Coexistence of localized and itinerant electrons in the double-perovskite Ba2Fe2/3Mo4/3O6

Article abstract of DOI:10.1002/ejic.201402241

A polycrystalline Ba₂Fe_2/3Mo_4/3O ₆ double-perovskite has been prepared by decomposition of citrate precursors and subsequent thermal treatment under a reducing atmosphere. The crystal structure has been studied by combined X-ray and neutron powder diffraction (XRPD and NPD). At room temperature, the structure is cubic (Fm3?m) with lattice parameter a = 8.0710(1) ?. It is unchanged between 3 and 320 K. The crystallographic formula is Ba₂[Fe _0.52(2)Mo_0.48(2)]_4a[Fe_0.14(2)Mo _0.86(2)]_4bO_5.9(2). NPD, electron spin resonance (ESR), and magnetization measurements show spontaneous magnetization below the Curie temperature (T_C) = 310 K. The ESR behavior is associated with the presence of localized Fe³⁺ ions, whereas the transport properties (electrical conductivity and Seebeck effect) suggest the presence of highly correlated electrons in a metallic band with disorder, which can be associated with the Mo t_2g electrons. The coexistence of localized and itinerant electrons leads to non-negligible magnetoresistance properties.

Full text of DOI:10.1002/ejic.201402241

FULL PAPER
DOI:10.1002/ejic.201402241
Coexistence of Localized and Itinerant Electrons in the
Double-Perovskite Ba2Fe2/3Mo4/3O6
Carlos A. López,*[a] Martín E. Saleta,[b] José C. Pedregosa,[a]
Rodolfo D. Sánchez,[b] José A. Alonso,[c] and
María T. Fernández-Díaz[d]
Keywords: Iron / Molybdenum / Mixed-valent compounds / Magnetic properties / Perovskite phases
A polycrystalline Ba2Fe2/3Mo4/3O6 double-perovskite has
been prepared by decomposition of citrate precursors and
subsequent thermal treatment under a reducing atmosphere.
The crystal structure has been studied by combined X-ray
and neutron powder diffraction (XRPD and NPD). At room
onance (ESR), and magnetization measurements show spon-
taneous magnetization below the Curie temperature (TC) =
310 K. The ESR behavior is associated with the presence of
localized Fe3+ ions, whereas the transport properties (electri-
cal conductivity and Seebeck effect) suggest the presence of
highly correlated electrons in a metallic band with disorder,
which can be associated with the Mo t2g electrons. The coex-
istence of localized and itinerant electrons leads to non-neg-
¯
temperature, the structure is cubic (Fm3m) with lattice pa-
rameter a = 8.0710(1) Å. It is unchanged between 3 and
320 K. The crystallographic formula is Ba2[Fe0.52(2)
-
Mo0.48(2) 4a[Fe0.14(2)Mo0.86(2)]4bO5.9(2). NPD, electron spin res- ligible magnetoresistance properties.
]
The magnetic and electron-transport properties in the
Introduction
A2FeMoO6 (for A = Ca, Sr, or Ba) system are strongly re-
lated to the degree of the Fe/Mo site ordering and the oxi-
dation state of Fe and Mo.[5,6] In view of this, a number of
experimental studies were directed to change this degree of
ordering of Fe/Mo cations and their oxidation state. These
experimental studies included: doping at the A site,[7–9] dop-
ing at Fe or Mo sites,[10–12] and changing the Fe/Mo ra-
tio.[13–16] It is possible to shift from the oxidized form
(Fe3+–Mo6+) to the reduced form (Fe2+–Mo5+) when the
Fe/Mo ratio takes the limit values of 2 and 0.5, respectively,
in A2Fe1–xMo1+xO6 stoichiometries. If Fe/Mo = 1, the stan-
dard A2FeMoO6 formula contains pairs of Fe2+–Mo6+ ions
in equilibrium with Fe3+–Mo5+ couples. When Fe/Mo = 2,
the stoichiometry would be A2Fe4/3Mo2/3O6 (equivalent to
A3Fe2MoO9) in which all Fe and Mo ions are in an Fe3+
The search for materials with improved magnetotrans-
port properties has been a topic of interest during the last
decades in the field of transition-metal oxides, stimulated
by their potential technological applications.[1] In particular,
owing to the observation of colossal magnetoresistance
(CMR), manganese perovskites rapidly became promising
materials.[2] A few years after the discovery of this effect,
attention also became focused on some double-perovskite
oxides. The revival of interest in this family was triggered
by a report on Sr2FeMoO6,[3] which demonstrated that in
the electronic structure only minority spins are present at
the Fermi level; this material was shown to exhibit intrinsic
tunneling-type magnetoresistance (TMR) at room tempera-
ture.[3,4] In a simple picture, FeO6 and MoO6 octahedra
alternate along the three directions of the crystal structure
of Sr2FeMoO6; the system can be described as an ordered
arrangement of parallel Fe3+ (S = 5/2) magnetic moments
antiferromagnetically coupled with Mo5+ (S = 1/2) spins.
2
(t2g3eg ) and Mo6+ (t2g0) state, respectively.[17–19] Finally, if
Fe/Mo = 0.5, the formula is A2Fe2/3Mo4/3O6 (equivalent to
A3FeMo2O9) with Fe and Mo ions in the states Fe2+ and
Mo5+. Also, all the intermediate situations are possible, in-
cluding those for which 0.5 Ͻ Fe/Mo Ͻ 2. Whereas the
cases with Fe/Mo = 1 and 2 are well known,[13–16] the perov-
skites for Fe/Mo = 0.5 (all Fe and Mo are as Fe2+ and
Mo5+ reduced forms) have not been reported, and therefore
their structure and physicochemical behaviors are un-
known. Interesting magnetic and electric properties can be
predicted in a system in which Mo5+ (t2g1) with one itiner-
[a] INTEQUI – Área de Química General e Inorgánica “Dr. G. F.
