WorksPhysics and Astronomy
Positive Temperature Coefficient of Anisotropy due to Groundstate in -FeO
Abstract
Positive anisotropy temperature coefficients () usually arise from hybridization between magnetic 3d states and strongly spin-orbit-coupled (SOC) subsystems. Yet -FeO shows (125-200~K) without an obvious SOC partner. We study pure and Cr-doped (with weakened Fe-O hybridization) -FeO across the transition from low temperature incommensurate to high-anisotropy magnetic phases. We identify a groundstate in -FeO while reduced hybridization yields in the Cr-doped system, highlighting metal-ligand hybridization as a route to tune magnetic, electronic and orbitronic properties.
Introduction– Magnetocrystalline anisotropy () is driven by spin-orbit coupling, where an atom’s electron spin is tied to its orbital motion, which is in turn shaped by the crystal’s electrostatic fields generated by the lattice. As a result, the magnetic orientation becomes linked to the lattice, generating a preferred spin orientation in the material. In the absence of defects, domain wall motion etc., the magnetic coercivity of a material () is proportional to . In most magnetic materials, magnetocrystalline anisotropy decreases with increasing temperature () below their magnetic ordering temperature. However, there are a number of exceptional materials with the opposite behaviour (), where anisotropy strengthens with increasing temperature. This unusual property makes them attractive for applications that need strong room-temperature coercivity, such as permanent magnets.
The most notable example of this is MnBi, a rare-earth element-free permanent magnet candidate, where the Mn sublattice shows the conventional while strong hybridization and spin-orbit coupling of the Bi sublattice with thermal lattice expansion results in [?, ?, ?]. Because this latter term is dominant at higher temperatures, for a large temperature range.
While MnBi provides the clearest example of as a result of intrinsic competing anisotropies, it is far from the only material with such behaviour. Many of the rare-earth-transition-metal intermetallics (RCo, RFeB where R=Nd, Sm etc.) show over a finite temperature range due to competition between the and systems[4, 5], while the spin reorientation transitions of rare-earth ferrites and chromites (RFeO, RCrO) suggest a similar competition[6, 7, 8]. Together these materials present a coherent picture where the behaviour is driven by hybridization between the magnetic subsystem and the strongly spin-orbit coupled “heavy” subsystem.
This model is challenged by -FeO, where between 125 and 250 K, but there is no clear “heavy” subsystem present. Instead, -FeO is a perovskite with four distinct Fe sites as shown in fig:1a. Of these sites, one has tetrahedral coordination (T) while three have octahedral coordination (O). Two O sites are structurally distorted (DO) with varying Fe-O bond lengths, while one is not (RO). Below 100 K, -FeO has an incommensurate magnetic state, which transforms to a hard (high anisotropy) ferrimagnetic state at the onset of with maximum in excess of 2 T (see fig:1b) – similar to MnBi with its special rare-earth behaviour. This hard ferrimagnetic state softens at high temperature (480 K<) with a of around 850 K[10, 11].
This unique temperature dependent ordering is the result of complex electronic and magnetic interplay between the four Fe sites; however, the electronic structure of -FeO has remained elusive. Fe L edge x-ray absorption spectroscopy (XAS) has provided only limited insights because all cations are nominally Fe, resulting in significant overlap between site contributions[12, 13]. As shown in Figure 1c, the XAS spectra exhibit minimal differences between the incommensurate and ferrimagnetic phases. By contrast, O K edge XAS (Figure 1d) spectra reveal a transformation in Fe-O bonding between the two phases via a change in line shape below 530 eV (pre-peak). Additionally, increased spectral weight at 548 eV indicates increased excitation into delocalized 4sp Fe bands[14], suggesting the onset of an itinerant electronic structure in the ferrimagnetic phase. However, the Fe sites responsible for these changes cannot be identified using XAS alone.

