Structure and cation distribution in perovskites with small cations at the A site: the case of ScCoO3.

We synthesize ScCoO3 perovskite and its solid solutions, ScCo1-x Fe x O3 and ScCo1-x Cr x O3, under high pressure (6 GPa) and high temperature (1570 K) conditions. We find noticeable shifts from the stoichiometric compositions, expressed as (Sc1-xMx )MO3 with x = 0.05-0.11 and M = Co, (Co, Fe) and (Co, Cr). The crystal structure of (Sc0.95Co0.05)CoO3 is refined using synchrotron x-ray powder diffraction data: space group Pnma (No. 62), Z = 4 and lattice parameters a = 5.26766(1) Å, b = 7.14027(2) Å and c = 4.92231(1) Å. (Sc0.95Co0.05)CoO3 crystallizes in the GdFeO3-type structure similar to other members of the perovskite cobaltite family, ACoO3 (A3+ = Y and Pr-Lu). There is evidence that (Sc0.95Co0.05)CoO3 has non-magnetic low-spin Co3+ ions at the B site and paramagnetic high-spin Co3+ ions at the A site. In the iron-doped samples (Sc1-xMx )MO3 with M = (Co, Fe), Fe3+ ions have a strong preference to occupy the A site of such perovskites at small doping levels.

Introduction

ABO3 perovskite-type compounds, where A3+ = Y and La-Lu and B3+ = V, Cr, Mn, Fe, Co, Ni and Ni0.5Mn0.5, and their solid solutions have been attracting a lot of attention for decades from the viewpoints of fundamental physics and practical applications [1-3]. For example, some ACrO3 compounds exhibit spin-reorientation transitions [4], and doped ACrO3 are good oxygen-ion conductors and show sensitivity toward methanol, ethanol, some gases and humidity [5]. ACoO3 compounds have been investigated a lot because of spin-state transitions in Co3+ ions and metal-insulator transitions [6]; ACoO3 also exhibit thermoelectric [7] and catalytic properties [8]. The large A site of ABO3 perovskites is usually occupied by larger cations (such as, rare earths), and the small B site by smaller cations (such as, transition metals). The A/B inter-site mixing is very rare in simple perovskites. In complex perovskites, the A/B inter-site and intra-site mixing can occur, and the cation distribution could significantly modify properties of materials, such as, magnetic and dielectric properties. For example, the appearance of Mg2+ at the A site in BaMg1/3Ta2/3O3 [9] and Mn2+ at the A site in SrTiO3 [10] results in increased dielectric loss, and the degree of Ni2+ and Mn4+ ordering at the B site in ANi0.5Mn0.5O3 changes magnetism of the system [11].

In recent years, ABO3 perovskites have been extensively expanded to smaller A cations, such as, Mn2+, Sc3+, and In3+ [12] with expectations to find new magneto-structural coupling behaviours because of large structural distortions. Unusual physical properties were indeed found, for example, in MnVO3 (incommensurate magnetic ordering and metallic conductivity) [13], In2NiMnO6 (spin-induced ferroelectricity) [12] and ScVO3 (distinct magnetic, orbital and structural properties from other members of the AVO3 (A3+ = Y and La-Lu) family [14]). Because the difference in the sizes of the A and B cations decreases the probability of the A/B inter-site mixing is increased in such perovskites. Noticeable cation mixing or, more precisely, shifts in the composition were found in ()MnO3 (1/9 ≤ y ≤ 1/3) [15] and ()Mn0.65Ni0.35O3 [16]; in such perovskites, small divalent transition metals are located at the A site.

In this work, we investigated ScCoO3 perovskite and its solid solutions ScCo1−FeO3 and ScCo1−CrO3. We found noticeable shifts in the composition of such perovskites from ScCoO3 to (Sc1−Co)CoO3 and the appearance of significant amounts of small trivalent cations (Co3+ and Fe3+) at the A site. To the best of our knowledge, the presence of Fe3+ at the A site was detected for the first time in ABO3 perovskites.