Puelles”, Facultad de Química, Bioquímica y Farmacia,
Universidad Nacional de San Luis,
Chacabuco y Pedernera, 5700 San Luis, Argentina
E-mail: calopez@unsl.edu.ar
http://www.sanluis-conicet.gob.ar/intequi/
[b] Centro Atómico Bariloche, Comisión Nacional de Energía
Atómica and Instituto Balseiro, Universidad Nacional de Cuyo,
8400 S. C. de Bariloche, Río Negro, Argentina
[c] Instituto de Ciencia de Materiales de Madrid, CSIC,
Cantoblanco, 28049 Madrid, Spain
2
ant electron and Fe2+ (t2g4eg ) cations are stabilized.
Furthermore, the fact that the end member perovskites
SrMoO3 and BaMoO3 (both with Fe/Mo = 0) exhibit met-
allic behavior with the highest conductivities ever observed
[d] Institut Laue-Langevin,
B. P. 156, 38042 Grenoble Cedex 9, France
Eur. J. Inorg. Chem. 2014, 3555–3563
3555
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
in transition-metal oxides is appealing. In these materials
the electrons of Mo4+ (t2g2), are completely delocalized.
Table 1 summarizes the intermediate and end situations of
Fe/Mo ratio and the possible oxidation states.
Table 1. A2Fe1–xMo1+xO6 double-perovskites with different Fe/Mo
ratios.
Fe/Mo ratio
A2Fe1–xMo1+xO6
Oxidation states[a]
2
1
0.5
0
A2Fe4/3Mo2/3O6
A2FeMoO6
A2Fe2/3Mo4/3O6
AMoO3
Fe3+/Mo6+
Fe2+/Mo6+ – Fe3+/Mo5+
Fe2+/Mo5+
Mo4+
[a] Expected oxidation states from the chemical point of view.
In the present work, we describe the synthesis of
Ba2Fe2/3Mo4/3O6 double-perovskite (Fe/Mo = 0.5) obtained
by soft-chemistry procedures followed by thermal treat-
ments under reducing conditions (H2/N2). A well-crys-
tallized sample was studied by X-ray and neutron powder
diffraction, and its crystal structure was refined by the Riet-
veld method. To obtain a better understanding of this new
system, we performed magnetic, electron-transport, mag-
neto-transport, and thermoelectric measurements. The pre-
dominant Fe3+ oxidation state was confirmed by electron
spin resonance (ESR).
Figure 1. Observed (circles), calculated (full line), and difference
(bottom) Rietveld profiles at room temperature for Ba2Fe2/3
-
Results and Discussion
Mo4/3O6 after a combined XRPD-NPD refinement. The second
and third series of tick marks correspond to the Bragg reflections
for the BaMoO4 and Fe impurity phases. (a) XRPD. Inset: sche-
matic view of crystallographic structure showing MoO6 and FeO6
octahedra. (b) NPD. The fourth series of tick marks in NPD corre-
spond to the Bragg reflections for the magnetic structure.
Crystallographic Structure
The structural refinement was performed with combined
X-ray powder diffraction (XRPD) and neutron powder dif-
fraction (NPD) data at room temperature. The good agree-
ment between the observed and calculated XRPD and
NPD patterns after the refinement is shown in Figure 1a the states Fe2+/Mo5+ or alternatively Fe3+/Mo4.5+. The de-
and b. The present sample was successfully refined as cubic gree of Fe/Mo disordering was experimentally refined, as
in the space group Fm3m based upon the model of the well as the occupancy of oxygen positions. The final
¯
Ba2FeMoO6 double-perovskite.[8–13]
refined stoichiometry is Ba2[Fe0.52(2)Mo0.48(2)]4a[Fe0.14(2)
-
In this space group (no. 225), the unit-cell parameter is Mo0.86(2)]4bO5.9(2). The most important structural param-
related to a0 (ideal cubic perovskite, a0 ≈ 3.9 Å) as a = 2a0. eters of the crystallographic structure and the discrepancy
Ba atoms were located at the 8c (1/4,1/4,1/4) positions, Fe/ factors after the refinements are listed in Table 1. Within
Mo at 4a (0,0,0) and at 4b (1/2,1/2,1/2) sites, and oxygen the experimental error, the Fe/Mo ratio is 0.5, and the oxy-
¯
atoms at the 24e (x,0,0) positions. For the Fm3m space gen content can be considered stoichiometric. The fact that
group there are two possible sites for B-type cations, some Fe partially occupies the 4b sites indicates the pres-
namely, 4a and 4b. Minor amounts of BaMoO4 and elemen- ence of some structural disordering beyond the intrinsic dis-
tal iron were detected as impurities and included in the re- order given for this stoichiometry. If we call α the amount
finement with their respective crystal structures defined in of Fe at 4b sites, we can define the degree of order (DO) as
[20,21]
From the scale fac- DO = 1 – 3α.[18] For maximum order DO = 1, and for
¯
the space groups I41/a and Im3m.
tor of the main and secondary phases we estimated the fol- random distribution DO = 0. In the present case we ob-
lowing impurity levels as: 1.0(4) and 0.6(4) wt.-% of served DO = 0.58.
BaMoO4 and Fe0, respectively. Additionally, as this phase
The thermal evolution of the crystallographic structure
is magnetic at room temperature, we included a model for was investigated from NPD data sequentially collected from
¯
the magnetic structure, as described below.
