Tuning Oxygen Reduction Kinetics in LaSrCoO₄ with Strained Epitaxial Thin Films and Wrinkled Freestanding Membranes

Abstract

Sluggish oxygen reduction reaction (ORR) remains a critical barrier to advancing intermediate-temperature electrochemical energy devices. Here, we demonstrate that strain engineering in two platforms, epitaxial thin films and freestanding membranes, systematically tunes ORR kinetics in Ruddlesden-Popper LaSrCoO₄. In epitaxial films, film thickness is varied to control in-plane tensile strain, whereas in freestanding membranes strain relaxation during the release step using water-soluble sacrificial layers produces flat or wrinkled architectures. Electrochemical impedance spectroscopy analysis reveals more than an order of magnitude increase in the oxygen surface exchange coefficient for tensile-strained films relative to relaxed films, together with a larger oxygen vacancy concentration. Wrinkled freestanding membranes provide a further increase in oxygen surface exchange kinetics and a lower activation energy, which are attributed to increased active surface area and local strain variation. These results identify epitaxial tensile strain and controlled wrinkling as practical design parameters for optimizing ORR activity in Ruddlesden-Popper oxides.

1. Introduction

Developing intermediate-temperature solid oxide fuel cells (IT-SOFCs), which are promising electrochemical devices for environmentally friendly, efficient energy production and conversion, requires overcoming sluggish cathodic oxygen reduction reaction (ORR) at intermediate temperatures [1,2,3,4,5,6,7]. Mixed ionic and electronic conducting ABO₃ perovskites, such as La_1−xSr_xCo_1−yFe_yO_3−δ (LSCF113) and La_1−xSr_xCoO_3−δ (LSC113), are widely used as electrodes because concurrent electronic conduction and mobile oxygen vacancies extend the electrochemically active region from the triple-phase boundary to the entire electrode surface [8,9,10,11,12,13]. However, these ABO₃ perovskites suffer from high thermal expansion coefficients and low chemical stability due to facile lattice oxygen loss, which causes undesirable lattice expansion in reducing atmospheres [14,15]. In addition, surface Sr segregation, a primary degradation pathway in LSCF113 and LSC113, suppresses ORR kinetics over time [11,16,17,18].

Ruddlesden–Popper (RP) oxides are attractive alternatives since they offer good electronic conductivity together with good thermal and chemical stability [19,20,21,22]. RP structures consist of n consecutive perovskite layers (ABO₃) alternating with rock-salt layers (AO) along the crystallographic c-axis direction [23,24]. Within this family, oxygen vacancy-dominated La_2−xSr_xCoO_4±δ is notable for stable reduction behaviors over a wide range of temperatures (300~1500 °C) [25,26,27] and oxygen surface exchange kinetics comparable to those of LSCF113 and LSC113 [28,29]. Moreover, when La_2−xSr_xCoO_4±δ is interfaced with perovskites such as LSCF113 or LSC113, the heterostructured interface can yield fast oxygen surface exchange [30,31]. Nevertheless, for stand-alone La_2−xSr_xCoO_4±δ, the intrinsic ORR activity at intermediate temperatures still requires further improvement for IT-SOFCs.

Epitaxial oxide thin films have shown remarkable enhancements in ORR kinetics by enabling controlled structure–property tuning not accessible in bulk [8,32]. In such films, epitaxial strain is a key design parameter that modifies oxygen vacancy formation energies and surface chemistry, thereby controlling both bulk oxygen transport and oxygen surface exchange kinetics [32,33,34,35]. For ABO₃ perovskites, tensile strain, introduced via substrate choice or reduced film thickness, is widely reported to enhance oxygen surface exchange kinetics [36,37]. By contrast, only a few studies have investigated strain effects in RP oxides, and the reported influence of epitaxial strain on ORR activity is inconsistent, with both enhancement and suppression observed [19,21,38,39,40]. For example, tensile strain along the c-axis has been reported to accelerate oxygen exchange in RP nickelate thin films, whereas other studies have reported minimal changes or opposite trends in oxygen transport and/or surface exchange depending on crystal orientation and microstructure. Consequently, the strain dependence of oxygen surface exchange kinetics in RP oxides remains unresolved.

