Sr₂FeO₄: A transition metal oxide at the insulator-metal borderline

Ferrates with iron in the formal oxidation state 4+ form a class of compounds showing a variety of electronic and magnetic properties. For instance, the cubic perovskite $SrFeO$₃ shows a high metallic-like conductivity down to 4 K, but nevertheless orders antiferromagnetically below $T$_N $=$ 130 K. As expected from the crystal structure Mössbauer spectra of $SrFeO$₃ reveal a single site in the paramagnetic as well as antiferromagnetic phase, respectively. On the other hand, the slightly distorted perovskite $CaFeO$₃ is an antiferromagnetic semiconductor ($T$_N $=$ 120 K), and the Mössbauer spectra evidence two sites with different isomer shifts and hyperfine fields. These observations have been interpreted in terms of a 2 Fe⁴⁺ ==> Fe³⁺ + Fe⁵⁺ valence disproportionation.

Figure 1: Mössbauer spectra of the paramagnetic (top) and antiferromagnetic phase (bottom) of $Sr$₂FeO₄ . The small intensity contribution in the upper spectrum arising from a magnetic hyperfine sextet corresponds to unreacted $Fe$₂O₃.

In order to improve the understanding of the factors determining the electronic behavior of ferrates(IV) we have investigated $Sr$₂FeO₄ employing a variety of techniques. Samples of $Sr$₂FeO₄ were prepared from SrO and $Fe$₂O₃ in flowing $O$₂ gas atmosphere. The compound has the $K$₂NiF₄ -type crystal structure with quasi-two-dimensional sheets of corner sharing $FeO$₆ octahedra. Electrical resistivity and magnetic susceptibility measurements reveal that $Sr$₂FeO₄ is an antiferromagnetic semiconductor with $T$_N $=$ 60 K. The room temperature Mössbauer spectrum of $Sr$₂FeO₄ representing the paramagnetic phase (Fig. 1) consists of a quadrupole doublet with an isomer shift of -0.02 $mm s-1$ relative to alpha-iron. These observations are consistent with $Fe4+$ at a tetragonal lattice site, as expected from the crystal structure. The Mössbauer spectra of the magnetically ordered phase are quite complicated and can be interpreted in terms of four $Fe4+$ hyperfine sextets. The isomer shifts of the four components, however, are identical and typical for tetravalent iron. Thus, a valence disproportionation of $Fe4+$ into $Fe3+$ and $Fe5+$ as well as a major $Fe3+$ fraction due to oxygen nonstoichiometry can be excluded. Possible origins for the complicated Mössbauer spectra are subtle structural changes accompanying the paramagnetic - antiferromagnetic phase transition and/or the formation of a complicated spin structure.

Figure 2: Optical reflectivity of $Sr$₂FeO₄ at room temperature for different pressures.

In view of the metallic properties of $SrFeO$₃ we have investigated the possibility of a pressure-driven semiconductor to metal transition in $Sr$₂FeO₄ by optical reflectivity studies (0.5 - 4 eV, T = 300 K) using diamond anvil cell techniques. Reflectivity spectra at zero pressure are consistent with the semiconducting behavior of the compound. For pressures above 6 GPa a major change is observed in the near-infrared reflectivity (see Fig. 2), which corresponds to a significant increase of the near-infrared oscillator strength. This behavior is attributed to a narrowing of the energy gap between ground and excited states resulting in a semiconductor-metal transition near 6 GPa.

Figure 3: Raman spectra of $Sr$₂FeO₄ at 20 K and different pressures. Raman bands at 208, 281, and 509 $cm-1$ (at zero pressure) correspond to $A$_1g(Sr), $E$_g(O),and $A$_1g(O) phonon modes, respectively, of the $K$₂NiF₄ structure. In addition, there is an intense Raman band at 380 $cm-1$.

Additional evidence for a phase transition near 6 GPa comes from Raman investigations. Fig. 3 shows low-temperature Raman spectra of $Sr$₂FeO₄ for different pressures. The observed Raman bands can be assigned to the Raman-active phonon modes in the $K$₂NiF₄ crystal structure, except for the intense band at about 380 $cm-1$. Upon heating to room temperature this additional Raman band broadens dramatically without showing any discontinuity at the magnetic ordering transition. Since the band shifts to lower frequency for an isotope exchanged $Sr$₂Fe¹⁸O₄ sample, it clearly is related to oxygen vibrations. Possible origins for the additional band in the low pressure range are a symmetry lowering by a small displacement of the $Fe4+$ ions from the centre position of the $FeO$₆ octahedra or a phonon Raman scattering induced by the spin degrees of freedom. With increasing pressure, the Raman band disappears in a rather narrow pressure range around 5.5 GPa, and above 6 GPa only the phonon Raman modes characteristic of the $K$₂NiF₄ structure are seen. Whether the disappearance of the additional Raman band is related to the change of the electronic behavior occurring in the same pressure range remains to be clarified.

The pressure-induced changes in electronic structure can be interpreted within a cluster configuration interaction model. Considering a single $[FeO$₆]^8- unit, the electronic ground state of $Fe4+$ can be described in terms of hybridization of the ionic $3d4$ and the charge transfer $3d5L-1$ configurations which are separated by the charge transfer energy Delta . For octahedral symmetry the ground state is a $5E$_g state. The general trend of a decreasing Delta with an increase in oxidation number and the large exchange stabilization of the $3d5$ configuration lead to a dominant weight of the $3d5L-1$ configuration in the ground state (negative Delta ). It is suggested that either a local hybridized $5E$ state or an itinerant $3d5L-1$ state may be the electronic ground state in ferrates(IV). The relative stability of the two states is largely determined by the oxygen-bandwidth $W$_p, which should be larger for the perovskite $SrFeO$₃ than for the $K$₂NiF₄ -type system $Sr$₂FeO₄ . Accordingly, the itinerant $3d5L-1$ is the ground state for the former, the local $5E$ for the latter. On the other hand, $W$_p is also increased by pressure via shortening of the Fe-O-Fe distances which explains the gap narrowing or closure seen in the reflectivity spectra. These considerations are similar to electronic structure models for oxocuprates with $Cu3+$ where localized $1A$₁ or itinerant $d9L-1$ states are found.

(P. Adler, A.F. Goncharov, K. Syassen and E. Schönherr)

From the yearbook of the institute ("Wissenschaftlicher Tätigkeitsbericht") 1994

webmaster