Anna-Lena Hansen1,2, Reinhard K Kremer3, Eva M Heppke4, Martin Lerch4, Wolfgang Bensch1. 1. Christian-Albrechts-Universität zu Kiel, Institut für Anorganische Chemie, Max-Eyth-Str. 2, 24118 Kiel, Germany. 2. Institute for Applied Materials-Energy Storage Systems-IAM-ESS, Karlsruhe Institute of Technology-KIT, Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany. 3. Max-Planck-Institut für Festkörperforschung, Heisenbergstraße 1, 70569 Stuttgart, Germany. 4. Technische Universität Berlin, Fakultät II, Institut für Chemie, Straße des 17. Juni 135, 10623 Berlin, Germany.
Abstract
Nanosized samples of the cubic thiospinel FeCr2S4 were synthesized by ball milling of FeS and Cr2S3 precursors followed by a distinct temperature treatment between 500 and 800 °C. Depending on the applied temperature, volume weighted mean (L vol) particle sizes of 56 nm (500 °C), 86 nm (600 °C), and 123 nm (800 °C) were obtained. All samples show a transition into the ferrimagnetic state at a Curie temperature T C of ∼ 167 K only slightly depending on the annealing temperature. Above T C, ferromagnetic spin clusters survive and Curie-Weiss behavior is observed only at T ≫ T C, with T depending on the heat treatments and the external magnetic field applied. Zero-field-cooled and field-cooled magnetic susceptibilities diverge significantly below T C in contrast to what is observed for conventionally solid-state-prepared polycrystalline samples. In the low-temperature region, all samples show a transition into the orbital ordered state at about 9 K, which is more pronounced for the samples heated to higher temperatures. This observation is a clear indication that the cation disorder is very low because a pronounced disorder would suppress this magnetic transition. The unusual magnetic properties of the samples at low temperatures and different external magnetic fields can be clearly related to different factors like structural microstrain and magnetocrystalline anisotropy.
Nanosized samples of the cubic thiospinel FeCr2S4 were synthesized by ball milling of FeS and Cr2S3 precursors followed by a distinct temperature treatment between 500 and 800 °C. Depending on the applied temperature, volume weighted mean (L vol) particle sizes of 56 nm (500 °C), 86 nm (600 °C), and 123 nm (800 °C) were obtained. All samples show a transition into the ferrimagnetic state at a Curie temperature T C of ∼ 167 K only slightly depending on the annealing temperature. Above T C, ferromagnetic spin clusters survive and Curie-Weiss behavior is observed only at T ≫ T C, with T depending on the heat treatments and the external magnetic field applied. Zero-field-cooled and field-cooled magnetic susceptibilities diverge significantly below T C in contrast to what is observed for conventionally solid-state-prepared polycrystalline samples. In the low-temperature region, all samples show a transition into the orbital ordered state at about 9 K, which is more pronounced for the samples heated to higher temperatures. This observation is a clear indication that the cation disorder is very low because a pronounced disorder would suppress this magnetic transition. The unusual magnetic properties of the samples at low temperatures and different external magnetic fields can be clearly related to different factors like structural microstrain and magnetocrystalline anisotropy.
