Nonstoichiometry, point defects and magnetic properties in Sr 2FeMoO6-δ double perovskites

Article abstract of DOI:10.1016/j.jssc.2011.10.043

The phase stability, nonstoichiometry and point defect chemistry of polycrystalline Sr₂FeMoO_6-δ (SFMO) was studied by thermogravimety at 1000, 1100, and 1200 °C. Single-phase SFMO exists between -10.2≤log pO₂≤-13.7 at 1200 °C. At lower oxygen partial pressure a mass loss signals reductive decomposition. At higher pO₂ a mass gain indicates oxidative decomposition into SrMoO₄ and SrFeO_3-x. The nonstoichiometry δ at 1000, 1100, and 1200 °C was determined as function of pO₂. SFMO is almost stoichiometric at the upper phase boundary (e.g. δ=0.006 at 1200 °C and log pO ₂=-10.2) and becomes more defective with decreasing oxygen partial pressure (e.g. δ=0.085 at 1200 °C and log pO₂=-13.5). Oxygen vacancies are shown to represent majority defects. From the temperature dependence of the oxygen vacancy concentration the defect formation enthalpy was estimated (ΔH_OV=253±8 kJ/mol). Samples of different nonstoichiometry δ were prepared by quenching from 1200 °C at various pO₂. An increase of the unit cell volume with increasing defect concentration δ was found. The saturation magnetization is reduced with increasing nonstoichiometry δ. This demonstrates that in addition to Fe/Mo site disorder, oxygen nonstoichiometry is another source of reduced magnetization values.

Full text of DOI:10.1016/j.jssc.2011.10.043

Journal of Solid State Chemistry 185 (2012) 76–81
Contents lists available at SciVerse ScienceDirect
Journal of Solid State Chemistry
journal homepage: www.elsevier.com/locate/jssc
Nonstoichiometry, point defects and magnetic properties in Sr2FeMoO6ꢀ
double perovskites
d
n
R. Kircheisen a,b, J. To¨pfer a,
a University of Applied Sciences Jena, Dept. SciTec, Carl-Zeiss-Promenade 2, 07745 Jena, Germany
b Fraunhofer Institute of Ceramic Technology and Systems (IKTS), Faradaystrasse 1, 07629 Hermsdorf, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
The phase stability, nonstoichiometry and point defect chemistry of polycrystalline Sr2FeMoO6ꢀ
d
Received 6 May 2011
Received in revised form
17 October 2011
(SFMO) was studied by thermogravimety at 1000, 1100, and 1200 1C. Single-phase SFMO exists
between ꢀ10.2rlog pO2r ꢀ13.7 at 1200 1C. At lower oxygen partial pressure a mass loss signals
reductive decomposition. At higher pO2 a mass gain indicates oxidative decomposition into SrMoO4 and
Accepted 29 October 2011
Available online 7 November 2011
SrFeO3ꢀx. The nonstoichiometry
is almost stoichiometric at the upper phase boundary (e.g.
and becomes more defective with decreasing oxygen partial pressure (e.g.
d
at 1000, 1100, and 1200 1C was determined as function of pO2. SFMO
d
¼0.006 at 1200 1C and log pO2¼ ꢀ10.2)
Keywords:
d
¼0.085 at 1200 1C and
Double perovskites
Nonstoichiometry
Magnetic properties
log pO2¼ ꢀ13.5). Oxygen vacancies are shown to represent majority defects. From the temperature
dependence of the oxygen vacancy concentration the defect formation enthalpy was estimated
(
D
HOV¼25378 kJ/mol). Samples of different nonstoichiometry
1200 1C at various pO2. An increase of the unit cell volume with increasing defect concentration
found. The saturation magnetization is reduced with increasing nonstoichiometry . This demonstrates
d
were prepared by quenching from
d
was
d
that in addition to Fe/Mo site disorder, oxygen nonstoichiometry is another source of reduced
magnetization values.
& 2011 Elsevier Inc. All rights reserved.
