First-Principles Study of Sr2–x La x FeO4 and Comparison with Sr2FeO4‑δ: The Nature of the Positively Charged Defect Determines the Character of the Electronic Structure Authors: Yuri A. Mastrikov, Denis Gryaznov, Andrew Chesnokov, Guntars Zvejnieks, Maksim Sokolov, Maija M. Kuklja, Rotraut Merkle, Eugene Kotomin Published: 2026 ABSTRACT Lanthanum doping of the first-order Ruddlesden–Popper (RP) phases Sr2–x La x FeO4 (0 ≤ x ≤ 1) is investigated by density functional theory (DFT) calculations. Excess volume and excess energy of this solid solution series are calculated. The evolution of the electronic structure with increasing La content is discussed based on the calculated Fe magnetic moment. It is found that for the same average Fe oxidation state, the electronic structure in Sr2–x La x FeO4 displays a more delocalized character than in corresponding Sr2FeO4‑δ composition. This is attributed to the fact that oxygen vacancies VO •• which are formed under Fe–O bond breaking (and which carry a doubly positive charge) represent a stronger local distortion than singly charged substitutional LaSr • dopants. Introduction The group of strontium ferrate perovskites and perovskite-related oxides is a fascinating material class. On one hand, it is interesting from a fundamental point of view, because different geometrical and electronic structures have very close energies so that delicate interactions finally decide which one becomes most stable. For the first-order Ruddlesden–Popper (RP) phase Sr2FeO4, magnetic structure measurements indicated a symmetry reduction from the tetragonal phase, and computational investigations also identified a symmetry reduction to orthorhombic and interpreted it as a Jahn–Teller (JT) distortion. , In the second-order RP phase Sr3Fe2O7, the energy difference between a charge-ordered state (with Fe3+ and Fe5+) and a JT-distorted state is on the order of only 34 meV/formula unit, as found by the hybrid r²SCANh calculations. Overall, the electronic structure of Sr n+1Fe n O3n+1 ± δ RP phases has a more localized character than that of the simple perovskite SrFeO3‑δ (indicated, e.g., by the fact that the vacancy formation energy is almost constant in Sr2FeO4‑δ, while it strongly increases with increasing δ in SrFeO3‑δ ). This is to some degree simply related to the transition from a 3- to a 2-dimensional system. However, the comparison of Sr2FeO4‑δ and Sr2–x La x FeO4 in the present investigation demonstrates that also the nature of the introduced positively charged defect (oxygen vacancies, or La3+ substituting for Sr2+) plays an important role. These defects not only decrease the average Fe oxidation state but also affect the character of the electronic structure. Introducing oxygen vacancies results in a rather localized character wherefrom a combination of Fe–O bond lengths and Fe magnetic momentsiron atoms with integer Fe3+ and Fe4+ oxidation states could be recognized. In the present study, we investigate the effects of La doping in Sr2–x La x FeO4, which will be shown to result in a more delocalized character. The character of the Fe–O bonding is not only of fundamental interest but also directly affects the material’s properties regarding proton incorporation. Redox-active perovskites and perovskite-related materials serve as hole-oxygen vacancy mixed conducting or even proton-hole-vacancy triple conducting air electrode materials in solid oxide cells and protonic ceramic cells. − Compared to ABO3 simple perovskites and AA’B 2O6 or A 2 BB’O6 double perovskites, RP phases comprise a higher fraction of A-site cations, which is expected to increase the overall basicity of the material. Further, the fraction of alkali earth (2+) and rare earth (3+) cations on the A-site affects the properties. Sr2FeO4 is an aristotype for the first-order RP phases utilized as mixed conducting electrode materials in ceramic electrochemical cells. − While Sr2FeO4 itself needs to be prepared under high oxygen pressure, higher-order RP phases Sr n+1Fe n O3n+1 ± δ or rare-earth doped compositions can be prepared under ambient conditions, e.g., Sr3Fe2O7 , or SrEu2Fe2O7. Besides Fe-based RP phases, also Co- and in