3 to 320 K. The cubic model (space group Fm3m) was suc-
Double-perovskites with stoichiometry A2BЈ1–xBЈЈ1+xO6, cessfully considered in the whole temperature range. The
in which BЈ/BЈЈ ϶ 1, exhibit an intrinsic octahedral disor- average thermal expansion coefficient from 150 to 320 K
der, since some BЈ ions occupy BЈЈ sites and vice versa. In is 11.3ϫ10–6 K–1. The thermal evolution of the a unit-cell
the present case with Fe/Mo = 0.5, the formula is parameter and distances ϽFe/Mo–OϾ are shown in Fig-
Ba2Fe2/3Mo4/3O6, and Fe and Mo ions nominally exhibit ure 2; they display a weak temperature dependence. Fig-
Eur. J. Inorg. Chem. 2014, 3555–3563
3556
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
ure 2 also includes the expected distances for different oxi-
dation states for Fe and Mo.
Figure 3. TGA curves of Ba2Fe2/3Mo4/3O6 sample (oxidation pro-
cess) obtained in air flow at 5 Kmin–1.
fore, the sample presents an Fe(3d6–δ)–Mo(4d1+δ/2) configu-
ration with 0 Ͻ δ Ͻ 1. From the structural information
described above, including the thermal evolution of the dis-
tances (Figure 2), we can assume that the δ value is close to
1.
Magnetic Structure
As previously mentioned, the NPD pattern at room tem-
perature shows a magnetic contribution to the scattering on
allowed Bragg positions (k = 0), mainly on the (111) reflec-
tion, which suggests a ferro- or ferrimagnetic long-range
ordering. The magnetic contribution is enhanced upon
cooling, as is clearly shown in the thermal evolution of the
NPD patterns (Figure 4). We have fit the magnetic contri-
butions to the neutron scattering by modeling a ferri-
magnetic structure that consists of a collinear arrangement
of two sets of spins with opposite directions at the 4a and
4b sites, which corresponds to the B positions of the per-
ovskite structure, with the same unit-cell parameters as the
crystallographic Ba2Fe2/3Mo4/3O6 phase. The magnetic mo-
ments of both sets have been constrained to the relationship
Figure 2. Top panel: a unit-cell parameters as a function of tem-
perature. Bottom panel: octahedral size ϽFe/Mo–OϾ as a function
of temperature for both crystallographic B sites. Dotted and dashed
lines indicate the expected distances for different oxidation states
for Fe and Mo.
Considering the stoichiometry Ba2Fe2/3Mo4/3O6, and
from the chemical point of view, we expected that the oxi-
dation states would be between Ba2Fe2+2/3Mo5+4/3O6 and
Ba3Fe3+2/3Mo4.5+4/3O6. The structural analysis might help
to ascertain the true configuration. As shown in Table 1,
the ϽFe/Mo–OϾ distances at room temperature are
2.045(1) and 1.990(1) Å for the 4a and 4b sites, respectively.
The comparison with the expected distances for different
possibilities for the oxidation states of Fe and Mo suggests
that in the electronic configurations for the 4b site the dis-
tances are close to those expected to Fe3+Mo5+, whereas for
the 4a site they are closer to Fe3+Mo4+ than to Fe2+Mo5+
.
Thermogravimetric Analysis
Figure 3 illustrates the TGA curves in air (oxidation pro-
cess) of Ba2Fe2/3Mo4/3O6. The samples are stable in an air
flow up to 575 K; above this temperature, an oxidation pro-
cess starts. The final products were identified by XRPD al-
lowing for the reaction according to [Equation (1)].
Ba2Fe2/3Mo4/3O6 + O2 Ǟ 1/2 Ba2Fe4/3Mo2/3O6 + BaMoO4
(1)
The experimental weight gain for these reactions (2.8%)
is in agreement with the theoretical ones (2.9%). These facts
indicate that, within the experimental error, Fe and Mo ions
are as Ba2Fe2+2/3Mo5+4/3O6 or the equivalent isoelectronic
Figure 4. Thermal evolution of NPD patterns, with λ = 2.492 Å,
state Ba3Fe3+2/3Mo4.5+4/3O6, which are indiscernible. There- collected sequentially.
Eur. J. Inorg. Chem. 2014, 3555–3563
3557
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
that exists between the Fe contents of 4a versus 4b sites
(0.52/0.14). After refinement, the ordered magnetic mo-
ments for 4a and 4b sites are 2.6 and 0.7 μB, respectively.
These values can be obtained by multiplying the magnetic
moment of Fe3+ (3d5) of 5 μB for its occupation of each
site. The good agreement between observed and calculated
NPD patterns at 3 K is shown in Figure 5a. The thermal
variation of the magnitude of the ordered 4a and 4b mag-
netic moments is displayed in Figure 5b.
Figure 6. Magnetization (left axis) and first derivative of the mag-
netization (right axis) as a function of temperature at 5 kOe. The
inset shows the magnetizations versus field loops at 5 and 298 K.
The magnetization versus magnetic field curve at 5 K
and room temperature is plotted in the inset of Figure 6.
At both temperatures, the maximum magnetization values
measured for the higher applied field of H = 10 kOe reach
ca. 0.7 and 1.3 μB per formula unit (f.u.) at 298 and 5 K,
respectively. In a completely localized moment picture, the
global ferromagnetism is the result of the microscopically
unbalanced occupancy of 4a and 4b sites by Fe and Mo
ions. Assuming that the coupling between neighbors
through common oxygen atoms (either Fe–O–Fe, or Fe–O–
Mo, or Mo–O–Mo) is always antiferromagnetic (AFM), the
coherence across the crystal is perfectly maintained, and the
saturation magnetization is the result of the different occu-
pation between the crystallographic 4a and 4b sites. For this
model, considering Fe3+ (S = 5/2) and equimolar amounts
of Mo4+ (S = 1) and Mo5+ (S = 1/2), we would obtain a
saturation magnetization of 1.33 μB per f.u., in perfect
agreement with the observed magnetization value at 5 K of
1.3 μB per f.u. However, when observed in detail, the M–H
loops have an elongated “S” shape with a small coercive
field (approximately Hc = 160–180 Oe at low temperatures).