1. Introduction
a ferrimagnetic arrangement with a saturation magnetization of
4 mB [1,2]. Mo¨ssbauer [5] and NMR [6] spectroscopy have revealed
Sr2FeMoO6ꢀ (SFMO) is a spin-polarized half-metal with a
a partial transfer of electron density resulting in a mixed valence
situation with high-spin Fe2.5þ (3d5.5) and Mo5.5þ (4d0.5) giving
rise to the same value of Ms. The magnetization was shown to be
sensitive to the degree of cation ordering. A correlation between
preparation temperature, cationic disorder and magnetization
was shown by Balcells et al. [2]. The kinetics of the ordering
process was monitored by Shimada et al. [7], and maximum
ordering could be achieved at prolonged sintering at around
1150 1C. Samples with a very high degree of cation order were
prepared through a complex gel route and sintering in silica
ampoules with a Fe getter; this synthesis procedure results in a
maximum Ms of 3.96 mB [8]. It was demonstrated, that the
magnitude of the LFMR also depends on the degree of ASD [9].
Disordered intergrowths were detected as another source of
reduced Ms and MR [10]. Moreover, it was shown that oxygena-
tion/reduction treatments can reversibly change the oxygen
stoichiometry and grain boundary composition and, hence, MR
properties [11]. Another model has been put forward recently
that allows to interpret the observed variation of the electronic
and magnetic properties with the degree of cation ordering as a
d
Curie temperature of about 410 K which exhibits large values of
low-field magneto-resistance (LFMR) at room temperature [1].
These properties qualify SFMO as a promising candidate material
for spintronics and sensor applications. The crystal structure of
SFMO is that of a B-site ordered double perovskite A2B0B00O6ꢀ
d
with an ordered arrangement of Fe and Mo on the perovskite
B-sites. However, in reality ideal ordering of B and B0 cations is not
reached and some degree of antisite disorder (ASD), i.e. Fe on Mo
sites and vice versa, is observed [2]. At room temperature SFMO
crystallizes in a tetragonal unit cell; the space group I4/mmm was
suggested by many authors, e.g. [2]. At around 400 K a phase
¯
transition to a cubic Fm3m structure occurs [3]. However, the
space group of the room temperature tetragonal structure has
been debated for quite a while. Neutron diffraction measure-
ments seem to indicate that I4/m most likely is the correct space
group [3,4]. The magnetism of SFMO is discussed as being due to
antiferromagnetic interactions between the localized Fe3þ ions
(t23g eg2, S¼5/2) and itinerant 4d1 Mo5þ ions (e1g, S¼½) resulting in
signature of
a polaronic–itinerant electronic transition with
n Corresponding author. Fax: þ49 3641 205451.
E-mail address: joerg.toepfer@fh-jena.de (J. To¨pfer).
well-ordered itinerant regions growing in a partially ordered
matrix phase [12].
0022-4596/$ - see front matter & 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jssc.2011.10.043

R. Kircheisen, J. To¨pfer / Journal of Solid State Chemistry 185 (2012) 76–81
77
Sr2FeMoO6ꢀ is commonly prepared through calcinations of
equation
d
the starting oxides in air and sintering under reducing gas atmo-
Dd
¼
d
ꢀ
d0 ¼ kpn
ꢀ
d0
ð1Þ
O2
sphere (Ar/H2 mixtures). However, precise knowledge about the
boundaries of the SFMO phase field is limited. Many studies
reported in the literature are performed on samples containing
different amounts of impurity phases. Moreover, precise experi-
mental sintering conditions, i.e. the oxygen partial pressure, are
not reported in most of the papers. Nakamura et al. [13] have
shown in an early study that SFMO exists at 1200 1C in a limited
stability range only at low pO2. The phase relations of the system
SrO–Fe3O4–MoO3 at 1200 1C in Ar/1%H2 were investigated [14].
with the nonstoichiometry at the reference oxygen partial pres-
sure 0. The parameters 0, k, and n were determined and values
of were calculated.
Samples with different values of the nonstoichiometry
d
d
d
d
were
prepared by annealing the sintered pellets in the thermobalance.
The pellets were held at 1200 1C and the reference oxygen partial
pressure until a constant mass indicates that equilibrium is
established (typically 10 h). Then, the oxygen partial pressure
was changed to another value corresponding to the desired
nonstoichiometry
24 h for complete equlibriation. Finally, the samples were rapidly
cooled in the balance with a rate of 35 K/min.