Experimental details

Samples were prepared from stoichiometric mixtures of Sc2O3 (99.9%), Co3O4 (99.9%), Cr2O3 (99.9%), Fe2O3 (99.999%) and KClO4 (as the source of oxygen). The mixtures were prepared in a glove box, placed in Au capsules (in the amount of about 0.5 g for each sample) and treated at 6 GPa in a belt-type high-pressure apparatus at 1570 K for 2 h (heating rate to the desired temperature was 10 min). After the heat treatment, the samples were quenched to room temperature (RT), and the pressure was slowly released. The samples were washed in water to remove KCl obtained after the decomposition of KClO4.

X-ray powder diffraction (XRPD) data were collected at RT on a RIGAKU Ultima III diffractometer using CuKα radiation (2θ range of 10–100°, a step width of 0.02°, and a counting time of 2s/step). Synchrotron XRPD data were measured at 293 K on a large Debye-Scherrer camera at the BL15XU beam line of SPring-8 [17]. The intensity data were collected between 1° and 61.5° at 0.003° intervals in 2θ; the incident beam was monochromatized at λ = 0.65298 Å. The samples were packed into Lindenmann glass capillaries (inner diameter: 0.1 mm), which were rotated during measurements. Absorption coefficients were also measured, and Rietveld analysis was performed using the RIETAN-2000 program [18].

Electron probe microanalysis (EPMA) was performed using a JEOL JXA-8500F instrument. The surface of the pellets was polished on a fine alumina (0.3 μm) coated film before the EPMA measurements; and Sc2O3 and Co3O4 were used as standard samples for Sc and Co, respectively.

DC magnetic susceptibilities (χ = M/H) were measured using SQUID magnetometers (Quantum Design, MPMS-XL and 1T) between 2 and 400 K in different applied magnetic fields under both zero-field-cooled (ZFC) and field-cooled (FC) conditions. FC measurements were performed on cooling (FCC) from high temperatures to 2 K after the ZFC measurements. In all ZFC measurements, samples were rapidly (within 3–5 min) inserted into a magnetometer, which was kept at 10 K; then, temperature was set to 2 K, and finally a measurement magnetic field was applied. Isothermal magnetization measurements (M versus H) were performed between −70 and 70 kOe at 2 K and 300 K. Specific heat, Cp, was recorded between 2 and 300 K on cooling at 0 and 90 kOe by a pulse relaxation method using a commercial calorimeter (Quantum Design PPMS). 57Fe Mössbauer spectra were recorded at 300 K using a conventional constant-acceleration spectrometer MS-1104Em in the transmission geometry. The radiation source 57Co(Rh) was kept at RT. All isomer shifts are referred to α-Fe at 300 K. The experimental spectra were processed and analysed using methods of spectral simulations implemented in the SpectrRelax program [19]. Differential scanning calorimetry (DSC) curves of (Sc0.95Co0.05)CoO3 powder were recorded on a Mettler Toledo DSC1 STARe system at a heating/cooling rate of 10 K min−1 between 290 K and 873 K in open Al capsules; no DSC anomalies were detected, and the sample remained single-phase after the DSC experiment.

Results and discussion

The stoichiometric ScCoO3 sample contained Sc2O3 impurity (figure 1(a)) suggesting that the composition of the main perovskite phase is shifted according to the scheme:

Figure 1.

(a) A portion of the experimental synchrotron XRPD pattern of ScCoO3. The bars show possible Bragg reflection positions for the perovskite phase and Sc2O3 impurity (from top to bottom); hkl indexes for some reflections are given. A star marks a reflection from Au (a contamination from an Au capsule). (b) Portions of experimental (black crosses), calculated (red line) and difference (black line) synchrotron XRPD patterns for (Sc0.95Co0.05)CoO3. The bars show possible Bragg reflection positions for the perovskite phase.

