Oxygen Nonstoichiometry and Valence State of Manganese in La1–xCaxMnO3+δ

Perovskites of the ABO3 type, such as LaMnO3, can be used as air electrodes in solid oxide fuel cells and electrolyzers. Their properties can be tuned by A- and B-site substitutions. The influence of La substitution by Ca on the oxygen nonstoichiometry has been investigated frequently, but the results depend highly on the synthesis and atmospheric conditions. In this work, a series of La1–xCaxMnO3+δ (x = 0–0.5) was synthesized using conventional solid-state synthesis under an air atmosphere. The structures of the materials were studied in detail with powder X-ray diffraction. The initial oxygen nonstoichiometries were determined using thermogravimetric reduction. The samples were subsequently analyzed in terms of defect chemistry in dependence of temperature, atmosphere, and Ca content via thermogravimetric analysis. The changes in the manganese charge states were investigated by X-ray absorption near-edge spectroscopy experiments. The influence of intrinsic and extrinsic effects on the Mn-valence state of the differently Ca-substituted samples as calculated from thermogravimetric analysis and as determined directly from X-ray absorption near-edge spectroscopy is presented.


■ INTRODUCTION
Lanthanum−manganese oxides are frequently discussed for application as oxygen electrodes, for example, in solid oxide cells, as catalysts or as oxygen sensors and supercapacitors. 1−5 Some of them are also well known for their colossal magnetoresistive effect, such as Ca-doped LaMnO 3 . 6−8 Substitutions at the lanthanum site, for example, by calcium, revealed the possibility to change the material structure and defect chemistry. Defect chemistry is mainly governed by oxygen nonstoichiometry and manganese charge state and plays an important role for the catalytic activity of the material. 9−12 Material synthesis methods and conditions affect the defect chemistry. 12−17 As they are exposed to varying atmospheres at high temperatures during their application, it is important to investigate how the defect chemistry changes under varying temperature and atmosphere. In contrast to the classical oxygen-deficient perovskites, such as La 1−x A′ x BO 3−δ (A′ = Ca or Sr and B = Fe or Co), LaMnO 3 and its Casubstituted derivatives are frequently described as oxygenexcess perovskites in the form of La 1−x Ca x MnO 3+δ . Investigations on the oxygen excess in La 1−x Ca x MnO 3+δ have been performed, and it appeared that oxygen excess in La 1−x Ca x MnO 3+δ is not possible by the introduction of oxygen into interstitial sites. Instead, it was found that A-and B-site cation vacancies in the form of (La, Ca) 1−ε Mn 1−ε O 3 (ε = δ/(3 + δ)) are established. 14,18−20 However, for comparability with the literature, in general, the oxygen excess is considered.
The material properties of the unsubstituted LaMnO 3+δ highly depend on its oxygen nonstoichiometry δ and the corresponding Mn-oxidation state. 18 Previous structural studies suggested three different crystal structures for room temperature LaMnO 3+δ . Stoichiometric LaMnO 3 has the pseudocubic O′-orthorhombic structure ( c a b 2 < < , Pnma) and can transform into a less distorted O-orthorhombic structure. 12,18,21 Bogush et al. 21

≈ ≈
The transition into a rhombohedral structure (R3̅ c) is also possible. 21 The perovskite structures are mainly dependent on the relative equilibrium bond lengths within the perovskite. The Goldschmidt tolerance factor gives an indication to which extent the bond lengths (B−O) and (A−O) deviate from the ideal cubic perovskite with the A-site cation on (0 0 0) and the B-site cation on (1/2 1/2 1/2). 17,22 For a perovskite under ambient conditions, it can be calculated using the ionic radii (e.g., after Shannon 23 ) as presented in the following eq 1.
For the cubic perovskites, the Goldschmidt tolerance factor is 1. Deviations represent a decrease in the symmetry of the materials. In unsubstituted LaMnO 3 , the Goldschmidt tolerance factor is t = 0.954, assuming a coordination number of 12 for La and a coordination number of 6 for all other ions. 18 With increasing Mn 4+ content, the Goldschmidt tolerance factor increases, as pointed out by Topfer and Goodenough. 18 An increase in the Mn 4+ concentration within LaMnO 3 can be obtained intrinsically by influencing the oxygen stoichiometry and the formation of defects or extrinsically by A-site substitution, for example, with Ca.
Starting with 100% Mn 3+ as in LaMnO 3 , the MnO 6 octahedra are strongly Jahn−Teller-distorted and three different Mn−O bond lengths can be found. This results in the strongly distorted O′-phase with c a b 2 < < as a result of the superposition of the Jahn−Teller distortion with the orthorhombic distortion introduced by the tolerance factor t < 1. 12,18,21 This distortion can be reduced by the incorporation of oxygen (LaMnO 3+δ ), which is accompanied by an increasing Mn 4+ content. At about 14% Mn 4+ , the Jahn− Teller distortion disappears and only one Mn−O bond length is present within the MnO 6 octahedra. In this O-phase, the orthorhombic distortion is only induced by the tolerance factor t < 1 and the octahedra are tilted around the a and b axis. 12,18,21 With even higher Mn 4+ contents (20−22%), the structure becomes rhombohedral (R-phase) and the octahedra are tilted around the cubic ⟨111⟩ directions. 21,24 The interrelation between the Mn-oxidation state and oxygen stoichiometry has been explained by the establishment of mixed oxidation state structures (Mn 3+ /Mn 4+ ) within LaMnO 3+δ , as shown in Scheme 1. 18 According to this, the Mn 4+ content is correlated with 2δ.