Figure 1. a) Crystal structure of -FeO. Temperature dependent b) magnetic coercivity (reproduced in part from [9]) and c) Fe L spectra of -FeO (black) and -(FeCr)O (red), offset for clarity. d) O K edge XAS spectra of -FeO (black) and -(FeCr)O (red) at 50 K (solid) and 250 K (dashed).
Mössbauer spectroscopy has emerged as a key technique for characterizing -FeO owing to its sensitivity to the local Fe environment. This technique measures recoil-free absorption and emission of -rays to directly quantify the nuclear energy states of Fe ions. These energy levels are altered by hyperfine interactions arising from the electric charge distribution and magnetic dipole moment of the Mössbauer atom and from neighbouring atoms. As a result, Mössbauer spectroscopy provides quantitative information about the local electronic and magnetic environment at the different Fe sites.
Temperature dependent studies have revealed that onset of and the hard ferrimagnetic state occurs with the transformation of the local environment[10, 13, 15, 16, 17, 18]. Most notably, the magnetic hyperfine field () of this site decreases to anomalously low values upon warming. To identify the origin of this behavior, we chose to examine both -FeO and a perturbed system of Cr-doped -FeO (-(FeCr)O). Cr-doping was chosen because relative small concentrations of Cr dopants have been shown to disrupt the electronic and magnetic structures of other strongly-correlated perovskite systems including manganites and ferrites[19, 20, 21, 22, 23].
In -(FeCr)O the Cr ions dope into DO sites and act as hole defects, perturbing the electronic and magnetic structures[9]. The doping concentration was chosen because there is no evidence of structural distortions or secondary phases, and the characteristic temperature dependence of -FeO (including the temperature regime of ) is maintained. Reductions in (see Figure 1) and magnetic saturation () at 300 K indicate altered Fe-Fe and Fe-O exchange interactions and hybridization, while Mössbauer spectra revealed that the Cr-doping causes two site local environments with hyperfine parameters that average to those of -FeO. Such results suggest that the Cr-doping disrupts the stabilization mechanism present in -FeO[9]. Like the parent compound, Fe L edge XAS of -(FeCr)O (see Figure 1) shows no significant differences between the incommensurate and ferrimagnetic phases. O K edge XAS at 50 K (see Figure 1) confirms that Cr-doping weakens Fe-O hybridization in the incommensurate phase, while differences in pre-peak lineshape between -FeO and -(FeCr)O reveal the hybridized states of the ferrimagnetic phase are significantly altered.
Methods– -FeO and -(FeCr)O nanoparticles were prepared using a metal ion-impregnated silica gel. Synthesis and characterization are described in references 9, 13. Temperature dependent L Fe edge X-ray absorption spectroscopy (XAS) was done at beamline 4.0.2 of the Advanced Light Source. Powder samples were mounted on carbon tape, and spectra were collected in total electron yield (TEY). Transmission Mössbauer spectra were acquired from 10 to 300 K using a Janis SHI-850 closed-cycle refrigeration system and WissEl constant acceleration spectrometer with a source. Source drive velocity calibration was done with a 6 m thick -Fe foil at room temperature. Non-linear least squares analysis based on first-order perturbation theory was used to fit the spectra and obtain hyperfine parameters. Resonant inelastic X-ray scattering was done at beamline ID32 of the European Synchrotron Radiation Facility at 300 K. The Fe L map of -FeO was acquired using the high throughput spectrometer at the High Field Magnet endstation[24] with 220 meV resolution at 707 eV, while the individual spectra were acquired with the high resolution spectrometer[25] (35 meV resolution). For all measurements circularly polarized light was used to maximize flux, the spectrometer was fixed at a scattering angle and acquisition was done in single-photon counting mode.