Beyond epitaxial strain, epitaxial oxide films provide a platform to create freestanding oxide membranes, enabling morphological control beyond substrate clamping [41]. Freestanding membranes fabricated by wet etching maintain high crystallinity and structural quality [42,43,44,45]. During dissolution of the sacrificial layer, strain relaxation can leave the membrane flat or produce wrinkled architectures [41]. Wrinkling introduces spatially varying local strain and curvature, and can increase the density of active ORR sites. Such wrinkled geometries have been used to enhance functional properties, including piezoelectric and optoelectronic properties, in other oxide systems [46,47,48]. However, the direct impact of controlled wrinkling on ORR kinetics in MIEC oxides has, to our knowledge, not been investigated, leaving a clear opportunity to couple increased surface area and local strain for catalytic benefit.

In this study, we use LaSrCoO₄ (LSC214) as a model RP oxide to investigate two design parameters, epitaxial strain, and membrane wrinkling, for accelerating ORR at intermediate temperatures. RP cobaltites can accommodate oxygen nonstoichiometry through both vacancy and interstitial defects [24]. Since epitaxial strain can shift the relative stability of these defects, LSC214 provides a well-suited platform to correlate strain-induced defect changes with oxygen surface exchange kinetics. First, we grow epitaxial LSC214 films on (001) single-crystal yttria-stabilized zirconia (YSZ) and systematically vary thickness to tune in-plane tensile strain. YSZ provides high oxide-ion conductivity with negligible electronic leakage while serving as a single-crystal growth platform, which in practice makes thickness-controlled strain the most general approach for films on YSZ [9,33,49]. Second, we fabricate freestanding LSC214 membranes via wet etching to induce controlled wrinkling, enabling simultaneous manipulation of active surface area and local strain fields in the absence of substrate clamping. Taken together, our results establish epitaxial strain and wrinkling as design strategies for optimizing ORR activity in RP oxides and provide practical design rules for next-generation IT-SOFC cathodes.

2. Materials and Methods

2.1. Synthesis of Epitaxial Thin Films

Epitaxial LSC214 thin films with four different thicknesses were grown on (001) single-crystal YSZ substrates using pulsed laser deposition (PLD). Prior to the LSC214 growth, a ~5 nm Gd_0.2Ce_0.8O₂ (GDC) buffer layer was deposited to prevent the formation of a La₂Zr₂O₇ layer at the interface [50]. The platinum ink (Pt) counter electrode (#6082, BASF Catalysts, Newark, NJ, USA) was applied to the back side of the YSZ substrate and dried in air at 900 °C for 1 h before the deposition of LSC214 and GDC layers. To ensure thermal contact, a small amount of silver paint (Leitsilber 200, Ted Pella, Redding, CA, USA) was used to affix the YSZ substrate to the PLD substrate holder.

PLD was carried out using a KrF excimer laser with a wavelength (λ) of 248 nm, a pulse rate of 10 Hz, and a laser fluence of 1 J/cm² under an oxygen partial pressure (pO₂) of 6.6 × 10⁻⁵ atm (50 mTorr). The same repetition rate and laser fluence were used for all deposited layers in this work. The GDC buffer (~5 nm; 500 pulses) was deposited at 550 °C, followed by 1000, 2000, 5000, and 10,000 pulses of LSC214 at 650 °C, resulting in film thicknesses of ~20, 37, 69, and 129 nm, respectively. Film thickness was determined from atomic force microscopy (AFM) step-height measurements across a masked edge. After LSC214 growth, samples were cooled to room temperature in the PLD chamber for ~1 h under a pO₂ of 6.6 × 10⁻⁵ atm (50 mTorr).

