Revealing Strain Effects on the Chemical Composition of Perovskite Oxide Thin Films Surface, Bulk, and Interfaces

Understanding the effects of lattice strain on oxygen surface and diffusion kinetics in oxides is a controversial subject that is critical for developing efficient energy storage and conversion materials. In this work, high‐quality epitaxial thin films of the model perovskite La0.5Sr0.5Mn0.5Co0.5O3−δ (LSMC), under compressive or tensile strain, are characterized with a combination of in situ and ex situ bulk and surface‐sensitive techniques. The results demonstrate a nonlinear correlation of mechanical and chemical properties as a function of the operation conditions. It is observed that the effect of strain on reducibility is dependent on the “effective strain” induced on the chemical bonds. In‐plain strain, and in particular the relative BO length bond, is the key factor controlling which of the B‐site cation can be reduced preferentially. Furthermore, the need to use a set of complimentary techniques to isolate different chemically induced strain effects is proven. With this, it is confirmed that tensile strain favors the stabilization of a more reduced lattice, accompanied by greater segregation of strontium secondary phases and a decrease of oxygen exchange kinetics on LSMC thin films.


Introduction
Developing new oxide materials with high ionic and electronic conductivity and fast oxygen surface exchange kinetics is crucial for optimizing performance in a range of device applications including solid oxide fuel cells (SOFCs), batteries, and permeation membranes. The structure and chemistry of attempted to predict perovskite behavior based on various indicators such as correlating electronic structure and reaction kinetics and relating conductivity changes to strain in thin films. [17] The link between electronic structure and reactivity is derived from the well-established relationship between reactivity and the d-band center relative to the Fermi level in strained metals. [20,21] For perovskites, the e g occupancy has been investigated as a descriptor for oxygen reduction reaction catalysis [22,23] and some reports demonstrated experimentally that catalytic activity of perovskite oxides is dependent on the concentration of oxygen vacancies, and the B-site cation valence state. [24][25][26] La 1−x Sr x CoO 3 (LSC) and La 1−x Sr x MnO 3 (LSM) are two of the most widely studied cathode materials for SOFCs. The investigation of thin film LSC by Klenov et al. [27] (x = 0.5) at temperatures up to 600 °C suggested the oxygen vacancy structure of LSC on SrTiO 3 (STO) and LaAlO 3 (LAO) substrates is related to the presence of both thermal and physical strain. More recently, oxygen surface exchange and diffusion kinetics studies by Kubicek et al. [28] (x = 0.2 and 0.4), in the temperature range 280 to 475 °C, as well as surface and electronic structure studies of Cai et al. [29] (x = 0.2), from room temperature to 450 °C, demonstrated that the activity of LSC can be enhanced with tensile strain. Cai et al. propose three mechanisms for the electronic structure differences they report in strained LSC thin films, namely, cobalt spin-state transitions, electronic bandwidth broadening due to structural changes, and formation of oxygen vacancies modifying the d-band structure of the cobalt, concluding that the oxygen defect structure is responsible.