Both sides of the loop, upon increasing and decreasing H,
can be described by a Langevin function [M/Ms = coth(x) –
1/x, in which x = μ(HϮHc)/kT and Ms is the saturation
Figure 5. (a) Rietveld plots after the refinement of the crystal struc-
ture and magnetic structure from NPD data at T = 3 K. The line
at the bottom is the difference between calculated and experimental
patterns. The four sets of Bragg reflections correspond to the main
crystallographic phase, BaMoO4, metal Fe, and the magnetic phase,
respectively. (b) Thermal dependence of the ordered magnetic mo-
ments refined from NPD data. The inset shows a schematic view
of the magnetic structure.
Magnetic Properties
The magnetization (M) versus temperature curve is dis- magnetization]. This behavior can be associated with the
played in Figure 6. The sample has a spontaneous in- rotation of the net magnetization of small regions (“FM
crement of the magnetization below 310 K. The Curie tem- clusters”), which can interact between them through the
perature (TC) was determined as the minimum value in the itinerant Mo (t2g) electrons. It is an alternative picture in
dM/dT versus T curve (see Figure 6). This temperature is in which the uncompensated localized Fe3+ ions produce a net
excellent agreement with that determined from tempera- magnetization in the material, which is reinforced by the
ture-dependent NPD data, which also shows that the long- presence of itinerant t2g electrons.
range magnetic ordering is established close to 310 K.
The ESR spectra were collected in the temperature range
Above TC, the temperature range is too narrow to per- 100 K Ͻ T Ͻ 350 K; some selected spectra are shown in
form a Curie–Weiss fit. In similar phases with high TC val- Figure 7. The obtained ESR line shape is Lorentzian, as
ues it was also difficult to obtain information on the para- expected for ceramic samples reduced to a fine powder. The
magnetic region owing to the existence of antisite defects spectra for T Ͻ TC are shown in Figure 7a. The magnetic
or the presence of spurious ferromagnetic impurities, which resonance field is shifted to the paramagnetic value, and the
mask the paramagnetic behavior.[22,23]
line is narrowed with increasing T. Yet the spectra for T Ͼ
Eur. J. Inorg. Chem. 2014, 3555–3563
3558
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
TC (shown in Figure 7b) present a constant value of the
The experimental peak-to-peak line width (ΔHPP) and
resonant magnetic field (HR) that corresponds to g = 2.05. the line intensity (IESR) for the crystalline powder of
We discard the possibility of assigning the observed reso- Ba2Fe2/3Mo4/3O6 as a function of temperature are shown
nance to localized Fe2+, because its ground state in an octa- in Figure 8. The ΔHPP shows a minimum of approximately
hedral environment is a triplet with an expected g factor of 380 Oe at Tmin ≈ TC. Below this temperature ΔHPP(T) in-
about 3.4, which is far from our measured value. However, creases with decreasing T to reach values of 1500 Oe at
as Fe2+ presents an usually short relaxation time, it would 130 K. Above Tmin, ΔHPP increases with T to reach 650 Oe
render their ESR line undetectable, as in the case of FeO. at 350 K. The IESR shows the same behavior as the macro-
This obtained value is close with a spin-only ground state scopic magnetization of the compound (Figures 5b and 6).
that principally corresponds to the localized 3d5 Fe3+ (1.997 All the ESR parameters present the expected behavior of a
Յ g Յ 2.003)[24] and it is similar to that reported for ferromagnetic material.
Sr2FeMoO6[22,23] perovskite, in which the itinerant electrons
of Mo ions contribute to shift the Fe3+ g value a little.
Figure 8. ESR intensity (IESR) (left axis) and peak-to-peak line-
width (ΔHPP) (right axis) as a function of temperature.
The Curie temperature is also observed in the ESR ex-
periments. The ESR lines reach a constant value of the res-
onant magnetic field at 310 K, as observed in Figure 7b.
The ΔHPP presents a minimum value close to TC in the
same way as other similar compounds.[22,23] One of the
common problems in the ESR results of samples in which
one of the cations is Fe3+ is the observation of spurious
magnetic impurities as ferromagnetic iron oxides. In gene-
ral, their strong ESR signal masks the response of the mate-
rial in focus. However, in this case, the ESR intensity as a
function of temperature (Figure 7) presents the same spon-
taneous magnetization and TC as that observed by other
two techniques: the magnetic moment per site (NPD),
which is represented in Figure 5b, and by the magnetization
curves as a function of temperature plotted in Figure 6.
This corroborates that the ESR signal is the response of
the magnetic ions contained by the double-perovskite under
study and is not from spurious magnetic phases.
Figure 7. (a) ESR spectra (absorption derivatives) at several tem-
peratures: (a) below TC and (b) in the paramagnetic region. The
Electron-Transport Properties
inset shows the resonance field (HR) versus temperature.