Binary and ternary chromium chalcogenides
exhibit a large variety
of chemical and physical properties explaining the intense research
of these materials. Examples for the diversity of properties are a
large linear negative thermal expansion,[1] the tuneable thermoelectric properties,[2−5] a large magnetocaloric effect,[6] the photocatalytic properties in dye degradation,[7] the tuning of magnetic properties by Li intercalation,[8] ferrimagnetic properties in nanosheets,[9] the application as stable electrode material
in sodium-ion batteries,[10] or the generation
of nanoscale networks.[11] One of the most
intensely investigated ternary chromium chalcogenides is the iron
chromium thiospinel FeCr2S4, which crystallizes
in the normal cubic spinel structure with the general formula AB2X4 (space group: 3m, no. 227). The S2– anions are
arranged in a close-packed face-centered cubic (fcc) lattice, creating
tetrahedral (A) and octahedral (B) voids, which are partially occupied
by the Fe and Cr cations. The Fe2+ cations occupy 1/8 of
the tetrahedral voids and the Cr3+ cation 1/2 of the octahedral
positions. The d-orbitals of the Fe2+ cations are split
into an energetically lower e doublet and an upper t2 triplet. The 3d6 electrons are
distributed over the orbitals according to Hund’s rule, yielding
a high-spin configuration with S = 2, and these cations
are Jahn–Teller active. The 3d3 electrons of the
Cr3+ cations occupy the t2g level with S = 3/2. The magnetic and especially the magnetoelectric
properties of FeCr2S4 attracted particular attention.[12−14] Three main exchange couplings affect the magnetic behavior of FeCr2S4: nearest-neighbor ferromagnetic B–B (Cr–Cr)
interactions, the more distant neighbor antiferromagnetic B–B
interactions, and the 120° A–B superexchange interaction
that is antiferromagnetic. By neutron powder diffraction, FeCr2S4 has been shown to order ferrimagnetically below
∼180 K with Fe and Cr moments aligned collinear antiparallel.[15] Already early on unusual physical properties
like a complex behavior of the Hall resistivity,[16] anisotropic resistivity and magnetoresistance effects due
to spin disorder were reported.[17] Mössbauer
spectroscopic investigations on polycrystalline FeCr2S4 demonstrated an unusual behavior of the electric field gradient
at the Fe2+ site at about 10 K.[18,19] Further, low-temperature Mössbauer spectroscopy studies indicated
a static cooperative Jahn–Teller (JT) distortion (c/a < 1) on the tetrahedral site, stabilizing
the 5Eg ground state of the Fe2+ cations.[20−22] Heat capacity measurements at low temperatures exhibited a λ-like
anomaly around this temperature but the particular shape of this anomaly
depends on the stoichiometry of the sample.[23,24] Colossal magnetoresistance (CMR) behavior was observed with a magnitude
comparable to that found in manganites.[25] However, adverse to the manganites, the CMR is not caused by double
exchange or a JT distortion.[26,27] It was proposed that
by increasing the temperature, long-range magnetic order gradually
breaks down by approaching TC accompanied
by the formation of nanosized spin clusters, which survive well above TC. In an external magnetic field above TC, the rapidly fluctuating spin clusters coalesce
and generate bulk-like ferromagnetism, which leads to the lowering
of electrical resistivity. Evidence for superexchange interactions
between Fe2+ and Cr3+ cations in FeCr2S4 was obtained using resonant inelastic X-ray scattering
(RIXS) investigations.[28]Below the
Curie temperature TC, which
varies between 177[29] and 167 K,[30−32] several unusual features were observed in the temperature-dependent
magnetization curves: A cusp-like anomaly occurs at Tm ∼ 60 K as well as a splitting of the zfc–fc susceptibility data below this temperature.[30,33−35] The divergence of the zfc–fc susceptibilities resembles a glassy behavior, which is exceptional
for a stoichiometric and ordered system. Spin-glass-like states were
also postulated to exist below 60 K due to fluctuating competing exchange
interactions in conjunction with high magnetocrystalline anisotropy
for the Fe2+ moments at the tetrahedral site being strongly
antiferromagnetically coupled with Cr3+ cations.[33] This unusual magnetic behavior was attributed
to changes in the domain structure[30] and
the appearance of noncubic magnetocrystalline anisotropy.[36] The latter was proposed to be the result of
the structure transformation, generating pinning centers for magnetic
domain walls. But neutron scattering investigations demonstrated that
the magnetic structure is of simple Néel type well down to
4 K with μ(Cr) = 2.9 μB and μ(Fe) = 4.2
μB, giving a total magnetic moment μ = 1.6
μB (TC = 180 K).[37] In other neutron diffraction studies, the values
1.92 μB/fu (fu = formula unit)[39] and 1.59 μB/fu were determined.[12] A broadening of Bragg reflections in X-ray powder
diffraction patterns below TC was attributed
to inhomogeneous lattice distortions.[40] Low-temperature high-resolution transmission electron microscopy
on oriented single crystals of FeCr2S4 showed
a cubic to triclinic structural phase transition within crystallographic
domains and an overall symmetry reduction from 3m to F4̅3m.[41] The proposed structural
transformation at Tm ≈ 60 K could
be revealed with ultrasonic measurements on single crystals.[42] An anomaly in the magnetization versus field