Studies of strontium-containing perovskites frequently highlight substantial strontium segregation, a primary cathode degradation mechanism in SOFCs due to a reduction of active surface sites and related phase changes, at thin film surfaces. [30] Dulli et al. [31] linked strontium segregation in the perovskite LSM (x = 0.35) to a surface layer restructuring to a Ruddlesden-Popper (RP) phase, after their films were annealed at 900 °C. The reverse has also been reported by Chen et al. [32] with a (La 1−x Sr x ) 2 CoO 4 (x = 0.25, 0.5) RP-phase surface restructuring of the first three unit cells, due to strontium segregation, to the perovskite LSC after annealing in an oxygen-enriched atmosphere. For perovskite LSM (x = 0.3), Jalili et al. [33] report enhancement of both strontium segregation and oxygen vacancy formation, at room temperature and 500 °C, in thin films under tensile strain. The enhancement is primarily attributed to structurally induced bond length changes, i.e., chemical expansion. Further works also concluded that lattice expansion, related to tensile strain, is a key factor for promoting surface cation segregation. [34,35] This work investigates thin films of La 0.5 Sr 0.5 Mn 0.5 Co 0.5 O 3−δ (LSMC), which has the highest reported oxygen stoichiometry range in a perovskite system. [36] LSMC can accommodate two distinct structures, a hexagonal R3 c stoichiometric (δ = 0) and an orthorhombic Pbnm hypo-stoichiometric (up to δ = 0.62). The catalytic activity and selectivity of the oxidized and reduced LSMC phases for the oxidation of hydrocarbons was suggested to be related to the oxygen content and transition metal oxidation states, indicating that control of oxygen stoichiometry allows the catalytic behavior of these phases to be tuned. [36] This difference in catalytic behavior, linked to the defect structure, makes LSMC an intriguing material for the study of oxygen stoichiometry control. Double substitution at both the A-and B-perovskite sites, provides scope for building on work of simpler perovskite systems with single occupancy of the B-site, for understanding key mechanisms at the surface of perovskite thin films.

Structural Characterization
To explore the effects of strain on the properties of LSMC, epitaxial thin films were deposited under oxidizing conditions by pulsed laser deposition (PLD) on (001) single crystal substrates: LAO, (LaAlO 3 ) 0.3 (Sr 2 AlTa 6 ) 0.7 (LSAT), and STO. The mismatch of the pseudocubic oxidized LSMC, relative to the single crystal substrates, resulted in tensile strain of 1.40% for STO and 0.44% for LSAT and compressive strain of −1.68% for LAO. For films on the three substrates, denoted as LSMC/LAO, LSMC/STO, and LSMC/LSAT, a combination of X-ray diffraction (XRD), reflectivity (XRR), and reciprocal space mapping (RSM) measurements was used to confirm high-quality, strained epitaxial growth, measure thin film thickness, and calculate the unit cell volume for the as-deposited thin films (see Figures S1-S3, Supporting Information). As expected for strained films, in-plane compression is accompanied by out-of-plane elongation, and vice versa, an elastic response often called the Poisson effect. [37] Figure 1 shows the inverse relationship between the in-plane and out-of-plane cell parameters of LSMC with strain and the total volume of the strained unit cells. The structural data confirm that lattice strain of LSMC thin films is directly linked to the average unit cell volume, with tensile strain resulting in the highest volume and compressive strain resulting in the lowest unit cell volume.
Previous work on other perovskite thin film systems has demonstrated that strain can be used to alter reaction kinetics. [38][39][40] XRD allows for very accurate determination of cell Adv. Mater. Interfaces 2020, 7,1901440 Figure 1. In-plane (squares) and out-of-plane (triangles) cell parameters and average unit cell volume (crosses) of LSMC thin films. Uncertainties represent the standard deviation in the measurements. Lines are to guide the eye and show the inverse relationship between the unit cell parameters (black) and the increase in volume (gray) with increasing tensile strain. www.advmatinterfaces.de parameters (below 10 × 10 −3 Å) and has been demonstrated as a measurement technique for in situ monitoring of thin film cell structure variations related to oxidation and reduction, facilitating direct correlation of structural changes with redox activity upon exposure to step changes between different atmospheres. [41][42][43] The detection limit of this in situ XRD technique is significantly smaller than expansion of the pseudocubic bulk LSMC cell parameters from 3.8511 to 3.8871 Å, for the oxidized and reduced phases, respectively, when heated under 5% H 2 /balance N 2 , a lattice parameter expansion of 3.6 × 10 −2 Å or 0.94%.