Information on the oxygen stoichiometry
scarce. Rager et al. [14] report -values determined by a hot
extraction method; for SFMO sintered at 1200 1C in Ar/1%H2 they
found
d of Sr2FeMoO6ꢀ are
d
d
d and the sample was held there for about
d
¼0.2. Yamamoto et al. measured d¼0.03 using a coulo-
metric titration technique for a sample synthesized at 1150 1C in
an evacuated silica ampoule with Fe getter and pO2¼2.6 ꢁ
10ꢀ13 atm [15]. Chang et al. [16] report on a large nonstoichio-
Powder X-ray diffraction measurements (XRD) were performed
with a Siemens D5000 using CuK
0.021; 10–751 2 ). Lattice parameters and antisite concentrations
a radiation (step time 8 s; step size
y
metry of
d¼0.16 for sintered samples which were annealed at
were refined using the TOPAS R software package (Bruker AXS,
Karlsruhe, Germany). The antisite defect concentration (AS per
formula unit) was determined by refining the occupation of Fe
and Mo positions; the concentration of antisites per f.u. corresponds
to the number of misplaced Fe or Mo ions (i.e. zero for perfect order
and 0.5 for complete disorder).
Magnetic properties were measured with a Quantum Design
SQUID magnetometer. Hysteresis loops M(H) were measured at
5 K. The magnetization at the maximum field of 50 kOe was taken
as saturation magnetization Ms.
850 1C in a N2/5%H2 atmosphere. Nonstoichiometry data of SFMO
are conflicting. To the best of our knowledge, no systematic study
on the variation of
d in Sr2FeMoO6ꢀ with T or pO2 has been
d
performed yet. On the other hand, it is expected that the
nonstoichiometry has a distinct effect on the magnetic and
magneto-resistive properties as well. It was pointed out, that
oxygen vacancies might reduce the magnetization significantly
[17]. In a very recent theoretical study using density functional
theory calculations it was shown that cation antisites and oxygen
vacancies are the dominating phenomena that reduce the mag-
netization [18].
We have already demonstrated in a preliminary study that the
3. Results and discussion
nonstoichiometry
and can be modeled based on oxygen vacancies as majority point
defects [19]. It was shown that samples with small display large
d of SFMO at 1200 1C varies as a function of pO2
3.1. Phase stability
d
saturation magnetizations, whereas a larger nonstoichiometry
results in reduced Ms and large magneto-resistance at room
temperature. In this paper we present results of the measurement
Sr2FeMoO6ꢀ is stable at low oxygen partial pressures only
d
and becomes unstable under more oxidizing or reducing condi-
tions, respectively. We have determined the location of the upper
and lower phase boundaries of SFMO at 1200 1C by thermogravi-
metry (TG) and XRD. The TG analysis for the characterization of
SFMO at 1200 1C in the vicinity of the upper phase boundary is
shown in Fig. 1. The sample was held at a reference oxygen partial
pressure (log pO2¼ ꢀ11.8) until equilibrium was reached as
signaled by constant mass. Then the pO2 was increased stepwise
(and decrease back to the reference pO2) as demonstrated in
of the nonstoichiometry
d of Sr2FeMoO6ꢀ at 1000, 1100, and
d
1200 1C. Structural and magnetic properties of samples with
different nonstoichiometries
d are discussed.
2. Experimental
SFMO samples were prepared with a standard mixed-oxide
route starting from SrCO3, MoO3, and Fe2O3. The raw materials
were homogenized in isopropanol. After calcination at 900 1C in
air, the powders were milled in isopropanol. After compaction the
pellets were sintered at 1200 1C for 20 h in an Ar/2%H2 atmo-
sphere. Measurements of the nonstoichiometry
d were performed
in a Setaram thermobalance TGA92 at 1000, 11000, and 1200 1C
and different partial pressures of oxygen. Gas atmospheres were
established with N2/CO/CO2 mixtures and the pO2 was measured
with zirconia oxygen sensors at the inlet/outlet positions of the
gas mixture into/out of the balance system. In a typical run, the
sample was held at the reference oxygen partial pressure (e.g.
log pO2¼ ꢀ11.84 at 1200 1C) until equilibrium was reached as
signaled by a constant mass. Then the pO2 was changed to
another value and the corresponding mass change recorded.