curve was observed for μ0H = 5.5
T, which was attributed to the development of a new magnetic phase,
most probably from the incommensurate noncollinear spin structure[43] to the commensurate collinear spin structure.[44] Spin reorientation at 60 K was also reported
from ultrasound studies on dense polycrystalline samples.[45] ac magnetic susceptibility measurements performed
on oriented single crystals showed a pronounced frequency dependence
between 90 and 20 K for both the real and imaginary parts of the ac
susceptibility. The observations made for the field and temperature
dependence were explained by domain wall pinning. Below 60 K, changes
in the domain structure and appearance of pinning centers caused by
structural changes were suggested to lead to spin-glass-like magnetic
anomalies below 60 K.[30]A second
anomaly in the magnetic susceptibility data was observed
at around 9 K and attributed to orbital ordering.[18,19,39,46−48] It was argued that this anomaly occurs due to a unit cell volume
contraction and that a cooperative JT effect and spin–orbit
coupling of the Fe2+ ion compete, leading to spin reorientation
at about 60 K, which gives rise to the onset of short-range orbital
ordering at this temperature.[49] It is noteworthy
that the anomaly at T ∼ 9 K can be suppressed
in polycrystalline specimens by applying large magnetic fields.[50] For single crystals, experimental evidence was
presented that an orbital glass phase is formed at TOO ∼ 9 K.[32,33,35,43,44,48] The Curie temperature and the anomaly at
9 K are only weakly shifted at high external fields up to 9 T. At
lower magnetic fields, the magnetic moments of Fe2+ and
Cr3+ are lower than the spin-only values, and only for
μ0Hext = 5.5 T, the expected
spin-only values are reached. The dielectric permeability shows a
linear correlation with the magnetization of the sample, which is
consistent with the existence of ferroelectric polarization and a
multiferroic ground state below 10 K.[12,13] The critical
exponents, β, γ, and δ, for the paramagnetic to
ferrimagnetic phase transition were determined using different approaches.
β and γ are found close to the mean field theory values,
whereas δ is higher than expected, which has been attributed
to incomplete ferrimagnetic transition and the presence of short-range
ordering above TC.[31]A perpetual problem with FeCr2S4 samples,
crystals and polycrystalline material, is that often the physical
properties depend essentially on the detailed synthesis conditions,
causing hardly controllable minute compositional mismatch and/or positional
disorder of the cations. Consequently, samples synthesized by conventional
high-temperature techniques often showed differences in the magnetization
and zfc–fc curves compared to, e.g., a material
obtained by the field-activated sintering technique (FAST). However,
for the FAST sample, the magnetic anomaly at TOO ∼ 9 K was found absent and the anomaly normally occurring
at Tm ∼ 60 K is shifted to a lower
temperature. In addition, the transition from paramagnetic to the
ferrimagnetic state is less sharp for the FAST sample. All observations
have been attributed to residual structural disorder in the samples
prepared with FAST.[51] To summarize, the
magnetic characteristics of single-crystalline and polycrystalline
FeCr2S4 samples are as follows: (i) a transition
from paramagnetic to the ferrimagnetic state occurs at TC ∼ 170 K; (ii) below the Curie temperature, structural
changes or lattice distortions as well as spin reorientation set in,
leading to magnetic anomalies at about 60 K; and (iii) orbital ordering
is observed at TOO ≈ 9 K.In the past, all investigations were performed on well-crystallized
polycrystalline samples of FeCr2S4 or on single
crystals, which were prepared by chemical vapor deposition. The only
exception is the study of FAST samples reported in ref (51), which showed a different
magnetic behavior compared to single crystals or microcrystalline
samples. A systematic study of the changes of the magnetic behavior
as a function of sizes of coherently scattering domains was not performed
until now on FeCr2S4. For the preparation of
nanoparticles, several synthetic approaches were well established.
Very often solvent-mediated methods are applied, which have the disadvantage
that the surfaces of the crystallites are covered by capping molecules.
To avoid such impurities, mechanochemistry is a promising approach.
While this synthetic approach is applied in many areas of inorganic
and organic chemistry, it has rarely been utilized for the preparation
of ternary sulfides. Hence, we decided using the mechanochemical technique
for the first time for the generation of nanosized FeCr2S4 materials. The main aim of the present study is to
get better insights into the variation of the magnetic properties
in the nanoregime and therefore we intentionally synthesized nanosized
FeCr2S4 polycrystalline samples with particle
size distributions ranging between 10 and 50 nm by reacting ball-milled
mixtures of FeS and Cr2S3 at temperatures between
500 and 800 °C. With increasing annealing temperature, we find
a continuous increase of the particle size, a moderate increase of
saturation magnetic moment, and a variation of magnetocrystalline
anisotropy. We attribute these observations to the variation of the
defect structure with the annealing temperature modifying the approach
to magnetic saturation.