In epitaxial thin films, matching between in-plane cell parameters of the substrate and the thin film inhibits changes in the thin film layer, therefore the in-plane cell parameter of LSMC is not expected to vary significantly due to changes of the atmosphere. [41] During in situ XRD measurements (see Figure S4a, Supporting Information), the films were heated (600 to 750 °C), aligned to observe the 103 asymmetric film reflection and cycled between a synthetic air (≈2.1 × 10 5 ppm O 2 , 20% O 2 /80% N 2 ) and nitrogen (≈10 ppm O 2 , 100% N 2 ) atmosphere. Within the scan range of Δ2θ = 2.5°, a 103 substrate reflection tail was also observed, which was used as an internal reference (see Figure S5, Supporting Information). Each 2.5° 2θ scan was fitted using a two-peak, pseudo-Voigt fit (see Figure S6, Supporting Information) and the LSMC peak position was used to calculate the out-of-plane cell parameter relative to the substrate to account for any film peak shift due to sample misalignment during the experiment.
For both compressive (LSMC/LAO) and tensile (LSMC/STO) strain, oxidation of the films on switching from low to high pO 2 is a consistently faster process than the reduction process, occurring when switching from high to low pO 2 . For example, for LSMC/LAO at 750 °C (see Figure 2), the film takes 75 to 80 min to reduce by ≈95% of the measured maximum reduction, within a given cycle. On oxidation at the same temperature, the film takes 70 to 100 s to reach ≈100% of the original cell parameter. Further, using a double exponential, with time  www.advmatinterfaces.de constants t 1 < t 2 , provides a better fit in the fitting of the reduction process, however the oxidation is best described by a single component, with time constant t 1 only. Part of the asymmetry between oxidation and reduction may be accounted for by the large pO 2 step change used in the in situ XRD experimental setup, necessary to obtain a measurable change in the cell parameter, as similar materials have different exchange values depending on the pO 2 . The difference in time for oxidation and reduction, as well as the number of fitting parameters, could also suggest that the oxygen reduction and evolution reactions follow different reaction pathways, possibly due to slower re-equilibration of the sample, including an equilibration of oxygen at the interface between the film and substrate, after a fast oxygen surface exchange. An additional equilibration step would depend on the energy barriers for oxygen transport and the energetics of oxygen vacancy formation across the interface of the film and substrate, as well as oxygen diffusion in the substrate. For application in SOFCs as a cathode, the material would be exposed to a high pO 2 and therefore would always have a larger exchange value. The difference in reaction rates has consequences for applications requiring the use of both oxygen evolution and reduction with large pO 2 changes, including metal-air batteries, as the rate of oxygen incorporation and evolution impacts charge and discharge rates.
The in situ XRD measurements were performed with reduction in a nitrogen atmosphere, resulting in the expansion of the out-of-plane cell parameter by up to 3.9 × 10 −3 Å or 0.10% for the film on LAO. Surface exchange coefficients were calculated by fitting cell parameter profiles, following the method described by Moreno and Santiso et al. [41][42][43] (see Figure S7, Supporting Information). The change between the relative, average oxidized and reduced cell parameters for each temperature shows a greater change in the magnitude of the cell parameter for films under compressive strain (see Figure 2c). This could be related to the STO film being more reduced from the beginning as proven by X-ray absorption near edge spectroscopy (XANES) analysis (see Section 2.2), and therefore, experiencing a lower driving force for reduction under the nitrogen atmosphere used during the experiment. When looking to the reaction kinetics, it seems that K oxi for the LAO films is bigger than for the STO film in the first cycle. This should be related to the lower energy required for oxidation in LSCM under compressive strain. The kinetics deterioration in the second cycle should be related to the passivation of the surface due to further Sr segregation as suggested by previous studies. It should be noted that the passivation effect could be bigger under reduction due to an enhancement of the Schottky defects in the perovskites for which the generation of oxygen vacancies can be accompanied by the creation of Sr vacancies in competition with the reduction of the lattice. This could be an additional factor to explain the slower reduction kinetics compared to the oxidation observed. The Arrhenius plots suggest a transition at 650 °C (Figure 2d,e), however the process responsible for this transition has not yet been identified. It has previously been noted that exchange coefficients for slow diffusers are relatively scattered, as in tracer diffusion experiments by De Souza and Kilner, [44,45] and contribute to large inaccuracies in k exchange values. The error bars in Figure 2d,e are based on fitting the data with an exponential factor and therefore do not fully account for experimental errors such as temperature stability.