After equilibrium was reached (typically several hours), the
pO2 was switched back to the reference pO2 to check the
process reversibility. The measured mass changes were trans-
formed into changes of nonstoichiometry, Dd, using the relation
Fig. 1. Variation of the sample mass of Sr2FeMoO6ꢀ with time at 1200 1C for
d
Dd¼(
as the molar mass of SFMO and oxygen. Absolute values of the
nonstoichiometry were obtained by fitting the data to the
DmMSFMO)/(m0Mo) with m0 as the initial sample mass and M
three consecutive oxidation steps induced by jumps of the oxygen partial
pressure; R as reference pO2, the oxygen partial pressures 1 and 2 are within
the SFMO stability field, whereas 3 is above the upper phase boundary.
d

78
R. Kircheisen, J. To¨pfer / Journal of Solid State Chemistry 185 (2012) 76–81
Fig. 3. Variation of the sample mass of Sr2FeMoO6ꢀ with time at 1200 1C for
d
three consecutive reduction steps induced by jumps of the oxygen partial
Fig. 2. Part of the X-ray diffraction patterns of SFMO samples quenched from
1200 1C and different oxygen partial pressures: log pO2¼ ꢀ11.8 as reference pO2
with single phase SFMO (bottom), log pO2¼ ꢀ10.1 (middle), and log pO2¼ ꢀ6.0
(top) above the upper phase boundary with formation of SrMoO4 (SM).
pressure; R as reference pO2, the oxygen partial pressures 1 and 2 are within
the SFMO stability field, whereas 3 is below the lower phase boundary.
Fig. 1. As long as the sample is at pO2 within the SFMO field, the
sample mass became constant after several hours indicating that
the corresponding nonstoichiometry
d has been established
(shown here for log pO2¼ ꢀ10.29 and ꢀ11.22, respectively, in
Fig. 1). In contrast to that, at slightly higher oxygen partial
pressure (log pO2¼ ꢀ10.19) the sample mass continuously
increased for a holding time of more than 20 h and did not
approach equilibrium. This indicates that the sample is outside
the SFMO stability field under these conditions and that decom-
position of SFMO proceeds. Consequently, the location of the
upper phase boundary at 1200 1C is at log pO2¼ ꢀ10.20. At larger
oxygen partial pressures SFMO is oxidized according to:
Sr2FeMoO6 þ1=4O2$SrMoO4 þSrFeO2:5
ð2Þ
The location of the upper phase boundary was confirmed by
XRD measurements on quenched samples (Fig. 2). For example, a
sample quenched from log pO2¼ ꢀ11.8 proved single-phase to
XRD, whereas a sample quenched from log pO2¼ ꢀ10.1 exhibits
weak reflections of SrMoO4, signaling the beginning of SFMO
decomposition. In Fig. 2 the (112) main peak of SrMoO4 at
Fig. 4. Part of the X-ray diffraction patterns of SFMO samples quenched from
1200 1C and different oxygen partial pressures: log pO2¼ ꢀ11.8 as reference pO2
with single phase SFMO (top), and log pO2¼ ꢀ14.15 below the lower phase
boundary with formation of Sr3FeMoO7 and Fe (bottom).
2
y¼27.61 appears with low intensity in this sample. Another
sample annealed at more oxidizing conditions of log pO2¼ ꢀ6
clearly exhibits reflections of SrMoO4 and SrFeO2.5 (not shown in
Fig. 2).
of the SFMO nonstoichiometry (see next chapter) are clearly
within the SFMO phase fields at both temperatures too.
The location of the upper phase boundary at 1200 1C is in fair
agreement with the value of log pO2¼ ꢀ9.8 as reported by
Nakamura et al. [13] and slightly higher compared to the upper
phase stability limit at log pO2¼ ꢀ10.65 measured by Sharma
et al. by coulometric titration [20].
The location of the lower phase boundary at 1200 1C was
determined by similar TG experiments (Fig. 3). It is shown that for
log pO2¼ ꢀ13.31 and ꢀ13.51 equilibrium within the SFMO phase
field is reached. At log pO2¼ ꢀ13.73 a much larger mass loss
indicates that the sample is reduced and equilibrium is not
reached after 20 h. Therefore the lower phase stability limit of
SFMO is in between log pO2¼ ꢀ13.51 and ꢀ13.73. This is in good
agreement with the only other value of log pO2¼ ꢀ13.5 from the
literature [13]. This is verified by XRD of a sample quenched from
T¼1200 1C and log pO2¼ ꢀ14.15 (Fig. 4). The diffractogram of
that sample clearly shows partial decomposition of SFMO and
formation of Sr3FeMoO7 and Fe.