particular Ni-based materials have been investigated as air electrode materials for protonic ceramic cells, e.g., (La,Sr3)­Co1.5Fe1.5O10, (Pr,Sr)2NiO4, (La,Sr)2NiO4, (Ba,Sr,La)2NiO4, or oxygen storage materials Sr3Fe2–x Ni x O7‑δ. Here, we concentrate on Fe-based Sr2–x La x FeO4, RP phases where we can compare with the preceding computational work with the same methodological approach for Sr2FeO4‑δ and SrFeO3‑δ. ,, While both oxygen vacancies and La dopants are positive defects (charge-compensated by a decrease of the average iron oxidation state), the present results indicate that they modify the electronic structure of the material in different ways. This comparison enhances the general understanding of the defect chemistry of mixed conducting oxides, particularly in delicate situations as encountered in the Ruddlesden–Popper oxides where the energetic difference between more or less localized electronic defects is small. Method and Model Doping Sr2FeO4‑δ by La has been performed by substitution LaSr •. La concentration x has been varied from 0.125 up to 1.00. Varying x continuously, we minimize the possibility of technical errors in calculations and ensure the integrity of the results within the range of interest. The supercell of eight Sr2FeO4 formula units (56 atoms) is suitable for modeling a relatively low La concentration (x = 0.125). Only one inequivalent configuration of LaSr is possible in a model supercell. For higher values of x, all possible spatial distributions of LaSr have been investigated (Table S1) by means of the Supercell code. Screening of this large number of configurations proved necessary because the lowest energy configuration for a given La content in the supercell does not necessarily occur for the intuitively guessed arrangements with maximum La distances (cf. discussion of Figures and S1). Note that representability of a particular configuration depends on its energy, multiplicity, and structural stability. Calculations have been performed by the computer code VASP. Effects of core electrons were emulated by PAW potentials (Table S2). Computational details are presented in Table . The Hubbard correction value U eff = 4 eV has been chosen based on an extensive analysis of structural and electronic parameters of SrFeO3 and Sr2FeO4. In addition to that, the value has been considered optimal for similar (Ba,Sr)­FeO3‑δ and (Pr,Sr)2FeO4‑δ perovskites. ,, 1Main Computational Parameterscomputer codeVASP 6.1.2cutoff energy520 eVspace group (SG), Settings64, Bbem transformation matrix from the crystallographic unit cell to the supercell (110−110001) K-pointsMonkhorst–Pack4 × 4 × 2exchange–correlation functionalPerdew–Burke–Ernzerhof U eff 4.0 eVresidual external pressure<1 kbarforces on atoms<0.02 eV/Å Main Computational Parameters Electron charge and magnetic moments were analyzed by the Bader method. Structure visualization was performed using the code VESTA. Considering one extra electron released to the system by each substitution of Sr2+ by La3+ (LaSr •), and two extra electrons released by an oxygen vacancy, the formal Fe3+/Fe ratio can be defined as follows: 1 Sr2−xLaxFe1−(2δ+x)4+Fe2δ+x3+O4−δ The high covalency of the Fe–O bonds in the system prevents a clear distinction between these two oxidation states of Fe. A direct charge calculation by the Bader method proved to be unreliable for assigning Fe oxidation states. Therefore, for oxidation state identification, we used magnetic moment combined with the lengths of Fe–O bonds, as proposed in ref . Results Excess Energy, Volume, and Spatial Distribution of LaSr • The energetic stability of a complex material like Sr2–x La x FeO4‑δ depends on defect (A-cation substitution, oxygen vacancy/interstitial) concentration as well as on their spatial distribution. Oxygen defects have been studied in detail in ref In the present work, we focus on LaSr • substitution, preserving the oxygen stoichiometry at δ = 0, except for x = 0.125 (discussed below). The energetic stability of Sr2–x La x FeO4 was calculated as excess energy relative to the linear interpolation between the La-free composition (x = 0) and the lowest energy state for x ref. = 1.00 (Table S3 and Figure S1d). 