Chemical Characterization
Unit cell volume and defect chemistry are closely connected through chemical expansion and changes in lattice parameter due to strain are expected to promote a change in defect concentration. [46] Cai et al. [29] suggest chemical expansion as the primary reason for the impact of strain on the electronic properties of strained LSC and Jalili et al. [33] suggest chemical expansion as a key driving force for differences in strontium segregation in LSM. Following the theory of chemical expansion, based on the structural data and in the absence of a dominant accommodation of the structural changes by compositional changes at the A-and B-sites, LSMC/LAO is expected to accommodate the highest transition metal oxidation states as it has the smallest unit cell volume (56.57 ± 0.29 Å 3 ). LSMC/STO has the largest unit cell volume (58.59 ± 0.13 Å 3 ), suggesting the lowest transition metal oxidation states. This assumption was investigated using a wide range of techniques to probe the chemical composition of the thin films. Oxidation states were analyzed using XANES. The elemental composition of the LSMC thin films was determined using ion scattering techniques: Rutherford backscattering (RBS) to determine the average elemental composition of the films and low-energy ion scattering (LEIS) to investigate the elemental depth profile of the film composition. The chemical composition of the LSMC thin film surfaces was investigated using X-ray photoelectron spectroscopy (XPS).
The measurement principle of RBS and LEIS involves an ion beam (He + and Ne + ) directed onto the solid sample surface from which part of the primary projectiles are backscattered, and the energy distribution of these ions is measured and converted to a mass spectrum by describing the ion-surface atom interaction as a binary collision with conservation of momentum. These are powerful techniques for determining film composition, with RBS able to measure oxygen content and LEIS providing information about the outermost atomic surface terminations. [47,48] In combination with a sputtering ion beam, a depth profile can also be obtained of the chemical composition through the film. While similar, these two techniques provide complementary information due to differences in cross sections and the electronic excitations and charge transfer processes. [48] RBS measurements, encompassing the full thickness of the LSMC thin films on LAO and STO substrates (see Figure 3), observed that the composition of the thin films is the same regardless of strain, within the uncertainty (due to statistical precision and systematic errors, primarily from calculated He energy loss values). The RBS spectra have flat tops, within the statistical error margins, and RUMP fitting [49] estimated the La/Sr ratio varies by less than 4% throughout the films (see Figure S8, Supporting Information). Target element atomic mass ratios play a key role in determining the composition of thin films deposited by PLD and previous work has demonstrated that lighter elements are often lower in content. [50,51] This accounts for the deviation in the LSMC stoichiometry www.advmatinterfaces.de which has a lower cobalt content compared to bulk LSMC for films under both compressive and tensile strain. However, under the PLD conditions used for these LSMC thin films, there is no corresponding decrease in the manganese concentration which is of both similar atomic mass and has a similar metal-oxygen species stability in the plasma. Variation of the Co/Mn ratio across the film thickness cannot be estimated from the RBS measurements due to the overlap of the B-site cations signals and a significant background contribution from the substrates. The stoichiometry similarity of the LSMC thin films supports the chemical similarity observed from the oxygen surface exchange determined by in situ XRD measurements.
LEIS was used to obtain a depth-resolved composition profile through the LSMC thin film, encompassing both the air-thin film and thin film-substrate interfaces, by alternating Ne + analysis and Ar + sputtering beams (see Figure S9, Supporting Information). Changes in the signal for each of the elements were analyzed to determine the relative composition of the film as a cross section through the film thickness. When the signal stabilizes to a constant value, this region is treated as the bulk of the film and taken to be an internal reference, with the stoichiometric ratios defined by RBS. By normalizing the composition of each spectra relative to the internal reference ratio, a depth profile of the surface and near surface regions can be compiled. Figure 4a shows a representative depth profile through a LSMC/LSAT thin film where the outermost surface is defined by a zero dose, and the interface between the film and substrate is indicated by the increasing tantalum signal, an element found only in the substrate. While the outermost surface can be represented by a single point, the interface between the thin film and substrate appears as a gradient as the interface is not precisely normal to the ion beam. Ion mixing and roughness below the surface can also broaden the signal.