The amount of Sc2O3 was estimated to be 3–5 weight % from the Rietveld fitting of different laboratory XRPD data. The Rietveld refinement of the synchrotron XRPD data (figure 1(a)) gave about 7.7 weight % of Sc2O3. A sample with the total chemical composition of Sc0.9CoO2.85 (≈(Sc0.95Co0.05)CoO3) was prepared without impurities. The EPMA showed that the Sc:Co ratio was 0.901(11):1 in Sc0.9CoO2.85, in very good agreement with the target chemical composition. No other chemical elements (such as, K and Cl) were detected in Sc0.9CoO2.85.

The structural analysis showed that all cation and oxygen sites are fully occupied in Sc0.9CoO2.85 suggesting the ()CoO3 composition with x ≈ 0.0526. Structural parameters of (Sc0.95Co0.05)CoO3 are summarized in table 1, and selected bond lengths, angles and bond-valence sums (BVS) [20] in table 2. The BVS values of all the cation sites are close to the formal ionic values of +3. Experimental, calculated and difference synchrotron XRPD profiles are shown in figure 1(b). (Sc0.95Co0.05)CoO3 crystallizes in the GdFeO3-type structure with space group Pnma similar to other members of the perovskite cobaltite family ACoO3 (A3+ = Y and Pr-Lu) (except for LaCoO3) [6, 12]. (Sc0.95Co0.05)CoO3 has just two cation positions: the first position is for the A cation, and the second for the B cation. The lattice parameters and unit cell volume of (Sc0.95Co0.05)CoO3 follow the general trends observed in the ACoO3 (A3+ = Y and Pr-Lu) family with low-spin Co3+ ions [12]. However, the unit cell volume of (Sc0.95Co0.05)CoO3 (V = 185.141 Å3) with the Co3+ radius of 0.545 Å [21] is smaller than that of ScAlO3 (V = 185.915 Å3) with the Al3+ radius of 0.535 Å probably because of the shift from the stoichiometric composition [12].

Table 1.

Structure parameters of (Sc0.95Co0.05)CoO3 and (Sc0.95M0.05)MO3 (M = Co0.75Cr0.25) at room temperature.

Site	Wyckoff position	X	y	z	B (Å2)
(Sc0.95Co0.05)CoO3
Sc/Co	4c	0.07704(8)	0.25	0.97603(12)	0.333(9)
Co	4b	0	0	0.5	0.158(7)
O1	4c	0.4468(3)	0.25	0.1313(3)	0.16(3)
O2	8d	0.3101(2)	0.0631(2)	0.6830(2)	0.22(2)
(Sc0.95M0.05)MO3 (M = Co0.75Cr0.25)
Sc/M	4c	0.07621(8)	0.25	0.97673(14)	0.511(11)
M	4b	0	0	0.5	0.311(8)
O1	4c	0.4465(3)	0.25	0.1307(3)	0.20(4)
O2	8d	0.3101(3)	0.06395(18)	0.6838(3)	0.35(3)

The occupation factor of all sites is unity. Statistical distribution according to the compositions was assumed for the Sc/Co, Sc/M and M sites.

Space group Pnma (No 62); Z = 4.

(Sc0.95Co0.05)CoO3: a = 5.26766(1) Å, b = 7.14027(2) Å, c = 4.92231(1) Å, and V = 185.141(1) Å3; Rwp = 3.72%, Rp = 2.43%, RB = 2.08%, and RF = 1.44%; ρcal = 5.474 g cm−3.

(Sc0.95M0.05)MO3: a = 5.29435(1) Å, b = 7.20430(2) Å, c = 4.95170(1) Å, and V = 188.868(1) Å3; Rwp = 3.04%, Rp = 2.17%, RB = 3.83%, and RF = 2.85%; ρcal = 5.296 g cm−3.

Table 2.

Selected bond lengths, l (Å) < 3.0 Å, bond angles (deg) and bond valence sums (BVS) in (Sc0.95Co0.05)CoO3 and (Sc0.95M0.05)MO3 (M = Co0.75Cr0.25).