Black squares ■ represent the cation vacancies on the Aand B-sites. Oxygen uptake (change in the δ-value) and simultaneously partial change in the Mn-oxidation state from Mn 3+ to Mn 4+ lead to a reduced distortion. These changes within stoichiometric LaMnO 3 are not influenced by substitution, and as such, we define them as intrinsic. Topfer et al. 18,25 described that if δ < 0.1, the structure is O′orthorhombic, while it is rhombohedral for δ > 0. 1 12,20 Substituting LaMnO 3 with divalent cations, such as Ca 2+ , leads to an extrinsically controlled transition of Mn 3+ to Mn 4+ , which stabilizes the structure from O′-orthorhombic to Oorthorhombic due to increasing Mn 4+ content. 12 Dabrowski et al. 12 showed that the O′-phase is stable only for very small δvalues close to zero and calcium contents x < 0.15.
Similar to Scheme 1 (without cation vacancies), the Mnoxidation state of the calcium substituted samples can be calculated as presented in Scheme 2.
In this, the influence of both, the change in the δ-value and the introduction of Ca 2+ (leading to a forced transition of Mn 3+ to Mn 4+ ), on the Mn 4+ content is considered.
Scheme 2 stresses the importance of experimentally derived δ-values for the determination of the Mn 4+ content. In this work, we present the results of the systematic investigation of the La 1−x Ca x MnO 3+δ system with respect to the changes in its oxygen stoichiometry and Mn-oxidation state in dependence of Ca content, atmosphere, and temperature by powder X-ray diffraction (PXRD), thermogravimetric analysis (TGA), and X-ray absorption near edge spectroscopy (XANES). Powder X-ray Diffraction and Rietveld Refinement. Qualitative characterization of the calcined powders was performed with PXRD on a PANalytical Empyrean Series 2 diffractometer equipped with a PANalytical X'Celerator detector, using a Cu Kα radiation source. The PXRD measurements for Rietveld refinement were performed using a Bruker AXS D8 Advance powder diffractometer with Cu Kα radiation (λ 1 = 154.0596 pm and λ 2 = 154.4493 pm) equipped with a LynxEye semiconductor strip detector (SSD) and Ni filter. θ/θ scans were performed for each sample in three ranges in steps of Δ2θ = 0.0105°: (i) 10−60°2θ, counting 0.8 s per step, (ii) 60−100°2θ, counting 1.7 s per step, and (iii) 100−140°2θ, counting 3.5 s per step, summing up to about 7 h scanning time per sample. Powders were prepared onto flat sample holders. Structure refinements with the Rietveld method 26 were performed using the TOPAS Academic suite. 27 The refinements were carried out simultaneously against all three measurement ranges taken for each of the compounds. A fundamental parameter approach was used to calculate the line profiles. 28 The lattice parameters of the perovskite-type phases, the fractional coordinates of the atoms (for those coordinates allowed to vary by symmetry), isotropic thermal displacement parameters for each element (or the site for mixed occupancies), the sample height, and size and strain contributions to the peak shape were refined simultaneously against all ranges as "global" parameters. A 6-to 10-coefficient background polynomial, a scaling factor, and a minor Cu Kβcontamination were refined individually for the ranges.
Determination of the Initial Oxygen Nonstoichiometry. The initial oxygen content in the studied materials was determined using a thermogravimetric analyzer (TGA 5500 from TA Instruments). The powdered sample (50 mg) was weighed into a platinum crucible. Pressurized air (99.999%) and forming gas (99.1%) consisting of argon and 4% hydrogen have been used for TGA experiments. The calcined materials were heated twice to 950°C (5 K/min) in a pressurized air atmosphere and then cooled (5 K/min) down, in order to stabilize the system. In an Ar/4%H 2 flow, the sample was slowly heated to 900°C with a rate of 0.5 K/min. Mass changes were continuously recorded, and the final masses after reduction (under Ar/4%H 2 atmosphere) were taken for calculation of the initial oxygen nonstoichiometry (eq 2). The materials are decomposed upon reduction as shown in Scheme 3. The decomposition into such a mixed structure was verified by PXRD (see Figure S2). As presented in Scheme 3, the knowledge of the decomposition products allows the calculation of the samples' initial delta values via their mass loss upon reduction according to eq 2. It also shows that the oxygen content in the final solid products is lower than that in here m i describes the initial sample mass of La 1−x Ca x MnO 3+δ , m e is the end mass of the decomposition products, M e is the molar mass of the decomposition products, M LCMO is the molar mass of La 1−x Ca x MnO 3.00 , and M O is the molar mass of oxygen.
Determination of Changes in Oxygen Nonstoichiometry in Various Atmospheres and Temperatures. Aircalcined La 1−x Ca x MnO 3+δ (x = 0−0.5) and additionally nitrogen-tempered LaMnO 3 and La 0.9 Ca 0.1 MnO 3+δ were measured in oxygen (99.9992%), nitrogen (99.9992%), and synthetic air (80% N 2 , 20% O 2 ) in a thermogravimetric analyzer (STA 449 F1 Jupiter from Netzsch) to determine changes in their oxygen stoichiometry. An aluminum oxide crucible containing 500 mg of the sample was used for these measurements. For buoyancy correction, measurements were conducted on empty crucibles for all applied gases and temperatures. Two cycles were performed: in each, the samples were heated with 5 K/min to 950°C, held for 5 h, and cooled down with the same rate to room temperature. In the first cycle, the desired atmosphere is applied to the air-calcined samples. Reversibility of the data obtained after cycle 1 was verified by performing a second measurement cycle. Experiments with a third cycle (not shown) revealed that the second cycle already delivers reversible results. With the help of the previously determined initial δ-values via reduction, the measured mass changes could be correlated with the samples' changes in their δ-value and, therefore, changes in their oxygen nonstoichiometry, which in the following will be written as 3+δ.
Temperature-dependent TGA of La 0.9 Ca 0.1 MnO 3+δ was performed in O 2 to determine whether the sample is in equilibrium when heated up to 950°C. Therefore, air-calcined La 0.9 Ca 0.1 MnO 3+δ was investigated from 550 to 950°C in 100°C steps. As described before, two cycles with a 5 h holding time at the respective temperature were performed. The resulting mass changes were correlated to the samples' specific oxygen nonstoichiometry.