RBS measurements of the overall film composition suggest a cobalt deficiency within the films that could be related to some Co evaporation during the PLD deposition. LEIS suggests subtle changes in the ratio of manganese to cobalt (see Figure 4d), with a lower ratio at the surface indicative of relatively more cobalt and a higher ratio at the interface indicative of more manganese. B-site enriched subsurface regions have previously been observed in similar systems and were related to A-site surface segregation. [32,52] For LSMC there is a subsurface enrichment of cobalt at the B-site in the outermost layers, but an overall B-site deficiency for approximately the top third of the film (see Figure 4c).
The A/B cation ratios established from the LEIS profiles show that all films, regardless of strain, were A-site terminated. Segregation of alkaline-earth dopants to the surface of perovskite oxides is a common occurrence and is usually considered a degradation mechanism, as surface A-cation enrichment can deactivate surface reaction sites and increase resistance. [53] Strontium segregation at perovskite oxide surfaces has been widely reported. [54][55][56] The strong strontium termination observed (see Figure 4b) is in agreement with recent publications [57,58] and previous work has shown alkaline-earth surface segregation to be associated with elastic and electrostatic interactions of the alkaline-earth with the surrounding lattice, with tensile strain promoting strontium segregation. [34,55] In this work, despite strain significantly altering the unit cell volume of LSMC, no relationship between strontium termination and strain was observed by LEIS or RBS.
In order to quantify the Sr in different chemical environments and understand better the effect of strain on the Sr segregation, XPS analyses were performed. This allowed differentiation of the Sr present in secondary phases such as Sr(OH) 2 or SrCO 3 from Sr enrichment within the perovskite lattice. In order to perform the XPS analysis, the samples were annealed at 300 °C to remove carbon contamination. Angle-resolved XPS was performed in order to quantify the different Sr   www.advmatinterfaces.de species as a function of depth. The XPS normal to the sample surface showed a slight increase in both oxygen-hydrogen (B.E. ≈ 531.5 eV) and carbon (B.E. ≈ 285 and 288.5 eV) for the films under tensile strain, which can be related to an increase in the strontium segregation in the form of secondary phases (see Figure S10, Supporting Information). In perovskites containing strontium, an insulating phase generally forms on the surface and has been shown to have a majority decomposition product of strontium oxide, SrO, as well as hydroxyl, Sr(OH) 2 and carbonate, SrCO 3 species. [59,60] Angle-resolved XPS (see Figure 5), at emission angles of 60° and 80° between the sample surface normal and the detector position shows clear differences in the strontium environment confirming a higher segregation of Sr in the form of secondary phases under tensile strain These differences are consistent with reported experimental and modeling results for related perovskite systems, where tensile strain has been linked to a surface enrichment of strontium carbonate. [29,30,34,55] With these results we can conclude that while RBS and LEIS prove the overall Sr content to be similar in the different films, the Sr chemical environment changes as a function of strain proving a higher tendency to form surface Sr-secondary phases with tensile strain.
XANES measurements of the transition metal edge position for the as-deposited LSMC thin films at room temperature (see Figure 6a,b) showed that all films were slightly oxygen deficient but that tensile strain favored a more reduced film. Interestingly, Adv. Mater. Interfaces 2020, 7,1901440

www.advmatinterfaces.de
we observed that, within the experimental error, the Co oxidation state did not seem to be affected by strain. However, the Mn oxidation state was substantially affected leading to a reduction in oxidation state from 4.1 ± 0.1 to 3.6 ± 0.1. This could be related to the smaller ionic size of Mn(IV) compared with Co(III, HS) in an octahedral environment (0.53 and 0.61 Å, respectively [61] ). The shorter MnO bond would experience a much greater tensile strain promoting a more facile oxygen release within the Mn octahedra resulting in a lower energy for the formation of oxygen vacancies and larger Mn reduction. Further in situ studies were performed with XANES (see Figures S4b and S11, Supporting Information) in which thin films were reduced by heating from room temperature to 500 °C in hydrogen (O 2 ≈ 0.1 ppm) to obtain further chemical information under a more reducing atmosphere. It is noted that the manganese K-edge (6539 eV) is noisy, particularly for LSMC/LAO samples, due to contributions from the lanthanum L 1 -edge (6266 eV) which reduce the signal intensity.