3.2. Nonstoichiometry and point defects
The variation of the nonstoichiometry
and 1200 1C is shown in Fig. 5. At 1200 1C SFMO exhibits a small
nonstoichiometry of
d vs. pO2 at 1000, 1100,
d
¼0.006 close to the upper phase boundary
at log pO2¼ ꢀ10.2. At 1100 and 1000 1C the nonstoichiometry at
higher oxygen partial pressures is smaller, i.e. SFMO is almost
stoichiometric in the vicinity of the upper phase boundary. At
lower oxygen partial pressures larger deviations from stoichio-
metry are observed; e.g.
d¼0.085 at 1200 1C close to lower phase
boundary at log pO2¼ ꢀ13.5. The nonstoichiometry values of
SFMO reported here agree well with the singular value of
d
¼0.03 at 1150 1C and log pO2¼ ꢀ12.58 reported by Yamamoto
et al. [15]. However, the large nonstoichiometry of
¼0.2
reported for a sample prepared at 1200 1C in Ar/1%H2 [14] is too
d
The phase boundaries at 1100 and 1000 1C were not precisely
determined. However, the pO2 values used for the measurement
large and outside the range of
present study.
d-values determined in the

R. Kircheisen, J. To¨pfer / Journal of Solid State Chemistry 185 (2012) 76–81
79
The nonstoichiometry
dated as oxygen vacancies:
d
in Sr2FeMoO6ꢀ can be accommo-
the nonstoichiometry data within the SFMO stability range
according to Eq. (1) results in k¼ ꢀ(1.470.4) ꢁ 10ꢀ6 and
n¼ ꢀ0.3570.01 for T¼1200 1C, k¼ ꢀ(2.071.3) ꢁ 10ꢀ7 and
n¼ ꢀ0.3770.02 for T¼1100 1C, and k¼ ꢀ(6.571.3) ꢁ 10ꢀ8
and n¼ ꢀ0.3570.01 for T¼1000 1C, respectively. The estimated
values of the exponent n significantly deviate from the expected
value of n¼ ꢀ0.5. This demonstrates that the simple picture of the
formation of isolated oxygen vacancies as majority point defects
is a crude approximation only. This discrepancy might be due
to interactions at lower pO2 between oxygen vacancies and
surrounding cations to form defect associates.
d
OxO$VOꢂꢂ þ2e0 þ1=2O2
ð3Þ
From the law of mass action and the electroneutrality condi-
Â
Ã
tion 2 VꢂOꢂ ¼ ½e0ꢃ the variation of
d
as a function of pO2 for oxygen
Â
Ã
vacancies formation is derived as
d
¼ VꢂOꢂ ꢄ pꢀO1=2. Evaluation of
2
Since the parameter k in Eq. (1) has the dimension of an
equilibrium constant, one can extract the defect formation
enthalpy from the temperature dependence of k (Fig. 6). We have
estimated the defect enthalpy based on values of k resulting from
fits of the
n¼ ꢀ0.3670.01. The resulting enthalpy for the formation of
oxygen vacancies in SFMO is
ovH¼(25378) kJ/mol. This
value is similar to the oxygen vacancy formation enthalpy
ovH¼(34976) kJ/mol of LaMnO3ꢀ in the low oxygen partial
d vs. pO2 curves with the mean value of
D
D
d
pressure regime [21]. The oxygen vacancy formation in SrFeO3ꢀ
d
has an enthalpy of
larger range of nonstoichiometry.