The electrical resistivity of the Ba2Fe2/3Mo4/3O6 perovsk-
ite was performed from 300 to 80 K. As shown in Figure 9,
The temperature dependence of HR is shown in the inset at room temperature the resistivity is 191 Ωcm–1. The resis-
of Figure 7, which also shows the magnetic-field resonance tivity increases quickly from 300 to 280 K, and then the
expected for paramagnetic Fe3+ ions. Below TC, HR values resistivity increases smoothly up to 80 K. The negative
diminish as T decreases owing to the internal magnetic field slope value of the resistivity versus temperature data could
that comes from the magnetic order inside the sample, thus be associated with a semiconductor response of the mate-
demonstrating that the Fe3+ ions are part of the magnetic rial, but the absolute ρ values are relatively low (ρ Ͻ
lattice. This result is also supported by NPD experiments, 102 Ωcm–1), which might indicate that the Fermi level of
in which the presence of Fe3+ is evidenced by the average the material, close to the mobility border of the electrons,
Fe/Mo–O distances.
occupies localized states in a metallic band owing to the
Eur. J. Inorg. Chem. 2014, 3555–3563
3559
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
presence of disorder. The theory predicts, for highly corre-
lated electrons in metals with disorder, that the thermal be-
havior of the electrical conductivity must follow σ(T) = σ0
+ A·T1/2 [25]
This thermal and linear behavior of the con-
.
ductivity with the square root of the temperature can be
viewed in the inset of Figure 9, thereby confirming that this
double-perovskite is metallic with highly correlated elec-
trons under a disorder potential. It is noteworthy that this
effect is observed in this double-perovskite at immediately
below TC, unlike V2O3 and doped semiconductors Si:P, in
which this effect is observable at low temperature, or also
unlike the clear metallic state observed in the perfectly or-
dered Sr2FeMoO6 double-perovskite.[25]
Figure 10. Seebeck coefficient as a function of temperature. The
dashed line represents the temperature dependence of a contri-
bution to the phonon drag thermopower, labeled Sg, from a typical
phonon in the acoustic branch. The diagram also shows a negative
diffusion contribution Sd (doted line) and illustrates how the ob-
served thermopower results from a sum of the two contributions
(solid line).
The electrical resistance was also measured as a function
of magnetic field at room temperature and at 90 K (see Fig-
ure 11a). The measurement at room temperature showed a
negligible effect with a magnetic field of 3 kOe. However, at
90 K a magnetoresistive effect was clearly observed. Fig-
ure 11 shows the magnetoresistance (MR) defined as ΔR/
R (%) = 100ϫ[ρ(H) – ρ(0)]/ρ(0) as a function of H/T. A
Figure 9. Electrical resistivity as a function of temperature. The negative magnetoresistance is observed; it reaches a maxi-
inset shows the linear behavior of σ = σ0 + T1/2
.
mum MR of –1.5% at 3 kOe. The ΔR/R versus H/T plot
also showed an interesting hysteresis, the maximum value
of which can be associated with the coercive field of the
sample measured in the hysteresis loop.
Figure 10 shows the thermal variation (cooling and
warming) of the Seebeck coefficient (S); S is a zero-current
transport coefficient, and it was measured using the stan-
dard temperature differential technique. It is a very useful
proof to discern between semiconductor and metallic be-
havior. The S values are negative and show a linear behav-
ior in the measured temperature range. This behavior has
also been observed in double-perovskite systems such as
A2FeMoO6,[26–28] and it is remarkably similar to that found
in the metallic phase of the high-temperature superconduct-
ing cuprates[29] and manganites.[30] In general, the diffusion
term of the thermoelectric power coming from an electron
gas is described by the Mott formula [Sd = (π2kB2T/
3e)(Ѩlnσ/Ѩε)εF].[31] The (Ѩlnσ/Ѩε)εF is a generalization,
which is associated with the energy at the Fermi surface
(εF). The Mott formula is linear with temperature; π, kB,
and e are constants and the last term can be reduced, ac-
cording to the model, to a constant value, too. The slope
of the S versus T data is –0.024 μVK–2. In the limit when
T Ǟ 0, also S Ǟ 0. An extra contribution should be present
at the Seebeck coefficient, which can be the phonon drag
contribution. It has a maximum value around a fraction of
the Debye temperature (ΘD/3), and the contribution de-
creases as 1/T. We use the model of Trodahl[29] to interpret
the S coefficients in terms of a phonon drag component
(dashed line) and a diffusion term (dotted line). This behav-
ior has also been observed in double-perovskite systems
such as A2FeMoO6.[26–28]
Figure 11. (a) Magnetoresistance (ΔR/R) as a function of magnetic
field at 90 K. (b) ΔR/R (left axis) and M/Ms (right axis) as a func-
tion of H/T.
Eur. J. Inorg. Chem. 2014, 3555–3563
3560
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
However, the electrical resistivity also presents an anom- electric properties can be understood in the context of a
alous behavior close to room temperature. The coexistence disordered metal that presents negative magnetoresistance,
of important electronic conductivity and FM cluster inter- which is dominated by the scattering between the itinerant
actions below TC accounts for the magnetotransport prop- electrons with the magnetic field alignment of the ferromag-
erties, as observed in other half-metallic double-perovskites. netic region polarization.
In Sr2FeMoO6 the extra itinerant Mo electrons occupy
down-spin bands and they are shared among Fe and Mo
Conclusion
atoms. In the Ba2Fe2/3Mo4/3O6 system there are itinerant
(associated with Mo) and localized (associated with Fe)
electrons, as shown from NPD data, whereby the localized
Fe spins contribute to the magnetic scattering and to the
EPR signal, and, on the other hand, the transport and See-
beck measurements, which show a clear metallicity and itin-
erancy of some electrons. But the parallelism with
Sr2FeMoO6 is not complete, since significant differences
can be noted from the results of the present study. One of
them is the magnetic behavior with H, in which the presence
of regions with FM clusters can be described with a Lange-
vin function.
The MR properties would be a result of the spin-depend-
ent scattering of the charge carriers through the magneti-
cally ordered regions; this scattering is reduced when an
external magnetic field is applied, thus aligning the orienta-
tion of the isolated clusters. The maximum value of MR
(–1.5%) was measured at 90 K (see Figure 11a). The ob-
served increase in MR as the magnetic field increased is
similar to that exhibited in other double-perovskites,
namely, Sr2FeMoO6, in which the main component of MR
is attributable to grain boundary tunneling effects that take
place at temperatures well below TC; however, in our case
the magnetoresistive effect is lower than in Sr2FeMoO6 or
Ba2FeMoO6.[32,33] In general, the MR is proportional to the
square of the magnetization, and in this case the experimen-
tal MR versus H/T curve can be successfully reproduced by
considering the square of the Langevin equation and the
hysteresis process of 150 Oe approximately. This can be seen
in Figure 11b, in which the ΔR/R versus H/T plot is cor-
rectly described with this model.