For both LSMC/LAO and LSMC/STO, the cobalt and manganese oxidation states reduce as the temperature is increased, as expected for heating under strongly reducing conditions. The differences between the most oxidized and most reduced recorded values are: Δ Co = 0.78 and Δ Mn = 1.08 for LSMC/LAO and Δ Co = 0.85 and Δ Mn = 1.15 for LSMC/STO. For both transition metals, the oxidation states in the LSMC/LAO thin film are relatively less reduced than for LSMC/STO, but for both films there is a greater reduction of the manganese oxidation state supporting our theory of a higher strain induced in the MnO bonds. It is also worth noting that Co and Mn seem to be unevenly distributed within the films showing a relatively larger quantity of manganese at the film-substrate interface and a higher segregation of Co in the near-surface. Generally, thin films will have the highest strain at an interface with a single crystal substrate, and relax further from the substrate interface, suggesting that manganese experiences greater strain effects in the strained LSMC films.

Conclusion
A fundamental understanding of the effect of strain on the structural and chemical composition of perovskites is crucial to prove the potential of strain-induced oxygen nonstoichiometry systems for practical applications. This work has shown that compressive and tensile strain of the perovskite LSMC result in a linear trend in unit cell volume with strain, demonstrating the impact of strain on the structural properties of LSMC thin films. Extensive characterization of the chemical properties of the LSMC thin films confirmed that these structural changes are also associated with a change in the oxygen content and the surface chemistry of the films. Specifically, we prove that tensile strain favors lower oxygen content that is associated with a higher reduction in the Mn oxidation state. We correlate this to a higher tensile strain-induced debilitation of the shorter MnO bond. Furthermore, tensile strain is proven to induce a higher segregation of Sr secondary phases in the www.advmatinterfaces.de surfaces as suggested by previous studies. Under strong reducing conditions, the role of tensile strain was evident in the greater reduction and stability of the transition metal oxidation states at high temperature, weaker reducing environments could however lead to misleading conclusions if the driving force for further reduction of films is not reached. These results highlight the importance of using ex situ and in situ complementary techniques to really isolate and understand the effect of strain in the chemistry of thin films.
Thin Film Growth: Thin films of LSMC were deposited onto (001) single crystals of lanthanum aluminate (LAO), lanthanum aluminatestrontium aluminum tantalate (LSAT), and strontium titanate (STO) from CrysTec GmbH Kristalltechnologie (see Table S1, Supporting Information, for lattice parameters). Mismatch between the pseudocubic oxidized LSMC lattice parameter and the substrates was calculated using [37] LSMC/substrate substrate L SMC,ox where f is the lattice mismatch and a is the lattice parameter. Substrates were single-side epi-polished, single crystals cleaned by sonication in acetone, isopropanol, and distilled water prior to deposition. Films were grown by PLD using a KrF excimer laser (λ = 248 nm, Lambda Physik COMPex 201) with a target-to-substrate distance of 128 mm in a modified Neocera combinatorial PLD chamber. Before growth, the substrates were exposed to vacuum (≤5 × 10 −5 Torr) and heated (800 °C, ramp: 30 °C min −1 ). The deposition pressure (30 mTorr) was achieved by the introduction of oxygen (BOC). The target was pre-ablated (300 pulses at 4 Hz) to ensure cleanliness of the target surface. LSMC films were deposited at 5 Hz and subsequently cooled (10 °C min −1 ) in an oxygen-rich atmosphere (600 Torr).