DovH¼80 kJ/mol only [22] and, hence, a much
3.3. Structure and magnetism of nonstoichiometric samples
To study the influence of nonstoichiometry
d on the magnetic
properties of SFMO, we prepared a set of eight samples at 1200 1C
Fig. 5. Nonstoichiometry
d of Sr2FeMoO6ꢀ as function of oxygen partial pressure
d
at 1000, 1100, and 1200 1C; lines are fits to Eq. (1).
and different pO2. The preparation conditions, corresponding
nonstoichiometry values
d and lattice parameters are given in
Table 1. XRD showed all samples to be single phase double
perovskites, no reflections of secondary phases were detected
(Fig. 7). The unit cell parameters, as determined from Rietveld
refinements of the diffractograms in space group I4/m, are shown
as function of
whereas c0 slightly decreases with increasing defect concentra-
tion . The unit cell parameters are in good agreement with those
reported for an almost stoichiometric sample [23], or those
reported by Vasala et al. for a sample with E0.04 [24] (although
the data were refined in a triclinic space group I-1). However, the
unit cell volume expands with increasing (Fig. 8b). This is due to
d in Fig. 8a. The lattice parameter a0 increases,
d
d
d
the formation of lower valence cations as result of charge
compensation. The relative intensity of the (101) reflection as
function of the defect concentration
d is shown in Fig. 9. The
intensity of this superstructure peak is an indication of the degree
of ordering of Fe and Mo cations within the unit cell. Balcells et al.
[2] defined the ratio I101/(I200þI112) as semi-quantitative order
parameter. For our samples the intensity ratio exhibits a slight
decrease with increasing d. This is attributed to a modest increase
of disorder connected with the formation of oxygen vacancies.
Accordingly, the concentration of antisite defects (as determined
Fig. 6. Variation of ln k vs. the inverse temperature.
from Rietveld refinements) increases with
d (Fig. 9).
Table 1
Preparation conditions (1200 1C, pO2), lattice parameters, superstructure peak intensity ratios, Fe occupations on Fe position, antisite concentration per formula unit, and
saturation magnetizations of Sr2FeMoO6ꢀ samples.
d
3
No.
log pO2
d
I101/(I200þI112
)
Fe occ
AS per f.u.
Ms (lB)
˚
˚
˚
a0 (A)
c0 (A)
Vol. (A )
1
2
3
4
5
6
7
8
ꢀ10.12
ꢀ10.48
ꢀ11.66
ꢀ12.48
ꢀ12.91
ꢀ12.98
ꢀ13.07
ꢀ13.58
0.0055
0.0073
0.0191
0.0373
0.0529
0.0560
0.0603
0.0913
5.5704(1)
5.5709(1)
5.5711(1)
5.5728(3)
5.5732(1)
5.5745(2)
5.5733(1)
5.5764(2)
7.9028(2)
7.9032(2)
7.9023(2)
7.8989(6)
7.8997(2)
7.8955(7)
7.9001(2)
7.8925(3)
245.22(1)
245.28(1)
245.27(1)
245.31(3)
245.37(1)
245.35(3)
245.38(1)
245.43(3)
0.0570
0.0414
0.0332
0.0381
0.0360
0.0257
0.0319
0.0307
0.935(4)
0.940(4)
0.927(4)
0.925(5)
–
0.065(4)
0.060(4)
0.073(4)
0.075(5)
–
3.57
3.36
3.31
3.43
–
0.889(5)
0.905(5)
0.888(4)
0.111(5)
0.095(5)
0.112(4)
3.22
3.30
3.13

80
R. Kircheisen, J. To¨pfer / Journal of Solid State Chemistry 185 (2012) 76–81
Fig. 9. Intensity ratio of XRD reflections (110)/[(200)þ(112)] and refined antisite
concentration (AS per formula unit) as function of nonstoichiometry for
Sr2FeMoO6ꢀ
Fig. 7. X-ray diffraction patterns of single phase SFMO samples prepared at
1200 1C and log pO2¼ ꢀ10.48 with ¼0.007 (bottom), at log pO2¼ ꢀ13.07 with
¼0.060 (middle), and at log pO2¼ ꢀ13.58 with ¼0.091 (top).
d
d
.
d
d
d
Fig. 8. Unit cell parameters (a) and volume (b) as function of nonstoichiometry
for Sr2FeMoO6ꢀ
d
Fig. 10. Magnetization vs. field at T¼5 K (a) for SFMO samples prepared at
1200 1C and log pO2¼ ꢀ10.12 with ¼0.006 (top), at log pO2¼ ꢀ11.66 with
.
d
d
d
¼0.019 (middle), and at log pO2¼ ꢀ13.58 with d¼0.091 (bottom), and satura-
tion magnetization Ms at T¼5 K as function of nonstoichiometry
d for
Sr2FeMoO6ꢀ (b).