Results & discussion – The temperature dependent evolution of the Mössbauer spectra of -FeO and -(FeCr)O are shown in Figure 2a,b. While fit parameters are essential to extract meaningful information about the individual local environments, such maps highlight the similarities and differences between the two systems. At low temperatures (10 to 125 K), the -FeO and -(FeCr)O have remarkably similar spectra with sharp spectral lines. Between 125 and 140 K, the transition to the high anisotropy phase occurs, evidenced by discontinuities in the spectral lines as the local environments transform. Above 140 K in the high anisotropy phase, -(FeCr)O shows more diffuse spectral lines – a signal of additional complexity in the local Fe environments. Figure 2c,d show the fitted 200 K Mössbauer spectra of -FeO and -(FeCr)O, respectively, with residuals. The black lines show the fit discussed hereafter in the text, while the grey line in Figure 2d is the fit

Figure 2. Maps of temperature dependent Mössbauer spectra of a) -FeO and b) -(FeCr)O. Colours correspond to transmission. 200 K Mössbauer spectrum of c) -FeO and d) -(FeCr)O with fit (black line) and residuals. Grey line in d) shows the fit and corresponding residuals without the T component.
without the component, showing its importance for fitting the broadened lines of -.
Keeping the overall trends in mind, we use the temperature dependent hyperfine parameters: isomer shift (), quadrupole splitting () and magnetic hyperfine field (), shown in Figure 3, to understand the evolution of the Fe local environments of - and -. We start in the low temperature regime ( K) to evaluate the impact of Cr-doping. The isomer shifts (), a measure of the -electron density around the nuclei, of all - components (see fig:3a,fig:3b) are in agreement with their - counterparts. However, the effect of the Cr ions is evidenced by the average isomer shift of the component decreasing, while that of the component increases. Such changes are consistent with -orbital electrons shifting towards the site from the site. Decreased quadrupole splittings (), which quantify nuclear electric field gradients, for - (see fig:3c) reveal that the shift in electrons is a consequence of the Cr ions creating a more isotropic electronic gradient around all Fe sites[?].

Figure 3. Temperature dependent isomer shift () of a) and b) components, c) quadrupole splitting () and d) hyperfine field () of -Fe₂O₃ (black) and -(Fe₀.₉₆Cr₀.₀₄)₂O₃ (red).
The reduced hyperfine field () of the site (see fig:3d) provides further evidence of the more localized environment due to the Cr-doping as this quantifies the effective magnetic field acting on the nuclei at a particular site and is primarily due to the isotropic Fermi contact field resulting from the spin-polarization of -electrons by unpaired valence electrons. However, the anisotropic contributions to from the dipole interactions between the valence electrons and the nuclei, and the orbital moment () also play a role. As the non-spherical electron density decreases, the dipolar contribution to is reduced. While reductions in for - are observed in all three components, only is reduced at all temperatures. This indicates that the incorporation of ions into the site alters the site most strongly, and points to strong - site exchange.
Furthermore, while and of - show the same temperature dependence as - in the low temperature incommensurate phase, diverges at 75 K. As shown in fig:3d, of all three components in - are reduced from those of -. This decrease coincides with the onset of the reduction in that precedes the transition to the high anisotropy phase[12]. The - components also exhibit a decrease in at 100 K; however, the lower onset temperature and larger relative decrease of highlight the reduction in -. This difference is caused by the reduced Fe-O hybridization from hole doping in - because it lowers the thermal energy for the reduction of . Such a shift in transition temperature is consistent with the shown in Figure 1, where of -(Fe₀.₉₆Cr₀.₀₄)₂O₃ is greater than that of Fe₂O₃ between 75 and 200 K[9].
Above 125 K, both -Fe₂O₃ and -(Fe₀.₉₆Cr₀.₀₄)₂O₃ reveal via as the systems transition into the high anisotropy ferrimagnetic phase. In Mössbauer spectroscopy this onset is observed via a two-stage rearrangement of the local electronic structure at 127.5 K in -Fe₂O₃ and 130 K in -(Fe₀.₉₆Cr₀.₀₄)₂O₃. The most pronounced changes occur in the components, strongly indicating that this coordination environment drives the enhanced magnetic anisotropy. Additional insight into this high anisotropy phase are revealed in the -(Fe₀.₉₆Cr₀.₀₄)₂O₃ system through the appearance of a second tetrahedral component. (), indicating increased complexity within the ferrimagnetic phase.