The same two cycles up to 950°C were performed for LaMnO 3 that was either calcined in air or additionally tempered in N 2 before measurement. These measurements were performed in O 2 , N 2 , and synthetic air.
Investigation of the Mn Electronic State Using XANES. Mn K-edge X-ray absorption spectra were recorded at beamline P65 at PETRA III, Deutsches Elektronen-Synchrotron (DESY) in Hamburg, Germany. 29 The undulator X-ray beam is focused using a water-cooled fixed exit Si double-crystal monochromator, using the Si 111 lattice plane. In order to avoid contributions from higher harmonics, two plane mirrors with variable angles of incidence are placed in front of the monochromator. For the measurements, the powdered samples were mixed with appropriate amounts of cellulose powder and pressed into thin pellets, which were placed in a sample holder that is mounted in the beam path. Absorption spectra were recorded in transmission geometry, with the first ionization chamber placed in front of the sample collecting incident beam intensity and a second ionization chamber collecting sample intensity to yield the absorption coefficient μ(E).
The photon energy used for measurements ranged from 6400 to 7500 eV. The spectra were normalized using a linear pre-edge and a second-order polynomic post-edge utilizing the program Athena. 30 O K-edge and Mn L 32 -edge spectra were collected on powdered samples at the UE56/1-SGM beamline at Helmholtz-Zentrum Berlin, in the total electron yield detection mode. Spectra are two-point normalized to the post-and preedge.

■ RESULTS AND DISCUSSION
Powder X-ray Diffraction. PXRD measurements of the Ca-substituted La 1−x Ca x MnO 3+δ (x = 0−0.5) series synthesized at 1100°C in air are shown in Figure 1a. They indicate a change in the crystal structure from a rhombohedral R3̅ c structure of the unsubstituted LaMnO 3+δ sample to an orthorhombic structure Pnma of the Ca-substituted samples. LaMnO 3+δ shows a change in the space group to orthorhombic Pnma when tempered for 4 h at 1000°C in a nitrogen atmosphere ( Figure 1b).
Comparison of LaMnO 3 and La 1−x Ca x MnO 3+δ reveals a shift of the main reflection position toward higher angles with increasing Ca content. This indicates that the unit cell volume shrinks on substituting La 3+ by Ca 2+ . This is crystal-chemically explainable, as the ionic radius of Ca 2+ (134 pm) is slightly smaller than that of La 3+ (136 pm). 23 In addition to this, the increasing oxidation of Mn 3+ to Mn 4+ and the resulting changes toward smaller ionic radii are expected to bias the lattice parameters as well. 31,32 Figure 1c shows an exemplary refinement of La 0.7 Ca 0.3 MnO 3+δ using the Rietveld method. Rietveld refine-Scheme 3. Decomposition reaction for thermogravimetric reduction ments were performed in detail against all diffraction data shown in Figure S1a. The obtained lattice parameters and unit cell volumes are presented in Table S1 and Figure 2a . 12 Only pure LaMnO 3+δ is either rhombohedral R when calcined in air (δ = 0.137, Table 2) or in the strongly distorted O′-phase with c a b 2 < < when calcined in nitrogen (δ = 0.076, Table 2). This agrees well with the values published by Topfer et al., 18,25 who state that at room temperature, the O′-phase is stable for 0 ≤ δ ≤ 0.11 and the rhombohedral phase for δ ≥ 0.11. Bogush et al. 21 find similar conditions for the rhombohedral phase (Mn 4+ ≥ 22%, δ ≥ 0.11), but they found the O-phase for 0.07 ≤ δ ≤ 0.11 and the O′-phase for 0.00 ≤ δ ≤ 0.07. The table shown by Dabrowski et al. 12 gives similar values for pure LaMnO 3+δ (R for δ ≥ 0.11, O for 0.04 ≤ δ ≤ 0.11, and O′ for 0.00 ≤ δ ≤ 0.04).
The unit cell volume of rhombohedral LaMnO 3+δ , synthesized in air, is comparable to that of La 0.9 Ca 0.1 MnO 3+δ . When LaMnO 3+δ is tempered in nitrogen, the unit cell volume increases drastically (>0.3%) from 58.8582(4) to 60.8625(8) Å 3 . According to the literature, LaMnO 3 experiences significant structural deformation in a nitrogen atmosphere, as the amount of Mn 3+ in the structure increases, leading to stronger Jahn−Teller-distortion. 16 This leads to a highly distorted structure with a larger cell volume (orthorhombic). When synthesized in air, Mn 3+ is increasingly oxidized to Mn 4+ , which is not as affected by Jahn−Teller distortions. This diminishes the distortion within the structures in accordance with the formation of a rhombohedral structure, which exhibits a smaller unit cell volume. The shrinkage of the unit cell has been explained by Miyoshi et al. 31,33,34 as a result of oxygen incorporation, which indeed occurs by the formation of additional unit cells and cation vacancies. The latter is compensated by the transition of Mn 3+ to Mn 4+ which also reduces the cationic radius. With increasing Ca content, a decrease in the lattice parameters a, b, and c is observed. This trend corresponds to a shrinkage of the cell volume with increasing Ca content and has also been reported in the literature by Sagdeo et al. 32 The explained and discussed trends from Table S1 are presented in Figure  . For LaMnO 3+δ , a visible difference appears.
While c still follows the trend as before, b 2 decreases and a increases, and therefore, the structural transition to the O′phase with c a b 2 < < can be observed for the sample treated in N 2 . The air-calcined sample is rhombohedral with the trigonal lattice parameter a (α = 60.68°) being slightly smaller than all orthorhombic parameters for La 0.9 Ca 0.1 MnO 3+δ .
