The magnetoelectric perovskite Sr3Fe2TeO9: An insight from magnetic measurements and neutron powder diffraction

Article abstract of DOI:10.1016/j.materresbull.2006.07.016

A study of the crystallographic and magnetic structures of the double perovskite Sr₃Fe₂TeO₉ has been carried out on a polycrystalline sample using neutron powder diffraction (NPD) data between 10 and 650 K. An analysis of the NPD patterns at room temperature has shown that this compound crystallises in the tetragonal space group I4/m with a = 5.5614(7) ? and c = 7.867(1) ? and has a partially ordered arrangement of Fe and Te at the B-sites. The compound undergoes an I4/m → Fm-3m improper ferroelectric phase transition near 460 K. A low-temperature ferrimagnetic ordering (below T_N = 260 K) has been followed from the magnetisation measurements and sequential NPD data analysis. In good agreement with magnetic measurements the ferrimagnetic structure with very weak magnetisation is defined by the propagation vector k = (0, 0, 0). In addition to the obtained experimental results on magnetic and electric properties some aspects of magnetoelectricity in this perovskite are also discussed and compared with those of another quaternary oxide Sr₃Fe₂B⁶⁺O₉.

Full text of DOI:10.1016/j.materresbull.2006.07.016

Materials Research Bulletin 42 (2007) 776–789
www.elsevier.com/locate/matresbu
The magnetoelectric perovskite Sr3Fe2TeO9: An insight from
magnetic measurements and neutron powder diffraction
d
S.A. Ivanova,1, P. Nordblad b, S.-G. Eriksson c, R. Tellgren d, , H. Rundlof
*
¨
a Department of Inorganic Materials, Karpov’ Institute of Physical Chemistry,
Ul. Vorontsovo Pole, 10 105064 Moscow K-64, Russia
b Department of Engineering Sciences, University of Uppsala, Uppsala, Sweden
c Department of Inorganic Chemistry, University of Gothenburg, Sweden
d
˚
Department of Materials Chemistry, The Angstrom Laboratory, Uppsala University, Uppsala, Sweden
Received 5 June 2006; accepted 19 July 2006
Available online 18 September 2006
Abstract
A study of the crystallographic and magnetic structures of the double perovskite Sr3Fe2TeO9 has been carried out on a
polycrystalline sample using neutron powder diffraction (NPD) data between 10 and 650 K. An analysis of the NPD patterns at
˚
room temperature has shown that this compound crystallises in the tetragonal space group I4/m with a = 5.5614(7) A and
˚
c = 7.867(1) A and has a partially ordered arrangement of Fe and Te at the B-sites. The compound undergoes an I4/m ! Fm-3m
improper ferroelectric phase transition near 460 K. A low-temperature ferrimagnetic ordering (below TN = 260 K) has been
followed from the magnetisation measurements and sequential NPD data analysis. In good agreement with magnetic measurements
the ferrimagnetic structure with very weak magnetisation is defined by the propagation vector k = (0, 0, 0). In addition to the
obtained experimental results on magnetic and electric properties some aspects of magnetoelectricity in this perovskite are also
discussed and compared with those of another quaternary oxide Sr3Fe2B6+O9.
# 2006 Elsevier Ltd. All rights reserved.
Keywords: A. Ceramics; C. Neutron scattering; D. Crystal structure; D. Magnetic structure
1. Introduction
Magnetoelectric compounds (ME), which simultaneously manifest ferroelectric and ferromagnetic properties, have
gained renewed and ever increasing research interest during the past years [1,2]. Search and preparation of ME
materials would be a milestone for modern electronics and functionalised materials taking an advantage of two
coupled degrees of freedom based on local off-centered polar distortion and electron spin [3,4]. Many efforts have
been devoted to prepare new ME materials and to find ME properties in known compounds. It is very important to
understand under which thermodynamic conditions and chemical environment the coexistence of magnetic and
* Corresponding author. Tel.: +46 184713776; fax: +46 18513548.
E-mail addresses: ivan@cc.nifhi.ac.ru, Sergey.Ivanov@mkem.uu.se (S.A. Ivanov), per.nordblad@angstrom.uu.se (P. Nordblad),
stene@inoc.chalmers.se (S.G. Eriksson), rte@mkem.uu.se (R. Tellgren).
1
Fax: +7 95 975 24 50.
Available online 18 September 2006.
0025-5408/$ – see front matter # 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.materresbull.2006.07.016