(Sc0.95Co0.05)CoO3		(Sc0.95M0.05)MO3 (M = Co0.75Cr0.25)
Sc—O1	2.051(2)	Sc—O1	2.061(2)
Sc—O2	2.091(1) × 2	Sc—O2	2.101(1) × 2
Sc—O1	2.093(2)	Sc—O1	2.104(2)
Sc—O2	2.317(1) × 2	Sc—O2	2.331(1) × 2
Sc—O2	2.528(1) × 2	Sc—O2	2.555(1) × 2
BVS(Sc3+)	3.02	BVS(Sc3+)	2.92
Co—O2	1.908(1) × 2	M—O2	1.917(1) × 2
Co—O2	1.919(1) × 2	M—O2	1.933(1) × 2
Co—O1	1.919(1) × 2	M—O1	1.935(1) × 2
BVS(Co3+)	3.35	BVS(Co3+)	3.24
Co—O1—Co	136.93(5)	M—O1—M	137.16(5)
Co—O2—Co	140.79(5) × 2	M—O2—M	140.61(5) × 2

BVS = ν = exp[(R0—l)/B], N is the coordination number, B = 0.37, R0(Sc3+) = 1.849, and R0(Co3+) = 1.70 [20].

The Mössbauer spectrum of (Sc0.95M0.05)MO3 ( at 300 K is shown on figure 2(a). It clearly consists of two doublets, Fe1 and Fe2, whose isomer shift (δFe1 < δFe2) and quadrupole splitting (ΔFe1 < ΔFe2) values indicate that the high-spin (HS) Fe3+ ions occupy two positions with different oxygen surrounding. The existence of these doublets could only originate from 57Fe3+ ions in positions corresponding to the A and B sublattices. The δFe1 (=0.32(1) mm s−1) and ΔFe1 (=0.42(1) mm s−1) values for the first Fe1 doublet are in good agreement with the δ = 0.31–0.33 mm s−1 and Δ = 0.38–0.50 mm s−1 values for Fe3+ ions located in the B site of Fe0.02O3 (R = Y, Eu and Lu) perovskites [22]. Taking into account that an increase in the average 〈Fe-O〉 distances leads generally to an increase in δ values [23], the Fe2 doublet with the larger isomer shift of δFe2 = 0.45(1) mm s−1 should correspond to 57Fe3+ ions located at the larger A site, and the larger quadrupole splitting of ΔFe2 = 1.26(1) mm s−1 indicates that Fe3+ ions have highly asymmetric coordination at the A site. It is expected that smaller 3d transition metals (Cr3+, Fe3+ and Co3+) should be displaced off the position occupied by the larger Sc3+ ions similar to the displacement of Mn2+ ions found in Sr0.98Mn0.02TiO3 [10]. Based on experimental values of the areas of the two doublets (IFe values in table 3), the distribution of 57Fe3+ ions is not statistical (the statistical distribution would result in about 5% of 57Fe3+ ions at the A site), but 57Fe3+ ions preferably occupy the A site (about 30%).

Figure 2.

57Fe Mössbauer spectra at 300 K and fitting results for (a) (Sc0.95M0.05)MO3 ( and (b) (Sc0.89M0.11)MO3 (M = Co0.6Fe0.4). The difference between the experimental and calculated spectra is shown at the bottom. (c) Distribution functions p(ΔFe1) of the quadrupole splitting ΔFe1 for the Fe1 subspectrum in (Sc0.95M0.05)MO3 ( (green crosses) and (Sc0.89M0.11)MO3 (M = Co0.6Fe0.4) (red squares).

Table 3.

Hyperfine parameters of the 57Fe Mössbauer spectra of (Sc0.95M0.05)MO3 ( and (Sc0.89M0.11)MO3 (M = Co0.6Fe0.4) at 300 K.