2 Eexcess=Ex−(Ex=0+(Emin⁡x=1−Ex=0)x) The nonzero excess energies in Figure a indicate moderate deviations from an ideal solution. The negative excess energy values indicate an overall driving force for mixing between Sr2FeO4 and SrLaFeO4 (the distribution of La and Sr within the cell may still have a preference for the La segregation, see later). Comparison with other perovskite-type materials (Table ) indicates that a smaller size mismatch and more flexibility in the structural and chemical bonding properties (e.g., presence of more loosely bound rocksalt layers in RP phases) favor the formation of stable solid solutions. The rocksalt AO layer in the RP phase offers even more flexibility to accommodate differently sized A-cations (for RP phases of higher orders even with crystallographically different A-sites). (a) Excess energy (per formula unit) of Sr2–x La x FeO4. (b) Excess volume (per formula unit) of Sr2–x La x FeO4. The data correspond to the configurations with the lowest total energy for each La content. 2Excess Energy for Various Composition Changes in Perovskite-Related Structures: A 2 BO4 and ABO3 materialstructuremixing cation, size difference, Åvariationexcess energysourceSr2–x La x FeO4 1st order RP A 0.08atomic radius, oxidation state of A- and B-cationsnegativethis studyLa1–x Sr x CoO3 perovskite A 0.08 La1–x Sr x MnO3 perovskite A 0.08 Ba1–x Sr x TiO3 perovskite A 0.17atomic radiuspositive BaFe1–x Ce x O3 perovskite B 0.28 Excess Energy for Various Composition Changes in Perovskite-Related Structures: A 2 BO4 and ABO3 Interestingly, the excess energy curve of Sr2–x La x FeO4 has an asymmetric shape. Compositions with intermediate La content x = 0.5 and x = 0.75 have the most negative excess energies, i.e., the minimum is shifted toward higher La contents. Also, the volume shows moderate deviations from the ideal behavior Figure b, but in contrast to the excess energy, the excess volume shows an approximately symmetrical shape. The absolute molar volume increases with increasing La substitution despite La3+ (1.36 Å) having a smaller Shannon radius than Sr2+ (1.44 Å, both in 12-fold coordination), in line with experimental data. This volume expansion is caused by a decrease in the average Fe oxidation state and an increase in the corresponding Fe ionic radius. The volume expansion between Sr2FeO4 and SrLaFeO4 amounts to 2.7%, which is significantly larger than the respective expansion by 1% between Sr2FeO4 and Sr2FeO3.5 (also changing from average Fe4+ to Fe3+). This results from the fact that in Sr2–xLa x FeO4 the Fe reduction is not accompanied by removal of oxygen ions. Figure shows that for each La content the lowest energy is found for the configurations with the shortest average La–O distance (configurations shown in Figure S1). These arrangements lead to the most beneficial La3+–O2– electrostatic interactions. A more detailed analysis (Figure S2) indicates the sensitivity for certain detailed structural features (which together finally determine the average La–O distance). Figure S2a demonstrates that for a given x, a larger number of LaSr •–LaSr • bonds (1NN) across the perovskite layer slightly increases the energy. A larger number of 2NN LaSr •–LaSr • bonds lowers the energy (Figure S2b,e,h). Both criteria together correspond to accumulating La within the same rocksalt layer (but distributed over both AO planes of that layer). If several LaSr • are located within the same AO plane, then the configurations with more 3NN pairs are more favorable (Figure S2c). This can be traced back to the fact that this arrangement allows for shorter La–O and longer Sr–O distances reflecting their size as well as charge difference (Figure b), in contrast to the case of Figure c, where the topology enforces equalbut energetically less favorableLa–O and Sr–O lengths. So, the spatial arrangement does not simply follow the principle of maximizing or minimizing the distance between LaSr • substitutions which carry a relative positive charge compared to Sr. Energy per f.u. relative to the lowest value for each value of x depending on the averaged LaSr •–O distance, which is obtained by averaging