S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
777
ferroelectric orderings can take place. One direction in searching for new ME compositions is to combine
ferroelectrics with magnetic materials creating some solid solutions. Another direction is to combine Pb2+ and Bi3+
cations with lone electron pairs and magnetic cations with d-electrons in the same phase.
However, it is clear now [4] that ferroelectric ferromagnets are rare and mostly exhibit rather weak ferromagnetism.
Why and under what circumstances a magnetoelectric coupling will appear is a major open question. This problem has
proved difficult to tackle owing to the absence of magnetoelectrics that show such coupling. Much experimental work
still needs to be done in order to reach a clear understanding of this coexistence, and eventually to search and design
ME compositions. The fundamental physics mechanism behind ME compositions is rich and fascinating to be
explored. Although investigation on ME compounds can be traced back to 1950s [5,6], the renaissance of this class of
materials was just heralded by the discovery of so-called colossal magnetic resistance (CMR) effect in Sr2FeMoO6 [7],
signalling a robust interplay between the magnetic and dielectric properties.
Preliminary studies of double perovskites seem to indicate that the occurrence of magnetoelectric properties is a
common feature in some of them [8]. Recently, BiMnO3 and Bi2NiMnO6 were unravelled to be a supplement to the
scarce ME perovskite family [9,10]. In the past few years, there has been renewed interest in studying the iron-based
ME perovskites which have higher Curie temperatures and relatively large ME effect [11]. However, very often ME
coupling is usually enhanced around magnetic or ferroelectric phase transition temperatures and one strategy for
obtaining an attractive ME material is to bring these transitions to around room temperature. Despite the recent
progress in understanding the physics of the double perovskites the origin of magnetism and ferroic polar distortion is
still controversial. But it is clear now that these physical properties are very sensitive to even small structural changes.
In the course of our recent search for new iron-based perovskite magnetoelectrics A3Fe2BO9 [12–15], the complex
metal oxide Sr3Fe2TeO9 (SFTO) has been studied. The importance of this class of perovskites in many applications has
encouraged us to throw light on the role of different non-magnetic B0-type cations. In the current case we have studied
the crystal structure, magnetic and dielectric properties of Sr3Fe2BO9 with B = Te6+ cation with 4d105s05p0 electron
configuration in an attempt to establish possible correlations between structural features and ME behaviour. Some
reported characteristic features of SFTO are quite author-dependent [16–21]. This complex metal oxide was
synthesised for the first time almost 30 years ago by Harari et al. [16] and then forgotten for many years. It belongs to
the class of perovskite compounds, which simultaneously can exhibit the desired long-range ordering of electrical and
magnetic dipoles [18–20]. Later it was reported to be ferrimagnetic with TN = 240 K [17] and an improper
ferroelectric with TC = 480 K [20]. The crystal structure at room temperature has been reported several times [16–21]
but there are some differences in the structural details reported by different authors and still many points remain
unclear. In the first reports on the preparations of SFTO [16,17] the crystal structure of this perovskite was studied by
X-ray diffraction. The results of the XRD measurements presented in [16–20] led to the conclusion that the symmetry
of the SFTO structure is compatible with s.g. Pm-3m. However, some authors [21] did not agree with this conclusion,
and affirmed that their XRD data show undoubtedly, at least for the sample they studied, that SFTO compound has a
partial ordered perovskite structure with the tetragonal s.g. I4/m. Such confusion is considered to arise from the small
deviation from cubic symmetry, possible different composition and temperature treatment. The ME properties of
SFTO, which not yet are well understood, need an accurate structural studies. The purpose of the present work is to
report new diffraction data (both XRPD and NPD) that confirm some partial ordered arrangement of the Fe and Te
cations over the B-sites of the double perovskite structure. The nuclear and magnetic structures were analysed at
different temperatures to find out if the compound undergoes any phase transition at high temperatures, like many
other perovskites with symmetry I4/m (see, for example [22]). The evolution of both the nuclear and the magnetic
structure is analysed from NPD data, complemented by magnetic measurements. It is well known, that NPD is the
most useful in the case of structural analysis of ME perovskite oxides because of the high scattering power of the
oxygen atoms and the possibility to study the magnetic moments.
2. Experimental
According to the procedure described in [20] a high quality polycrystalline sample of SFTO was prepared by the
conventional solid-state reaction from oxides SrO, Fe2O3 and TeO2 (the SrO powder was obtained from a thermal
decomposition of SrCO3 at 1500 8C under vacuum). They were first weighed in an appropriate metal ratio, mixed
together and ground to a uniform slurry under acetone in an agate mortar protected in a glove-box. The sample was
pressed into pellets at 100 MPa and then synthesised at 930–950 8C for two days in air with several intermediate