Sample	Sites	δ (mm s−1)	Δ (mm s−1)	W (mm s−1)	I (%)
(Sc0.95M0.05)MO3	Fe1	0.32(1)a	0.42(1)a	0.31(2)	69(1)
(M = Co0.95Fe0.05)	Fe2	0.45(2)	1.26(2)	0.34(1)	31(2)
(Sc0.89M0.11)MO3	Fe1	0.33(2)a	0.55(3)a	0.25(2)	88(2)
(M = Co0.6Fe0.4)	Fe2	0.43(2)	1.30(2)	0.28(1)	12(2)

These are the average 〈δFe1〉 and 〈ΔFe1〉 values obtained from the distribution functions p(δFe1) and p(ΔFe1).

δ is an isomer shift, Δ is quadrupole splitting, W is linewidth, and I is a relative intensity.

To verify the correctness of the doublet assignment to the A and B positions, we calculated a lattice contribution (lat) to the electric field gradient (EFG) tensor at 57Fe at the A and B positions, using the experimental crystallographic data of (Sc0.95Co0.05)CoO3 (table 1). After diagonalization, the main EFG tensor components (|VZZ| ≥ |VXX| ≥ |VYY|) were used to estimate the theoretical quadrupole splitting Δtheor = eQVZZ/2(1 + η2/3)1/2, where η ≡ (VXX–VYY)/VZZ is the parameter of asymmetry of EFG. The best agreement between the theoretical (Δtheor = 0.32 mm s−1 and Δtheor = 0.56 mm s−1) and experimental values (table 3) of quadrupole splitting (Δiexp) was obtained for the oxygen dipole polarizability of αО ≈ 0.6 Å3 (for nominal charges of ZO = −2, ZSc = +3 and ZCo = +3, and the quadrupole moment of 57Fe nuclei of Q = 0.21 barns [24]). The obtained high value of αО agrees well with the data for other oxides [25]. The main factors, which can be responsible for the observed discrepancy between the Δitheor and Δiexp values, are the uncertainty in choosing the effective charges on the ions (Sc, Co and O) and the nucleus quadrupole moment eQ for 57Fe nuclei [24]. However, our calculations qualitatively correctly predict the ratio of the Δi values ( and see table 3) thus confirming that our model, which was used for the fitting and interpretation of the experimental spectrum, is reliable, and the Fe1 and Fe2 doublets are correctly assigned to the B and A positions, respectively, in the structure of (Sc0.95M0.05)MO3 ().

We observed no difference between the ZFC and FCC curves measured at low magnetic fields (e.g., 0.1 kOe) and high magnetic fields (e.g., 70 kOe) (figure 3(a)). At high temperatures, almost no difference was found in magnetic susceptibilities measured at 0.1 and 70 kOe; however, at low temperatures, magnetic susceptibilities were suppressed by high magnetic fields in agreement with the isothermal M versus H curves (figure 4(a)). (Sc0.95Co0.05)CoO3 exhibits paramagnetic behaviour (figure 3(a)) with a relatively large effective magnetic moment of μeff = 1.749(6)μB/f.u. (μB is the Bohr magneton and f.u. is the formula unit) and the Curie–Weiss temperature of θ = −130(3) K. It is expected that Co3+ ions at the B site should be in the non-magnetic low-spin (LS) state similar to other members of the ACoO3 (A3+ = Y and Pr-Lu) family [6, 26]; the temperature of the spin-state (LS-to-HS) transition increases sharply with decreasing the size of the A type cation [6]. Therefore, a large effective magnetic moment should originate from the high-spin Co3+ ions located at the A site. The expected calculated effective magnetic moment is 1.124μB (for 0.0526Co3+), which is close to the experimentally obtained value. Large effective magnetic moments and Curie–Weiss temperatures were also observed in LaCo1−xMxO3 (M = Rh and Ir) [27]; μeff for the impurity-related magnetism is usually one order of magnitude smaller [28]. Magnetic properties of (Sc0.95M0.05)MO3 ( were very similar with those of (Sc0.95Co0.05)CoO3 (figure 3(b)), with a slightly larger μeff = 2.050(4)μB/f.u. because of the presence of Fe3+ ions (the expected μeff is about 1.63μB). Note that the intrinsic magnetic moment of (Sc0.95Co0.05)CoO3 is quite small at high temperatures; therefore, diamagnetic contributions (from sample holders and core diamagnetism) have a significant influence on the μeff and θ values (figure 3) making it difficult to discuss them.