Results and Discussion
Particle Size
The XRPD patterns of the three samples
and the results of the Rietveld refinements assuming the space group 3m (no. 227) (Fe: 8a;
Cr: 16d; S: 32e, 0.383) are displayed in Figure . All samples are phase pure and the cubic
lattice parameters were refined to 9.9832(1) Å (500 °C),
9.9922(1) Å (600 °C), and 9.9923(1) Å (800 °C).
The cubic lattice parameter a of the 500 °C
sample is significantly smaller than values reported in the literature.
There data scattering in a very narrow band around 10 Å[13,51−56] can be found, while in ref (33), the authors obtained a = 9.983 Å
for one sample, mentioning that the lattice parameter depends on the
stoichiometry, but unfortunately the exact composition was not reported.
Even for a sample with crystallite sizes between 40 and 45 nm, a was determined to 10.008(1) Å.[51] We note that the cubic lattice parameter a is significantly reduced at low temperatures, from 9.9894(3) Å
at 180 K to 9.9808(3) Å at 5 K.[49] In
a further study, a decrease from 9.9813(3) Å (175 K) to 9.9756(3)
Å (10 K) was reported.[39]
Figure 1
Results of
the Rietveld refinements of the powder patterns of the
three samples. Bragg indices of some prominent reflections are given.
Black: collected data; red: refined data; and blue: difference curves.
The vertical bars indicate the positions of Bragg reflections used
to calculate the patterns. The insets show an enlarged view of the
311 reflection.
Results of
the Rietveld refinements of the powder patterns of the
three samples. Bragg indices of some prominent reflections are given.
Black: collected data; red: refined data; and blue: difference curves.
The vertical bars indicate the positions of Bragg reflections used
to calculate the patterns. The insets show an enlarged view of the
311 reflection.Increasing the annealing temperature
of our samples has two effects
revealed by the XRPD patterns:The broad bulge below ∼30°
in 2θ decreases significantly in intensity and almost vanishes
for the 800 °C sample, which is assigned to a reduction of incoherently
scattering defects.As demonstrated for the 311 Bragg reflections
(the right side of Figure ), increasing the annealing temperature leads to a linewidth
reduction, indicating an increase in the sizes of coherently scattering
domains.This effect is also clearly
seen in the parameters of fitted log-normal
distribution of the coherently scattering domains (Figure ). The mean particle size value
μ and the variance σ vary continuously with the annealing
temperature. Whereas μ increases by more than a factor of 2,
σ decreases as shown in Figure . The volume weighted mean (Lvol) increases from 56 nm (500 °C) to 86 nm (600 °C)
ending at 123 nm for the sample heated at 800 °C.
Figure 2
Size distribution of
the coherently scattering domains analyzed
from the XRPD patterns according to eq as a function of annealing temperature, as indicated.
The shaded areas indicate log-normal distributions with parameters
given in the inset.
Figure 3
Mean value μ and
variance σ of the log-normal distributions
(eq ) of the coherently
scattering domains as a function of annealing temperature.
Size distribution of
the coherently scattering domains analyzed
from the XRPD patterns according to eq as a function of annealing temperature, as indicated.
The shaded areas indicate log-normal distributions with parameters
given in the inset.Mean value μ and
variance σ of the log-normal distributions
(eq ) of the coherently
scattering domains as a function of annealing temperature.The strain values ε0 obtained from the Rietveld
refinements decrease with the annealing temperature from 7.4 ×
10–4 (500 °C) to 1.8 × 10–4 (600 °C) to finally reach 9 × 10–5 for
the samples annealed at 800 °C. The smaller value for the lattice
parameter of the sample prepared at 500 °C may be caused by the
larger number of nanocrystallites compared to the materials heated
at higher temperatures. In the literature, mainly a lattice expansion
was reported for oxides even for particles larger than 100 nm,[57] while a lattice contraction was observed for
metallic nanoparticles.[58] However, there
are also observations that the lattice contracts with decreasing particle
size, e.g., for α-Fe2O3.[59] This phenomenon was explained by an increased covalency
of the Fe–O bond on the nanoscale. Because the Cr–S
and Fe–S bonds are much more covalent than in oxides, we assume
that the unit cell contraction is based on a similar phenomenon.