778
S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
grindings and pelletings. The maximum heating temperature was set to 950 8C because TeO2 volatilise at higher
temperatures. The samples were heated and cooled at the rate of 5 8C/min. After being allowed to cool to room
temperature, the black powder product was removed from the crucible and characterised by X-ray powder diffraction
(XRPD) using a Bruker D8/Advance instrument with Cu Ka radiation. A step scan mode was employed with a step
width of 2u = 0.028 and a sampling time of 5 s per step. The XRPD pattern showed that the prepared sample was a
single phase without any detectable impurity including SrTeO4.
The lattice parameters were calculated using a silicon powder as an internal standard. The cation composition of the
SFTO sample was determined by X-ray energy-dispersion spectrometry (EDS) using well-characterised standards.
Electron probe microanalyses were performed on 20 different crystallites. The result did not differ significantly; the
metal composition of the studied SFTO sample was close to the expected cation ratio, which was estimated to
Sr3.03(3)Fe1.98(2)Te0.99(2)O9 if the sum of the elements is assumed to be 6. The size of the crystallites in the obtained
powder does not exceed 15 mm.
Medium-resolution NPD data were collected at the Swedish Research reactor R2 in Studsvik with a Huber two-
3
circle diffractometer equipped with an array of 35 He detectors. The intensity from each detector was statistically
analysed and summed. A monochromator system with two copper crystals (2 2 0) in a parallel arrangement was used.
The wavelength was 1.470(1) A and the neutron flux at the sample position was approximately 10 neutrons cmꢀ2 sꢀ1
6
˚
.
Diffractograms were registered at different temperatures between 10 and 650 K. The powder sample of SFTO
(approximately 3 g) was loaded into a thin-walled vanadium container and held in a cryostat. Absorption effects were
later corrected for in the Rietveld refinements using experimentally determined value mR = 0.0703. The step-scan
covered a 2u-range 4–139.928 with a step-length 0.088, each experiment taking approximately 15 h. The diffraction
data were refined by the Rietveld method using the FULLPROF 2000 software [23]. Scattering lengths and form
factors for Sr, Fe, Te and O were taken from the library of the program.
The neutron scattering lengths of Fe and Te are different (bFe = 9.94 fm, bTe = 5.80 fm), thus the chemical
composition of the B-site cations can be observed by NPD with high accuracy. The neutron scattering length of oxygen
is comparable to those of the heavy atoms and NPD provide accurate information concerning its position and
stoichiometry.
The diffraction peaks were described by a pseudo-Voigt function and a Lorentzian contribution to the Gaussian
peak shape was refined. Peak asymmetry correction was made for angles below 358 in 2u. The background was
described by a six parameter polynomial. During the refinements, the two octahedrally coordinated metal cations (Fe
and Te) were allowed to vary their proportions on the two possible metal sites. Each structural model was refined to
convergence and the best result was chosen on the basis of agreement factors and stability of the refinement.
Magnetisation was measured with a Quantum Design SQUID magnetometer at the different applied fields in the
temperature range 5–400 K. Zero field cooled measurements (ZFC) were performed by cooling the samples to 10 K,
applying the field, and measuring the magnetisation as the samples were warmed. Magnetisation versus field curves
was measured at some temperatures between 10 and 300 K.
An E8-2 bridge was used for the measurement of the dielectric properties (the frequency was 1 kHz and 1 MHz).
¨
57
The transmission Fe Mossbauer spectra of CFWO were obtained by means of a standard constant acceleration
spectrometer using a 57Co/Pd source.
3. Results
The dielectric measurements were carried out in the 90–600 K range. The maximum in the dielectric permittivity
was found near 460 K, which is in a good agreement with earlier result [20]. A Fe3+ charge state for the iron
coordinated octahedrally has been clearly confirmed by the isomer shift and hyperfine field determined by Fe57
¨
Mossbauer spectroscopy at room temperature.
Figs. 1 and 2 show results from the magnetisation measurements; in Fig. 1 the thermal evolution of the ZFC and FC
susceptibility (M/H) is plotted under applied fields of 100 Oe and 30 kOe and in Fig. 2 the magnetisation versus field is
shown at some different temperatures between 10 and 300 K. The high temperature (300 K) M versus H curve shows a
linear response indicating a paramagnetic character of the sample. At lower temperatures a faster increase occurs at
lower fields followed by a quasi-linear regime at higher fields. This behaviour indicates that the sample has attained a
weak ferromagnetic moment. From a linear extrapolation to zero field of the curve at 50 K the value amounts to about
0.03 mB per iron atom. It is however not possible to distinguish an anomaly that clearly indicates that and where a

S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
779
Fig. 1. Temperature dependence of the magnetisation of Sr3Fe2TeO9 (H = 10 Oe and 30 kOe).
magnetic transition from paramagnetic to a magnetically ordered state occurs from the M/H versus T curves of Fig. 1.
A hint that ordering occurs is given by an enhanced irreversibility between the ZFC and FC susceptibility below about
260 K in the 100 Oe measurements. Combining this anomaly with the results from the NPD measurements discussed
below, one concludes that ferrimagnetic order, at least on short length scales, occurs below a transition temperature TN
of about 260 K. This ferrimagnetic order is from the NPD data found to sustain to the lowest temperatures of our
experiments. In addition, the low temperature behaviour of the ZFC and FC magnetisation indicates that a re-entrant
spin glass (SG) like transition occurs at about 60 K, this is suggested by the maximum in the low field ZFC
magnetisation and the flattening off of the FC magnetisation at temperatures below 60 K. Also the suppression of the
magnitude of M/H and suppression to lower temperature of the maximum in the ZFC susceptibility and the onset of
Fig. 2. M vs. H measurements at different temperatures for Sr3Fe2TeO9.