Figure 3.

ZFC (filled symbols) and FCC (empty symbols) uncorrected magnetic susceptibility (χ = M/H) curves at 100 Oe and 70 kOe for (a) (Sc0.95Co0.05)CoO3 and (b) (Sc0.95M0.05)MO3 ( The right-hand axes give inverse FCC curves (χ−1 versus T) at 70 kOe. Parameters (μeff and θ) of the Curie–Weiss fits (bold lines) between 300 and 400 K are given. The thin lines show the same FCC χ−1 versus T curves at 70 kOe corrected for contributions from diamagnetic sample holders and core diamagnetism.

Figure 4.

(a) Uncorrected M versus H curves of (Sc0.95Co0.05)CoO3 at 2 K and 300 K (symbols with the line). The line shows the Brillouin function with g = 2 and S = 2 at 2 K, multiplied by 0.028. (b) M versus H curves of (Sc0.95M0.05)MO3 ( at 2 K and 300 K (symbols with the line). The line shows the Brillouin function with g = 2 and S = 5/2 at 2 K, multiplied by 0.036. Broken lines show M versus H curves corrected for diamagnetic sample holders and core diamagnetism.

The isothermal M versus H curves of (Sc0.95Co0.05)CoO3 and (Sc0.95M0.05)MO3 ( showed no hysteresis and passed through the origin (figure 4); no saturation behaviour was also observed at 2 K, in contrast with the expected property for free ions, that is, the Brillouin function behaviour. The M versus H curve of (Sc0.95Co0.05)CoO3 was linear at 300 K up to 70 kOe. Deviations from the Brillouin function behaviour was observed in some doped LaCoO3 samples [29]. Specific heat of (Sc0.95Co0.05)CoO3 is given on figure 5; between 9 and 31 K, the data follow the equation Cp/T = γ + β1T2 with γ = 7.86(8) mJmol−1 K−2 and β1 = 0.05452(17) mJmol−1 K−4 (the line in the inset of figure 5). Taking into account the fact that (Sc0.95Co0.05)CoO3 is an insulator, the upturn of the Cp/T values below 9 K and the apparent electronic contribution γ could originate from Schottky-type contributions or single-ion excitations. The β1 value of (Sc0.95Co0.05)CoO3 was close to that of ScRhO3 (β1 = 0.0589 mJmol−1 K−4) [28].

Figure 5.

Specific heat data of (Sc0.95Co0.05)CoO3 (powder was washed from KCl and then pressed into pellets at 3 GPa) at a zero magnetic field (white circles) and 90 kOe (blue diamonds) plotted as Cp/T versus T. The inset shows the details below 40 K; the red line is the fit with the equation Cp/T = γ + β1T2 between 9 and 31 K.