over the nine La–O bonds around La and all LaSr • cations. Nearest neighbors (NN) of the A-cation sublattice in the Sr2FeO4 RP 1st-order perovskite structure in an 8-formula unit supercell (a). (Note that the periodic boundary conditions are neglected solely for showing the 7NN). LaSr cations in the Sr1.75La0.25FeO4 structure as 3NN (b) and 5NN (c). We also briefly explored the energetics of oxygen-deficient composition Sr1.875La0.125FeO3.875 (Table S4). Since LaSr • and VO •• are both positive defects, it is not surprising to find a moderate positive interaction energy (Figure S4), in the range from 0.2 to 0.8 eV. Since the interaction energy depends only slightly on the actual LaSr •–VO •• distance, the presence of VO •• is not expected to significantly change the relative stability of different LaSr •–LaSr • configurational arrangements for the higher La concentrations. Just like in the case of Sr2FeO4, VO •• forms preferentially in the FeO2 planes. The lowest VO •• formation energies in Sr1.875La0.125FeO4 are only slightly higher than in Sr2FeO4 (2.60 and 2.35 eV for Sr–VO ••-Fe and Fe–VO ••-Fe, respectively), which is in line with the fact that for Sr2FeO4‑δ only a minor variation of VO •• formation energy with an iron oxidation state has been observed. Identification of Fe Oxidation States in Sr2–xLa x FeO4 The high degree of covalency of the Fe–O bonds prevents a clear distinguishing between the 4+ and the 3+ oxidation states of Fe. As discussed in more detail in, the Fe–O bonds in the simple perovskite SrFeO3 have a higher charge transfer from the oxide ions to iron and more delocalized character than in Sr2FeO4 with the same formal Fe4+ oxidation state. Nevertheless, the Fe–O bonds in Sr2FeO4 are still far from purely ionic, and thus, the question whether fixed Fe oxidation states can be assigned is not trivial (attempting to use the Bader charges yields inconsistent results). As proposed in, the Fe magnetic moment combined with the lengths of Fe–O bonds can serve as an indirect indicator of the oxidation state (Figure S5). Already for x = 0.125 (one extra electron in a supercell), we observed quite a broad distribution of the magnetic moments (Figure a); it is not possible to recognize a single Fe cation with a clear 3+ oxidation state where the excess electron is localized (the data points at high magnetic moment ≈3.84 μB represent a total of five (2NN + 4 × 1NN) Fe atoms, Table S5). Reference cells which contain only Fe3+ have much higher magnetic moments: (i) in SrLaFeO4, the moments amount to 4.25–4.30 μB, (ii) in Sr2FeO4‑δ, the moments cover a larger range of 4.05–4.25 μB (larger moments for larger δ 2). In Figure a, the corresponding ranges are indicated in dark gray (comparison with SrLaFeO4) and light gray (lower end interpolated according to the La content between SrLaFeO4 and Sr2FeO4‑δ with low δ). None of the Fe in Sr1.875La0.125FeO4 comes close to the 3+ range. Some Fe are in the lower part of the Fe4+ reference zone (3.65–3.8 μB for Sr2FeO4 and Sr2FeO3.875, beige area), while others attain values around 3.85 μB. The latter are regarded as a result of delocalization of the excess electron over several iron cations. Magnetic moment as a function of averaged Fe–Oeq distance (average of the Fe–Oeq around each individual Fe in the supercell) for δ = 0 and (a) x = 0.125, (b) x = 0.25, (c) x = 0.5, and (d) x = 0.75. Solid symbols denote the values for the lowest energy configurations. The shaded areas indicate the zones where integer Fe3+ and Fe4+ oxidation states were assigned for reference configurations. Figure S3 shows the quite similar plots with Fe–Oap and Fe–O (all) on the horizontal axis. For x = 0.25, the situation is closely related (Figure b). The configuration with the lowest energy (solid symbols) has only Fe with moments higher than the 4+ reference, i.e., the two excess electrons are homogeneously distributed among all Fe in the cell. However, there are configurations at only slightly higher energy with some Fe4+ besides other Fe with decreased oxidation state. For x = 0.5, we see a broad distribution of moments, not touching 3+ and only barely touching the 4+ reference values (Figure