In both - and -( ) and ([]) (#fig:3) decrease above the transition. These decreases are larger in -( ), but offset by the new component where and are increased. In all three components the temperature dependence of and are correlated. This is important, because while is commonly associated with oxidation state, it is only an indirect probe of the valence electrons through shielding effects. More direct contributions to originate from changing population of the orbitals due to bond formation as well as the compression of and orbitals as a result of changing Fe-O bond lengths –the effects of which also impact .
Because the temperature dependent lineshape of Fe L edge XAS spectra does not support a change in oxidation state for either system (see [])(#fig:1c)), the unusual parameters in the high anisotropy phase, including the two distinct local environments of -( ) must be a consequence of altered Fe-O bonding. In this scenario, the smaller is associated with shorter more covalent bonds, while those of the larger are longer and more ionic[27]. Corresponding behave similarly, stabilizing to -0.54(7) and 0.2(1) mm/s for and , respectively, with the signs revealing prolate and oblate electric field gradients. In both and , the average of the two components in -( ) is in agreement with those of -.
The sharp decrease of of the tetrahedral components is the Mössbauer signature of the transition from the incommensurate to the high anisotropy phase. Above this transition, of both - and -( ) become only weakly temperature dependent with average values of 30.5(1) T and 29.7(4) T, respectively. The component of -( ) in unique as it shows a second decrease in at 145 K, accompanied by concurrent reductions in and . The decrease in indicates a reduced electric field gradient, which weakens the dipole contribution to . Meanwhile, radial movement of the valence electrons affects both the isotropic Fermi contact term of and -electron density at the nucleus (). This second transition, occurring 10 K above the first, acts like a correction to the initial electronic rearrangement. Notably, this secondary transition observed in of -( ) mirrors the two-stage transition in the single component previously reported in -[13].
In contrast to the tetrahedrally coordinated sites, of the octahedrally coordinated components of both systems are only weakly affected by the characteristic transition and remain consistent with Fe in coordination. Instead the effect of the high anisotropy phase is reflected in changes to . At the transition, and of -FeO and -(FeCr)O become more negative (increasing in magnitude), indicating an anisotropic shift of the local electronic charge distribution. Upon further warming, the two octahedral components behave differently: decreases in magnitude to the near-zero value expected for undistorted Fe, while becomes more negative. The near zero value of in -FeO indicates an isotropic electric field gradient, while in -(FeCr)O is more asymmetric. Conversely, of -FeO is more negative than -(FeCr)O. From these parameters, we find that reducing Fe-O hybridization by doping Cr into the DOh site causes the ROh (DOh) site to become less (more) isotropic in the high anisotropy phase. Effectively, the Cr ions prevent the -(FeCr)O Fe local environments from recovering –a sign of strong intersite exchange interactions.
This behaviour is further identified in the temperature dependence of , which exhibit local minima at the transition temperatures of 127.5 K and 130 K for -FeO and -(FeCr)O, respectively, then increase with further warming to 160 K. The local minima is a consequence of spin-orbit coupling contribution being minimized as is minimized at the magnetic phase transition[12]. The agreement between these of -FeO and -(FeCr)O is further evidence that the Fermi contact terms are not impacted by Cr-doping. The subsequent increase in the is caused by the increase of and spin-orbit coupling as the high-coercivity magnetic phase is established and the anisotropic contributions to increase. While of the tetrahedral component of -(FeCr)O was reduced from that of -FeO in the low temperature phase due to the incorporation of Cr ions into the DOh site reducing Fe-O hybridization at the site, in the high anisotropy phase, all sites are impacted. This effect shows that the hard ferrimagnetic phase of -FeO is a consequence of strong hybridization between all the Fe sites. Furthermore, since -(FeCr)O maintains the unique temperature dependent behaviour, it confirms that the Cr ions have a perturbative effect on -FeO –important as we shift focus to the origin of the unusual environments.