The magnetization measured as a function of the magnetic
d
field (T¼5 K) is saturated at a field of 50 kOe (Fig. 10a). Saturation
magnetizations (Fig. 10b) were found to be maximum (3.6
mB) for
measured for a well ordered sample [2]. It was demonstrated
that the use of reactive precursors and prolonged heating at
intermediate temperatures improves the cation ordering and,
hence, the magnetization; maximum Ms values of almost 4 mB
were reported [7,8]. Another model that focuses on the partial
decomposition of SFMO upon oxygenation (change of the 1:1 Fe/
Mo ratio and formation of grain boundary phases) was proposed
to explain reduced Ms values [11]. In this study we show, that the
pO2 of the surrounding gas atmosphere is another important
parameter, since the formation of oxygen vacancies leads to the
formation of lower valence cations (e.g. formation of ferrous ions
with S¼4/2, or Mo4þ with S¼1), thus reducing Ms. This situation
an almost stoichiometric sample. Increasing oxygen vacancy
concentrations tend to reduce Ms. Saturation magnetizations
smaller than 4 mB were reported in many other studies. Usually,
samples are prepared in Ar/H2 gas mixtures; if the hydrogen
concentration in the gas mixture is large (41%) this causes
extensive oxygen vacancy formation or even partial sample
decomposition. Balcells et al. [2] and many others have shown
that the synthesis temperature is an important factor that
influences cation ordering and, hence, the saturation magnetiza-
tion. A strong correlation between the antisite defect concentra-
tion and Ms was found. The magnetization is reduced at a rate of
Ms¼4 mBꢀ8/ASD concentration. A maximum Ms¼3.7 mB was
results in a reduction of Ms with a rate of 4 mBꢀ2d. This model

R. Kircheisen, J. To¨pfer / Journal of Solid State Chemistry 185 (2012) 76–81
81
was recently proposed based on ab-initio calculations [17]. The
magnetization data in Fig. 10b indicate that Ms is reduced from
the Bundesministerium fu¨r Bildung und Forschung (Germany) is
acknowledged (Grant 03WKF22E).
Ms¼3.5 mB at
The fact that the nearly stoichiometric samples already exhibit a
B is attributed to the presence of a
d¼0 to Ms¼3.1 mB at d¼0.1 (linear fit in Fig. 10b).
reduced magnetization of o4
m
certain level of antisite disorder in these samples. Hence the
magnetization reduction due to defect formation now is defined
References
by 3.5
vacancies only, a Ms¼3.3 would be expected for
observed reduction of Ms is somewhat larger than the predicted
rate of ꢀ2/ . This might be explained by the simultaneous
increase of the cation antisite disorder (ASD) with increasing
concentration of oxygen vacancies , as shown in Fig. 9. However,
it is demonstrated that nonstoichiometry and the formation of
oxygen vacancies is another source for reduced magnetization
values.
m
B–2
d
. If the reduction of Ms would be due to oxygen
[1] K.I. Kobayashi, T. Kimura, H. Sawada, K. Terakura, Y. Tokura, Nature 395
(1998) 677.
[2] L. Balcells, J. Navarro, M. Bibes, A. Roig, B. Martinez, J. Fontcuberta, Appl. Phys.
Lett. 78 (3) (2001) 781.
[3] O. Chmaissem, R. Kruk, B. Dabrowski, D.E. Brown, X. Xiong, S. Kolesnik,
J.D. Jorgensen, C.W. Kimball, Phys. Rev. B 62 (2000) 14197 #.
[4] S. Nakamura, K. Oikawa, J. Phys. Soc. Jpn. 72 (12) (2003) 3123–3127.
[5] J. Linden, T. Yamamoto, M. Karppinen, H. Yamauchi, Appl. Phys. Lett. 76
(2000) 2925.
[6] C. Kapusta, P.C. Riedi, D. Zajac, M. Sikora, J.M. De Teresa, L. Morellon,
M.R. Ibarra, J. Magn. Magn. Mater. 242–245 (2002) 242–245.