c). At a high La content of x = 0.75 (average Fe oxidation state 3.25+), the distribution further shifts to higher μB values and becomes again narrower (Figure d). Finally, for x = 1 (all iron Fe3+), all moments are within 4.25–4.3 μB. So, for most of these cases, it is not possible to assign integer 3+ and 4+ oxidation states, and even if some Fe fall into the region of the 4+ state, the excess electron(s) may still be distributed over several other Fe cations. For x = 0.125 (the composition with the highest tendency to form localized states), we ran extended test calculations attempting to obtain a fully localized polaron (a single Fe3+) by using starting geometries with deliberately distortion Fe–O bond lengths primarily within the DFT + U methodology. However, Table S6 demonstrates that such a solution was not obtained, and further that the energies of partially localized and fully delocalized states are nearly indistinguishable within the accuracy of the method (∼4 meV). To further check the trend of the increasingly delocalized character of the electronic structure with high La content, we also used the HSE06 functional to calculate the configurations with the lowest energy (Figure S1) for various La contents (details are given in the Supporting Information). Figure S6 compares the data for the lowest-energy configurations obtained with DFT + U and HSE06. The overall trend of increasing delocalization of the excess electron(s) with increasing La content is observed for both methods. For HSE06, even for the lowest La content, no single Fe3+ is found (the maximum magnetic moment remains much below the value expected for Fe3+, and four Fe have an increased magnetic moment, i.e., the excess electron is shared between the 4 × 1NN Fe). A comparison of the electronic density of states (DOS) calculated using DFT + U and the HSE approach is given in Figures S7–S9. Also, here the main trends with varying La content agree well. This is encouraging to see, because HSE calculations are more computationally demanding and more delicate to properly converge. Comparison with Sr2FeO4 A direct comparison with the earlier studied Sr2FeO4‑δ 2 can be made for combinations of δ and x that yield identical average Fe oxidation states (i.e., δequiv = x/2). Considering one extra electron released to the system by the LaSr 3+ cation and two extra electrons released by oxygen vacancy, formally, the Fe3+/Fe4+ concentrations can be defined as follows: 3 Sr2−xLaxFe1−(2δ+x)4+Fe2δ+x3+O4−δ Particularly, we can compare two cases x = 0.25 and x = 0.5 for δ = 0 with δ = 0.125 and δ = 0.25 for x = 0, respectively. Figure a,b compares the situation for the averaged Fe3.75+ state. In Sr2FeO3.875, one can recognize several Fe which are identified as 4+ based on magnetic moments (3.65–3.8 μB) and Fe–O bond length distribution and others attributable as Fe3+. This is in pronounced contrast to the more continuous moment distribution for Sr1.75La0.25FeO4, where in particular no Fe reaches the Fe3+ zone. A similar situation is found for Fe3.5+ in Figure c,d, where in Sr2FeO3.75 clearly a large number of Fe3+ and Fe4+ can be recognized (with larger δ, the range attributable to 4+ expands up to 3.8 μB, shaded light red). In contrast, Sr1.5La0.5FeO4 presents a very smooth moment distribution which does not reach the 3+ and hardly touches the 4+ zone. So, the overall conclusion is that Sr2–xLa x FeO4 presents a much more delocalized situation than Sr2FeO4‑δ. Analogously, experimental Mössbauer spectra for Sr2–x La x FeO4 samples with 0.7 ≤ x ≤ 1.0 were fitted with a significant fraction of a mixed 3+/4+ oxidation state. Magnetic moment as a function of averaged Fe–Oeq for average Fe3.75+: (a) x = 0; δ = 0.125; and (b) x = 0.25 δ = 0. For average Fe3.5+: (c) x = 0; δ = 0.25; and x = 0.5; δ = 0. For the Sr2FeO4‑δ cells with δ > 0, only the energetically more stable Fe–VO ••–Fe vacancy configurations are considered. Solid symbols denote the values for the lowest energy states. This difference can be rationalized from the following considerations. The presence of oxygen vacancies in Sr2FeO4‑δ constitutes a much stronger distortion than the substitution of some Sr2+ by