To summarize, Ba2Fe2/3Mo4/3O6 is a partially disordered
perovskite that presents FM-cluster or small-FM-region be-
havior below 310 K and the coexistence of localized Fe3+
with itinerant electrons from the Mo ions. As in
Sr2FeMoO6, the electron itinerancy arises from down-spin
Mo 4d1+δ/2 electrons of t2g character. The decrease in the
Fe content, realized in the perovskite end-member SrMoO3,
leads to a metallic state with a total electron delocalization.
In the present case, we are dealing with intermediate Fe
content in which Fe electrons occupy localized states,
whereas majority Mo electrons are itinerant and account
for the transport properties. The high magnetic-ordering
temperature is therefore the result of direct antiferromag-
netic superexchange Fe–O–Fe interactions combined with
the presence of itinerant Mo t2g electrons. This magnetic
state is promoted by a structural disorder (DO = 0.58),
which promotes the occurrence of Fe atoms with different
environments, from Fe–O–(Fe6) to Fe–O–(Mo6), and all the
intermediate situations.[18] The net magnetization of this
A new double-perovskite, Ba2Fe2/3Mo4/3O6, with Fe/Mo
mixed-valence states and a configuration close to Fe3+
–
Mo4.5+ has been obtained by citrate techniques followed by
an annealing in an H2/N2 flow at 800 °C. The crystal struc-
¯
ture is cubic with space group Fm3m in the temperature
range 3–320 K; the average thermal expansion coefficient
is 11.3ϫ10–6 K–1. The crystallographic formula obtained
at room temperature, Ba2[Fe0.52(2)Mo0.48(2)]4a[Fe0.14(2)
-
Mo0.86(2)]4bO5.9(2), involves an important antisite disorder-
ing, which accounts for the observed magnetic properties.
The magnetic ordering temperature is TC = 310 K, as
shown from magnetization and NPD data. This high mag-
netic-ordering temperature is the result of direct antiferro-
magnetic superexchange Fe–O–Fe interactions combined
with the presence of itinerant Mo t2g electrons. The mag-
netic structure at 3 K has been successfully refined by mod-
eling a ferrimagnetic-like arrangement of two sets of spins
with opposite directions at the 4a and 4b sites that corre-
spond to the B positions of the perovskite structure. The
uncompensated magnetization of the crystallographic sites
results in a net ferromagnetic component. The magnetiza-
tion behavior can be described with a Langevin function.
The electric conductivity shows a square-root tempera-
ture dependence, which is characteristic of disordered met-
als and is observed in deficient vanadium oxide, doped
semiconductors, and La1–xSrxCoO3. A magnetoresistance
as high as –1.5% for H = 3 kOe is observed at 90 K. The
Seebeck effect is the result of two contributions: one of
them is the phonon drag component, and the second is
characteristic of high conductivity samples with itinerant
electrons in a conduction band.
Experimental Section
The Ba2Fe2/3Mo4/3O6 double-perovskite was prepared as a black
polycrystalline powder from citrate precursors obtained by soft-
chemistry procedures under a controlled atmosphere. Stoichiomet-
ric amounts of analytical-grade Ba(NO3)2, Fe(NO3)3·9H2O, and
(NH4)6Mo7O24·4H2O were dissolved in citric acid. The polymeriza-
tion of the citrate solution upon gentle heating led to organic resins
that contained a random distribution of the cations involved at an
atomic level. These resins were first dried at 180 °C, and then all
the organic materials and nitrates were eliminated in a subsequent
treatment at 600 °C in air for 12 h, thus giving rise to highly reac-
tive precursor materials.
The conditions for the Ba2Fe2/3Mo4/3O6 synthesis were established
from several trials under different conditions of temperature and
time, in H2/N2 (5:95) atmosphere. The tested final temperatures
were 700, 750, 800, 850, 900, and 950 °C. The optimal synthesis
compound can be described by a Langevin function. The conditions were found after treatment under H2/N2 (5:95) atmo-
Eur. J. Inorg. Chem. 2014, 3555–3563
3561
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
sphere at 800 °C for 12 h. The rate of heating and cooling was
5 °Cmin–1. Thermal treatments longer than 12 h did not produce
changes in the quality of the samples, thus suggesting that the equi-
librium with the atmosphere is reached before this treatment time.
In Table 2 we summarize the different temperatures tested and the
phases obtained in each case. It is worth commenting that trials to
obtain the same phase with Sr instead of Ba did not lead to single
perovskite phases but a mixture of oxides. Different aliquots, in
powder form, from a unique batch obtained from the synthesis
were used for the measurements.
Table 3. Crystallographic data for the Ba2Fe2/3Mo4/3O6 phase. Sys-
tem: cubic; space group: Fm3m.
¯
Crystallographic data from combined XRPD and NPD
(λ = 1.91 Å) data at r.t.; a = 8.0710(1) Å, V = 525.76(1) Å3
(a) Positions, thermal and occupancy factors
Atom Wyckoff site
x/a
y/b
z/c Biso [Å2]
Occ
Ba
Fe1
Mo1
Fe2
Mo2
O
8c
4a
4a
4b
4b
0.25
0
0
0.5
0.5
0.25 0.25
0.29
0.25
0.25
0.16
0.16
0.50
1
0
0
0
0
0.52(2)
0.48(2)
0.14(2)
0.86(2)
0.99(3)
0.5
0.5
0
0.5
0.5
0
Table 2. Phases observed by XRPD (indicated with crosses) at dif-
ferent synthesis temperatures.