2.2. Synthesis of Freestanding Membranes

Freestanding LSC214 membranes were fabricated by depositing LSC214 on water-soluble sacrificial layers, namely Sr₂CaAl₂O₆ (S₂C₁AO) and SrCa₂Al₂O₆ (S₁C₂AO), grown on (001) single-crystal SrTiO₃ (STO) substrates by PLD. A ~5 nm thick GDC buffer layer was inserted between LSC214 and the sacrificial layers to improve lattice coherence and prevent interdiffusion. S₂C₁AO was deposited under vacuum, whereas S₁C₂AO was deposited at a pO₂ of 1.31 × 10⁻⁴ atm (100 mTorr). The PLD conditions were 1500 pulses for the 30 nm thick sacrificial layer at 750 °C, 500 pulses for the GDC buffer at 550 °C, followed by 8000 pulses of LSC214 at 650 °C, resulting in a total LSC214 film thickness of about 120 nm.

A 100 μm thick polypropylene carbonate (PPC) layer was spin-coated on LSC214 to serve as mechanical support. The PPC/LSC214/GDC/S₂C₁AO or S₁C₂AO/STO stack was then immersed in deionized water to dissolve the sacrificial layer, releasing the freestanding PPC/LSC214/GDC membrane from the STO substrate. The detached membrane was transferred onto a YSZ substrate pre-coated with a Pt counter electrode, after which the PPC support layer was removed using chloroform etching to yield a pristine freestanding LSC214 membrane for electrochemical measurements.

2.3. Thin Film and Freestanding Membrane Characteristics

High-resolution X-ray diffraction (HRXRD) was performed using a four-circle diffractometer to examine the phase purity and crystal orientation of the thin film systems. Measurements were conducted in both in-plane and out-of-plane configurations. YSZ has a cubic structure (lattice parameter a = b = c) [51], and thus the lattice parameter, a, was directly determined from the YSZ (202) reflection. LSC214 has tetragonal symmetry with space group I4/mmm [52], where the c-axis lattice parameter was determined from the LSC214 (002) reflection, and assuming a- and b-axis lattice parameters are equivalent, the in-plane lattice parameter was determined from the LSC214 (103) reflection. Peak positions (2θ) were determined by fitting the corresponding reflections with a Gaussian line shape to precisely determine the peak centers, following standard thin film XRD analysis procedures [53]. â and ĉ, which are the relaxed lattice parameters for films in an unstrained state, were obtained from the following equation [10,49,54],

$${\displaystyle \frac{(c-\hat{c})}{\hat{c}}}={\displaystyle \frac{-2\mathsf{\nu }}{1-\mathsf{\nu }}}{\displaystyle \frac{(a-\hat{a})}{\hat{a}}},$$

(1)

assuming ĉ/â = 3.285 at 298 K [52] and ν = 0.3 [10,49,54,55]. Because a quantitative â(T,δ) and ĉ(T,δ) relationship is not available for LSC214, Equation (1) was used as a self-consistent approximation to define the relaxed reference for comparing relative strain across the film series. This approach is commonly used for RP thin films when a fully independent relaxed reference is not directly available [19,20].

In-plane and out-of-plane strains were calculated from the following equations. The in-plane strain is given by ${\mathsf{\epsilon }}_{\mathit{aa}}$, and the out-of-plane strain by ${\mathsf{\epsilon }}_{\mathit{cc}}$.

$${\mathsf{\epsilon }}_{aa}={\displaystyle \frac{(a-\hat{a})}{\hat{a}}}, {\mathsf{\epsilon }}_{\mathit{cc}}={\displaystyle \frac{\left(c-\hat{c}\right)}{\hat{c}}}.$$

(2)

Surface morphology of the freestanding membranes was examined using optical microscopy (Carl Zeiss AG, Oberkochen, Germany).

