Magnetic and electronic properties of CaFeO₂Cl

2.1 Synthesis

CaFeO₂Cl was synthesized by a modified procedure of the protocol reported by J. Ackermann [19]. The starting materials Fe₂O₃ (Alfa Aesar, 99.99%), and CaCl₂ (Abcr, 99.99%), were mixed in a ratio 1:100 and ground in an agate mortar in air. The samples were prepared in alumina crucibles under air by heating to 850°C for 12 h, then cooled to 350°C for 5 h, and finally cooled to room temperature (step 1). After grounding the samples obtained in step one, the procedure from step one was repeated with a slower cooling rate of 20 K h⁻¹ (step 2). Washing the product several times with distilled water removed all excess CaCl₂. The resulting samples were dried under vacuum and stored under argon. This approach yielded a polycrystalline red-brown powder with brown plate-like single crystals. The samples contained the desired product CaFeO₂Cl and some unreacted Fe₂O₃. Through this optimized synthesis, yields of up to 92% were achieved.

2.2 X-ray diffraction and structure refinement

The powder X-ray patterns were obtained with a STOE Stadi-P diffractometer (MoKα₁ radiation) equipped with a STOE Mythen 1k detector. Rietveld refinement were done with Topas [21]. Single-crystal diffraction data was recorded with a STOE IPDS-I diffractometer at room temperature using MoKα₁ radiation (λ=0.71069 Å). Reflection intensity integration, data reductions, and multi-scan absorption corrections were performed with Apex2 [22] and Sadabs [23]. The structure was solved with Jana2006 [24] and refined with the Shelxl crystallographic software package [25].

CCDC 1957141 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

2.3 Magnetic measurements

Magnetic measurements with CaFeO₂Cl single crystals were carried out with a vibrating sample magnetometer (VSM) option in a Quantum Design Physical-Property-Measurement-System (PPMS-9). Crystals were selected manually from the samples, aligned parallel to the ab crystal plane of CaFeO₂Cl in a parquet-like pattern and glued with low temperature varnish. Magnetization measurements were carried out with the ab crystal plane parallel and perpendicular to the magnetic field.

2.4 Electronic structure DFT calculations

Electronic structure calculations were performed using the Vienna ab initio simulation package (Vasp) [26], [27] which is based on density functional theory (DFT) and plane wave basis sets. Projector-augmented waves (PAW) [28] were used and contributions of correlation and exchange were treated in the generalized-gradient approximation (GGA) [29]. The strongly correlated Fe 3d states were corrected using the LDA+U method in the rotationally invariant approach by Dudarev et al. [30].

2.5 Optical measurements

Diffuse reflectance spectra were measured with powder samples on a UV/Vis Jasco V-650 spectrophotometer (200–800 nm). The spectra were converted to absorption spectra based on the Kubelka-Munk [31] theory to determine the optical band gap.

2.6 Mössbauer spectroscopic measurements

A ⁵⁷Co source in a Rh matrix was used for the ⁵⁷Fe Mössbauer spectroscopic investigation of CaFeO₂Cl. The measurement was performed in a continuous-flow cryostat system (Janis Research Co LLC). The usual transmission geometry was used. The source was kept at room temperature while the sample was cooled to T=6 K. The optimal absorber thickness was calculated according to the work of Long et al. [32]. The sample was placed in a thin-walled PMMA container and diluted with potassium chloride for a complete distribution of the sample within the container volume. Fitting of the experimental data was performed with the WinNormos for Igor6 program package [33].

3.1 Synthesis and crystal structure

CaFeO₂Cl was synthesized in a CaCl₂ flux in an open system from Fe₂O₃ as a red-brown, moisture and air stable product. Single-crystal X-ray diffraction data confirmed the structure in the monoclinic centrosymmetric space group C2/m (No. 12) with the lattice parameters a=9.969(2), b=3.811(8), c=8.735(17) Å and β=103.62(3)°. The single-crystal X-ray data was used to refine the powder X-ray pattern (Fig. 2). The Rietveld fit yielded CaFeO₂Cl as the main component (ca. 92 wt.%) with Fe₂O₃ as impurity.

Fig. 2:

X-ray powder pattern (MoKα₁) of the CaFeO₂Cl sample (blue) with Rietveld fit (red) and difference line (grey line).

Figure 1 shows the crystal structure of CaFeO₂Cl. Layers of edge sharing FeO_2/2O_3/3 polyhedra alternate with regions of CaCl. The FeO₅ pyramid is slightly distorted with four different Fe–O distances ranging from 1.89 to 2.00 Å. The Addison τ parameter (τ=(β–α)/60; α, β=largest bond angles) [34] of 0.1 confirms the square pyramidal coordination. Calcium has three oxygen and four chlorine neighbours. The iron substructure is a planar but slightly elongated honeycomb net with Fe–Fe distances of 2.743 and 2.959 Å.

3.2 Optical properties

Figure 3 shows the results of an optical reflection measurement of the CaFeO₂Cl powder (λ=500–800 nm). Because weak signals in absorbance spectra are enhanced in reflected spectra, reflectance spectra in absorbance units cannot be compared directly. Analyzing reflection spectra using the Kubelka-Munk function makes them similar to absorbance spectra and allows the determination of the band gap energy E_g [31].

Fig. 3:

Band gap calculation using the Kubelka-Munk function.

F(R∞)=1−R∞R∞=K(λ)s(λ)∝α=(hν−Eg)2hν

R_∞ is the diffuse reflectance of an infinitely thick sample, K(λ) is the absorption coefficient and s(λ) is the scattering coefficient. The band gap is determined by the extrapolation from the linear portion of the plot of [F(R_∞)hν]² against hν (Tauc plot). The estimated band gap is 1.3 eV (Fig. 3), which is consistent with the red-brown color of CaFeO₂Cl.