Magnetic Properties
Figure displays the inverse magnetic susceptibilities
collected at 1, 2, and 4 T. At sufficiently high temperatures and
significantly above the Curie temperature, the inverse susceptibilities
converge to Curie–Weiss behavior with the same slope for all
three samples similar to that observed by Gibart et al.[29] On approaching the ferrimagnetic Curie temperature,
the susceptibilities become field-dependent with the largest field
splitting seen for the 800 °C sample. We attribute this splitting
to ferrimagnetic spin clusters, which can be saturated with increasing
field, leading to an increase of the inverse susceptibilities with
larger measuring fields.
Figure 4
Inverse molar magnetic susceptibilities of the
three FeCr2S4 samples prepared at different
temperatures: (a) 500
°C, (b) 600 °C, and (c) 800 °C. Blue, red, and black
colored traces refer to data collected at 1, 2, and 4 T, respectively.
The dashed lines indicate Curie–Weiss laws approaching the
high-temperature behavior of the data collected at 4 T.
Inverse molar magnetic susceptibilities of the
three FeCr2S4 samples prepared at different
temperatures: (a) 500
°C, (b) 600 °C, and (c) 800 °C. Blue, red, and black
colored traces refer to data collected at 1, 2, and 4 T, respectively.
The dashed lines indicate Curie–Weiss laws approaching the
high-temperature behavior of the data collected at 4 T.zfc–fc magnetic susceptibilities
measured at 0.1 T. The arrows mark the transition into the orbital
ordered state at ∼9 K.For all samples, a transition to the ferrimagnetic state is determined
at TC ∼ 167 K, in agreement with
the literature where values between 177[29] and 167 K were reported.[30−32] Bifurcation of the zfc–fc data occurs at 120, 110, and 100 K (Figure ) for the samples
prepared at 500, 600, and 800 °C, respectively. Below the bifurcation,
the zfc curves continuously decrease and settle into
a plateau below ∼25 K. A clear step-like drop of the magnetic
susceptibility occurs at ∼9 K for the 800 and 600 °C samples,
which is caused by the long-range orbital ordering as discussed in
the literature. This anomaly is less well developed for the sample
prepared at 500 °C. The decrease of the zfc susceptibility
with decreasing temperature can be attributed to microscopic modifications
in the long-range ordered magnetic structure and/or different magnetocrystalline
anisotropy due to varying amounts of structural defects, affecting
the temperature dependence of the domain dynamics. Notable is the
different behavior of the samples. Whereas for the 800 °C sample,
“melting” of the domain structure on approaching from
low temperatures starts at about 50 K, a sizeable increase of the zfc susceptibility of the 500 °C sample begins only
at about 75 K, indicating more effective domain pinning. Below the
bifurcation and before settling into saturation, the fc susceptibilities pass through shallow maxima, most pronounced for
the 800 °C sample.
Figure 5
zfc–fc magnetic susceptibilities
measured at 0.1 T. The arrows mark the transition into the orbital
ordered state at ∼9 K.
Figure displays
magnetization curves collected at 4 K in external fields up to 9 T.
The saturated moment increases with the preparation temperature from
1.76 μB, 1.80 μB, to 1.83 μB at 9 T for the 500, 600, and 800 °C samples, respectively.
For the collinear ferrimagnetic structure, a saturated magnetic moment
of 1.6 μB was reported from high-magnetic-field experiments.[44] DFT-based calculations predicted saturation
values of 2.0 μB[60] and
1.92 μB.[61] For polycrystalline
compounds, the data reported in the literature are, e.g., 1.55 μB given in ref (61) or ∼1.6 μB reported in refs (12, 37). For the samples investigated here, the
values are noticeably larger but do not fully reach the theoretically
predicted values. The increase of the magnetic moment with the annealing
temperature may be explained by reduced surface spin canting, a smaller
magnetic dead-layer and/or a reduced number of defects on the Cr and
Fe sublattices. One may speculate that the larger values obtained
here are caused by an imperfect collinear magnetic structure, which
may be due to microstrain that is significantly reduced with increasing
annealing temperature (see above). We note that the saturation value
of the sample annealed at 800 °C matches that measured on a single
crystal for the <111> direction (1.84 μB),
which
is the hard magnetic axis.[33]
Figure 6
Magnetization
at 4 K of the three samples prepared at 500, 600,
and 800 °C, as indicated. The inset displays an enlarged view
of the low-field regime.