The presence of Co3+ ions in the LS and HS states is in qualitative agreement with the energy diagrams of Co3+ in crystal fields with local symmetries of O (for the B position, in the first approximation) and D4 (for the A position, in the first approximation) (figure 6) [30]. In the case of the same average bond distances 〈Co-O〉, the crystal field splitting, 5/3α4 (where α4 ∼ 1/[〈Co-O〉]5 is a radial integral), of Co3+ orbitals for the O octahedral site is higher than the crystal field splitting for the D4 site (16/27α′4). Moreover, the average 〈Co-O〉 bond distances are longer in the A position in comparison with the B position (table 2), thus, further reducing the 16/27α′4 value and the crystal field splitting. In the case of the O octahedral site, where the LS state of Co3+ is experimentally realized, it gives α4 > 6/5JH, where JH is the intraatomic Hund energy, [ELS–EHS = (6×(−2/3α4) + 2 × (−3JH))–(1×(−2/3α4) + (−10JH)) < 0]. In the case of the D4 site, the LS state of Co3+ will only be realized at α′4 > 27/10JH, [ELS–EHS = (2×(−8/9α′4) + 4×(−4/27α′4) + 2×(−3JH)) –(1×() + (−10JH)) < 0]. Considering that α′4 should be smaller than α4 (and with the same JH for Co3+), the above conditions result in the HS state of Co3+ at the D4 site. Note that the HS state of Co3+ was experimentally found in BiCoO3 [31], where Co3+ ions are located in a square pyramidal coordination.

Figure 6.

Schematic presentations of the energy diagrams for the Co3+ ion in the idealized A site with the D4 symmetry and in the idealized B site with the O symmetry in (Sc0.95Co0.05)CoO3 (where α0 and α4 are radial integrals, JH is the intra-atomic Hund energy).

By the analogy with Sc0.9CoO2.85, we prepared solid solutions with the total composition of Sc0.9Co1−FeO2.85 (x = 0.2, 0.4, 0.6 and 0.8). However, the samples with x = 0.2, 0.4 and 0.6 contained Sc2O3 impurity (figure 7) suggesting that the chemical composition of the perovskite phases is further shifted. Sc0.9Co0.2Fe0.8O2.85 already contained a large amount of ScFeO3 impurity with the corundum structure [32]. The lattice parameters of the solid solutions are shown on figure 8. Monotonic changes of the lattice parameters were found with a deviation for the two-phase sample with x = 0.8; this fact suggests that the solid solution limit is near x = 0.7.

Figure 7.

Portions of XRPD patterns of samples with the total composition of Sc0.9Co1−FeO2.85 with x = 0.2, 0.4, 0.6 and 0.8. The bars show possible Bragg reflection positions for the perovskite phase and Sc2O3 impurity (from top to bottom) on a–c. A star marks a reflection from KCl in unwashed samples. On d, the bars show possible Bragg reflection positions for the perovskite phase and ScFeO3 impurity (from top to bottom).

Figure 8.

Lattice parameters versus the composition for Sc0.9Co1− CrO2.85 (x = 0, 0.25 and 0.5) and ScCrO3 [36] (empty symbols) and Sc0.9Co1−FeO2.85 (x = 0, 0.05, 0.2, 0.4, 0.6 and 0.8) (filled symbols).

Almost single-phase Sc0.8Co0.6Fe0.4O2.7 (≈(Sc0.89M0.11)MO3 with M = Co0.6Fe0.4) was prepared whose Mössbauer spectrum at 300 K is shown on figure 2(b). The spectrum also consisted of two quadrupole doublets, Fe1 and Fe2, but in contrast to (Sc0.95M0.05)MO3 ( the most intense doublet Fe1 had broadened and asymmetrical components that could be caused by the existence of different configurations {(6-m)Co3+, mFe3+} in the local surrounding of the Fe3+ ions within the B sublattice. We fitted the experimental spectrum as a superposition of a discrete doublet Fe2 and a distribution p(ΔFe1) of the quadrupole splittings (ΔFe1), assuming a linear relation between ΔFe1 and δFe1 [33]. For comparison, a similar fitting analysis was carried out for (Sc0.95M0.05)MO3 ( (table 3); note that the Mössbauer parameters for this sample in two models (the first model is two discrete doublets, and the second one with a distribution for Fe1) were almost identical. The obtained p(ΔFe1) distributions are shown in figure 2(c), and the best-fit hyperfine parameters (average 〈δFe1〉 and 〈ΔFe1〉 values for the Fe1 subspectra) and relative intensities (Ii) of the partial spectra are listed in table 3. A comparison of these data shows that changing the iron content in the samples does not significantly affect hyperfine parameters of the Fe1 and Fe2 doublets, while their relative intensities undergo some changes. According to the experimental intensity ratio of the partial spectra, IFe1/IFe2 (table 3), in the case of (Sc0.89M0.11)MO3 (M = Co0.6Fe0.4), Fe3+ ions were distributed almost statistically between the A and B sites (statistical distribution would give 10% of Fe3+ at the A site, and the experimental doublet area is 12(2)%). The resulting distribution p(ΔFe1) for (Sc0.95M0.05)MO3 ( is narrow and has symmetrical profile, thus, indicating a uniform nearest surrounding of Fe3+ ions.