780
S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
irreversibility between ZFC and FC curves support this suggestion. This re-entrant spin glass behaviour should be a
consequence of the existence of both ferro- and anti-ferromagnetic interactions [24] and disorder amongst these that
give rise to the frustration necessary to create as spin glass like state [25]. It should also be pointed out that one of the
important characteristics and problem with the re-entrant spin glass state is the apparent coexistence of remaining
magnetic order observed by, e.g. neutron diffraction and the simultaneous occurrence of spin glass characteristic
macroscopic response in magnetisation and susceptibility measurements. It should also be noted that the M versus H
curves in Fig. 2 have been measured after cooling the sample in zero field, applying 50 kOe and measuring the
magnetisation on decreasing the field to zero. This explains why the 10 K curve lies well below the 25 and 50 K curves
and also indicates that a comparably large remanent magnetic moment (of order 0.5 emu/g or 0.03 mB per iron atom) is
achieved after removing a large field after field cooling the sample to 10 K.
The coexistence of SG and ferrimagnetic order in SFTO may lead to an additional amorphous contribution
scattering to the background which was registered in our NPD patterns at low temperatures. As a conclusion, SFTO
exhibits a whole range of magnetic states and starting from the paramagnetic range it has a ferrimagnetic character
below 260 K, which continues down to some temperature point near 60 K where the ‘‘freezing’’ of the spins starts to
grow.
According to the X-ray analysis, the sample was perfectly phase pure (see Fig. 3). The XRD pattern at 295 K shows
the splitting of certain peaks, characteristic of a tetragonal distortion of the cubic perovskite structure. XRD pattern
˚
˚
could be indexed on the basis of an I-centered tetragonal cell with lattice parameters a = 5.5626(5) A, c = 7.8634(7) A.
This indexing is only in slight disagreement with the cell parameters reported in [21] but different from the primitive
cubic cell parameters found in [16–20]. Such confusion is considered to arise from the small tetragonal distortion of
the lattice.
Inspection of the NPD pattern at 650 K provided some evidence for several low intensity superstructure reflections
suggesting a doubling of the perovskite unit cell because of the ordering between Fe and Te cations at the B-sites (see
Fig. 4). Owing to the existence of these weak peaks in this pattern, possible models based on disordered B-cation
distributions were rejected from future consideration for the high temperature phase. The high temperature NPD data
˚
collected at 650 K (see Fig. 5a) was refined in space group Fm-3m with a = 7.9128(4) A (see Table 1). Initially the
fractional occupancies of Fe and Te were defined as 67% Fe and 33% Te in accordance with the results of chemical
analysis. The Fe/Te arrangement over the two crystallographically distinct B-sites was allowed to vary, with the
constraint that the 2:1 composition ratio was maintained. The site occupancy refined by the Rietveld method supports
the chosen structural model and indicates some degree of ordering between Fe and Te at the 4a and 4b sites. The
contrast in scattering lengths for Fe and Te allows the distribution of these cations over these positions to be
determined with a high accuracy. Refinements of the site occupancy factors of the Sr and O atoms did not reveal
Fig. 3. XRD powder pattern for Sr3Fe2TeO9 at room temperature.

S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
781
Fig. 4. Temperature evolution of NPD patterns for Sr3Fe2TeO9. The superstructure and magnetic peaks are marked.
significant changes from full occupancy (within one standard deviation) and was therefore fixed at unity. The final
Fourier difference synthesis calculation using (Fo–Fc) neutron data was without any significant maxima. However, the
same structural model could not account for the NPD data starting with 400 K. When the temperature was decreased
down to 400 K, some diffraction peaks were split (see Fig. 6). It was found that SFTO adopt a tetragonal distorted
perovskite-type structure in the temperature range between 10 and 400 K. It was pointed out in [26] that the most
possible space group for this type of double perovskites with tetragonal symmetry is I4/m. We have tried to fit our data
at 400 K using other possible space groups but it was evident that the space group I4/m provided the best fit to our
powder pattern. The I4/mmm model is less suitable having bad R-factors and negative thermal parameter for several
cations. The final Fourier difference synthesis calculation showed no significant maxima. Table 1 specifies the refined
atomic positions, isotropic displacement parameters, occupancy factors, and agreement factors for SFTO at different
temperatures. The quality of the fit is illustrated in Fig. 5. From the Rietveld analysis, Fe and Te were found to occupy
the (2a) and (2b) sites with different probability. The Fe2/Te2 octahedra are elongated along the c-axis. On the other
hand, the Fe1/Te1 octahedra are compressed along the c-axis (see Table 2). Temperature evolution of Fe1/Te1 and Fe2/
Te2 bonds is illustrated in Fig. 7.
The improper ferroelectric phase transition has been observed previously in Sr3Fe2TeO9 near 480 K [20]. There are
only three structural degrees of freedom, which may be responsible for this phase transformation: (a) cation shifts, (b)
distortions of the oxygen polyhedra coordinating the cations in the A and B-sublattices and (c) tilting of the oxygen
octahedra. If one takes the values of the polyhedra distortions into account the ferroic properties of SFTO may be
connected with the Sr sublattice. At room temperature it was found that the distortion of the SrO12 polyhedra is
significantly larger (19.4 ꢁ 10ꢀ4) than for the Fe1/Te1 and Fe2/Te2 ones (0.17 ꢁ 10ꢀ4 and 0.08 ꢁ 10ꢀ4, respectively).
In the many cases a phase transition with the Fm-3m ! I4/m symmetry change is related to an out-of-phase tilting of
the BO6-octahedra about the c-axis (the so-called a0 a0 cꢀ tilt system [26]) owing to some mismatch between the size
of the Sr cation and the cubooctahedral interstitial space between these octahedral units.
Fig. 8 shows a view of the crystal structure of SFTO. Both Fe and Te cations are located at alternate B-sites as a
result of the difference between the sizes and valence charges of these cations. Fig. 9 represents the evolution of the
lattice parameters for SFTO. Thermal variations of the unit-cell parameters would be related to the contraction of the
lattice due to the magnetic transition at 260 K. The thermal expansion in the paramagnetic phase is significantly larger
than in the ferrimagnetic phase. There is also clear evidence of structural phase transition I4/m ! Fm-3m between 400
and 500 K.
Two major factors will determine the arrangement of different cations at the B-sites in the perovskite structure. The
cations can be either ordered or statistically distributed (disordered), and the controlling factors are the difference in ionic
charge and in ionic size. The ionic radii of Fe3+ and Te6+ are similar (0.645 and 0.560 A, respectively) and therefore we
˚

782
S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
Fig. 5. The observed, calculated and difference plots for the fit to the NPD patterns of Sr3Fe2TeO9 after Rietveld refinement of the nuclear and
magnetic structure at different temperatures: 650 K (a), 295 K (b) and 10 K (c).