Thin Film out-of-Plane X-Ray Diffraction: Thin film out-of-plane XRD data were collected using a PANalytical X'Pert Pro MRD diffractometer with parallel beam optics, copper Kα radiation (λ = 1.5418 Å), a 1/16° slit, an X'Celerator detector, and a four-angle goniometer with a crystal monochromator operated at 40 kV and 40 mA. High-resolution XRD was collected using a PANalytical Empyrean diffractometer using a hybrid monochromator (2 × Ge (220)-type channel-cut monochromator), a 4 mm fixed mask, a 1/32° divergence slit, and PIXcel3D Detector. A zeroshift measurement was be carried out before all thin film measurements on the PANalytical instruments to ensure no instrumental offsets were present.
In Situ X-Ray Diffraction: In Situ XRD measurements were performed and data fitted following the method previously described by Moreno et al. [41,42] A Panalytical X'Pert PRO MRD diffractometer was used with a multichannel, fast, linear, solid-state PIXcel detector and a goniometer with a radius of 320 mm with the sample mounted inside an Anton Paar DHS-1100C chamber. The films were heated to 600, 650, 700, and 750 °C and cycled between a synthetic air (≈2.1 × 10 5 ppm, 20% O 2 /80% N 2 ) and nitrogen (O 2 ≈ 10 ppm, 100% N 2 ) atmosphere. The cycle times were adjusted depending on the reaction kinetics (from 1 to 4 h), with shorter measurement times at higher temperatures. Gas flushing times were sufficiently fast (≤3 s) to neglect this contribution when determining the overall time response of the cell parameters. 2θ scans were collected with an acquisition time interval of 10 s over a range of 2.51° in static mode (255 channels with a resolution of ≈0.01°) of asymmetrical reflections (ω ≠ θ), namely, the cubic 103 of both the thin film and the substrate. The recorded patterns were fitted with two pseudo-Voigt curves, for the thin film and substrate trace. The thin film peak position was corrected by fixing the position of the substrate (to account for slight variations in the temperature) and the relative c-axis, out-of-plane, parameter was calculated using the following equation [37] 2sin cos where θ is the peak angle, ω is the incident angle, λ is the wavelength of incident X-rays, and l is the Miller index of the chosen reflection. The exchange coefficients were extracted by fitting the changes in cell parameter with a two-part exponential (reduction, t 1 < t 2 ) or single (oxidation, t 1 only) exponential component in Origin using a Levenberg-Marquart algorithm to find the damped least-squares fit of the (double) exponential decay given by where y 0 and x 0 are offsets, A is amplitude, and t is the decay constant. Reciprocal Space Mapping: High-resolution RSM was measured using both a PANalytical X'Pert MRD diffractometer and PANalytical Empyrean, both with a hybrid monochromator (2 × Ge (220)type channel-cut monochromator). With the X'Pert MRD, the monochromator was mounted with a nickel filter (0.020 mm), a fixed mask (2 mm), a divergence slit (1/2°), and a soller slit (0.4 rad) on the incident beam path. In the diffracted path, a PIXcel detector was used as a 1D linear detector with an antiscatter slit (7.5 mm). With the Empyrean diffractometer, a fixed mask (4 mm) and a divergence slit (1/32°) were used with a PIXcel3D detector.
X-Ray Reflectivity: Thickness of the thin films was determined using X-ray reflectivity data collected with a PANalytical Empyrean diffractometer using an incident beam pathway, consisting of a hybrid monochromator (2 × Ge (220)-type channel-cut monochromator), a PreFIX triple-axis analyzer crystal with a proportional detector, and an automatic attenuator (Ni 0.125 mm). The thin films were scanned from ≈0.1° to 2.0° with a step size of ≈0.005°. PANalytical Reflectivity software, using the Parratt reflectivity formalism, [62] was used to fit a 1D model to the data for the purposes of extracting the thin film thickness. The thickness of the LSMC layer from the thin films used in this study was between ≈10 and 85 nm.