Comparison of the component hyperfine parameters in the high anisotropy phase reveals that the average and of the and components in -(FeCr)O map onto those of the component in -FeO. This suggests the environments observed in the Cr-doped samples are a decomposition of the single environment measured in -FeO.

Figure 4. Schematic of electronic coupling for different electronic configurations where filled (empty) boxes describe filled (empty) Fe-O hybrid orbitals. Distance between orbitals increases (decreases) as the coupling strength increases (decreases). Increased covalency with ligand hole formation highlighted by thicker Fe-O bonds. In the bond-disproportionated () picture, the more (less) ionic configuration with reduced (increased) coupling strength has longer (shorter) bonds.
Since the splitting occurs with disruption to the Fe-O hybridization network, the single environment relies on ligand interactions for stability. Based on the hyperfine parameters of the three components, we propose an intermediate/high spin groundstate (where is a ligand hole) instead of the nominal high spin . Such a character is consistent with the measured of -FeO above the characteristic transition, which is decreased from that expected for in coordination. While is true in general, the increased covalency of the configuration causes the electron density to increase and to decrease. Anisotropy of this hybridization creates non-zero and . Most significantly, this groundstate explains the characteristically low of the high anisotropy phase due to reduced Fermi contact contributions as well as the low magnetic moment () reported for the site[10].
A similar groundstate, with a strong ligand hole component has been linked to the perovskite SrCoO, where a intermediate state is stabilized by negative charge transfer energy and a relatively large Coulomb repulsion [28, 29, 30]. Other such materials with ligand-hole dominated groundstates include CaFeO ()[31], LaCoO ()[32] and the rare earth () perovskite nickelates RNiO ()[33, 34]. These compounds are notable because they stabilize from into a configurations. In the case of the nickelates, the state is formed through bond-disproportionation[35, 36]. Unlike charge disproportionation, which describes a transfer of charge from one cation environment to the other, here the physical charge density around each environment is expected to remain constant while differences in bond lengths are the result of charge rearrangement around the cation[37]. Such behaviour is the result of a charge self-regulation mechanism, where rearrangement to the electronic density of the cation is compensated by that of the ligand[38]. This mechanism explains the two tetrahedral local environments in -(FeCr)O where the and components describe and , respectively. This bond disproportionated state forms because Cr-doping disrupts the Fe-O hybridization that stabilizes the groundstate in -FeO.

Figure 5. a) High throughput Fe RIXS map of -FeO at 300 K. Orange circles mark excitations, while black and white lines highlight charge transfer and fluorescence, respectively. (b) High resolution RIXS spectra of -FeO (black) and -(FeCr)O (red). Spectra are offset for clarity
For further evidence of this negative charge transfer model, we turn to Fe resonant inelastic x-ray scattering (RIXS) at 300 K in the ferrimagnetic phase. The high throughput energy map of -FeO shown in fig:5a shows localized excitations, broad charge transfer excitations, and fluorescence, as marked on the figure. Here the charge transfer excitations provide further evidence of strong O and Fe hybridization while the fluorescent signal extends to unusually low loss energies. High resolution RIXS spectra of both -FeO and -(FeCr)O (fig:5b) confirm that a continuum of spectral weight extends from the elastic line to the excitations for incident energies up to 707.0 eV. Such a signal is a sign of negative charge transfer as the partially occupied O DOS extends across the Fermi level[34].