2.4. Oxygen Surface Exchange Kinetics

In situ electrochemical impedance spectroscopy (EIS) was performed at open circuit with a microprobe station (SUSS MicroTec SE, Garching, Germany) connected to a frequency response analyzer (Solartron 1260, AMETEK, Oak Ridge, TN, USA) and dielectric interface (Solartron 1296, AMETEK, Oak Ridge, TN, USA). For epitaxial films, circular microelectrodes of ~200 µm diameter were used. For freestanding membranes, 5 × 5 mm² LSC214 squares with a central Ag contact dot (~200 µm diameter) were prepared. The temperature was controlled at 550 °C on a heating stage (Linkam TS1500, Linkam Scientific Instruments, Salfords, UK) and verified by a thermocouple in contact with the film surface (±5 °C variation). Impedance spectra were collected with 10 mV amplitude from 1 MHz to 1 mHz over pO₂ spanning from 10⁻³ to 1 atm using calibrated O₂/N₂ mixtures. The ZView software (Version 3.5g, Scribner Associates, Southern Pines, NC, USA) was used to construct an equivalent circuit and perform complex least squares fitting. An equivalent circuit, R₁-(R₂‖CPE₂)-(R_ORR‖CPE_ORR), was used to analyze the EIS data, where R₁ is the ohmic resistance, R₂‖CPE₂ represents the mid-frequency process, and R_ORR‖CPE_ORR represents the low-frequency ORR process. This model was used to extract the ORR resistance and the oxygen surface exchange coefficient. The electrical surface exchange coefficient (k^q), which is comparable to k* [56], was calculated as [57,58]

$$k^{\mathrm{q}}=\mathrm{RT}/4{\mathrm{F}}^2{\mathrm{R}}_{\mathrm{ORR}}{\mathrm{A}}_{\mathrm{electrode}}{\mathrm{c}}_{\mathrm{o}},$$

(3)

where R is the universal gas constant (8.314 J·mol⁻¹·K⁻¹), T is the absolute temperature, F is Faraday’s constant (96,500 C·mol⁻¹), ${\mathrm{R}}_{\mathrm{ORR}}$ is the resistance from the low-frequency feature, ${\mathrm{A}}_{\mathrm{electrode}}$ is the electrode area, and ${\mathrm{c}}_{\mathrm{o}}$ is the lattice oxygen concentration in LSC214, with

$${\mathrm{c}}_{\mathrm{o}}=(4+\mathsf{\delta })/{\mathrm{V}}_{\mathrm{m}},$$

(4)

where V_m is the molar volume of LSC214 at room temperature. In this study, ${\mathrm{c}}_{\mathrm{o}}$ was calculated with δ extracted from previously reported values [59].

The volume-specific capacitance (VSC), indicative of changes in δ induced by changes in the electrical potential, was obtained from EIS data via the expression,

$$\mathit{VSC}=[1/({\mathrm{A}}_{\mathrm{electrode}} \times \mathrm{t}\left)\right]{({\left({\mathrm{R}}_{\mathrm{ORR}}\right)}^{1-n}\mathrm{Q})}^{1/n},$$

(5)

where t is the electrode thickness, Q is the non-ideal “capacitance”, and n is the non-ideality factor of constant phase element (CPE). The fitted n values for semicircle CPE_ORR were found to range from ~0.96 to 1.0 over the entire pO₂ range examined (n = 1, ideal).

2.5. Surface Analysis of Thin Films

Surface chemistry of the LSC214 films before and after heat treatment was analyzed using ex situ Auger electron spectroscopy (AES) with the Physical Electronics 700 Scanning Auger Nanoprobe (PHI, Chanhassen, MN, USA). The AES measurements were performed at an accelerating voltage of 10 kV. To investigate the effects of heat treatment, the LSC214 films were annealed at 550 °C for 6 h under two different pO₂ of 10⁻³ and 1 atm. For AES data collection, three randomly selected spots with dimensions of 10 μm × 10 μm were chosen on each sample within an ultra-high vacuum chamber.