Rutherford Backscattering Spectrometry: RBS measurements were performed using a 2 MeV He ion beam and a silicon PIN diode detector mounted at 168° to the incident beam direction. The results were analyzed using the RUMP simulation code. [49] Residual planar channeling in the substrate was partially compensated in the simulation to obtain a better backgroçund fit. The composition was normalized using the condition: La + Sr = 1.
Low Energy Ion Scattering: LEIS data were obtained using an ION-TOF GmbH Qtac instrument under UHV conditions (1.1 × 10 −8 to 1.8 × 10 −8 mbar) on as-deposited thin films. Incident ions were produced by a heated filament in a He and Ne ion mixed gas atmosphere, accelerated through a potential to achieve energies of 3 and 5 keV, for www.advmatinterfaces.de He + and Ne + , respectively. An argon ion beam of 0.5 keV was used for low-energy sputtering with 10 s sputtering intervals and 1 s breaks between sputtering and analysis. Scattered ions were detected at 145° by a double toroidal analyzer. An unfocused electron shower was used to avoid charging of the surface. For the He spectrum of LSMC/STO, the dose density used was 2.75 × 10 14 ions cm −2 over an area of 1000 µm 2 . For the Ne depth profile of LSMC/STO, the dose density of the Ne beam was 2.81 × 10 15 ions cm −2 over an area of 1000 µm 2 , and the total dose of the Ar + sputtering beam was 4.81 × 10 17 ions. For the Ne depth profile of LSMC/LAO, the dose density of the Ne beam was 1.30 × 10 15 ions cm −2 over an area of 1000 µm 2 , and the total dose of the Ar + sputtering beam was 2.19 × 10 14 ions. Data were presented after the removal of an exponential background and all analysis was performed using Gaussian fits in IONTOF SurfaceLab 6. The data were collected from multiple areas of measurement on thin films deposited under comparable conditions, with a LSMC layer thickness of ≈18 nm.
X-Ray Photoelectron Spectroscopy: Spectra were recorded on a Thermo Scientific K-Alpha + X-ray photoelectron spectrometer operating at 2 × 10 −9 mbar base pressure. This system incorporates a monochromated, microfocused Al Kα X-ray source (hν = 1486.6 eV) and a 180° double focusing hemispherical analyzer with a 2D detector. The X-ray source was operated at 6 mA emission current and 12 kV anode bias providing an X-ray spot size of up to 400 µm 2 . Survey spectra were recorded at 200 eV pass energy, 20 eV pass energy for core level. A flood gun was used to minimize the sample charging.
X-Ray Absorption Near Edge Spectroscopy: XANES measurements were obtained on Beamline B18 at Diamond Light Source, Harwell Science and Innovation Campus, Oxfordshire, operating with a ring energy of 3 GeV and at a current of 300 mA. Manganese and cobalt K-edges were measured in the ranges 6400 to 7200 eV and 7600 to 8300 eV, respectively. Calibration of the monochromator was performed using manganese and cobalt metal foils prior to measurement. Thin films were measured in fluorescence with a nine-element germanium detector. The data were analyzed using Athena from the Demeter suite, which implements the FEFF6 and IFEFFIT codes. [63,64] In situ measurements were conducted inside a furnace on the Beamline between room temperature and 500 °C. The atmosphere in the furnace was evacuated, filled with an inert atmosphere before being placed on the beamline with an atmosphere consisting of hydrogen (Air Products, H 2 4.018 mol%, N 2 95.981 mol%) diluted in nitrogen (Air Products, N 2 BIP, O 2 < 10 ppb, H 2 O < 20 ppb, CO+CO 2 < 0.5 ppm, CH 4 < 100 ppb, H 2 < 1 ppm) to 3% H 2 /balance N 2 . The furnace had a large interior volume and a porous lining, therefore the experiments were conducted with the furnace overpressurized to prevent leakage into the furnace.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.