The formation of cation vacancies, as well as Ca substitution, also affects the Goldschmidt tolerance factor. To determine this factor, the exact stoichiometry of La 1−x Ca x MnO 3 needs to be determined. Therefore, oxygen excess obtained from the TGA was converted into cation defects, and the corresponding quantity of Mn 3+ and Mn 4+ was determined, as described above in Scheme 1. The as-calculated contents of La, Ca, Mn, and O within the compositions were used for the Goldschmidt factor calculations. The ionic radii of the cation vacancies themselves were not explicitly included. The results are shown in Table 1.
For the air-synthesized samples, an increasing Goldschmidt tolerance factor with increasing Ca substitution is observed.
The Goldschmidt tolerance factor ranges from t = 0.961 (x = 0) to t = 0.979 (x = 0.5), thus moving closer to the ideal cubic value (t = 1). This hints at an increasing symmetry and stability of the crystal structures with higher Ca contents. If the air-synthesized LaMnO 3+δ is treated additionally in N 2 , the Goldschmidt tolerance factor is calculated to be 0.958 (x = 0, N 2 ), showing a decreased symmetry within this sample. This transition from the O′-to R-structure with an increasing tolerance factor for pure LaMnO 3+δ has already been postulated by Topfer and Goodenough. 18 Calculation of the Oxygen Nonstoichiometry Using Thermogravimetry under a Reducing Atmosphere. Thermogravimetric reduction of La 1−x Ca x MnO 3+δ (x = 0.1− 0.5) in an Ar/4%H 2 atmosphere allows for the determination of the initial oxygen nonstoichiometries of the samples.
The reduction of the materials within TGA experiments delivers the start mass (m i ) and end mass (m e ) of the samples before and after reduction and therefore is the basis for the calculation of δ.
It is assumed that only oxygen is released from the sample, and thus, the mass difference is only caused by oxygen. An example of how the δ-value is calculated from the mass loss measured during reduction of La 0.9 Ca 0.1 MnO 3+δ is shown in Figure 3, in which, according to Scheme 3, the amount of From the relative end mass m e /m i (Table 2), the molar mass of stoichiometric La 0.9 Ca 0.1 MnO 3 , and the molar mass of its decomposition products (La 2 O 3 and Ca 0.1/1.1 Mn 1/1.1 O), the respective δ-value can be calculated using eq 2. The decomposition products were characterized by PXRD measurements, as shown in Figure S2. For the specific Ca contents, the decomposition products are assumed to form according to Scheme 3. In the case of La 0.9 Ca 0.1 MnO 3 , a value of δ = 0.088 has been determined. The reduction of all airsynthesized samples  in Figure S3. The corresponding relative end masses and the calculated δ-values are listed in Table 2.
From these data, it can be read that the amount of released oxygen increases along with increasing Ca content in the studied materials. These higher mass losses due to oxygen release do not correlate with higher nonstoichiometry. In contrast, higher Ca contents in the materials, although inducing higher mass losses upon reduction, result in smaller nonstoichiometry. This is especially the case for the samples with x ≥ 0.3. The reason for this lies in the different decomposition products that are included in the δ-calculation. LaMnO 3+δ decomposes to La 2 O 3 and MnO, whereas La 1−x Ca x MnO 3+δ (x = 0.1−0.5) decomposes to La 2 O 3 and Ca x/(x+1) Mn 1/(x+1) O, as can be seen from Scheme 3 and Figure  S2. This also introduces additional stages in the reduction reaction ( Figure S3). Oxygen incorporated in La 2 O 3 and Ca x/(x+1) Mn 1/(x+1) O is not released and therefore does not appear as an absolute mass loss but is considered in the δcalculations.
Ca substitution for La correspondingly decreases the amount of La 2 O 3 in the decomposition products. At the same time, an increase in the decomposition product Ca x/(x+1) Mn 1/(x+1) O occurs, which contains less oxygen per metal than La 2 O 3 , resulting in higher absolute oxygen release.
The obtained δ-values in this work are in agreement with the previously reported results. The δ-value of rhombohedral LaMnO 3+δ with δ > 0.13 is significantly above the limiting value described in the literature, at which the orthorhombic structure occurs (δ < 0.1). 18,25 The structural change under a nitrogen atmosphere to an orthorhombic structure happens due to increased amounts of Mn 3+ compensating for the lower oxygen nonstoichiometry and resulting in Jahn−Tellerdistortions, which also was reported by Topfer et al. 18,24 La 0.9 Ca 0.1 MnO 3+δ exists in a less-distorted, orthorhombic structure (O). The literature describes this structure to be present at 0.077 < δ < 0.159. 12 The observed value of δ = 0.088 fits well with the literature values. For the samples x ≥ 0.3, the orthorhombic O structure and δ-values near zero have been also observed by Dabrowski et al. 12 Thermogravimetric Experiments on La 1−x Ca x MnO 3+δ as a Function of Atmosphere (O 2 , Air, and N 2 ). In this section, we investigated the changes in oxygen nonstoichiometry within the La 1−x Ca x MnO 3+δ series as a function of atmosphere and temperature. TGA of La 1−x Ca x MnO 3+δ (x = 0−0.5), calcined at 1100°C in air, was performed at 950°C in synthetic air (80% N 2 , 20% O 2 ), O 2 , and N 2 atmosphere. Always, two cycles were measured.    Figure 4 shows the results of two samples (x = 0 and 0.5) cycled in a pure oxygen atmosphere. The first of the cycles gives the mass changes along with the heating of the materials, previously synthesized in air, under an oxygen atmosphere. The material is not in the equilibrium state. The second cycle is designed to detect reversible temperature-induced mass changes. Mass increases and decreases are associated with oxygen uptake and release. Different ranges I−V were introduced in order to describe the occurring processes. For x = 0, five ranges can be recognized. In range I, below 550°C, an irreversible mass loss occurs. In range II, between 550 and 750−800°C, the mass increases. In range III, a reversible mass loss occurs for T > 750−800°C. Range IV describes the dwell time at the maximum temperature applied (950°C), in which the mass stays constant. On cooling, summarized as range V, the process III reproduces in the reversed form, while the partially irreversible processes I and II may modify the cooling profile but cannot be distinguished clearly.