S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
783
Table 1
The nuclear and magnetic structure of Sr3Fe2TeO9 at different temperatures
2
B (A )
˚
Atom Position
x
y
z
n
T = 650 K, s.g. Fm-3m, a = 7.9128(4) A, Rp = 3.48, Rwp = 4.63, RB = 4.25, x2 = 1.58
˚
Sr
(8c)
(4a)
(4b)
(24e)
1/4
1/4
0
1/4
0
1.25(2)
0.47(1)
0.41(1)
1.45(3)
1.0
Fe1/Te1
Fe2/Te2
O
0
0.72/0.28(1)
0.61/0.39(1)
1.0
1/2
1/2
0
1/2
0
0.2551(8)
T = 400 K, s.g. I4/m, a = 5.5709(8) A, c = 7.880(1) A, Rp = 3.93, Rwp = 5.06, RB = 4.56, x2 = 1.49
˚
˚
Sr
(4d)
(2a)
(2b)
(8h)
(4e)
0
1/2
1/4
1.12(2)
0.43(2)
0.35(2)
1.26(3)
1.34(3)
1.0
Fe1/Te1
Fe2/Te2
O1
0
0
0
0.72/0.28(1)
0.61/0.39(1)
1.0
0
0
1/2
0.281(1)
0
0.228(1)
0
0
O2
0.251(1)
1.0
T = 295 K, s.g. I4/m, a = 5.5614(7) A, c = 7.867(1) A, Rp = 3.51, Rwp = 4.73, RB = 4.09, x2 = 1.52
˚
˚
Sr
(4d)
(2a)
(2b)
(8h)
(4e)
0
1/2
1/4
0.87(2)
0.38(2)
0.32(2)
1.03(3)
1.18(3)
1.0
Fe1/Te1
Fe2/Te2
O1
0
0
0
0.72/0.28(1)
0.61/0.39(1)
1.0
0
0
1/2
0.282(1)
0
0.228(1)
0
0
O2
0.252(1)
1.0
T = 200 K, s.g. I4/m, a = 5.5549(8) A, c = 7.862(1) A, Rp = 3.27, Rwp = 4.53, RB = 4.13, Rmag = 11.2, x2 = 1.52
˚
˚
Sr
(4d) 0
(2a)
1/2
1/4
0.66(2)
1.0
Fe1/Te1
Fe2/Te2
O1
0
0
0
0.35(2)
0.29(2)
0.86(3)
1.04(3)
0.72/0.28(1)
0.61/0.39(1)
1.0
(2b)
(8h)
(4e)
0
0
1/2
0.281(1)
0
0.228(1)
0
0
O2
0.251(1)
1.0
T = 100 K, space group I4/m, a = 5.5514(8) A, c = 7.857(1) A, Rp = 3.39, Rwp = 4.67, RB = 3.27, Rmag = 10.4, x2 = 1.55
˚
˚
Sr
(4d)
(2a)
(2b)
(8h)
(4e)
0
1/2
1/4
0.52(2)
0.32(2)
0.26(2)
0.74(3)
1.00(3)
1.0
Fe1/Te1
Fe2/Te2
O1
0
0
0
0.72/0.28(1)
0.61/0.39(1)
1.0
0
0
1/2
0.280(1)
0
0.227(1)
0
0
O2
0.251(1)
1.0
T = 10 K, space group I4/m, a = 5.5508(8) A, c = 7.854(1) A, Rp = 3.21, Rwp = 4.70, RB = 3.38, Rmag = 9.8, x2 = 1.64
˚
˚
Sr
(4d)
(2a)
(2b)
(8h)
(4e)
0
1/2
1/4
0.42(2)
0.27(2)
0.24(2)
0.71(3)
0.96(3)
1.0
Fe1/Te1
Fe2/Te2
O1
0
0
0
0.72/0.28(1)
0.61/0.39(1)
1.0
0
0
1/2
0.280(1)
0
0.229(1)
0
0
O2
0.249(1)
1.0
may say that this combination of cations in general favours a disordered arrangement. By using a long-range order
parameter h, introduced in [27] for characterisation of the degree of order in the arrangement in B-sites of double
perovskite lattices, the ionic distribution of Fe3+ and Te6+ cations for SFTO may be represented by the formula:
3þ
6þ
3þ
6þ
Sr½Fe
Te
ꢂ ½Fe
Te
ꢂ O3
ð2þhÞ=6
ð1ꢀhÞ=6
ð2ꢀhÞ=6
ð1þhÞ=6
B1
B2
The Fe3+ cations at the B1 site are surrounded by six B2 sites, which are occupied on average 61% by Fe3+ and 39%
by Te6+ ions on average. A Fe3+ cation on a B2 site is surrounded by six B1 sites, which are occupied on average 72%
by Fe3+ ions and 28% by Te6+. The Fe3+ cations provide much greater internal magnetic field to the neighbouring iron
sites than Te6+ ions, and therefore the magnetic moments are expected to differ for Fe3+ ions on B1 and B2 sites,
respectively. From the results obtained from analysis of our NPD data a long-range parameter h was estimated as 0.16
and the atomic arrangement in our SFTO sample can be represented as follows:
6þ
3þ
6þ
Te
0:194
Sr½Fe30þ:360Te
ꢂ
½Fe
ꢂ O3
0:140
0:306
B1
B2
The value of the magnetic transition temperature can be calculated following the hypothesis [27,28], that a magnetic
ion which has no magnetic neighbours or only one magnetic neighbour does not take part in magnetic ordering. In our
case there are two magnetic sublattices with octahedral sites, and k1 and k2 are the fractions of non-magnetic ions in the