The location of small Fe3+ ions at the A site of classical ABO3 perovskites is quite unusual, especially their strong preference to occupy the A site at small doping levels. To the best of our knowledge, (Sc1−M)MO3 compounds are the first example of such behaviour. It should be noted that Fe3+ ions were found by the Mössbauer spectroscopy at the A′ site of A-site ordered perovskites with the general composition of AA′3B4O12, for example, in CaCu3Fe4O12 [34] and CaMn3Mn4O12 [35]. However, Fe3+ ions substitute for Cu2+ or Mn3+ ions—other transition metals—in a special A′ position, whose coordination environment (square-coordinated A′O4) is quite different from a typical coordination of the A site in perovskites (AO8-AO12).

We also prepared solid solutions with the total composition of Sc0.9Co1−CrO2.85 (x = 0.25 and 0.5). The structural parameters of (Sc0.95M0.05)MO3 (M = Co0.75Cr0.25) are given in tables 1 and 2; and the compositional dependence of the lattice parameters is shown on figure 8. Almost linear changes of the lattice parameters suggest that the solid solutions are formed in the whole compositional range. Inverse magnetic susceptibilities of (Sc0.95M0.05)MO3 (M = Co0.75Cr0.25 and Co0.5Cr0.5) are given on figure 9. The χ−1 values were almost field-independent above 70 K for (Sc0.95M0.05)MO3 (M = Co0.75Cr0.25) and above 100 K for (Sc0.95M0.05)MO3 (M = Co0.5Cr0.5) suggesting an impurity-free paramagnetic behaviour. The Curie–Weiss fits of the data corrected for diamagnetic contributions gave μeff = 2.47μB/f.u. and θ = −144 K for (Sc0.95M0.05)MO3 (M = Co0.75Cr0.25) with the expected μeff = 2.20μB/f.u and μeff = 2.99μB/f.u. and θ = −107 K for (Sc0.95M0.05)BO3 (M = Co0.5Cr0.5) with the expected μeff = 2.91μB/f.u. The anomalies at 100 Oe below 70 K in (Sc0.95M0.05)MO3 (M = Co0.75Cr0.25) and below 100 K in (Sc0.95M0.05)MO3 (M = Co0.5Cr0.5) could originate from the onset of short-range or long-range magnetic interactions.

Figure 9.

Uncorrected inverse FCC magnetic susceptibility curves (χ−1 versus T) of (Sc0.95M0.05)MO3 (M = Co1−Cr with x = 0.25 and 0.5) at 100 Oe and 10 kOe. Parameters (μeff and θ) of the Curie–Weiss fits (lines) are given.

Conclusions

We found that ‘ScCoO3’-based perovskites are formed as non-stoichiometric (Sc1−M)MO3 with x = 0.05–0.11 and M = Co, (Co, Fe) and (Co, Cr) under high pressure (6 GPa) and high temperature (1570 K) conditions. There is evidence that (Sc0.95Co0.05)CoO3 has non-magnetic low-spin Co3+ ions at the B site and paramagnetic high-spin Co3+ ions at the A site. In the iron-doped samples (Sc1−M)MO3 with M = (Co, Fe), Fe3+ ions have strong preference to occupy the A site of such perovskites that is quite unusual for perovskites.