3.3 Magnetism

Selected plate-like single crystals of CaFeO₂Cl were arranged parallel on the sample holder in order to measure the magnetization in fields applied either perpendicular (⊥) or parallel (//) to the layers. Note that the crystals were not aligned along the directions perpendicular to the layers. Figure 4 shows magnetization isotherms at temperatures between 5 and 305 K.

Fig. 4:

Magnetization isotherms of CaFeO₂Cl between T=5 and 305 K with fields aligned perpendicular (top) and parallel to the FeO₂ layers (bottom). (fu=formula unit; 1 kOe=7.96×10⁴ A m⁻¹)

The magnetization increases linearly with the field strength but remains below 0.05 μ_B per formula unit CaFeO₂Cl, which indicates that the magnetic moments of the Fe³⁺ ions (S=5/2) are antiferromagnetically ordered in the considered temperature range (5–300 K). The magnetization increases at lower temperatures, which is contrary to the expected either constant values if the field is perpendicular or a decrease if it is parallel to the magnetization direction (easy axis) of an antiferromagnet [35]. Given that the magnetization remains very small even at low temperatures, the increase may be the result of traces of paramagnetic impurities. The anisotropy of the magnetization is relatively small with respect to the layered crystal structure. Assuming that the paramagnetic contributions (from impurities) are isotropic, we find that the magnetization at T=5 K is about 25% higher if the field is parallel to the FeO_2/2O_3/3 layers. This indicates that the alignment of the magnetic moments, or easy axis, should be perpendicular to the layers.

3.4 Mössbauer spectroscopy

The ⁵⁷Fe spectrum of polycrystalline CaFeO₂Cl at T=6 K is shown in Fig. 5. The corresponding fitting parameters for the most reliable fit are listed in Table 1. Figure 5 shows two different fitting approaches. Two overlapping sextet signals are visible in the recorded spectrum. The less intense one corresponds to the by-product Fe₂O₃ and the other one to CaFeO₂Cl. The top spectrum of Fig. 5 shows a fit with two regular sextets, the bottom one a fitting with a full Hamiltonian. No impurity phases other than Fe₂O₃ can be observed in the experimental spectrum, in accordance with the PXRD data. It is clearly apparent that the fit for the top spectrum is slightly off-centered from the normal sextet splitting. The full Hamiltonian fit with the inclusion of the parameter θ, which describes the angle between the direction of the magnetic hyperfine field and the tensor of the electrical field gradient, gives a more satisfactory fitting. Similar results have recently been reported for the brownmillerite phase Sr₂Fe₂O₅ [36].

Fig. 5:

Experimental (black dots) and simulated (colored lines) ⁵⁷Fe Mössbauer spectra of CaFeO₂Cl at T=6 K. Top: CaFeO₂Cl signal fitted with a sextet model; bottom: CaFeO₂Cl fitted with a complete Hamiltonian.

The isomer shift of 0.30 mm s⁻¹ indicates iron in the oxidation state +3 as can be expected from the empirical formula. This is in agreement with a large collection of ⁵⁷Fe data of iron oxides compiled by Menil [37]. The square pyramidal coordination of Fe(III) is rare; however, our ⁵⁷Fe Mössbauer spectroscopic data is in good agreement with the series of iron substituted chromates RETiCr_1−xFe_xO₅ [38].

The slightly negative quadrupole splitting of –0.97 mm s⁻¹ results from the asymmetric coordination environment of the iron site (distorted square pyramid). Full magnetic hyperfine field splitting is observed with a field of 42.6 T. The parameter θ mentioned above leads to the conclusion that the magnetic moments are aligned along a preferred crystallographic direction.

3.5 DFT calculations

The electronic band structure was calculated using the VASP package. Due to the actually unknown magnetic structure, the total energies of different trial magnetic ordering patterns were compared in order to identify a probable magnetic state. The primitive unit cell of CaFeO₂Cl contains four iron positions. The trial magnetic structures were either ferromagnetic (FM ↑↑↑↑), or antiferromagnetic (AF1 ↑↓↑↓; AF2 ↑↓↓↑; AF3 ↑↑↓↓). The calculations have revealed that AF1 is the most stable configuration, followed by AF2 (+0.1 eV), AF3 (+0.36 eV) and FM (+0.4 eV). AF1 corresponds to a Néel-type magnetic structure where all moments of the iron atoms within the honeycomb layer are alternating spin up and down as shown in Fig. 6.

Fig. 6:

Most stable trial AF magnetic structure in the honeycomb Fe layer of CaFeO₂Cl. Filled and open circles mark spin up and down, respectively.

The spin-polarized band structure shown in Fig. 7 was calculated using this model of the antiferromagnetic ordering and the LDA+U approach with a U_eff (= U–J; J=0) of 4 eV in order to account for the strongly correlated Fe-3d states. The calculated band gap E_g occurs between the O-2p valence bands and Fe-3d conduction bands. E_g correlates linearly with the U_eff parameter and results in 1.96 eV for U=4 eV and 1.59 eV for U=3 eV. The calculated ordered magnetic moment of 4.2–4.3 μ_B per Fe atom is smaller than the expected 5 μ_B per Fe atom for the S=5/2 state. This may have its origin in the special magnetic properties of the honeycomb substructure with only three neighbors. It has been shown that the magnetization in a threefold connected S=1/2 system is reduced by a factor of 0.87 compared to the square lattice [39]. A reduced ordered magnetic moment of 2.3 μ_B has also been found in the S=3/2 antiferromagnetic honeycomb lattice of Li₂MnO₃ [40].

Fig. 7:

Spin polarized LDA+U band structure of antiferromagnetic CaFeO₂Cl.