In the first cycle, the unsubstituted sample, after a slight initial mass decrease in range I, exhibits a mass increase of ≈0.03% in range II. Upon further heating, a mass decrease of about 0.2% occurs (range III). During the dwell time at 950°C, the mass remains constant (range IV), followed by a mass increase of 0.26% upon cooling. It finally results in an overall mass gain of the room temperature structure. At this elevated oxygen content, the second heating cycle shows a symmetric profile and is leading to the same mass as after cycle 1.
For x = 0.5, in the first cycle, only the irreversible mass loss in range I occurs and is much more pronounced (≈0.06%). In the second cycle, a small mass loss in range I is still recognizable, yet the final mass after two cycles is similar to the mass obtained after cycle 1. Figure 5 displays the thermogravimetric measurements of all air-synthesized samples in three different atmospheres (O 2 , synthetic air, and N 2 ). For the measurements in an oxygen atmosphere (Figure 5a), it can be observed that the samples with Ca contents below x = 0.3 behave similar to the behavior described above for the unsubstituted sample LaMnO 3+δ , while the increasingly substituted samples behave similar to La 0.5 Ca 0.5 MnO 3+δ .
In the first measurement cycle, the slight mass loss in range I becomes stronger with the growing amount of Ca. It is assumed to occur due to not fully equilibrated samples after synthesis, possibly due to the different temperature profiles and the fast cooling rate standing in contrast to the slow equilibration of the samples. Up to x = 0.3, this mass loss is followed by a slight increase in range II, which becomes weaker with increasing Ca content. It vanishes for x ≥ 0.4. The mass gain suddenly is terminated at around 750°C at the start of range III. This is caused most likely by a structural change. 31 The reversible mass change in range III increases from x = 0 to x = 0.1 and then decreases until it does not occur for x ≥ 0.4 anymore. The process behind this is the equilibration of the sample to temperature and atmosphere in the high-temperature regime as described by Mizusaki et al. 20 and Miyoshi et al. 31 and only occurs for samples containing excess oxygen.
In the second cycle, the mass loss in range I and the mass gain in range II are almost absent. The reversible mass loss in range III has been calculated. As the sample is in equilibrium only after cycle 1, the mass loss from RT (after cycle I) to 950°C (cycle 2) has been calculated. This mass loss amounts to −0.26, −0.25, and −0.19% for the samples x = 0, x = 0.1, and x = 0.2, respectively. It is much less for x = 0.3 and vanishes for x = 0.4 and x = 0.5. In a contour plot, showing the mass change, the effect of this oxygen release is summarized ( Figure 6). The mass difference is calculated from the mass at RT after cycle 1, when the sample is in equilibrium, and from the masses during the heating cycle up to 950°C (cycle 2). It makes clear that in region III, cycle 2, the oxygen release decreases with the Ca content x.
In a synthetic air atmosphere (Figure 5b), the samples' behavior is comparable, although the extent of mass changes differs as the oxygen partial pressure is reduced and is basically similar to the one during synthesis. An overall mass loss within the room temperature structure can be observed after two cycles for all studied samples. This is in contrast to the behavior of LaMnO 3+δ in a pure oxygen atmosphere, for which an overall mass gain was observed. This can be attributed to the process in range III and its reverse process in range V, which does not recover the initial mass. It can be explained by the equilibration in the high-temperature regime, this time under lower oxygen partial pressure. 20,31 The studied materials behave differently under a nitrogen atmosphere (Figure 5c). In the first cycle, upon heating, a significant mass loss in range III (0.8% for of LaMnO 3+δ ) can be observed for the samples containing up to 20% of calcium and only a slight mass loss (−0.05 to −0.07%) can be observed for higher calcium contents, which can be attributed to range I. The mass increase in range II is not observed for any of the samples as no oxygen is provided. Compared to the initial mass, there is an overall mass loss for all samples after two measurement cycles. It is assumed that in nitrogen, the samples are reduced toward their stoichiometric compositions, leading to a massive release of oxygen for those samples with high oxygen nonstoichiometry δ, but is less pronounced for samples without or very small oxygen nonstoichiometries. LaMnO 3+δ and La 0.9 Ca 0.1 O 3+δ still lose oxygen in the second cycle, meaning δ may not have reached zero yet.
Determination of Changes in Oxygen Nonstoichiometry in Various Atmospheres. The mass changes shown in Figure 5 could be further correlated with changes in the oxygen-nonstoichiometry Δδ of the samples according to eq S1. Masses were taken from the three differently named ranges (m 0,initial , m 950°C,2nd , and m end ) in Figure 4.
For all samples, the initial stoichiometry calculated from reduction experiments was taken as a starting point, that is, 100% of the mass (m 0,initial ). The mass changes shown in Figure  5 were correlated with changes in the oxygen-nonstoichiometry Δδ of the samples according to eq S1.
The initial oxygen nonstoichiometries are plotted as the blue stars in Figure 7. The behavior with x is similar to that reported for La 1−x Sr x MnO 3+δ . 20 The changes are displayed for the different atmospheres at 950°C in cycle 2 in Figure 7a and at RT after cycle 2 in Figure 7b. All the relative mass changes and the δ-value calculated from these for Figure 7 are listed in Table S2.
In general, heating up to 950°C (Figure 7a) leads to a decrease in the initial oxygen nonstoichiometry, which can be mainly attributed to range III. The extent strongly depends on the Ca content and atmosphere. For small Ca contents (x = 0−0.2), significant δ-value changes can be observed. For high Ca contents (x = 0.3−0.5), the atmospheric and temperature effects are very small and hardly distinguishable, as the initial oxygen stoichiometry is close to 3.