Curiously, not just the low energy fluorescence but the excitations of the RIXS spectra are well matched between -FeO and -(FeCr)O –signalling the same Fe electronic structure. This apparent conflict with the Mössbauer spectroscopy results is due to the different timescales between the RIXS ( fs) and Mössbauer processes ( ns). The environment that we observe in the Mössbauer spectra of -FeO is likely a dynamic fluctuation of environments, resolved at the RIXS timescale.
For both -FeO and -(FeCr)O, the transformation of the site local environment signals the onset of behaviour. This behaviour occurs because in the new groundstates, changes in orbital occupancy increase the orbital momentum and strengthen spin-orbit coupling –fundamentally altering the magnetocrystalline anisotropy[39]. While in other materials is a consequence of coupling between magnetic and spin-orbit coupled (often ) subsystems, our results reveal that in -FeO the tetrahedral site becomes the strongly spin-orbit coupled subsystem at the magnetic phase transition. These results point to the ability of strong metal-ligand hybridization to induce spin-orbit coupling. Being able to tune this behaviour would have implications for magnetic, electronic and orbitronic applications, and provides a clear direction for future studies.
Acknowledgements– The authors acknowledge funding from the Natural Sciences and Engineering Research Council of Canada (Vanier Scholarship program, and RGPIN-2018-05012) and the Canada Foundation for Innovation. Use of the Advanced Light Source, a US Department of Energy (DOE) Office of Science User Facility was supported by the US DOE under contract DE-AC02-05CH11231. The authors thank the European Synchrotron Radiation Facility for providing beam time under proposals IHHC4037, IHHC4143, and IHHC4252.
Data availability– The RIXS data are openly available at https://doi.esrf.fr/10.15151/ESRF-DC-2493094440. The other data that support the findings of this article are not publicly available. The data are available from the authors upon request.
References
- [4]C. E. Patrick and J. B. Staunton, Physical Review Materials 3, 101401 (2019).
- [5]R. Grössinger, X. Sun, R. Eibler, K. Buschow, and H. Kirchmayr, Journal of magnetism and magnetic materials 58, 55 (1986).
- [6]J.-S. Zhou, L. Marshall, Z.-Y. Li, X. Li, and J.-M. He, Physical Review B 102, 104420 (2020).
- [7]A. Moskvin, Magnetochemistry 7, 111 (2021).
- [8]H. J. Shin, J. S. Kim, K. W. Jeong, J. H. Kim, N. Lee, and Y. J. Choi, Scientific reports 13, 7105 (2023).
- [9]R. Nickel, C. Sun, D. M. Meira, P. Shafer, and J. van Lierop, Physical Review Materials 8, 024407 (2024).
- [10]M. Gich, C. Frontera, A. Roig, E. Taboada, E. Molins, H. Rechenberg, J. Ardisson, W. Macedo, C. Ritter, V. Hardy, et al., Chemistry of Materials 18, 3889 (2006).
- [11]J. L. García-Muñoz, A. Romaguera, F. Fauth, J. Nogués, and M. Gich, Chemistry of Materials 29, 9705 (2017).
- [12]Y.-C. Tseng, N. M. Souza-Neto, D. Haskel, M. Gich, C. Frontera, A. Roig, M. Van Veenendaal, and J. Nogués, Physical Review B 79, 094404 (2009).
- [13]R. Nickel, J. Gibbs, J. Burgess, P. Shafer, D. M. Meira, C. Sun, and J. van Lierop, Nano Letters 23, 7845 (2023), pMID: 37625017, https://doi.org/10.1021/acs.nanolett.3c01512.
- [14]I. Leonov, A. Yaresko, V. Antonov, and V. Anisimov, Physical Review B—Condensed Matter and Materials Physics 74, 165117 (2006).
- [15]E. Tronc, C. Chanéac, J. Jolivet, and J. Greneche, Journal of applied physics 98, 053901 (2005).
- [16]J.-L. Rehspringer, S. Vilminot, D. Niznansky, K. Zaveta, C. Estournès, and M. Kurmoo, in ICAME 2005 (Springer, 2006) pp. 475–481.