784
S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
Fig. 6. Details of selected portions of the NPD patterns at different temperatures. Splitting of some reflections at 400 K indicates the cubic to
tetragonal structure change.
sublattices B1 and B2, respectively. E(k1) (or E(k2) are the probabilities that a given ion of the sublattice B2 (or B1) has not
more than one magnetic neighbour in the sublattice B1(B2).
The number of magnetically active ions and the number of effective magnetic interactions per molecular unit are
given as:
N ¼ 0:5ð1 ꢀ k1Þ½1 ꢀ Eðk2Þꢂ þ 0:5ð1 ꢀ k2Þ½1 ꢀ Eðk1Þꢂ
n ¼ 6ð1 ꢀ k1Þ½1 ꢀ Eðk2Þꢂð1 ꢀ k2Þ½1 ꢀ Eðk1Þꢂ
where 6 is the coordination number for a cation at the B-site of the perovskite lattice. The magnetic transition
temperature of our SFTO composition TN(k1, k2) is related to the standard Neel temperature TN(0, 0) by
1
n
TNðk1; k2Þ ¼
ꢁ
ꢁ TNð0; 0Þ
6
N
where we take TN(0, 0) = 643 K is the Neel temperature of perovskite with only Fe3+ in the B-site (BiFeO3) in
agreement with [27]. The comparison of the calculated (313(25) K) and observed (260 K) values of TCM computed for
our SFTO sample, on the basis of the iron content obtained from our NPD data, shows relatively good agreement.
Following the results of dielectric and magnetic measurements below 260 K SFTO exists in antiferroelectric and
ferrimagnetic states. As the SFTO sample show ferrimagnetic order the nuclear and magnetic periodicity is the same,
i.e., the two contributions occur at the same scattering angle. In the simplest case of ferrimagnets there are two
sublattice systems. The magnetic moments at each sublattice differ in magnitude, and they are oppositely oriented. In
our NPD spectra registered below 260 K the set of magnetic reflections was observed. Because the selected perovskite
Table 2
˚
Selected interatomic distances (A) for Sr3Fe2TeO9 at different temperatures
Bond
650 K
295 K
200 K
10 K
2.635(2) ꢁ 4
2.781(2) ꢁ 4
2.935(3) ꢁ 4
2.636(2) ꢁ 4
2.778(2) ꢁ 4
2.930(3) ꢁ 4
2.637(2) ꢁ 4
2.775(2) ꢁ 4
2.923(3) ꢁ 4
˚
Sr–O (A)
2.799(2) ꢁ 12
2.018(3) ꢁ 6
1.939(3) ꢁ 6
˚
Fe1/Te1–O (A)
2.017(3) ꢁ 4
2.011(3) ꢁ 4
2.005(3) ꢁ 4
1.981(3) ꢁ 2
1.975(3) ꢁ 2
1.972(3) ꢁ 2
˚
Fe2/Te2–O (A)
1.954(3) ꢁ 2
1.939(3) ꢁ 4
1.959(3) ꢁ 2
1.941(3) ꢁ 4
1.956(3) ꢁ 2
1.938(3) ꢁ 4

S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
785
Fig. 7. Temperature behaviour of Fe/Te–O bond lengths for Sr3Fe2TeO9.
unit cell is already doubled by partial ordering of Fe/Te cations, all magnetic reflections are coincident with nuclear
reflections. Primarily attempts were made to refine the NPD patterns at these temperatures using a model assumed in
accordance with space group I4/m. These calculations resulted in non-acceptable high R-factors: in particular, big
differences were found between observed and calculated intensities in the low angle 2u region. The agreement between
the observed and calculated intensities did not improve due to shifting of atoms from their idealized positions. The
higher angle data covering 708 < 2u < 139.928 were refined using the initial model in s.g I4/m resulting in satisfactory
R-factors. It also gave quite reasonable structural parameters, including the isotropic displacement parameters. These
results indicate that the extra intensities decrease rapidly with increasing 2u-angle and are most likely due to magnetic
ordering of the iron cations. The additional scattering was found to disappear above 260 K. To verify that the observed
Fig. 8. A polyhedral view of the tetragonal crystal structure of Sr3Fe2TeO9 at room temperature.

786
S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
Fig. 9. The temperature dependence of the lattice parameters for Sr3Fe2TeO9.
extra intensities indeed were magnetic to its origin, X-ray diffraction measurements were carried out at low
temperatures.
Worthwhile to notice is that below 260 K there are no further changes in the NPD patterns except a progressive
increase in intensity of weak magnetic reflections. This indicates that the magnetic structure remains stable across the
temperature range 10–260 K. The NPD patterns of SFTO below 260 K were refined with both a nuclear and magnetic
structure simultaneously. The parameters varied in the refinement of the magnetic structure were the magnitudes and
the canting angles of the moments.
A ferrimagnetic structure was modeled with magnetic moments at two possible Fe positions. Magnetic moments
were refined both along the a- and the c-axis. A significantly better fit to the data was obtained when refined in the c-
direction. The ordered moment was constrained to lie along the c-axis, but a small component (in the frame of one
standard deviation) along the a- or b-axis is possibly given within the resolution of our data. The best fit was obtained
for a G-type magnetic ordering. Within this model a given moment is coupled antiparallel to its six nearest neighbours
Fig. 10. A sketch of the magnetic structure for Sr3Fe2TeO9 at 10K. Diamagnetic ions are omitted.

