Xu, I. Lee, C.P. Bergmann, N. Pryds, V. Esposito,J. Mater. Chem. A 6 (2018) 7887e7896.

[11] K.-Y. Liu, L. Fan, C.-C. Yu, P.-C. Su, Electrochem. Commun. 56 (2015) 65e69.[12] K.-Y. Liu, Y.J. Yoon, S.H. Lee, P.-C. Su, Electrochim. Acta 247 (2017) 558e563.[13] H.-S. Noh, H. Lee, B.-K. Kim, H.-W. Lee, J.-H. Lee, J.-W. Son, J. Power Sources

196 (2011) 7169e7174.[14] H.-S. Noh, J.-W. Son, H. Lee, H.-S. Song, H.-W. Lee, J.-H. Lee, J. Electrochem. Soc.

156 (2009) B1484eB1490.[15] H.-S. Noh, K.J. Yoon, B.-K. Kim, H.-J. Je, H.-W. Lee, J.-H. Lee, J.-W. Son, J. Power

Sources 247 (2014) 105e111.[16] H.-S. Noh, K.J. Yoon, B.-K. Kim, H.-J. Je, H.-W. Lee, J.-H. Lee, J.-W. Son, J. Power

Sources 249 (2014) 125e130.[17] J.H. Park, S.M. Han, K.J. Yoon, H. Kim, J. Hong, B.-K. Kim, J.-H. Lee, J.-W. Son,

J. Power Sources 315 (2016) 324e330.[18] K. Bae, D.Y. Jang, H.J. Choi, D. Kim, J. Hong, B.-K. Kim, J.-H. Lee, J.-W. Son,

J.H. Shim, Nat. Commun. 8 (2017) 14553.

[19] K. Bae, H.-S. Noh, D.Y. Jang, J. Hong, H. Kim, K.J. Yoon, J.-H. Lee, B.-K. Kim,J.H. Shim, J.-W. Son, J. Mater. Chem. A 4 (2016) 6395e6403.

[20] C.-A. Thieu, J. Hong, H. Kim, K.J. Yoon, J.-H. Lee, B.-K. Kim, J.-W. Son, J. Mater.Chem. A 5 (2017) 7433e7444.

[21] H.-S. Noh, J. Hong, H. Kim, K.J. Yoon, B.-K. Kim, H.-W. Lee, J.-H. Lee, J.-W. Son,J. Electrochem. Soc. 163 (2016) F613eF617.

[22] M.R. Weimar, L.A. Chick, D.W. Gotthold, G.A. Whyatt, Cost Study forManufacturing of Solid Oxide Fuel Cell Power Systems, DOE Report, PNNL,2013.

[23] B.-K. Lai, A.C. Johnson, H. Xiong, S. Ramanathan, J. Power Sources 186 (2009)115e122.

[24] D.B. Chrisey, G.K. Hubler, Pulsed Laser Deposition of Thin Films, 1994.[25] H.-S. Noh, H. Lee, H.-I. Ji, H.-W. Lee, J.-H. Lee, J.-W. Son, J. Electrochem. Soc. 158

(2011) B1eB4.[26] J.-H. Park, W.-S. Hong, K.J. Yoon, J.-H. Lee, H.-W. Lee, J.-W. Son, J. Electrochem.

Soc. 161 (2014) F16eF22.[27] T. Ishihara, J. Kor. Ceram. Soc 53 (2016) 469e477.[28] F. Han, R. Mücke, T. Van Gestel, A. Leonide, N.H. Menzler, H.P. Buchkremer,

D. St€over, J. Power Sources 218 (2012) 157e162.[29] E.-O. Oh, C.-M. Whang, Y.-R. Lee, S.-Y. Park, D.H. Prasad, K.J. Yoon, J.-W. Son, J.-

H. Lee, H.-W. Lee, Adv. Mater. 24 (2012) 3373e3377.[30] D. Udomsilp, D. Roehrens, N.H. Menzler, A.K. Opitz, O. Guillon, M. Bram, Mater.

Lett. 192 (2017) 173e176.[31] R. Hui, Z. Wang, S. Yick, R. Maric, D. Ghosh, J. Power Sources 172 (2007)

840e844.[32] Y.-Y. Chen, W.-C.J. Wei, Solid State Ionics 177 (2006) 351e357.[33] K.J. Yoon, M. Biswas, H.-J. Kim, M. Park, J. Hong, H. Kim, J.-W. Son, J.-H. Lee, B.-

K. Kim, H.-W. Lee, Nano Energy 36 (2017) 9e20.[34] H.-G. Jung, Y.-K. Sun, H.-Y. Jung, J.-S. Park, H.-R. Kim, G.-H. Kim, H.-W. Lee, J.-

H. Lee, Solid State Ionics 179 (2008) 1535e1539.[35] H.-Y. Jung, K.-S. Hong, H.-G. Jung, H. Kim, H.-R. Kim, J.-W. Son, J. Kim, H.-

W. Lee, J.-H. Lee, J. Electrochem. Soc. 154 (2007) B480eB485.[36] H.-S. Noh, J.-W. Son, H. Lee, H.-I. Ji, J.-H. Lee, H.-W. Lee, J. Eur. Ceram. Soc. 30

(2010) 3415e3423.[37] H.-S. Noh, J. Hwang, K. Yoon, B.-K. Kim, H.-W. Lee, J.-H. Lee, J.-W. Son, J. Power

Sources 230 (2013) 109e114.[38] H.-Y. Jung, Hierarchical Control of Nano-composite Cathode for the Perfor-

mance Improvement of Intermediate Temperature Solid Oxide Fuel Cell, Ph. D.Thesis in Dept, Materials Science and Engineering, Seoul National University,Seoul, 2011.

[39] H.-S. Noh, J.-S. Park, H. Lee, H.-W. Lee, J.-H. Lee, J.-W. Son, Electrochem. SolidState Lett. 14 (2011) B26eB29.

[40] H.-S. Noh, J.-W. Son, H. Lee, J.-S. Park, H.-W. Lee, J.-H. Lee, Fuel Cells 10 (2010)1057e1065.

[41] J.-H. Park, W.-S. Hong, G.C. Kim, H.J. Chang, J.-H. Lee, K.J. Yoon, J.-W. Son,J. Electrochem. Soc. 160 (2013) F1027eF1032.

[42] S.-Y. Park, H.-I. Ji, H.-R. Kim, K.J. Yoon, J.-W. Son, B.-K. Kim, H.-J. Je, H.-W. Lee,J.-H. Lee, J. Power Sources 228 (2013) 97e103.

[43] K. Kerman, B.-K. Lai, S. Ramanathan, J. Power Sources 196 (2011) 2608e2614.