La3+ (although both defects lead to a decrease of the average Fe oxidation state). An oxygen vacancy causes broken Fe–O bonds, which leads to significant structural relaxations. Further, it was shown in ref that the presence of an oxygen vacancy affects the electronic structure of iron in a way similar to a JT distortion (expanding two axial Fe–O bonds is not too different from completely removing one O ligand). This can stabilize localized Fe4+ states, and then in turn also the states accommodating the excess electrons caused by the oxygen deficiency acquire a more localized character. In contrast, such a specific driving force for forming localized Fe4+ states is absent in Sr2–x La x FeO4. An additional reason for the more localized character of Sr2FeO4‑δ might be the fact that oxygen vacancies are doubly positively charged defects (while substitutional La3+ has only a single relative positive charge compared to Sr2+), which is expected to cause more pronounced local distortions. The strong effects of oxygen vacancies and related structural distortions have also been observed in the SrFeO3‑δ simple perovskites where the electronic conductivity changes from metallic character for the cubic δ = 0 phase to polaron hopping in vacancy-ordered noncubic phases for δ > 0.03. So, the material class of strontium ferrate perovskite and Ruddlesden–Popper phases is an intriguing example of transition-metal oxides where materials with more or less pronounced delocalized character of the electronic structure are quite close in energy. Correspondingly, small changes suffice to switch the system into one or the other direction. These may be the lattice parameter expansion in the Sr1–x Ba x FeO3 series which induces a change from delocalized character, the absence of JT distortion, and metallic-type conductivity to a more localized character with polaron hopping for Sr0.5Ba0.5FeO3 (or possibly already for lower Ba contents). The differences of the Fe–O bonding character are not only of fundamental interest. Preceding DFT investigations have shown that the propensity of oxides for proton uptake by water incorporation is closely coupled to the material’s proton affinity and further related to the character of the O 2p states. Within a BaFeO3‑δ supercell, O atoms opposite to an oxygen vacancy exhibit an increased covalency of the significantly shortened Fe–O bond with a decreased center of gravity of the O 2p states and correspondingly are less favorable for protonation. It remains to be investigated whether a material such as Sr2FeO4‑δ with distinctly different oxygen atoms (i.e., some strongly preferred protonation sites), or Sr2–x La x FeO4 with more equivalent oxygens yields a higher total proton uptake (for comparable average Fe oxidation state). However, in the former, quite pronounced proton trapping effects are anticipated, which will be detrimental for a long-range proton transport. Conclusions This computational investigation of the Sr2–x La x FeO4 first-order RP phase indicates that the La/Sr configurations with the shortest average La–O bond length yield the lowest total energy, and that excess energy as well as excess volume in the Sr2FeO4–SrLaFeO4 solid solution series are negative. While La substitution decreases the average (formal) Fe oxidation state in Sr2–x La x FeO4, it is not possible to identify any localized Fe3+ states based on magnetic moments. This is in contrast to oxygen-deficient Sr2FeO4‑δ where localized Fe3+ species had been identified. This indicates thatfor the same average Fe oxidation statethe electronic structure exhibits a more delocalized character in Sr2–x La x FeO4 compared to Sr2FeO4‑δ. The different behavior is attributed to oxygen vacancies VO •• (with doubly positive charge and breaking of Fe–O bonds) exerting a stronger local distortion than singly charged LaSr • and thus shifting the delicate energy balance between delocalized and more localized states toward the latter. Such differences in the electronic structure also modify the basicity of individual oxide ions and may thus affect the energetics of proton incorporation. This remains to be investigated in the future. Supplementary Material