[7] T. Shimada, J. Nakamura, T. Motohashi, H. Yamauchi, M. Karpinnen,
Chem. Mater. 15 (2003) 4494.
d
¼0.1, but the
d
d
[8] Y.H. Huang, J. Linden, H. Yamauchi, M. Karppinen, Chem. Mater. 16 (2004)
4337–4342.
4. Conclusions
[9] M. Garcia-Hernandez, J.L. Martinez, M.J. Martinez-Lope, M.T. Casais,
J.A. Alonso, Phys. Rev. Lett. 86 (2001) 2443.
[10] X.Z. Liao, A. Sharma, M. Wei, J.L. MacManus-Driscoll, W. Branford, L. Cohen,
Y. Bugoslavsky, Y.T. Zhu, D. Peterson, Y. Jiang, H. Xu, J. Appl. Phys. 96 (2004)
7747.
The double perovskite Sr2FeMoO6ꢀ (SFMO) is stable only at low
d
oxygen partial pressure, e.g. between ꢀ10.2rlog pO2r ꢀ13.7 at
1200 1C. SFMO exhibits a significant nonstoichiometry
d as measured
by thermogravimety at 1000, 1100, and 1200 1C. At higher pO2 close
to the upper phase boundary SFMO is almost stoichiometric (e.g.
[11] J. MacManus-Driscoll, A. Sharma, Y. Bugoslavsky, W. Branford, L.F. Cohen,
M. Wei, Adv. Mater. 18 (2006) 900.
[12] Y.H. Huang, M. Karppinen, H. Yamauchi, J.B. Goodenough, Phys. Rev. B 73
(2006) 104408.
[13] T. Nakamura, K. Kunihara, Y. Hirose, Mat. Res. Bull. 16 (1981) 321.
[14] J. Rager, M. Zipperle, A. Sharma, J.L. MacManus-Driscoll, J. Am. Ceram. Soc. 87
(2004) 1330.
[15] T. Yamamoto, J. Liimatainen, J. Linden, M. Karppinen, H. Yamauchi, J. Mater.
Chem. 10 (2000) 2342–2345.
d
¼0.006 at 1200 1C and log pO2¼ ꢀ10.2), but
decreasing oxygen partial pressure (e.g.
¼0.085 at 1200 1C and
log pO2¼ ꢀ13.5). The nonstoichiometry is accommodated by oxygen
vacancies. The defect formation enthalpy is
d increases with
d
D
HOV¼25378 kJ/mol.
The oxygen vacancies might be associated to form clusters; this
requires further investigations.
The unit cell volume and the antisite concentration of non-
stoichiometric SFMO increases with increasing oxygen vacancy
concentrations. The saturation magnetization decreases with
This demonstrates that in addition to Fe/Mo antisite disorder
[16] H. Chang, M. Garcia-Hernandez, J.A. Alonso, Appl. Phys. Lett. 89 (2006)
182501.
[17] S. Collis, D. Stoeffler, C. Meny, T. Fix, C. Leuvrey, G. Pourroy, A. Dinia,
P. Panissod, J. Appl. Phys. 98 (2005) 033905.
[18] R. Mishra, O.D. Restrepo, P.M. Woodward, W. Windl, Chem. Mater. 22 (2010)
6092–6102.
d.
[19] J. To¨pfer, R. Kircheisen, S. Barth, J. Appl. Phys. 105 (2009) 07D712.
[20] A. Sharma, J.L. MacManus-Driscoll, W. Branford, Y. Bugoslavsky, L.F. Cohen,
J. Rager, Appl. Phys. Lett. 87 (2005) 112505.
oxygen nonstoichiometry
tization values.
d is another source of reduced magne-
[21] J.A.M. van Rossmalen, E.H.P. Cordfunke, J. Solid Stae. Chem 110 (1994)
113–117.
[22] A. Holt, T. Norby, R. Glenne, Ionics 5 (1999) 434–443.
[23] D. Niebieskikwiat, A. Caneiro, R.D. Sanchez, J. Fontcuberta, Phys. Rev. B 64
(2001) 180406.
[24] S. Vasala, M. Lehtima¨ki, Y.H. Huang, H. Yamauchi, J.B. Goodenough,
M. Karppinen, J. Solid State Chem. 183 (2010) 1007–1012.
Acknowledgments
The authors thank S. Barth (IKTS Hermsdorf, Germany) for the
preparation of the starting SFMO material. Financial support by