Magnetization
at 4 K of the three samples prepared at 500, 600,
and 800 °C, as indicated. The inset displays an enlarged view
of the low-field regime.Also apparent from the
magnetization curves is the difference in
the behavior at small fields where domain wall motions are the prevailing
process of magnetization. For the 800 °C sample, magnetization
grows rapidly, whereas for the 500 °C sample, after a pedestal
at low fields, the magnetization increases with about the same slope
as for the other samples. The magnetization behavior reflecting domain
wall motion at low fields correlates with the magnitude of the microstrain
derived from the XRPD measurements. The different magnetization versus
field behavior at low temperatures can be attributed to varying crystalline
anisotropy energies for the different samples.The work W(M) done when magnetizing
to a certain field μ0H is given
by the area between the M–H curve and the magnetization axis. It can be obtained by integration
according to[62]For cubic
crystals, the anisotropy energy is usually described
by the three anisotropy coefficients K (i = 0, 1, 2), which weighs the products of even
powers of the directional cosines measured with respect to the distinguished
magnetization axes system. For a polycrystalline sample with a random
orientation of the crystallites with respect to the external field,
the anisotropy work W can only be taken as a qualitative
measure of the anisotropy energy. Integration of the area of the 4
K magnetization curves, as indicated by the shaded areas in Figure , indicates that
the anisotropy energy (see the inset of Figure ) decreases with increasing preparation temperature
by about 15%, paralleling the decrease of the microstrain for higher
preparation temperatures. In addition, higher preparation temperatures
appear to reduce imperfections in the samples and the crystal anisotropy.
Figure 7
Magnetic
moment per fu as a function of external magnetic field
at 4 K. The shaded areas highlight the integral ∫μ0H dM′. The
inset displays the integral ∫0μ0H(M′) dM′ (black dots) corresponding to the color shaded
areas in the main frame and the microstrain (red dots, see the text
above) as a function of preparation temperatures of the three samples
prepared at the temperatures, as indicated. The dashed and dotted
lines in the inset are guides to the eye.
Magnetic
moment per fu as a function of external magnetic field
at 4 K. The shaded areas highlight the integral ∫μ0H dM′. The
inset displays the integral ∫0μ0H(M′) dM′ (black dots) corresponding to the color shaded
areas in the main frame and the microstrain (red dots, see the text
above) as a function of preparation temperatures of the three samples
prepared at the temperatures, as indicated. The dashed and dotted
lines in the inset are guides to the eye.The magnetization of the samples was recorded at different temperatures
up to T = 175 K somewhat higher than TC (Figure ). As expected, the magnetic moment obtained at the highest magnetic
field decreases with increasing temperature. But even at 175 K, the
magnetization is not linear as expected for a paramagnetic material,
and a magnetic moment of ≈0.75 μB is observed
for all samples, which is again an indication for the presence of
spin clusters, which survived at T > TC.
Figure 8
Magnetization (per fu) of the 500, 600, and 800 °C
samples
recorded at different temperatures, as indicated.
Magnetization (per fu) of the 500, 600, and 800 °C
samples
recorded at different temperatures, as indicated.At sufficiently high fields (see Figure ) in the regime where magnetization processes
by domain rotation (μ0H ≳
2 T) prevail, it has been found that the field dependence of the magnetization
provides information about the type of defects in the sample (“law
of approach”).[62−66] Generally, the approach to saturation is expanded in a polynomial
of powers of the inverse magnetic field according toThe coefficient a1 is interpreted
to
describe inclusions and/or microstress and a2 is due to crystal anisotropy. At temperatures well below
the Curie temperature, the term a0μ0H, which represents the field-induced increase
in the spontaneous magnetization of the domains or forced magnetization,
is usually very small and neglected. A plot of the magnetization of
the three FeCr2S4 samples as a function of inverse
external field (Figure ), 1/μ0H, shows that the magnetizations
at 4 K for very high fields follow very well straight lines with no
significant slope change. Considering the field range from 0.5 to
9 T and fitting the experimental data to eq reveals that the deviations toward lower
fields can be favorably taken care of by additionally considering
the term ∝ 1/(μ0H)2, as displayed in Figure .
Figure 9
High-field magnetization plotted as a function of 1/μ0H. The dashed lines represent extrapolations
of the high-field behavior.
High-field magnetization plotted as a function of 1/μ0H. The dashed lines represent extrapolations
of the high-field behavior.Depending on the preparation temperature, the coefficients a1 and a2 exhibit
different behaviors (see the inset of Figure ). Whereas a1 is almost constant or moderately increasing toward 800 °C, a2 drops by a factor of 4 for the specimen prepared
at 800 °C, indicating that crystal anisotropy decreases with
the preparation temperature. This finding is in line with the results
of the analysis of the anisotropy work experiment. The observation
that a1 is mildly growing for the sample
prepared at 800 °C could possibly attribute to an increase of
sulfur defects with increasing preparation temperature.