S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
787
Fig. 11. The temperature dependence of the magnetic moments of the Fe1 and Fe2 sub-lattices in Sr3Fe2TeO9.
(see Fig. 10). Temperature evolution of the refined magnetic moments of B1 and B2 sites are shown in Fig. 11. The
observed moments are related to the antisite disorder and could be evaluated for the simplest ferrimagnetic
arrangements between Fe1 and Fe2 sublattices which produce a net magnetisation Ms = mBFe1 ꢀ mBFe2
.
4. Discussion
The refined atomic coordinates and bond distances for SFTO (see Tables 1 and 2) confirm the basic structural
features of the proposed tetragonal perovskite structure. Fe and Te were found to occupy the (2a) and (2b) sites with
different probability. The average Fe1–O bond length is very similar to that found in other oxides with Fe3+ cations [29]
and close to Fe3+–O length calculated from Shannon0s ionic radii [30] for high spin electron configuration in
octahedral coordination. The Fe–O–Fe superexchange pathways deviates from linear (Fe–O–Fe angle changes from
1688 (10 K) up to 1728 (400 K)). The observed magnetic moments of Fe are smaller than what is expected for Fe3+ (5.0
mB). In order to get some insight into the cation distribution, we carried out a bond-valence calculation by means of the
Brown model [31], which gives a relationship between the formal valence of a bond and the corresponding bond
Fig. 12. Details of composition dependence of NPD patterns at 10 K for selected samples in Sr3Fe2W1ꢀxTexO9 solid solutions. The decrease of
integrated intensities for some magnetic reflections with Te doping is shown.

788
S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
Fig. 13. Dependence of the magnetic phase transition temperature TN from ionic radii for the series Sr3Fe2B6+O9. The uncertainty of TN was
estimated for all published data.
lengths. In non-distorted structures, the bond-valence sum rule states that the valence of the cation (Vi) is equal to the
sum of the bond-valences ðvi jÞ around this cation. From individual cation–anion distances at room temperature the
bond-valence sums were calculated. Fe1 and Fe2 cations exhibit valences 3.13(1) and 3.35(1), slightly different from
¨
those expected in this compound, of +3. Following the results of Mossbauer measurements we can conclude that the Fe
cations have formally +3 valence in both oxygen environments.
The important question concerns the striking difference of NPD patterns of Sr3Fe2TeO9 (h = 0.16) and Sr3Fe2WO9
(h = 0.33) which was studied earlier [12] (see Fig. 12). Strong decreasing of magnetic scattering could arise from the
difference of electronic states in addition to the ionic radius between W6+ (5d06s0) and Te6+(4d105s05p0). The same
degradation of magnetic properties of Te compound in comparison with W analogue was previously reported for
perovskites Sr2CuBO6 [36] and Sr2NiBO6 [37] (B = W, Te).
It can be noted that other oxides Sr3Fe2B6+O9 (B6+ = Mo, W, Te, Re, U) with perovskite structure also show the
magnetic properties [12,32–35] (see Fig. 13). It is important to understand the possible influence of the B-site cations
on TN remembering that it is strongly related with the value of the B3+–O–B6+ angle and with lattice distortions. In
order to establish correlation between the structural and magnetic properties the TN values are shown in Fig. 13 as a
function of the ionic radius. It should be emphasised as a tendency that TN increases with an increasing size of the
different B6+ cations. At the same time the influence of the degree of crystallographic order/disorder in Fe-sites and/or
the oxygen deficiency on TN still remains to be solved because the available physical and structural data are very
restricted. At this stage, one cannot draw definite conclusions about this dependence and additional experimental data
should be obtained. It is well known that the magnetisation of double perovskites decreases systematically with the
anti-site defect concentration [38]. The actual degree of Fe/Te order depends on synthesis conditions and as a rule an
increased order may be obtained by different treatment conditions [38,39]. More investigative work is needed to
resolve this issue in the case of Sr3Fe2B6+O9 oxides. Studies of heat treatment effects on the structural and magnetic
properties of SFTO are in progress and we shall return to this theme in our future publication.
5. Concluding remarks
We have revised some experiments that may provide key ingredients for clear understanding of the magnetoelectric
properties of SFTO. The most striking feature of this compound is the coexistence of ferrimagnetic and
antiferroelectric types of ordering below TN = 260 K. Our NPD investigation has shown that the previous description
of Sr3Fe2TeO9, as being a simple cubic perovskite with s.g. Pm-3m at room temperature [16–20], was wrong. The