3. Results

3.1. Crystallinity and Strain of the LSC214 Thin Films and Freestanding Membranes

Figure 1 shows the HRXRD data of the epitaxial LSC214 films. The out-of-plane XRD pattern revealed the presence of (00l)_tetragonal peaks of LSC214 (where l is even) and (00l)_cubic peaks of GDC and YSZ, indicating the successful synthesis of c-axis-oriented epitaxial LSC214 films with a crystallographic relationship of (002)_tetragonal LSC214//(002)_cubic GDC//(002)_cubic YSZ (Figure 1a). The ϕ scans of the LSC214 films confirmed that LSC214 (103)_tetragonal, GDC (202)_cubic, and YSZ (202)_cubic have strong peaks with four-fold cubic symmetry (Figure 1b), which reveals the in-plane crystallographic relationships between GDC and YSZ (cube-on-cube alignment) and LSC214 and GDC (in-plane 45° rotation) as shown in Figure 1c. The measured lattice parameters a and c of the LSC214 films were found to be a = 3.806 Å and c = 12.51 Å for 129 nm at RT, which is in good agreement with the literature data (a = 3.807 Å and c = 12.507 Å for the c-axis oriented epitaxial LSC214 thin film with ∼100 nm) [60]. Note that perovskite oxides often reveal a gradual strain relaxation as the film thickness is increased, which is most likely due to the formation of complex defect structures (such as dislocations) at the film/substrate interface to relieve the built-up elastic energy resulting from the strain. Such a phenomenon is also observed in this work, providing more than one strain state through the use of multiple film thicknesses. At room temperature, the 20, 37, and 69 nm thick films were tensile-strained with the strain magnitude decreasing as the film thickness increased. At the measurement temperature (550 °C), the absolute strain may shift due to thermal mismatch, but this offset should be similar for all films on the same substrate/buffer stack and thus does not affect the relative strain trend with thickness. However, the 129 nm thick film was fully relaxed. These strains arise due to significant variations in atomic spacing between LSC214 and GDC in the (001) LSC214 orientation.

Table 1. Summary of calculated experimental and relaxed lattice parameters of epitaxial LSC214 thin films on a YSZ substrate with a GDC as the buffer layer obtained from XRD patterns.

Film Thickness (nm)	a (Å)	c (Å)	â (Å)	ĉ (Å)	a Strain (%)	c Strain (%)
20	3.85	12.45	3.82	12.54	0.85	−0.73
37	3.83	12.47	3.81	12.52	0.48	−0.41
69	3.81	12.49	3.81	12.50	0.11	−0.10
129	3.81	12.50	3.81	12.51	0.07	−0.06

shows the relaxed lattice parameters (â and ĉ) of LSC214 films, ranging from 3.81 to 3.82 Å and from 12.50 to 12.54 Å, respectively. The small residual strain values for the 129 nm thick film originated from limited numerical precision and rounding of the lattice parameters and thus were effectively zero within experimental uncertainty.

Figure 2a schematically illustrates the wet etching process used to fabricate freestanding LSC214 membranes. Figure 2b shows the out-of-plane HRXRD scans of as-grown LSC214/GDC stacks on water-soluble sacrificial layers. Similarly to the epitaxial LSC214 films directly grown on YSZ, only (00l)tetragonal peaks of LSC214 (where l is even) were observed together with (00l)cubic peaks of GDC, S1C2AO, or S2C1AO, and STO, confirming c-axis-oriented LSC214 films with a crystallographic relationship of (002)tetragonal LSC214//(002)cubic GDC//(008)cubic S1C2AO or S2C1AO//(002)cubic STO. The cubic GDC buffer (5.451 Å) rotated by 45° provides an effective in-plane spacing (${\displaystyle \frac{\sqrt{2}}{2}}$ a_GDC = 3.86 Å), which closely matches the in-plane lattice of the S1C2AO sacrificial layer (3.852 Å), but is substantially mismatched with S2C1AO (3.907 Å) [62]. Because LSC214 was grown on the thin (~5 nm) GDC buffer, the interfacial strain stored at the GDC/sacrificial-layer interface primarily determines the elastic energy stored in the released LSC214/GDC bilayer and thus whether the membrane remains flat or wrinkles. In the case of S1C2AO, the interfacial strain was nearly zero (~0.01%), minimizing the stored elastic energy. Accordingly, upon dissolving S1C2AO, strain relaxation or buckling did not occur, and the released LSC214/GDC bilayer remained flat and defect-free (Figure 2d). In contrast, for S2C1AO, the interfacial strain was much larger (~1.7%), and when the sacrificial layer was removed and substrate clamping was lost, the accumulated strain relaxed and the bilayer lowered its elastic energy by developing out-of-plane wrinkles (Figure 2e). It is worth noting that the post-transfer HRXRD peaks for both the flat and wrinkled membranes (Figure 2c) retained the same LSC214 and GDC reflections without additional peaks, indicating that crystallinity is preserved after release and transfer. The LSC214 (004) rocking curves (Figure 2f) showed a broader full-width half-maximum (FWHM) for the wrinkled membrane than the flat membrane, consistent with increased mosaic spread (local tilt distribution) introduced by the wrinkled geometry. The resulting localized deformation effectively relieved residual strain in the freestanding LSC214/GDC membrane [41].