Figure 10
Fit of eq to the
high-field magnetizations (per fu) (“law of approach”)
considering the terms ∝ 1/(μ0H) and ∝1/(μ0H)2 (solid lines). The temperature dependence of the coefficients a1 and a2 is displayed
in the inset.
Fit of eq to the
high-field magnetizations (per fu) (“law of approach”)
considering the terms ∝ 1/(μ0H) and ∝1/(μ0H)2 (solid lines). The temperature dependence of the coefficients a1 and a2 is displayed
in the inset.Next, we discuss the coercive
fields (the inset of Figure ) derived from the low-field
hysteresis curves displayed in Figure . The coercive fields range between ∼200
and ∼350 Oe. However, they show a nonuniform behavior with
the annealing temperature and after an initial drop from 500 to 600
°C, the coercive field increases again for the sample prepared
at 800 °C. This parallels the behavior of the parameter a1 of the law-of-approach fits (see the inset
of Figure ), indicating
that the coercive field is primarily determined by inclusions, some
disorder and/or microstrain, and less by anisotropy.
Figure 11
Low-field hysteresis
curves at T = 4.2 K for the
three samples prepared at the temperatures, as indicated. The upper
inset shows the coercive field as a function of preparation temperature.
Low-field hysteresis
curves at T = 4.2 K for the
three samples prepared at the temperatures, as indicated. The upper
inset shows the coercive field as a function of preparation temperature.The dynamics of domain wall motion was studied
by frequency-dependent
magnetic susceptibility experiments. Figure displays the real part of the ac susceptibilities
measured at 1 kHz as a function of temperature between 50 and 200
K. As for the dc susceptibilities, there is no indication of unusual
behavior at or around 60 K. With increasing preparation temperature,
the ac susceptibilities increase. On approaching TC, indicated by a peak and a steep decrease above, the
susceptibilities exhibit a bulge, which we ascribed to dissipation
due to domain reorientation dynamics, similarly seen in the imaginary
part of the ac susceptibilities (inset b of the Figure ).
Figure 12
Real part of the ac
susceptibilities collected at a frequency of
1 kHz.
Figure 13
(a) Temperature derivatives of the real
part, χ′,
of the ac susceptibilities taken at a frequency of 1 kHz. The inset
(c) highlights the steep decrease above TC. The arrows mark the positions of the minima, revealing a continuous
downshift of TC with increasing preparation
temperature. (b) Imaginary part of the ac susceptibility of the 500
°C sample measured with frequencies of 1, 5, and 10 kHz, as indicated.
Real part of the ac
susceptibilities collected at a frequency of
1 kHz.(a) Temperature derivatives of the real
part, χ′,
of the ac susceptibilities taken at a frequency of 1 kHz. The inset
(c) highlights the steep decrease above TC. The arrows mark the positions of the minima, revealing a continuous
downshift of TC with increasing preparation
temperature. (b) Imaginary part of the ac susceptibility of the 500
°C sample measured with frequencies of 1, 5, and 10 kHz, as indicated.The bulge shifts nonlinearly with the preparation
temperature as
is well seen in the temperature derivative of the real part of the
ac susceptibilities (Figure a). The maximum of the bulge is at the lowest temperature
for the sample prepared at 600 °C, correlating with the behavior
of the coercive fields (see the inset of Figure ). The ferrimagnetic transition sharpens
continuously with increasing preparation temperature. For identifying
the minimum of the derivative dχ′/dT with TC, the derivatives also reveal
that by increasing the preparation temperature a steady, however minute
downshift of TC from 168.6(5) to 167.8(5)
and 167.1(3) K takes place for sample preparation temperatures of
500, 600, and 800 °C, respectively.
Conclusions
FeCr2S4 samples synthesized by mechanochemistry
methods followed by subsequent heat treatment led to nanocrystalline
products exhibiting different domain size distributions. Increasing
the preparation temperature generates larger coherently scattering
domains and lower microstrain. dc and ac magnetic susceptibility investigations
evidence a phase transition from paramagnetic to the ferrimagnetic
state at a Curie temperature TC of ∼
167 K, slightly dependent on the preparation temperature. Above TC, the magnetic susceptibility data significantly
deviate from Curie–Weiss behavior, and only at temperatures
high enough, the linear behavior of the inverse magnetization data
is observed. This unusual observation is ascribed to the survival
of easily polarizable spin clusters well above TC. The zfc–fc curves show a bifurcation
at distinct T < TC, with T depending on the preparation temperatures
of the samples. Below this bifurcation, the zfc data
exhibit a sharp drop and reveal a magnetic anomaly at T ∼ 9 K, which is related to the transition into the orbital
ordered state. This transition is well pronounced for the 600 and
800 °C samples and less clearly visible for the 500 °C sample.