S.A. Ivanov et al. / Materials Research Bulletin 42 (2007) 776–789
789
symmetry was overestimated. The ideal elpasolite-type structure with s.g. Fm-3m was found for SFTO only above
TC = 460 K. The phase-pure sample of the perovskite Sr3Fe2TeO9 was synthesised, magnetically and structurally
characterised applying the Rietveld analysis of NPD data at different temperatures. This compound shows one
structural phase transition Fm-3m ! I4/m, and the tetragonal superstructure arises from the combination of partial
cation ordering and small tilting of the oxygen octahedral (about 68). The Fe and Te cations are partially ordered into
B-sites (2a) and (2b). The precise metal–oxygen distances derived from NPD data made it possible to calculate the
valences of the cations and the distortion of their polyhedra. These results suggest the electronic configuration to be
Fe3+–Te6+. The proposed model for the ferrimagnetic structure (below TN) can be modelled with magnetic moments at
two Fe positions (2a) and (2b) aligned along the [0 0 1] direction. Structural and magnetic features of magnetoelectric
Sr3Fe2TeO9 are considered and compared with those of other quaternary complex perovskite oxides.
Acknowledgements
Financial support of this research from the Royal Swedish Academy of Sciences and the Swedish Strategic
Foundation (SSF) within the frames of the program ‘‘Complex oxides for advanced applications’’ is gratefully
acknowledged.
References
[1] W. Prellier, M.P. Singh, P. Murugavel, J. Phys.: Condens. Matter 117 (2005) R803.
[2] M. Fiebig, J. Phys. D: Appl. Phys. 38 (2005) R123.
[3] N.A. Hill, A. Filippetti, J. Magn. Magn. Mat. 242 (2002) 976.
[4] N.A. Hill, Ann. Rev. Mater. 32 (2002) 1.
[5] G.A. Smolensky, I.E. Chupis, Sov. Phys.-Usp. 25 (1982) 475.
[6] Yu.N. Venevtsev, V.V. Gagulin, Ferroelectrics 162 (1994) 23.
[7] K.I. Kobayashi, T. Kimura, H. Sawada, K. Terakura, Y. Tokura, Nature 395 (1998) 677.
[8] A.J. Freeman, H. Schmid (Eds.), Magnetoelectric Interaction Phenomema in Crystals, Gordon and Breach, NY, 1975, p. 228.
[9] N.A. Hill, J. Phys. Chem. B104 (2000) 6694.
[10] M. Azuma, K. Tanaka, T. Saito, S. Ishiwata, Y. Shimakawa, M. Takano, J. Am. Chem. Soc. 127 (2005) 8889.
[11] C.H. Yang, T.Y. Koo, Y.H. Jeong, Solid State Commun. 134 (2005) 299.
[12] S.A. Ivanov, S.G. Eriksson, R. Tellgren, H. Rundlof, Mater. Res. Bull. 36 (2001) 2585.
[13] S.A. Ivanov, S.G. Eriksson, R. Tellgren, H. Rundlof, Mater. Res. Bull. 39 (2004) 615.
[15] S.A. Ivanov, S.G. Eriksson, R. Tellgren, H. Rundlof, Mater. Res. Bull. 39 (2004) 2317.
[16] D. Harari, J.C. Bernier, P. Poix, J. Solid State Chem. 5 (1972) 382.
[17] D. Harari, Thesis, Orsay, France, 1975.
[18] I.D. Zhitomirskii, V.V. Gagulin, N.V. Fadeeva, S.P. Soloviev, O.I. Chechernikova, Yu.N. Venevtsev, Fiz. Tverdogo Tela (USSR) 27 (1985)
2863.
[19] I.D. Zhitomirskii, O.I. Chechernikova, V.V. Ivanova, Yu.N. Venevtsev, Izv. Akad. Nauk SSSR Neorg. Mater. 19 (1983) 943.
[20] I.D. Zhitomirskii, Thesis, Moscow State University, Moscow, Russia, 1985.
[21] M.C. Viola, M.S. Augsburger, R.M. Pinacca, J.C. Pedregosa, R.E. Carbonio, R.C. Merkader, J. Solid State Chem. 175 (2003) 252.
[22] R. Mitchell, Perovskites: Modern and Ancient, Almaz Press, Canada, 2003.
[23] J. Rodriguez-Carvajal, J. Phys. 192 (1993) 55.
[24] J.A. Mydosh, Spin Glasses: An Experimental Introduction, Taylor and Francis, London, Washington, DC, 1993.
[25] A.P. Young (Ed.), Spin Glasses and Random Fields, World Scientific, Singapore, 1997.
[26] P.M. Woodward, Acta Cryst. B53 (1997) 32.
[27] K. Uchino, S. Nomura, Ferroelectrics 17 (1978) 505.
[28] M.A. Gileo, J. Phys. Chem. Solids 13 (1960) 33.
[29] Yu.N. Venevtsev, E.D. Politova, S.A. Ivanov, Ferro- and Antiferroelectrics of Barium Titanate Family, Chemistry Publishers, Moscow, 1985.
[30] R.D. Shannon, Acta Crystallogr., Sect. A 32 (1976) 751.
[31] N.E. Brese, M. O’Keefe, Acta Crystallogr., Sect. B 47 (1991) 192.
[32] I.D. Zhitomirskii, N.V. Fadeeva, V.V. Gagulin, S.P. Soloviev, O.I. Chechernikova, Yu.N. Venevtsev, Fiz. Tverdogo Tela (USSR) 26 (1984) 3696.
[33] M.C. Viola, J.A. Alonso, J.C. Pedregosa, R.E. Carbonio, Eur. J. Inorg. Chem. (2005) 1559.
[34] R.M. Pinacca, M.C. Viola, J.A. Alonso, J.C. Pedregosa, R.E. Carbonio, J. Mater. Chem. 15 (2005) 4648.
[35] L.A. Baum, S.J. Stewart, R.C. Mercander, J.M. Greneche, Hyperfine Interact. 156/157 (2004) 157.
[36] D. Iwanaga, Y. Inaguma, M. Itoh, J. Solid State Chem. 147 (1999) 291.
[37] D. Iwanaga, Y. Inaguma, M. Itoh, Mater. Res. Bull. 35 (2000) 449.
[38] D. Sanchez, J.A. Alonso, M. Garcia-Hernandez, M.J. Martinez-Lope, J.L. Martinez, Phys. Rev. B65 (2002) 104426.
[39] P.M. Woodward, R.D. Hoffman, A.W. Sleight, J. Mater. Res. 9 (1994) 2118.