3.2. Electrochemical Properties of the Epitaxial LSC214 Thin Films and Freestanding Membranes

Figure 3a shows representative Nyquist plots for the 20 nm thick LSC214 film at 550 °C as a function of pO₂. Based on the pO₂ dependence of the three features in the equivalent circuit, each frequency range was determined [63,64,65,66]. The high-frequency feature did not change with pO₂, and its magnitude and activation energy (~1.15 eV) matched previously reported values for oxygen ion conduction in YSZ [66]. The mid-frequency feature was independent of pO₂ and is attributed to oxygen ion transport at the LSC214/GDC interface. Its capacitance was also much smaller (~10⁻⁶ F) than that of the low-frequency feature (~10⁻³ F). In thin film electrodes, the capacitance mainly reflects changes in oxygen content. Accordingly, the low-frequency feature corresponds to oxygen surface reaction, and R_ORR was taken from the corresponding low-frequency element in the equivalent circuit fit. Importantly, k^q and VSC were extracted from this low-frequency element (R_ORR‖CPE_ORR), while the mid-frequency interfacial contribution was captured separately (R₂‖CPE₂). Thus, interfacial transport does not bias the ORR kinetics reported here. The strong pO₂ dependence of this feature indicates that oxygen surface exchange kinetics governs the ORR in the LSC214 films. Furthermore, the dominant role of surface exchange is further supported by the fact that all film thicknesses are much smaller than the critical thickness for bulk transport limitation (estimated to ∼5000 µm for the bulk La₂CoO₄ [27]).

Figure 3b compares impedance spectra at 550 °C and pO₂ = 1 atm for the LSC214 thin films of different thicknesses (20, 37, 69, and 129 nm) and for freestanding membranes (flat and wrinkled, both ~120 nm). The thickest, largely relaxed film (129 nm) was found to exhibit the largest low-frequency semicircle, whereas the thinnest, most tensile-strained film (20 nm) showed the smallest low-frequency semicircle. As thickness increased and epitaxial strain relaxed, the low-frequency semicircle increased systematically. For freestanding membranes of identical thickness (≈120 nm), morphology was decisive. The flat membrane showed a low-frequency semicircle comparable to the relaxed 129 nm thick epitaxial film, whereas the wrinkled membrane exhibited a substantially smaller low-frequency semicircle comparable to the strained 20 nm thick film.

To investigate the ORR activity of the LSC214 thin films and freestanding membranes, we extracted the k^q and VSC from the EIS data (Figure 4). Figure 4a shows the temperature dependence of k^q at 1 atm for the LSC214 films of different thicknesses and for flat and wrinkled membranes. Introducing in-plane tensile strain increased the k^q values of the LSC214 thin films more than one order of magnitude. In contrast, the wrinkled membrane induced by strain relaxation significantly enhanced the k^q values of LSC214 compared to the thin films, whereas the flat membrane exhibited comparable k^q values to the film with an equivalent thickness (~120 nm). It is worth noting that the wrinkled membrane exhibited a higher k^q value compared to the most strained thin film. In addition, the activation energy for oxygen surface exchange of the films was measured at 1 atm, resulting in values of 1.59, 1.52, 1.20, and 1.10 eV for films with thicknesses of 20, 37, 69, and 129 nm, respectively. The pO₂ dependence of k^q ($k^{\mathrm{q}}\propto {\mathrm{P}}_{{\mathrm{O}}_2}^{\mathrm{m}}$) was found to have an exponent m = ~0.65 for the representative 20 nm thick film (Figure 3a), indicating that the reaction rate-limiting step is a molecular oxygen dissociative adsorption process rather than a charge-transfer process [67,68]. This behavior is consistent with prior reports for RP cobaltite cathodes [31,60,69,70].

Figure 4b shows VSC as a function of temperature for the LSC214 thin films of different thicknesses and freestanding membranes obtained from the EIS data. VSC increased systematically as the film thickness decreased (i.e., tensile strain increased), indicating a higher density of redox-active defects and charge carriers in the strained films. Since VSC in mixed-conducting cobaltites scales with the concentration of oxygen vacancies and electronic carriers, these trends suggest that tensile strain promotes oxygen vacancy formation in LSC214. Under oxygen-deficient conditions, oxygen transport in La_2−xSr_xCoO_4±δ is known to proceed predominantly via oxygen vacancies in the perovskite layers, that is, through vacancy-assisted diffusion [71,72]. These prior results, together with our observed increase in VSC with strain, support our hypothesis that tensile strain in LSC214 enhances the concentration and/or mobility of oxygen vacancies. Combined with our k^q values, these trends are consistent with a vacancy-assisted oxygen reduction mechanism [73], in which adsorbed oxygen species are incorporated into the lattice via oxygen vacancies at the LSC214 surface.

3.3. Surface Chemistry of the Epitaxial LSC214 Thin Films

To assess thickness-dependent surface chemistry of the LSC214 thin films, area mode Auger electron spectroscopy (AES) was performed on the 37 and 69 nm thick LSC214 films under three different conditions, namely pristine, annealed at 1 atm, and annealed at 10⁻³ atm, all at 550 °C for 6 h (Figure 5). For both thicknesses, Co and Sr spectral features overlapped across all conditions, indicating negligible changes in cation surface composition upon annealing. Relative ratios extracted from AES (Figure 5e,f) showed low Co/(La + Sr) consistent with an A-site terminated RP surface (SrO termination) [25] and no systematic variation in Sr/(La + Sr) with thickness or annealing atmosphere. These results suggest that, within the sensitivity and depth of resolution of AES, the surface cation stoichiometry is effectively invariant across thicknesses and annealing conditions. While AES is semi-quantitative, it probes the near-surface region (few nm information depth) and has been shown to sensitively capture Sr enrichment/segregation in related cobaltite cathodes when quantified using relative sensitivity factors with inelastic-mean-free-paths (IMFP)-based attenuation corrections [19,20,31,74]. In the present study, the invariant Sr/(La + Sr) within experimental scatter indicates no detectable net Sr enrichment in the near-surface region. Consequently, the absence of measurable thickness-dependent surface chemistry implies that the observed trends in k^q and E_a are governed primarily by strain-controlled defect thermodynamics (e.g., oxygen vacancy population), rather than by changes in surface cation composition.

4. Discussion

As shown in Figure 4, thinner (more tensile-strained) LSC214 films showed larger k^q values, whereas thicker, more relaxed films exhibited smaller k^q values. This trend is consistent with tensile strain lowering the formation energy of oxygen vacancies and shifting the defect equilibrium toward higher vacancy concentrations [22,33,34]. The resulting increase in available carriers and surface defect sites raises the pre-exponential factor for oxygen surface exchange, thereby reducing the ORR polarization resistance. In addition, strained films were found to have higher k^q and higher E_a than relaxed films.

For RP oxides, the bulk oxygen 2p band center has been used as a useful descriptor that correlates with both oxygen surface exchange rates and activation barriers. Across oxygen-excess RP oxides, a lower oxygen 2p level aligns with higher surface exchange rates and lower barriers, whereas ABO₃ perovskites show the opposite trend [60]. In vacancy-dominated RP oxides such as LSC214, the behavior is more consistent with ABO₃ perovskites, where an upward shift in the oxygen 2p level that stabilizes vacancies is associated with enhanced surface exchange. In RP oxides, oxygen defect formation energies vary approximately linearly with the oxygen 2p band center, and shifting the oxygen 2p level upward stabilizes vacancies by lowering their formation energy while destabilizing interstitials [72]. A strain-driven upshift in the oxygen 2p band center, accompanied by enhanced oxygen surface exchange kinetics and increased oxygen vacancy concentration, has been demonstrated in tensile-strained La_0.6Sr_0.4CoO_3−δ epitaxial films on YSZ [33]. Accordingly, our observed increase in k^q with tensile strain is consistent with tensile-strain-induced stabilization of oxygen vacancies, while direct electronic-structure validation is beyond the scope of this work.

For the freestanding LSC214 membranes, the flat membrane behaves similarly to the relaxed epitaxial film (comparable polarization resistance and E_a ≈ 1.10 eV). The wrinkled membrane showed a markedly smaller ORR arc, higher k^q, and a lower E_a (≈0.89 eV). Because thickness and composition were fixed, two morphology specific factors are expected to contribute: (i) effective area amplification, which increases the number of accessible exchange sites per projected footprint, and (ii) local variations in strain and curvature across the wrinkles, which can influence adsorption and near-surface vacancy equilibria. We note that the wrinkled morphology necessarily introduces curvature in the freestanding bilayer. Such curvature is expected to generate spatially varying local strain and strain gradients (tensile at the outer surface and compressive at the inner surface), which can vary between wrinkle peaks and valleys. Indeed, prior studies on freestanding perovskite oxide membranes have directly visualized alternating tensile/compressive strain zones within individual wrinkles using atomic-resolution scanning transmission electron microscopy strain mapping, linking peak/valley curvature to the magnitude and sign of local strain/strain-gradient fields [46]. Similarly, bent freestanding perovskite oxides can sustain very large strain gradients (up to ~10⁷ m⁻¹) across the membrane thickness [75]. Accordingly, the enhanced ORR activity of the wrinkled membranes is consistent with a combination of increased effective surface area and curvature-induced local strain/strain-gradient fields, which is also consistent with the broader rocking-curve FWHM observed for the wrinkled membrane (Figure 2f), reflecting an increased distribution of local tilt (mosaic spread) associated with the wrinkled geometry. The slightly lower E_a is also consistent with these morphology-enabled effects, although direct nanoscale strain mapping of freestanding LSC214 membranes will be required to separate their contributions.

Taken together, these findings establish epitaxial strain and membrane wrinkling as practical, orthogonal levers for tuning ORR in RP oxides. For device-relevant platforms (films on YSZ), thickness-controlled strain is a robust route to enhance k^q. For flexible or transfer architectures, controlled wrinkling further boosts activity while modestly lowering E_a. Future work that quantifies the wrinkle-induced roughness factor and correlates it with k^q and E_a would help separate geometric from electronic contributions and refine predictive design rules.

5. Conclusions

In summary, this study demonstrates that the ORR activity of LSC214 thin films is strongly influenced by epitaxial strain and wrinkle formation. Compared to relaxed films, tensile-strained LSC214 exhibited higher k^q values, reflecting enhanced ORR kinetics, but also showed increased activation energy. This behavior is consistent with an increase in oxygen vacancy concentration and/or mobility under in-plane tensile strain. Wrinkle formation in freestanding LSC membranes further enhanced ORR activity, increasing the k^q value while lowering the activation energy due to a larger active surface area generated upon strain relaxation. VSC analysis further supports the correlation between oxygen vacancy concentration and the improved ORR performance in both strained films and wrinkled membranes. AES measurements indicate that the surface cation stoichiometry remains effectively invariant across thicknesses and annealing conditions. These findings highlight the pivotal role of epitaxial strain in modulating oxygen surface exchange and emphasize the necessity of integrating surface chemical analysis with electrochemical metrics to design highly active oxide catalysts for energy conversion applications.