It appears that the transition into the orbital ordered state is a
local phenomenon and does not markedly depend on the size of coherently
scattering domains. The analysis of the magnetization behavior of
the samples at low temperatures as a function of magnetic field clearly
reveals a relation between microstrain and different crystalline anisotropy
energies. Such a relation also holds for the coercive fields. Domain
dynamics and associated anomalies in the real parts of the ac susceptibilities
are consistent with these findings.
Experimental Section
The samples were prepared by mechanochemical milling in a high
energy planetary mill (Pulverisette 7, Fritsch, Idar-Oberstein, Germany).
Stoichiometric amounts of the corresponding binary sulfides, FeS and
Cr2S3, were filled in a zirconia grinding beaker
furnished with six zirconia spheres with a diameter of 15 mm and milled
at a rotational speed of 350 rpm for 6 h. To increase the crystallinity,
the powders were subsequently annealed for 2 h in a tube furnace under
flowing H2S gas at temperatures of 500, 600, and 800 °C.
FeS and Cr2S3 were prepared by a solid-state
reaction from the elements, Fe, S, and Cr, respectively, in evacuated
(10–4 mbar) and sealed silica ampoules. Due to the
significant vapor pressure of sulfur, the sealed ampoules were first
slowly heated up to 400 °C in 960 min and held at this temperature
for 1440 min. The temperature then was increased to the final temperature
of 1000 °C in 2880 min and held for 4320 min. After the reaction
was complete, the ampoules were slowly cooled down to room temperature.X-ray diffraction (XRPD) experiments were performed on a PANalytical
Empyrean (Cu Kα1,2), equipped with a PIXcel 1D detector
using 0.1 mm outer diameter glass capillaries (Hilgenberg GmbH). Rietveld
refinements were performed using TOPAS Academic software version 6.0.[67,68] To account for instrumental broadening, the profile function and
axial divergence were determined using a LaB6 standard
sample (NIST 660c) measured under the same conditions. The mean domain
size distribution was obtained by whole powder pattern modeling (WPPM)
assuming a log-normal distribution g(D) for the domain size according to[69,70]where D is the domain size
and μ and σ are the mean and variance, respectively.In addition, the volume weighted average domain size (Lvol) and an upper-limit strain (ε0 from
the strain) were determined.[71,72]Magnetic susceptibility
measurements were performed by the extraction
method in PPMS-9T system in a temperature range from 2 to 300 K and
in magnetic fields up to 9 T.
Authors: R Fichtl; V Tsurkan; P Lunkenheimer; J Hemberger; V Fritsch; H-A Krug von Nidda; E-W Scheidt; A Loidl Journal: Phys Rev Lett Date: 2005-01-18 Impact factor: 9.161
Authors: Joseph Wontcheu; Wolfgang Bensch; Martin Wilkening; Paul Heitjans; Sylvio Indris; Paul Sideris; Clare P Grey; Sergiy Mankovsky; Hubert Ebert Journal: J Am Chem Soc Date: 2007-12-13 Impact factor: 15.419
Authors: G M Kalvius; A Krimmel; O Hartmann; R Wäppling; F E Wagner; F J Litterst; V Tsurkan; A Loidl Journal: J Phys Condens Matter Date: 2010-01-19 Impact factor: 2.333
Authors: V Felea; S Yasin; A Günther; J Deisenhofer; H-A Krug von Nidda; E-W Scheidt; D V Quach; J R Groza; S Zherlitsyn; V Tsurkan; P Lemmens; J Wosnitza; A Loidl Journal: J Phys Condens Matter Date: 2014-11-04 Impact factor: 2.333
Authors: J Bertinshaw; C Ulrich; A Günther; F Schrettle; M Wohlauer; S Krohns; M Reehuis; A J Studer; M Avdeev; D V Quach; J R Groza; V Tsurkan; A Loidl; J Deisenhofer Journal: Sci Rep Date: 2014-08-15 Impact factor: 4.379