24e
0.2534(7)
NPD: Rp = 4.63%. Rwp = 5.70%. Rexp = 5.53%. χ2 = 1.06. RBragg
=
T [°C] (12 h)
Main phase
Spurious phases
1.70%, XRPD: Rp = 6.63%. Rwp = 8.78%. Rexp = 10.13%. χ2 = 0.75.
RBragg = 2.16%.
Ba2Fe2/3Mo4/3O6 BaMoO4 Fe0
BaMoO3
700
750
x
x
x
x
(b) Main distances
800[a]
850
x
x
x
x
x
x
x
[Fe/Mo]O6 octahedral
AO12 polyhedral
(Fe/Mo)4a–O(ϫ6) 2.045(1) Å
(Fe/Mo)4b–O(ϫ6) 1.990(1) Å
Ba–O(ϫ12) 2.854(4) Å
900
950
x
x
Crystallographic data from NPD (λ = 2.49 Å) data at 3 K, a =
8.0537(2) Å, V = 522.38(2) Å3
[a] Best synthesis conditions; sample selected for the study.
(a) Positions and thermal factors
The initial identification and characterization of the samples as well
as the crystallographic study were carried out from XRPD patterns
collected at room temperature. The experimental XRPD patterns
were obtained with a Rigaku D-MAX-IIIC diffractometer with
Cu-Kα (λ = 1.5418 Å) radiation. The data were collected over a
range of 10–120° (2θ) with a step of 0.02°; the effective counting
time was 5 s per step (Table 3).
Atom Wyckoff site
x/a
y/b
z/c Biso [Å2]
Occ
Ba
Fe1
Mo1
Fe2
Mo2
O
8c
4a
4a
4b
4b
0.25
0
0
0.5
0.5
0.25 0.25
0.70
0.74
0.74
0.40
0.40
0.40
1
0
0
0
0
0.52
0.48
0.14
0.86
1
0.5
0.5
0
0.5
0.5
0
24e
0.2466(8)
NPD: Rp = 2.41%. Rwp = 3.29%. Rexp = 0.83%. χ2 = 15.0.
RBragg = 8.21%.
For the structural refinements, NPD patterns were collected at
room temperature with a D1A neutron diffractometer at λ =
1.910 Å. The thermal evolution of the crystallographic and mag-
netic structures was carried out from NPD patterns sequentially
collected with a D20 diffractometer at a wavelength of 2.492 Å;
both instruments (D1A and D20) belong to the Institute Laue Lan-
gevin (ILL), Grenoble, France. The patterns were refined by the
Rietveld method[34] using the FULLPROF refinement program.[35]
A pseudo-Voigt function was chosen to generate the line shape of
the diffraction peaks. In the final runs, the following parameters
were refined: scale factors for the main and impurity phases, back-
ground coefficients, zero-point error, unit-cell parameters, pseudo-
Voigt-corrected asymmetry parameters, positional coordinates, an-
isotropic thermal factors, and antisite disorder of Fe/Mo and occu-
pancy factors for oxygen atoms. The coherent scattering lengths for
Ba, Fe, Mo, and O were 5.05, 9.45, 6.72, and 5.803 fm, respectively.
(b) Main distances
[Fe/Mo]O6 octahedral
(Fe/Mo)4a–O(ϫ6) 2.038(4) Å
(Fe/Mo)4b–O(ϫ6) 1.989(4) Å
AO12 polyhedral
Ba–O(ϫ12) 2.848(3) Å
temperature of one end of the sample. For these electron-transport
experiments, polycrystalline powder was pressed in bar-shaped pel-
lets and sintered at 800 °C for 6 h in an Ar flow. These sintering
conditions were used to obtain a ceramic sample equivalent to the
powdered samples in such a way that either the chemical composi-
tion or the physical properties are unaltered by the process.
The ESR experiments were performed at 9.5 GHz (X-band) with a
Bruker ESP-300 spectrometer from 100 to 350 K. The powder sam-
ple was reground and diluted with KCl (approx. 1:5) to fill the
whole length of the resonator to warrant a uniform microwave
field.
Thermogravimetric analysis was performed with a Shimadzu TG-
50H thermal analyzer apparatus using flowing air at 50 mLmin–1
from 298 to 1273 K at a heating rate of 5 Kmin–1. About 18 mg of
the sample was used in the experiment.
Magnetic susceptibility was measured with a commercial SQUID
magnetometer on powdered samples in the temperature range of
5–330 K and in a commercial VSM magnetometer in the range of
298–400 K.
Acknowledgments
J. C. P. thanks Consejo Nacional de Investigaciones Científicas y
Técnicas (CONICET) (project PIP 112-200801-01360), Secretaría
de Ciencia y Tecnología de la Universidad Nacional de San Luis
(SECyT-UNSL) (project PROICO 2-7707-22/Q823), and Agencia
Nacional de Promoción Científica y Tecnológica (ANPCyT) (pro-
ject PICT 6/25459). R. D. S. thanks Universidad Nacional de Cuyo
(UNCuyo), CONICET, and ANPCyT (Argentina) under grants 06/
C389, PIP-490, and PICT2007-0832/PICT2011-0752. J. C. P.,
R. D. S., C. A. L., and M. E. S. are members of CONICET. J. A. A.
The electrical-transport properties were determined with home-
made equipment. The experiments were carried out between 80 and
300 K. In the case of the electrical resistivity, it was measured by
the conventional four-probe technique. To obtain the magne-
totransport response, the electrical resistance was taken under mag-
netic fields up to 3 kOe, and the thermoelectric power was obtained
from the slope of the voltage generated by a thermal variation using
a temperature ramp of 3 K in temperature with respect to the fixed
Eur. J. Inorg. Chem. 2014, 3555–3563
3562
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.eurjic.org
FULL PAPER
[17] M. C. Viola, J. A. Alonso, J. C. Pedregosa, R. E. Carbonio,
Eur. J. Inorg. Chem. 2005, 1559–1564.