The temperature-related decrease in the oxygen nonstoichiometry at 950°C for samples with small Ca contents (0 ≤ x ≤ 0.2) could be attributed to the structural change at 750°C and the following equilibration to temperature and atmosphere in the range III. 20,31 For the other samples (x = 0.3−0.5), the observed changes can simply be attributed to the irreversible mass loss during the first heating ramp below 550°C (range I).
The extent of these temperature-related effects differs depending on the applied atmosphere. Taking the initial sample nonstoichiometry as a basis, the smallest changes can be observed when the samples are heated to 950°C in an In air, a change by Δδ = 0.05 (for both x = 0 and x = 0.1) can be observed, which was expected due to the decreased oxygen partial pressure in air. This dependence on oxygen partial pressure at temperatures above 700°C was reported by Miyoshi et al. 31 and Mizusaki et al. 20 The biggest decrease in the δ-value can be seen under a nitrogen atmosphere (with Δδ = 0.13 and 0.10 for x = 0 and 0.1, respectively). This can be attributed to the oxygen partial pressure being drastically reduced, leading to final 3 + δ close to the stoichiometric value of 3, as reported by Mizusaki et al. 20 for La (1−x) Sr x MnO 3+δ (LSM).
Thermogravimetric measurements after cooling the samples back to room temperature (Figure 7b) show that the samples do not only undergo the reported temperature-related changes but also atmosphere-related changes in the room temperature materials occur. Changing the atmosphere from air to oxygen, the synthetic air or nitrogen atmosphere has an influence on the oxygen nonstoichiometry. For materials with calcium contents higher than x = 0.3, the δ-value decreases irreversibly in all atmospheres and remains constant afterward. For lower calcium contents, the δ-value increases in an oxygen atmosphere, as more oxygen is available than during synthesis.
Under the air and nitrogen atmosphere, a decrease in oxygen nonstoichiometry has been observed. No change is expected in the air atmosphere, as the sample was synthesized in air. As can be seen from Figure 5b, while heating up, all samples undergo an irreversible mass loss in range I. We attribute this to a different temperature profile during calcination and the present measurement. In a nitrogen atmosphere, δ decreases toward zero, meaning the samples are almost stoichiometric afterward. Only for pure LaMnO 3+δ , δ remains 0.01, which might be due to slow oxygen release rates at low oxygen partial pressures and 950°C. For materials with small contents of calcium, it is possible to vary the oxygen nonstoichiometry with a changing atmosphere. This is in agreement with the δ/x phase diagram published by Dabrowski et al., 12 which shows less susceptibility to changes in δ for samples with higher x.
Thermogravimetric Analysis of La 0.9 Ca 0.1 MnO 3+δ in Dependence on the Maximum Cycling Temperature. To clarify, in how far the sample x = 0.1 is in equilibrium at various temperatures during the previous heating experiment to 950°C, La 0.9 Ca 0.1 MnO 3+δ was investigated thermogravimetrically with a similar profile, but this time, the maximum temperature was varied in 100°C steps from 550 to 950°C. For each temperature, two cycles were measured in an oxygen atmosphere, with a dwell time of 5 h at the respective  temperatures (Figure 8). The initial structure is always the aircalcined sample.
It can be observed that up to T ∼ 750°C, an irreversible oxygen uptake occurs, reflected by a mass increase, which remains present after cooling to room temperature. The highest mass increase can be observed at 650°C. Heating to 850°C and higher leads to a rapid mass loss and oxygen release at 750°C, which is taken up reversibly upon cooling. This behavior can also be observed qualitatively in the second cycle at 750°C and was discussed before as a reversible structural change occurring at around 750°C. 31 The results show that the processes below 750°C are slow, as expected for lower temperature. The equilibration takes a long time and is still not completed under the presented conditions. This applies especially for 650°C, which is also the temperature where the highest mass gain is achieved. This and the termination of the oxygen uptake at around 750°C are hints for a structural change occurring at T > 750°C. 20,31 Figure 9 shows the corresponding oxygen nonstoichiometry calculated from the mass changes seen in Figure 8, during the second cycle at elevated temperatures (red triangles) and the residual change after cooling down to room temperature (black triangles). Thus, when the sample La 0.9 Ca 0.1 MnO 3+δ is heated to 650°C, a maximum in oxygen uptake with Δδ = 0.02 can be observed for both temperatures (650°C and RT) as it is an irreversible change. Up to a temperature of 750°C, the incorporated oxygen is contained in the structure when the sample is cooled down. In both cases, the oxygen uptake can be attributed to range II. From the TGA curves in Figure 8, it can be inferred that the oxygen uptake is a slow process. When heated to temperatures higher than 750°C, the sample equilibrates with the temperature and atmosphere as described above. The amount of excess oxygen decreases with T. 20 Compared to the initial room temperature value, the δ-value after two cycles is essentially identical.
Thermogravimetric Study of Rhombohedral and Orthorhombic LaMnO 3+δ . The rhombohedral, air-synthesized LaMnO 3+δ (δ = 0.137) was tempered for 4 h at 1100°C in a N 2 atmosphere. The structure changed to the highly distorted orthorhombic O′ modification, as could be confirmed with PXRD and as has been previously reported in the literature. 16 Relating this to the δ-value of 0.076 as measured by thermogravimetric decomposition, this is in agreement with most of the previous studies. 16,18,21 In contrast, Dabrowski et al. 12 have reported the less distorted O structure for this δ-value. Thermal behavior of both polymorphic modifications under two different atmospheres (O 2 and N 2 ) was compared using TGA, and the measurements are displayed in Figure 10. Up to 800°C under an oxygen atmosphere, the N 2 -tempered sample, which is present in the orthorhombic modification, takes up oxygen strongly, leading to a mass change of about Δm = 0.8%. This mass change is much stronger than that observed for the in-air-synthesized sample (Δm = 0.03%). In the oxygen atmosphere, the previously described mass loss due to an expected structural change is visible at around 750−800°C and with a similar decrease (Δm = ±0.22%) as observed for the in-air-synthesized sample. Under the nitrogen atmosphere, slight oxygen uptake and release occurs to some amount, which might be explained by slow equilibration, but the end mass is very close to the initial one.
The rhombohedral structure of the air-synthesized sample heated up in a nitrogen atmosphere, which has been described before, does not show oxygen uptake during the first heating cycle. Starting at around 700°C, it releases a large amount of Figure 9. 3+δ-value of La 0.9 Ca 0.1 MnO 3+δ in an oxygen atmosphere during the second cycle at 550, 650, 750, 850, and 950°C (red triangles) and after subsequent cooling to room temperature (black triangles). As a comparison, the initial state at room temperature (blue stars) is shown. Figure 10. Thermogravimetric investigation of variously synthesized LaMnO 3+δ in O 2 and N 2 atmospheres. Heating and cooling were performed with a rate of 5 K/min, and the holding time at 950°C was set to 5 h. For each measurement, two cycles were performed. oxygen upon heating under a nitrogen atmosphere. It can be seen that with the choice of the atmosphere and temperature profile, the state of the sample can be changed. This shows that differently synthesized or treated samples (air and N 2 ), starting with different δ-values (δ = 0.137 vs δ = 0.076), are in the same state after exposure to the oxygen atmosphere at temperatures above 750°C. The sample with a lower δ-value at the start takes up significantly higher amounts of oxygen compared to the one with a higher δ-value. In a nitrogen atmosphere, the sample with a higher δ-value correspondingly releases much more oxygen than the sample with lower initial δ.
Calculation of the Mn-Oxidation State in the La 1−x Ca x MnO 3+δ Series Using Thermogravimetric Analysis. If the δ-value is known, the Mn-oxidation states can be calculated from Scheme 2, where the Mn 4+ content mainly corresponds to 2δ + x. In general, the defect chemistry of La 1−x Ca x MnO 3+δ can be separated into intrinsic and extrinsic effects, leading to a change in the Mn-oxidation state. The content of Mn 4+ within the samples can be influenced by either the change of the oxygen content within the sample as a reaction to temperature and atmosphere, which we call intrinsic, or by the introduction of Ca 2+ , which we call extrinsic. 20,35 Hence, for the unsubstituted LaMnO 3+δ , the Mnoxidation state is solely intrinsically determined as a reaction of the sample with the surrounding atmosphere. For low amounts of Ca substitution, the Mn oxidation state is influenced intrinsically, by its reaction with the atmosphere, and extrinsically, by Ca substitution. At high Ca contents, the extrinsic effect dominates and the Mn oxidation is mainly influenced by cation replacement. When looking at the Mn 4+ content within the room temperature structures, shown as red squares in Figure 11, it can be seen that LaMnO 3+δ (x = 0) contains ∼26% of Mn 4+ (2δ). Heating to 950°C in various atmospheres ( Figure 5) decreases the Mn 4+ concentration to about 2% (in N 2 ). The Mn 4+ concentration is highly dependent on the atmosphere and the corresponding oxygen partial pressure. The described intrinsic effect can be clearly defined here, as the effects cannot be attributed to Ca substitution.
The substitution of La by increasing amounts of Ca is accompanied by a general increase in the Mn 4+ content.
Similar to the unsubstituted sample, the Mn-oxidation state decreases with decreasing oxygen partial pressure for low Ca contents (x = 0−0.2). The extent of Mn 4+ changes becomes smaller with increasing Ca content. For high Ca contents (x = 0.3−0.5), the increase in Mn 4+ is independent of the atmosphere and scales almost linearly with x.
In general, the lowest Mn 4+ contents are achieved in the nitrogen atmosphere, where the Mn-oxidation state increases almost perfectly linearly up to around 50% Mn 4+ for x = 0.5 over the whole sample series (x = 0−0.5). In the nitrogen atmosphere, most of the intrinsic effects due to reaction with oxygen in the surrounding atmosphere are absent. Extrinsic effects dominate, which can be validated by the almost linear increase in the Mn 4+ content in N 2 , which is nearly perfectly scaling with the Ca-content x. Deviations show contributions from intrinsic effects due to reaction with surrounding oxygen. These deviations should scale with the oxygen partial pressure and can be observed for low Ca contents up to x = 0.2. Therefore, it is assumed that the Mn oxidation in this range is influenced by intrinsic and extrinsic effects. For higher Ca contents (x = 0.3−0.5), only extrinsic effects are visible.
Investigation of the Mn Electronic State Using XANES. To provide experimental evidence on the electronic state of the Mn cations, XANES experiments were performed in both the hard and soft X-ray regimes. The near-edge region of the normalized Mn K-edges from air-sintered La 1−x Ca x MnO 3+δ (x = 0−0.5) is shown in Figure 12a. For the air-synthesized samples, including rhombohedral x = 0, it can be observed that the Mn K-edge shifts to higher energies with increasing Ca content. This shift can be attributed to a change in the formal oxidation state of Mn, namely, an increased Mn 4+ content, within the sample. 36 Figure 12b shows the edge positions, E 0 , here defined as the value at half edge rise, as a function of the Ca content (x). It can be observed that between x = 0.2 and x = 0.3, a change in the electronic state occurs. This fits the results previously obtained in the TGA, where a change in defect chemistry behavior was deduced between x = 0.2 and x = 0.3. Compositions of the La 1−x Ca x MnO 3+δ series up to x = 0.2 show again that Mn oxidation is influenced by both extrinsic and intrinsic effects, while only extrinsic effects are present in samples with higher Ca contents (x = 0.3−0.5). The hard X-ray XANES results, shown in Figure 12, agree with these assumptions.
However, further quantification of these changes is difficult, as octahedral distortions due to a Jahn−Teller-effect, occurring preferably in low Ca regimes, bias the edge positions. Ramos et al. 37 found for LaMnO 3 that a shift in the K-edge position might occur due to a contraction of the long Mn−O bonds even if there is no change in the Mn-oxidation state.
In Figure 13, the Mn L 32 -edge spectra are displayed, which show a behavior similar to the La 1−x Sr x MnO 3−δ (LSM) material system, 38 where the position of the L 3 white line shifts to higher energies with increasing Sr content. The O Kedge (Figure 14a), in turn, is also similar to what was observed for LSM: 38 with increasing substitution with the divalent A-site cation, minority e g -states emerge at ∼533 eV. This is explained by smaller overlap of this energy region with La-derived electronic states (La 5d−O 2p-hybridized states) at 535 eV, making a clear distinction possible. Most revealing, although, is the leading edge, which consists of overlapping e g majority and t 2g minority states formed by O 2p−Mn 3d hybridization. These states are, especially for the Ca lean samples, indistinguishable because exchange splitting Δ ex and crystal- field splitting 10Dq are of similar magnitude as it is the case in the LSM system. 38 The intensity of this signal e g (max) + t 2g (min) is a direct measure of the hole concentration 38 and thus can be used to assess the Mn 3+ /Mn 4+ concentration semiquantitatively. Here also, the intensity (can be approximated by the peak height, as only small variations in peak width and shape are observed) as a function of Ca content (Figure 14b) indicates the same change in defect chemistry as TGA and hard X-ray XANES between x = 0.2 and 0.3. Additionally, for high Ca substitutions, (x ≥ 0.3) e g (max) + t 2g (min) is split, which can be explained by changes in the electronic structure and ligand field, so that the values of Δ ex and 10Dq are of different magnitudes, and e g (max) and t 2g (min) are distinguishable again.
■ CONCLUSIONS La 1−x Ca x MnO 3+δ (x = 0−0.5) was synthesized in an air atmosphere using solid-state synthesis techniques. The structure of LaMnO 3+δ was found to be dependent on the atmosphere and changed from a rhombohedral modification in air to the strongly Jahn−Teller distorted orthorhombic polymorph after exposure to the nitrogen atmosphere. The Ca-substituted compositions were all orthorhombic without Jahn−Teller distortion. The lattice parameters decreased with the Ca content, as well as the cell volumes, which was mainly explained by the increase in Mn 4+ content, being smaller than Mn 3+ and an increase in divalent Ca 2+ , which is smaller than La 3+ .
TGA of La 1−x Ca x MnO 3+δ in a strongly reducing atmosphere allowed the determination of the initial oxygen nonstoichiometry δ, which decreases with increasing Ca contents. With this calibrated initial nonstoichiometry, significant δ-value changes were observed for small Ca contents (x = 0−0.2) in further thermogravimetric experiments performed in various atmospheres (O 2 , air, and N 2 ). Higher Ca contents (x > 0.3) did not show significant changes in their δ-values, independent of the applied atmosphere. In the TGA experiments, three processes were observed during heating. A slow irreversible loss appears independent of the gas atmosphere at temperatures below 550°C (in range I), weak for x = 0 and becoming stronger with x. Above 550°C, a slow irreversible oxygen uptake becomes active up to 750°C (range II), which is the strongest for x = 0 and disappears for x = 0.4. This process is absent under the nitrogen atmosphere. At temperatures higher than 750−800°C (range III), the samples with x ≤ 0.3 equilibrate and nonstoichiometry decreases with T, while the extent depends on p(O 2 ).
The Mn-oxidation state was calculated from the δ-values determined by TGA, assuming the formation of equal A-and B-site vacancies. The influence on the Mn oxidation within the sample series was subdivided into three regions. In pure LaMnO 3+δ , Mn oxidation was attributed solely to the intrinsic effects, and the oxygen nonstoichiometry is only adapting to the surrounding atmosphere. Low calcium contents up to x = 0.2 always contain some amount of Mn 4+ extrinsically determined by the Ca-content and an additional intrinsic reaction with the atmosphere. In a range of 0.3 ≤ x ≤ 0.5, the Mn-oxidation state mainly is extrinsically influenced by the Ca substitution, and the samples show almost no reaction with the  atmosphere. In the N 2 atmosphere, the oxygen nonstoichiometry was reduced to 3 and the Mn 4+ content scales nearly perfectly with the Ca-content x.
Soft and hard X-ray XANES experiments were performed for the series (x = 0−0.5) to provide experimental evidence on Mn-oxidation state calculations. A shift of the Mn K-edge toward higher energies was observed with increasing Ca content. This shift was attributed to an increasing change in the formal oxidation state of Mn. From Mn L-edge and O Kedge spectra, similar results could be observed. Between x = 0.2 and x = 0.3, a change in the electronic state of Mn was detected. It is assumed that at this point, the main influence on the Mn-oxidation state changes from intrinsic effects in x = 0− 0.2 toward extrinsic effects for x ≥ 0.3.
The TGA experiments show that for samples with low Ca content, the oxygen nonstoichiometry 3 + δ and therefore Mn 4+ content react quickly and reversibly above 750°C. This susceptibility is decreasing with x, but the increasing Ca substitution stabilizes the structure. For low Ca contents, the operation temperature and atmosphere are expected to play an important role. Exposure at temperatures below 750°C enables one, to some extent, to irreversibly change the oxygen nonstoichiometry by the control of temperature and gas atmosphere. Knowledge of the defect chemistry obtained from the results of this study can be applied for design and development of a suitable air electrode material for solid oxide cell operation. Our results show that the choice of operation temperature, below or above 750°C, is expected to show a different behavior, as the reaction of the material with the atmosphere changes.