- [17]J. Kohout, P. Brázda, K. Závěta, D. Kubániová, T. Kmjeć, L. Kubíčková, M. Klementová, E. Šantavá, and A. Lančok, Journal of Applied Physics 117, 17D505 (2015).
- [18]R. Jones, R. Nickel, P. K. Manna, J. Hilman, and J. van Lierop, Physical Review B 100, 094425 (2019).
- [19]T. Kimura, R. Kumai, Y. Okimoto, Y. Tomioka, and Y. Tokura, Physical Review B 62, 15021 (2000).
- [20]C. S. Hong, N. H. Hur, and Y. N. Choi, Solid state communications 133, 531 (2005).
- [21]I. Raies, S. A. Al Dulmani, L. Ben Farhat, E. E. Fadlallah, and M. Amami, Journal of Asian Ceramic Societies 8, 1095 (2020).
- [22]A. Kumar, M. K. Warshi, A. Sagdeo, and P. Sagdeo, The Journal of Physical Chemistry C 125, 14048 (2021).
- [23]J. Tiburcio, E. Sacari, J. Chacaltana, J. Medina, F. Gamarra, C. Polo, E. Mamani, and A. Quispe, Energies 16, 786 (2023).
- [24]K. Kummer, A. Fondacaro, E. Jimenez, E. Velez-Fort, A. Amorese, M. Aspbury, F. Yakhou-Harris, P. Van Der Linden, and N. Brookes, Synchrotron Radiation 23, 464 (2016).
- [25]N. Brookes, F. Yakhou-Harris, K. Kummer, A. Fondacaro, J. Cezar, D. Betto, E. Velez-Fort, A. Amorese, G. Ghiringhelli, L. Braicovich, et al., Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 903, 175 (2018).
- [27]P. Gütlich, E. Bill, and A. X. Trautwein, Mössbauer spectroscopy and transition metal chemistry: Fundamentals and applications (Springer Science & Business Media, 2010).
- [28]R. Potze, G. Sawatzky, and M. Abbate, Physical Review B 51, 11501 (1995).
- [29]M. Zhuang, W. Zhang, A. Hu, and N. Ming, Physical Review B 57, 13655 (1998).
- [30]M. Pouchard, A. Villesuzanne, and J.-P. Doumerc, Journal of Solid State Chemistry 162, 282 (2001).
- [31]P. C. Rogge, R. U. Chandrasena, A. Cammarata, R. J. Green, P. Shafer, B. M. Lefler, A. Huon, A. Arab, E. Arenholz, H. N. Lee, et al., Physical review materials 2, 015002 (2018).
- [32]M. Korotin, S. Y. Ezhov, I. Solovyev, V. Anisimov, D. Khomskii, and G. Sawatzky, Physical Review B 54, 5309 (1996).
- [33]S. Johnston, A. Mukherjee, I. Elfimov, M. Berciu, and G. A. Sawatzky, Physical review letters 112, 106404 (2014).
- [34]V. Bisogni, S. Catalano, R. J. Green, M. Gibert, R. Scherwitzl, Y. Huang, V. N. Strocov, P. Zubko, S. Balandeh, J.-M. Triscone, et al., Nature Communications 7, 13017 (2016).
- [35]T. Mizokawa, D. Khomskii, and G. Sawatzky, Physical Review B 61, 11263 (2000).
- [36]R. J. Green, M. W. Haverkort, and G. A. Sawatzky, Physical Review B 94, 195127 (2016).
- [37]G. M. Dalpian, Q. Liu, J. Varignon, M. Bibes, and A. Zunger, Physical Review B 98, 075135 (2018).
- [38]H. Raebiger, S. Lany, and A. Zunger, Nature 453, 763 (2008).
- [39]P. Bruno, Physical Review B 39, 865 (1989).