[18] C. A. López, M. C. Viola, J. C. Pedregosa, J. A. Alonso, M. T.
Fernández-Díaz, Eur. J. Inorg. Chem. 2010, 26, 4110–4120.
[19] C. A. López, M. C. Viola, J. C. Pedregosa, R. C. Mercader,
Mater. Res. Bull. 2010, 45, 1520–1526.
acknowledges the financial support of the Spanish Ministry of
Education on the project MAT2007-60536. We are grateful to ILL
for making all facilities available.
[1]
[2]
[3]
J. L. Simonds, Phys. Today 1995, 48, 26.
M. B. Salamon, M. Jaime, Rev. Mod. Phys. 2001, 73, 583.
K.-I. Kobayashi, T. Kimura, H. Sawada, K. Terakura, Y. Tok-
[20] V. Nassif, R. E. Carbonio, J. A. Alonso, J. Solid State Chem.
1999, 146, 266–270.
[21] E. A. Owen, G. I. Williams, J. Sci. Instrum. 1954, 31, 49.
[22] D. Niebieskikwiat, R. D. Sánchez, A. Caneiro, L. Morales, M.
Vásquez-Mansilla, F. Rivadulla, L. E. Hueso, Phys. Rev. B
2000, 62, 3340–3345.
ura, Nature 1998, 395, 677.
[4]
[5]
[6]
[7]
T. H. Kim, M. Uehara, S.-W. Cheong, S. Lee, Appl. Phys. Lett.
1999, 74, 1737.
M. Garcia-Hernandez, J. L. Martinez, M. J. Martinez-Lope,
M. T. Casais, J. A. Alonso, Phys. Rev. Lett. 2001, 86, 2443.
L. Balcells, J. Navarro, M. Bibes, A. Roig, B. Martíınez, J.
[23] M. Tovar, M. T. Causa, A. Butera, J. Navarro, B. Martínez, J.
Fontcuberta, M. C. G. Passeggi, Phys. Rev.
0244091-5.
B 2002, 66,
[24] K. D. Bowers, J. Owen, Rep. Prog. Phys. 1955, 18, 304–373.
[25] T. F. Rosenbaum, S. A. Carter, J. Solid State Chem. 1990, 88,
94–103.
[26] B. Chattopadhyay, A. Poddar, S. Das, R. Majumder, R. Ran-
ganathan, J. Alloys Compd. 2004, 366, 28–33.
Fontcuberta, Appl. Phys. Lett. 2001, 78, 781.
D. Sánchez, J. A. Alonso, M. García-Hernández, M. J. Martí-
nez-Lope, M. T. Casais, J. L. Martínez, J. Mater. Chem. 2003,
13, 1771–1777.
[8]
H. M. Yang, W. Y. Lee, H. Han, B. W. Lee, C. S. Kim, J. Appl. [27] E. K. Hemery, G. V. M. Williams, H. J. Trodahl, Phys. Rev. B
Phys. 2003, 93, 6987–6989.
2006, 74, 0544231-5.
[28] E. K. Hemery, G. V. M. Williamsa, H. J. Trodahl, J. Magn.
Magn. Mater. 2007, 310, 1958–1960.
[9] M. Retuerto, M. J. Martínez-Lope, M. García-Hernández,
J. A. Alonso, Eur. J. Inorg. Chem. 2009, 8, 1103–1109.
[10] K.-I. Kobayashi, T. Okuda, Y. Tomioka, T. Kimura, Y. Tokura,
J. Magn. Magn. Mater. 2000, 218, 17.
[29] H. J. Trodahl, Phys. Rev. B 1995, 51, 6175–6178.
[30] N. Kaurav, K. K. Choudhary, D. Varshney, J. Phys. Conf. Ser.
2007, 92, 0121281–4.
[31] N. F. Mott, H. Jones, The Theory of the Properties of Metals
and Alloys, Dove Publications, New York, 1958, p. 310.
[32] Y. Yasukawa, J. Lindén, T. S. Chan, R. S. Liu, H. Yamauchi,
M. Karppinen, J. Solid State Chem. 2004, 177, 2655–2662.
[33] A. Maignan, B. Raveau, C. Martin, M. Hervieu, J. Solid State
Chem. 1999, 144, 224–227.
[11] S. Ray, A. Kumar, S. Majumdar, E. V. Sampathkumaran, D. D.
Sarma, J. Phys. Condens. Matter 2001, 13, 607.
[12] H. Chang, M. García-Hernández, M. Retuerto, J. A. Alonso,
Phys. Rev. B 2006, 73, 1044171-6.
[13] J. B. Shi, P. H. Peng, Y. Y. Fan, C. Y. Lee, W. T. Lai, H. C. Ku,
H. Z. Chen, S. L. Young, Phys. B 2003, 329–333, 731–732.
[14] T.-T. Fang, T.-F. Ko, J. Am. Ceram. Soc. 2003, 86, 1453–1455.
[15] G. Y. Liu, G. H. Rao, X. M. Feng, H. F. Yang, Z. W. Ouyang,
W. F. Liu, J. K. Liang, J. Alloys Compd. 2003, 353, 42–47.
[16] Y. Markandeya, K. Suresh, G. Bhikshamaiah, J. Alloys Compd.
2011, 509, 9598–9603.
[34] H. M. Rietveld, J. Appl. Crystallogr. 1969, 2, 65–71.
[35] J. Rodríguez-Carvajal, Phys. B 1993, 192, 55–69.
Received: March 31, 2014
Published Online: June 27, 2014
Eur. J. Inorg. Chem. 2014, 3555–3563
3563
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim