Available via license: CC BY-NC 4.0
Content may be subject to copyright.
ORIGINAL PAPER
Charge Trapping www.ann-phys.org
Suppression of the Verwey Transition by Charge Trapping
Carolin Schmitz-Antoniak,* Detlef Schmitz, Anne Warland, Masih Darbandi,
Soumyajyoti Haldar, Sumanta Bhandary, Biplab Sanyal, Olle Eriksson, and Heiko Wende
The Verwey transition in Fe3O4nanoparticles with a mean diameter of 6.3 nm
is suppressed after capping the particles with a 3.5 nm thick shell of SiO2.By
X-ray absorption spectroscopy and its associated X-ray magnetic circular
dichroism this suppression can be correlated to localized Fe2+states and a
reduced double exchange visible in different site-specific magnetization
behavior in high magnetic fields. The results are discussed in terms of charge
trapping at defects in the Fe3O4/SiO
2interface and the consequent
difficulties in the formation of the common phases of Fe3O4. By comparison
to X-ray absorption spectra of bare Fe3O4nanoparticles in course of the
Verwey transition, particular changes in the spectral shape could be correlated
to changes in the number of unoccupied d states for Fe ions at different lattice
sites. These findings are supported by density functional theory calculations.
1. Introduction
Magnetite (Fe3O4) nanoparticles are the subject of intense
research activities driven both by basic research and their
possible use in various applications. They can be used e.g. as
contrast agents in magnetic resonance imaging[1–3] for hyperther-
mia cancer treatment, targeted drug delivery, rotary shaft seal-
ing, oscillation damping, position sensing,[4] magnetic inks for
jet printing,[5] and to remove heavy metals from wastewater.[6,7] In
addition, due to the half-metallicity with a predicted negative spin
Dr. C. Schmitz-Antoniak
Peter-Gr¨
unberg-Institut (PGI-6)
Forschungszentrum J¨
ulich
52425 J¨
ulich, Germany
E-mail: c.schmitz-antoniak@fz-juelich.de
Dr. D. Schmitz
Helmholtz-Zentrum Berlin f¨
ur Materialien und Energie
Albert-Einstein-Str.15, 12489 Berlin, Germany
Dr. A. Warland, Dr. M. Darbandi, Prof. H. Wende
Fak ul t¨
at f¨
ur Physik and Center for Nanointegration Duisburg-Essen
(CENIDE)
Universit¨
at Duisburg-Essen
Lotharstr. 1, 47048 Duisburg, Germany
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/andp.201700363
C2018 The Authors. Published by WILEY-VCH Verlag GmbH & Co.
KGaA, Weinheim. This is an open access article under the terms of the
Creative Commons Attribution-NonCommercial License, which permits
use, distribution and reproduction in any medium, provided the original
work is properly cited and is not used for commercial purposes.
DOI: 10.1002/andp.201700363
polarization[8,9] in its cubic phase, mag-
netite is an interesting candidate for
spintronics[10] at room temperature and
above.
Magnetite is the Fe oxide with the
highest net magnetic moment and
crystallizes in a cubic inverse spinel
structure. In a simple picture, it consists
of Fe2+ions on octahedral (Oh) lattice
sites, Fe3+on octahedral and Fe3+on
tetrahedral (Td) lattice sites. Although it
is well known that due to hybridization
effects the charges of the ions differ
from the nominal values, we will use
this notation here to distinguish between
the different Fe species and denote them
Fe2+Oh ,Fe
3+Oh,andFe
3+Td , respectively.
Below its Curie temperature of about 860 K, magnetite is a
ferrimagnet due to an oxygen-mediated antiferromagnetic 125°
superexchange between Fe ions on Ohand Tdsites. A ferro-
magnetic order between Fe ions on the Ohsitesisaconse-
quence of superexchange, double exchange (between Fe2+Oh,
and Fe3+Oh ions), and direct exchange. At a temperature of
about 123 K magnetite undergoes a phase transition, the so-
called Verwey transition[11] accompanied by a spin re-orientation
transition at slightly higher temperature[12] characterized by
vanishing magnetocrystalline anisotropy. In the low tempera-
ture phase, magnetite has a monoclinic structure and exhibits
Dr. M. Darbandi
Chemistry faculty
University of Tabriz
Tabr iz, I r an
Dr.S.Haldar
Institute of Theoretical Physics & Astrophysics
University of Kiel
Leibnizstr. 15, 24098 Kiel, Germany
Dr.S.Bhandary
Centre de Physique Th´
eorique (CPHT)
Ecole Polytechnique
91128 Palaiseau cedex, France
Dr.B.Sanyal,Prof.O.Eriksson
Division of Materials Theory
Department of Physics and Astronomy
Uppsala University
Box-516, SE, 75120 Uppsala, Sweden
Prof. O. Eriksson
School of Science and Technology
¨
Orebro University
SE 70182 ¨
Orebro, Sweden
Ann. Phys. (Berlin)2018, 1700363 1700363 (1 of 7) C2018 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.
www.advancedsciencenews.com www.ann-phys.org
charge ordering[13,14] and orbital ordering on the Fe2+sites.[15,16]
In addition it was found that localized electrons are shared
between three Fe ions on Ohsites which can be viewed as
quasiparticles and were named trimerons.[17] In agreement to
this model, a reduction of the exchange interaction between
Fe ions on Ohlattice sites was reported.[18] In the Supporting
Information we present how this weakened double exchange
below the Verwey transition temperature influences the site-
specific spin canting in a 200 nm thick reference film.
In ensembles of nanoparticles, the Verwey transition
is difficult to be observed by diffraction methods, magne-
tometry or resistivity measurements as already discussed
elsewhere.[19] While tunnel spectroscopy[20] and scanning tun-
neling microspectroscopy[21,22] have been used to investigate
the Verwey transition in single nanoparticles, analysis of the
X-ray absorption near-edge structure (XANES) and its associ-
ated magnetic circular dichroism (XMCD) has turned out to
be a powerful tool to monitor the transition in ensembles of
magnetite nanoparticles.[23] Hence it is used in this work to
investigate the influence of a 3.5 nm thick amorphous silica
(SiO2) shell on the properties of magnetite nanoparticles with
a mean diameter of (6.3 ±0.9) nm. In particular, we focus on
the influence on the Verwey transition, spin canting effects and
(temperature-dependent) changes of the XANES that reflects
the unoccupied 3d density of states. The results are discussed
regarding charge trapping at the magnetite/silica interface and
related reduced double exchange.
2. Results
Temperature-dependent XANES and XMCD measurements are
shown in Figure 1 for bare magnetite nanoparticles (upper panel)
and silica capped magnetite nanoparticles (lower panel). For a
discussion of the fine structure the spectra are shown in a re-
duced energy range around the two absorption edges. Spectra
taken in the low temperature regime (4 K, 50 K) are plotted as blue
lines, spectra of the high-temperature phase (100 K, 115 K, 130 K,
150 K) in red. The violet colored line corresponds to an interme-
diate temperature of 75 K. Before turning to the temperature-
dependent changes of XANES and XMCD, the main spectral fea-
tures will be compared. At the L3absorption edge, the XANES
shows two small peaks in the pre-edge region denoted A and B
that can be assigned to contributions of Fe2+Oh. These peaks are
more pronounced and sharper for the silica capped magnetite
nanoparticles compared to the bare ones indicating a higher frac-
tion of Fe2+Oh ions and a stronger localization of the 3d states,
respectively. These ions are also mainly responsible for the first
main peak (C) in the XANES spectra together with small contri-
bution of Fe3+Oh. The second main peak (D) comes from Fe3+Oh
and Fe3+Td ions. From the ratio of the amplitudes of peak C to D,
the fraction of Fe2+Oh can be estimated. Since peak C is smaller
in comparison to peak D for the case of bare magnetite nanopar-
ticles, one can conclude that these nanoparticles are oxidized to-
wards γ-Fe2O3whereas the silica shell prevents the nanoparti-
cles from further oxidation in agreement to previously published
results.[24]
At the L2absorption edge, the first peak (E) originates again
from Fe2+Oh, while the two other peaks (F and G) can be as-
Figure 1. X-ray absorption near-edge structure (XANES) and X-ray mag-
netic circular dichroism (XMCD) for bare magnetite nanoparticles (upper
panel) and silica capped magnetite nanoparticles (lower panel) at various
temperatures. Blue lines refer to low temperatures, red lines to high tem-
peratures, and the violet line refers to an intermediate temperature of 75 K.
signed to Fe3+ions with an additional contribution from Fe2+
ions. Again, the amplitude of peak E and the ratio of peaks F
to G can be used for a rough estimation of the quality of mag-
netite. Particularly, peak F is higher than peak G for magnetite.
By comparison with reference data of non-stoichiometric Fe3-δO4
epitaxial films,[25] one can estimate 0.13 <δ<0.23 for our bare
magnetite nanoparticles and δࣈ0 for the silica capped ones.
The fine structure in the XMCD signal can be correlated to
the different Fe species in magnetite in a similar manner. Again,
the two small peaks (a and b) in the pre-edge region are caused by
contributions from Fe2+Oh. Peak c mainly originated from Fe2+Oh
as well together with a small contribution from Fe3+Oh. The two
other XMCD peaks (d and e) cannot be assigned to one single Fe
ion species in a straight-forward way. The positive values of the
XMCD are mainly caused by the antiferromagnetically aligned
Fe3+Td . However, their contribution is energetically not well sep-
arated from the XMCD of Fe3+Oh. In a simplified picture, peak
d is usually assigned to Fe3+Td ions and peak e to Fe3+Oh ions in
magnetite. But one should keep in mind, that all Fe ions con-
tribute to the XMCD signal in that energy region.
Having a look at the temperature dependence of XANES and
XMCD, it is obvious that there is a change of the intensities
between 50 K and 100 K for the bare magnetite nanoparticles
(Figure 1, upper panel). These changes clearly indicate the pres-
Ann. Phys. (Berlin)2018, 1700363 1700363 (2 of 7) C2018 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.
www.advancedsciencenews.com www.ann-phys.org
Figure 2. Temperature dependent effective spin magnetic moments at the
Fe sites of bare magnetite nanoparticles (black symbols) and silica capped
magnetite nanoparticles (red symbols) determined from XMCD. For all
values, the total uncertainty is about 20%.
ence of the Verwey transition as discussed in the literature.[19,23]
The transition is not visible in the spectra of the silica capped
nanoparticles (Figure 1, lower panel). A sum rule-based analy-
sis of the experimental data yields small average orbital mag-
netic moments for both samples that remain unchanged within
experimental uncertainties. The temperature-dependent effec-
tive spin magnetic moment is shown in Figure 2. For the bare
magnetite nanoparticles, a significant change in the effective spin
magnetic moment marks the phase transition of magnetite.[23]
For the silica capped nanoparticles, the temperature-dependent
effective spin magnetic moment follows a usual Mvs Tbehavior
of a ferro- or ferrimagnet without a clear indication of the phase
transition.
To measure the field-dependent magnetization of the Fe ions
site-specifically, the XMCD asymmetry at three different photon
energies was carefully analyzed for measurements in different
magnetic fields. The first distinct minimum of the XMCD spec-
trum at a photon energy of Eࣈ708.3 eV (denoted c in Figure 1)
is dominated by the magnetic signal from Fe2+Oh ions. Magne-
tization of Fe3+Oh was determined by the XMCD asymmetry of
the second distinct minimum at Eࣈ710.1 eV (denoted e in Fig-
ure 1), and the Fe3+Td mainly contribute to the maximum of the
XMCD asymmetry at E ࣈ709.4 eV (denoted d in Figure 1). Note
that at these energies, there is a (small) additional contribution
to the XMCD signal from the other Fe ions. To analyze the spin
canting, the site-specific magnetization was fitted to a power law
m/mSat =1−bH
n(1)
The exponent was fixed to n=−0.5 as it is expected for a compe-
tition between exchange and Zeeman energies.[26,27] The fitting
parameter bis a measure for the difficulty to saturate the sam-
ple magnetically. The experimental data and best fits are shown
in Figure 3. The extracted values for the parameter bare sum-
marized in Table 1. The normalized site-specific magnetization
values in a magnetic field of 5 T are presented in Table 2. For the
bare magnetite nanoparticles, the parameter bis approximately
the same for all Fe ions. Its sizeable value in high magnetic fields
clearly indicates spin canting effects. For silica capped magnetite
nanoparticles, we find a significant reduction of the canting pa-
rameter. Particularly for the case of Fe2+Oh ions, it is close to zero.
Figure 3. Normalized site-specific magnetization determined from XMCD
for bare magnetite nanoparticles (black symbols) and silica capped mag-
netite nanoparticles (red symbols) measured at 4 K. Solid lines show fits
to the experimental data.
Tabl e 1. Site-specific canting parameter bin T0.5 of the normalized
magnetic-field-dependent XMCD asymmetry for bare and silica capped
magnetite nanoparticles at T=4K.
b(Fe2+Oh)[T0.5]b(Fe3+Td)[T0.5]b(Fe3+Oh)[T0.5]
Bare 0.25 ±0.01 0.23 ±0.01 0.23 ±0.01
SiO2capped 0.09 ±0.01 0.14 ±0.02 0.14 ±0.02
Tabl e 2. Site-specific magnetization in a magnetic field of 5 T normalized
to the saturation value as obtained by fitting experimental data for bare
and silica capped magnetite nanoparticles at T=4K.
m/msat(Fe2+Oh)m/msat (Fe3+Td )m/msat(Fe3+Oh)
Bare (88 ±1) % (90 ±2) % (90 ±1) %
SiO2capped (96 ±1) % (94 ±2) % (94 ±1) %
This corresponds to a magnetization in 5 T close (96%) to the sat-
uration value while for the bare nanoparticles at these sites only
88% of the saturation magnetization could be reached.
3. Discussion
The discussion of the experimental results is split into three
parts. According to the two main observations, we discuss (i)
changes in the spin canting behavior of the silica capped mag-
Ann. Phys. (Berlin)2018, 1700363 1700363 (3 of 7) C2018 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.
www.advancedsciencenews.com www.ann-phys.org
netite nanoparticles compared to the bare ones and (ii) the sup-
pression of the Verwey transition. In addition, the suppression
of the Verwey transition in silica capped nanoparticles was used
to (iii) analyze changes in the electronic structure related to the
Verwey transition in the uncapped species.
3.1. Site-Specific Spin Canting
In order to discuss spin canting by analysis of the magnetic-field-
dependent site-specific magnetization of Fe ions, we used eq. (1)
with an exponent n=−0.5. This equation was derived by Di-
eny et al.[26] for an artificial ferrimagnet and has been adopted to
describe the ferrimagnetism of magnetite.[27,28] Although in the
low-field region there is an influence of magnetic hysteresis and
shape anisotropy originating from the magnetic dipole interac-
tion between the nanoparticles, the experimental data can be fit-
ted in the entire range between 0.1 T and 5 T. The bare magnetite
nanoparticles exhibit a sizeable spin canting that is reflected by
a large parameter b. The magnetization in an external magnetic
field of 5 T is only about 88–90% for all Fe ions. The canting is
slightly stronger for the Fe2+Oh ions in good agreement to results
obtained by M¨
ossbauer spectroscopy.[29] The silica capped mag-
netite nanoparticles can be easier magnetically saturated. The
magnetization in an external magnetic field of 5 T is already 94–
96%. Interestingly, the spin canting is much smaller (close to
zero) for the Fe2+Oh ions compared to the other Fe ions. The sig-
nificantly different canting behavior in silica capped magnetite
nanoparticles clearly indicates that the double exchange coupling
is weakened, since a strong double exchange would prevent the
different canting of Fe2+Oh and Fe3+Oh found here. This weaken-
ing can be explained by a localization of the additional electron
of Fe2+Oh that mediates the double exchange in a simple picture
and was discussed by McQueeney et al.[18]. For bulk material, this
behavior is also found below the Verwey temperature (see Sup-
porting Information).
In our work, this localization is visible in the sharp peaks of
the XANES (XMCD) for the silica capped nanoparticles denoted
A and B (a and b) in Figure 1 that do not change with tempera-
ture. The localization is related to trapped charges by defects at
the magnetite/silica interface. It is well-known that defects in sil-
ica can trap both holes and electrons. For instance, capture and
emission of carriers by defects in silica and silicon/silica inter-
faces is thought to be responsible for the 1/f noise in metal-oxide-
semiconductor devices, negative bias temperature instability, and
dielectric reliability and degradation.[30] The formation of stable
electron trapping defects was confirmed experimentally[31,32] and
by theoretical studies.[33,34] Trapped carriers were also found in
larger magnetite ellipsoid-like particles (100–240 nm) capped by
3–4 nm silica[35] and attributed to localized defect states of high
density in the interface region or intergranular boundaries facili-
tated by the large lattice mismatch between magnetite and silica.
In addition, the analysis of the site-specific canting shows that
one should be careful when estimating the degree of oxidation in
non-stoichiometric Fe3-δO4by comparison with XMCD data in
the literature. For the case of the bare magnetite nanoparticles,
the stronger spin canting at the Fe2+Oh sites yield an overestima-
tion of the oxidation δ, while for the silica capped magnetite it
leads to an underestimation. To avoid this problem, a compari-
son of the spectral shape of the white line is preferred.
3.2. Suppression of the Verwey Transition
The change in the effective spin magnetic moment around the
transition temperature determined from XANES and XMCD
spectra of bare magnetite nanoparticles is caused by a sizeable
contribution of the intra-atomic spin dipole term in the low
temperature phase of magnetite.[23] It is obvious from Figure 1
that the Verwey transition is suppressed in the case of the sil-
ica capped nanoparticles. In addition, Figure 2 shows that the
effective spin magnetic moment is higher for the silica capped
nanoparticles even at temperatures above the phase transition
temperature. Since in the cubic spinel phase of the bare nanopar-
ticles the intra-atomic spin dipole term is negligible, this differ-
ence reflects different spin magnetic moments. A reason for the
reduced spin magnetic moment of the bare magnetite nanopar-
ticles may be a slight oxidation towards a γ-Fe2O3-like oxide at
the surface which reduces the net magnetic moment. This is
supported by a different spectral shape of XANES of the bare
nanoparticles compared to the ones of the capped nanoparticles
as already mentioned before.
At 300 K, the effective spin magnetic moment is slightly
smaller for the silica capped nanoparticles. This may indicate a
reduced Curie temperature in agreement to a reduced double-
exchange coupling between the different Fe ions on octahedral
lattice sites that makes the different spin canting behavior possi-
ble. However, the difference in the effective spin magnetic mo-
ments are quite small and the total uncertainty is about 20%. For
technical limits, it was not possible to measure at higher tem-
peratures close to the Curie temperature in order to prove this
interpretation.
The driving mechanism behind the Verwey transition is not
well understood to date. It is connected to a change in both the
crystallographic and electronic structure. A change in the mag-
netic properties can be understood as a consequence of crys-
tallographic and electronic modifications. Assuming the crys-
tallographic change as a driving force, one may conclude that
the silica shell hampers the formation of a different crystal
structure in the magnetite core. However, the crystallographic
changes connected to the Verwey transition are very small[19]
and silica is quite a flexible material[36] so that this explanation
seems to be unlikely. In addition, no measurable influence of the
silica shell on the crystal structure of magnetite was found by
X-ray diffraction as presented in the Supporting Information.
Despite this, a small contribution due to strain effects evolving
at low temperatures cannot be excluded completely. As evident
effect remains the influence of silica on the electronic proper-
ties: In agreement to the conclusion deduced from spin canting
effects, trapped electrons impede the formation of the metallic
high-temperature phase of magnetite. In this respect, the silica
capped nanoparticles are trapped in a phase that has similar elec-
tronic properties as the low-temperature phase of magnetite. In
other words, although the silica capped magnetite nanoparticles
are free to change their crystallographic structure, the Verwey
transition does not take place because of the modified electronic
structure at the silica/magnetite interface.
Ann. Phys. (Berlin)2018, 1700363 1700363 (4 of 7) C2018 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.
www.advancedsciencenews.com www.ann-phys.org
Figure 4. Relative differences of XANES of silica capped magnetite
nanoparticles with respect to bare magnetite nanoparticles at different
temperatures below the transition (blue lines), slightly above the transi-
tion (red lines) and well above the transition at 300 K (green line). To relate
the particular changes of the spectral features of the XANES, a spectrum
was added (grey line, referring to the grey scale at the right-hand side).
The Verwey transition can also be suppressed for magnetite
nanoparticles covered by ligands as it is widely done to disperse
the particles in a solvent. A charge transfer from Fe towards the
ligand is expected to have the same effects, namely (i) reduced
spin canting, particularly at the Fe2+Oh, (ii) reduced double ex-
change coupling, and (iii) a suppression of the Verwey transition.
The former was already found e.g. for magnetite nanoparticles
capped by oleic acid.[37]
3.3. Changes of the Electronic Structure in Course of the Verwey
Transition
The suppression of the Verwey transition in silica capped
nanoparticles gives the possibility to study the changes of the
electronic structure in uncapped nanoparticles due to the transi-
tioninmoredetail.InFigure 4 we present the relative change in
the XANES of the silica capped nanoparticles with respect to the
bare nanoparticles for several temperatures. The relative changes
were calculated from the ratio between the normalized XANES
intensity of the silica capped nanoparticles and the normalized
XANES intensity of the bare nanoparticles as shown in Figure 1.
An additional offset of +1 avoids division by zero in the pre-edge
energy region:
(XANES) =Icapped
XANES +1
Ibare
XANES +1−1(2)
This analysis is advantageous over a simple comparison of the
XANES of bare magnetite in the low-temperature phase with the
XANES in the high-temperature phase because it minimizes the
influence of thermal electronic excitations to the XANES at dif-
ferent temperatures. Otherwise, it would be impossible to extract
the transition related changes in the electronic structure from ex-
perimental data in a straightforward way.
At temperatures below the transition (blue lines), one can
clearly see that the silica capped magnetite nanoparticles contain
a higher fraction of Fe2+Oh from the positive sign of the differ-
ence in the low energy regime of L3and L2absorption edges,
where mainly Fe2+Oh contribute (spectral features A, B, C at L3
edge and E at L2edge). At higher energies, the difference is nega-
tive corresponding to less Fe3+ions for the capped nanoparticles.
Interestingly, the transition can also be monitored in the differ-
ence (XANES). This can now be used to examine the changes
in the electronic configuration due to the transition in more de-
tail. Above the transition temperature (red lines in Figure 4), the
difference (XANES) is even larger in the energy range that is
unambiguously connected to Fe2+ions (spectral features A and
B) at the expense of intensity at higher energies. This is related to
a small shift of XANES intensity for the bare magnetite nanopar-
ticles towards slightly higher energies in this energy range. How-
ever, this shift is so small, that it cannot be seen directly in the
XANES spectra presented in Figure 1, but it is emphasized in
the plot of (XANES). In a simple picture, this shift of con-
tributions to the XANES intensity can be explained by a larger
number of unoccupied d states for the Fe2+Oh ions states above
the transition. If we stick to the oversimplified model of 2+and
3+valences of Fe ions at octahedral lattice for explanation pur-
poses, the shift will be related to a change from Fe2+Oh and Fe3+Oh
sites in the low temperature phase to Fe2.2+Oh and Fe2.8+Oh in the
high temperature phase. At temperatures well above the transi-
tion temperature (here: 300 K), this spectral change is very pro-
nounced and may indicate that the transition in the bare nanopar-
ticles takes place in a large temperature range. This reduction of
the number of unoccupied d states for the Fe3+ions is visible as
well in the energy range well above spectral feature D at the L3ab-
sorption edge and well above feature G at the L2absorption edge
related to Fe3+Oh ions. However, this effect seems to be smaller
and is not clearly visible due to a larger lifetime broadening of the
experimental spectra.
This finding is in qualitative agreement to band structure cal-
culations for bulk magnetite as presented in Table 3 for the
low temperature (monoclinic) phase and the high temperature
(cubic) phase. The calculated number of unoccupied d states
increases by about 1.3% for Fe2+Oh, while it remains largely con-
stant for Fe3+Td and only slightly decreases for Fe3+Oh. Interest-
ingly, the changes in the experimental data correspond to larger
changes in the occupation of 3d in the order of 10% rather than
1%. A possible reason is the influence of the surface that has not
been taken into account in the calculation, but may play an im-
portant role. A similar underestimation of the changes in course
of the Verwey transition by density functional theory were found
for the change of the intra-atomic spin dipole term in magnetite
nanoparticles.[23]
Tabl e 3. Calculated site-specific number of unoccupied d states in the mon-
oclinic (low temperature) and cubic (high temperature) phase of bulk mag-
netite and the relative change with increasing temperature.
nh3d(Fe2+Oh)nh3d (Fe3+Td )nh3d(Fe3+Oh)
Monoclinica) 3.908 4.092 4.167
Cubic 3.962 4.096 4.147
Relat. difference +1.3% +0.09% –0.16%
a) For detailed numbers of Fe ions on non-equivalent sites, see ref. [23].
Ann. Phys. (Berlin)2018, 1700363 1700363 (5 of 7) C2018 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.
www.advancedsciencenews.com www.ann-phys.org
The interpretation of a suppression of the Verwey transition
by modifications of the electronic structure is in line with the re-
cently published findings that electronic changes may occur at
different temperatures than the structural phase transition which
was found for magnetite[23] and for vanadium oxide, VO2.[38] In
the latter publication, a new monoclinic metallic phase was ob-
served, which may be stabilized by charge doping, similar to our
result.
4. Conclusion
By XANES and XMCD analyses we found that the spin canting
is different for different Fe ions in magnetite nanoparticles. Par-
ticularly the spin canting of Fe2+Oh ions is very sensitive to sur-
face modifications like capping with silica. As supported by the
changes in the XANES of bare magnetite nanoparticles in course
of the Verwey transition and density functional theory, these ions
play also a key role for the Verwey transition which is shown to be
suppressed by silica capping. These observations are explained
by charge-trapping at defects at the magnetite/silica interface
which leads to a reduced double exchange coupling and impedes
the formation of the common phases of magnetite demonstrat-
ing the important role of the electronic structure for the Verwey
transition.
Experimental Section
Nanoparticle Synthesis and Sample Preparation: Magnetite
nanoparticles with a narrow size distribution were synthesized
using a water-in-oil microemulsion technique using Igepal
CO-520 (polyoxyethylene nonylphenyl ether) as surfactant.[29]
Ferrous chloride (FeCl2) and ferric chloride (FeCl3) were used
as precursors for Fe2+and Fe3+respectively. By hydrolysis of
TEOS (tetraethylorthosilicate), the magnetite nanoparticles were
coated with a silica shell in a second step. For both bare and silica
coated magnetite nanoparticles, the surfactants were removed by
several washing steps with butanol, propanol, and ethanol. The
nanoparticles were collected subsequently by centrifugation and
redispersed in ethanol. The size distribution and shell thickness
are determined by transmission electron microscopy images.
The magnetite nanoparticles have a mean diameter of (6.3 ±
0.9) nm and the silica shell has a thickness of about 3.5 nm as
published before.[24] By M¨
ossbauer spectroscopy we could prove
that there is no relevant cation disorder.[29] For our experimental
work presented here, the nanoparticles were brought onto a
naturally oxidized p-doped Si wafer by drop coating. In order to
minimize oxidation of the bare Fe3O4nanoparticles, the particles
were freshly prepared and quickly inserted into the load lock
of the UHV system after the coating procedure. However, the
formation of a thin γ-Fe2O3-like surface layer cannot be avoided.
X-ray Absorption Spectroscopy: XANES and XMCD measure-
ments were carried out at the high-field end station of the vari-
able polarization beamline UE46-PGM1 at the HZB - BESSY II
synchrotron radiation facility in Berlin. Fe spectra were taken at
the L3,2 absorption edges (probing the spin-orbit split 2p3/2 →3d
and 2p1/2 →3d electron transitions) in the photon energy range
of 680 eV ࣘEࣘ780 eV. All spectra were taken in normal X-
ray incidence with the magnetic field parallel or antiparallel to
the wave vector of incoming X-rays. The absorption was detected
in total electron yield mode by measuring the sample drain cur-
rent. For temperature-dependent measurements, the sample was
cooled down to T=4 K in an applied magnetic field of −0.5 T.
The XMCD was measured at several temperatures in an applied
magnetic field of +3 T which was kept constant during heating.
The two samples – bare and silica capped magnetite nanoparti-
cles – were mounted on the same sample holder and measured
in quick succession at each stabilized temperature to ensure an
identical sample history and treatment.
Magnetic-field-dependent measurements were carried out at
T=4Kfrom+5Tto0Tandfrom−5Tto0T.Toobtainsite-
specific magnetization curves we measured full spectra at various
magnetic fields.
Density Functional Theory: Site-specific d charges (and mag-
netic moments as presented elsewhere[23]) in magnetite were cal-
culated using the projector augmented wave method[39,40] as im-
plemented in plane-wave based density functional code VASP[41]
within the generalized gradient approximation (GGA) as given
by Perdew, Burke and Ernzerhof.[42]
For all Fe 3d orbitals the exchange parameter J=0.89 eV[43]
was used and the on-site Coulomb interaction was included by
the parameter U=4.5 eV[44] to account for the strong electron
correlations.
The plane wave cut-off energy was set to 400 eV. 12 ×12 ×12
and 12 ×12 ×4 Monkhorst–Pack k-point grids in the Brillouin
zone were used for cubic and monoclinic structure, respectively.
The geometries were optimized until the force on all each atom
was reduced to 0.1 eV/nm.
Supporting Information
Supporting Information is available from the Wiley Online Library or from
the author.
Acknowledgements
The authors thank the Helmholtz-Zentrum Berlin (HZB) for beamtime al-
location and the HZB staff, in particular E. Schierle and E. Weschke, for
kind support. This work was funded by BMBF (05 ES3XBA/5), DFG (WE
2623/3-1), and the Helmholtz Association (Young Investigator’s Group
Borderline Magnetism under contract no. VH-NG-1031).
Conflict of Interest
The authors declare no conflict of interest.
Keywords
magnetite, nanoparticles, Verwey transition, X-ray absorption spec-
troscopy
Received: September 29, 2017
Revised: December 21, 2017
Published online: February 26, 2018
Ann. Phys. (Berlin)2018, 1700363 1700363 (6 of 7) C2018 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.
www.advancedsciencenews.com www.ann-phys.org
[1] A.K.Gupta,M.Gupta,Biomaterials 2005,26, 3995.
[2] S. Laurent, D. Forge, M. Port, A. Roch, C. Robic, L. Vander Elst, R. N.
Muller, Chem. Rev. 2008,108, 2064.
[3] M. Darbandi, S. Laurent, M. Busch, Z. Li, Y. Yuan, M. Kr¨
uger, M. Farle,
M. Winterer, L. Elst, R. Muller, H. Wende, J. Nanoparticle Res. 2013,
15, 1664.
[4] K. Raj, R. Moskowitz, J. Magn. Magn. Mater. 1990,85, 233.
[5] S. W. Charles, J. Popplewell, Endeavour 1982,6, 153.
[6] L. Carlos, F. S. Garc
´
ıa Einschlag, M. C. Gonz´
alez, D.O. M´
artire in
Waste Water - Treatment Technologies and Recent Analytical Develop-
ments, InTech, Rijeka, 2013, p. 63.
[7] P. Xu, G. Ming Zeng, D. Lian Huang, C. Ling Feng, S. Hu, M. Hua
Zhao, C. Lai, Z. Wei, C. Huang, G. Xin Xie, Z. Feng Liu, Sci. Total
Environ. 2012,424,1.
[8] Z. Zhang, S. Satpathy, Phys. Rev. B 1991,44, 13319.
[9] Y.S.Dedkov,U.R
¨
udiger, G. G¨
untherodt, Phys. Rev. B 2002,65,
064417.
[10] S. Parkin, X. Jiang, C. Kaiser, C. Panchula, K. Roche, M. Samant, Proc.
IEEE 2003,91, 661.
[11] E. J. W. Verwey, Nature 1939,144, 327.
[12] ¨
O. ¨
Ozdemir, D. J. Dunlop, Earth Planetary Sci. Lett. 1999,165, 229239.
[13] J. P. Wright, J. P. Attfield, P. G. Radaelli, Phys.Rev.Lett.2001,87,
266401.
[14] J. P. Wright, J. P. Attfield, P. G. Radaelli, Phys.Rev.B2002,66, 214422.
[15] H.-T. Jeng, G. Y. Guo, D. J. Huang, Phys. Rev. Lett. 2004,93,
156403.
[16] J. Schlappa, C. Sch¨
ußler-Langeheine, C. F. Chang, H. Ott, A. Tanaka,
Z. Hu, M. W. Haverkort, E. Schierle, E. Weschke, G. Kaindl, L. H.
Tjeng, Phys.Rev.Lett.2008,100, 026406.
[17] M. S. Senn, J. P. Wright, J. P. Attfield, Nature 2012,481, 173.
[18] R.J. McQueeney, M. Yethiraj, S. Chang, W. Montfrooij, T. G. Perring,
J. M. Honig, P. Metcalf, Phys.Rev.Lett.2007,99, 246401.
[19] C. Schmitz-Antoniak, D. Schmitz, A. Warland, N. Svechkina, S. Sala-
mon, C. Piamonteze, H. Wende, Sci. Rep. 2016,6, 20897.
[20] Q. Yu, A. Mottaghizadeh, H. Wang, C. Ulysse, A. Zimmers, V. Rebut-
tini, N. Pinna, H. Aubin, Phys. Rev. B 2014,90, 075122.
[21] P. Poddar, T. Fried, G. Markovich, A. Sharoni, D. Katz, T. Wizansky,
O. Millo, Europhys. Lett. 2003,64, 98.
[22] A. Hevroni, M. Bapna, S. Piotrowski, S. A. Majetich, G. Markovich,
Phys. Chem. Lett. 2016,7, 1661.
[23] D. Schmitz, C. Schmitz-Antoniak, A. Warland, M. Darbandi, S. Hal-
dar, S. Bhandary, O. Eriksson, B. Sanyal, H. Wende, Sci. Rep. 2014,4,
5760.
[24] A. Warland, C. Antoniak, M. Darbandi, C. Weis, J. Landers, W. Keune,
H. Wende, Phys.Rev.B2012,85, 235113.
[25] E. Pellegrin, M. Hagelstein, S. Doyle, H. O. Moser, J. Fuchs, D. Vol-
lath, S. Schuppler, M. A. James, S. S. Saxena, L. Niesen, O. Rogojanu,
G. A. Sawatzky, C. Ferrero, M. Borowski, O. Tjernberg, N. B. Brookes,
Phys. Status Solidi B 1999,215, 797.
[26] B. Dieny, D. Givord, J. M. B. Ndjaka, J. Magn. Magn. Mater. 1991,93,
503.
[27] D. T. Margulies, F. T. Parker, M. L. Rudee, F. E. Spada, J. N. Chapman,
P. R. Aitchison, A. E. Berkowitz, Phys.Rev.Lett.1997,79, 5162.
[28] J.-B. Moussy, S. Gota, A. Bataille, M.-J. Guittet, M. Gautier-Soyer, F.
Delille, B. Dieny, F. Ott, T. D. Doan, P. Warwin, P. Bayle-Guillemaud,
C. Gatel, E. Snoek, Phys.Rev.B2004,70, 174448.
[29] M. Darbandi, F. Stromberg, J. Landers, N. Reckers, B. Sanyal, W. Ke-
une, H. Wende, J. Phys. D: Appl. Phys. 2012,45, 195001.
[30] A. V. Kimmel, P. V. Sushko, A. L. Shluger, G. Bersuker, ECS Trans.
2009,19,3.
[31] M. Walters, A. Reisman, J. Electrochem. Soc. 1991,138, 2755.
[32] V. V. Afanasev, A. Stesmans, Phys. Rev. Lett. 1997,78, 2437.
[33] P. V. Sushko, S. Mukhopadhyay, A. M. Stoneham, A. L. Shluger, Mi-
croelectr. Eng. 2005,80, 292.
[34] A.-M. El-Sayed, M. B. Watkins, V. V. Afanasev, A. L. Shluger, Phys. Rev.
B2014,89, 125201.
[35] C.-C. Chang, L. Zhao, M.-K. Wu, J. Appl. Phys. 2010,108, 094105.
[36] S. Yamamoto, K. Sonobe, T. Miyashita, M. Mitsuishi, J. Mater. Chem.
C2015,3, 1286.
[37] A. G. Roca, D. Niznansky, J. Poltierova-Vejpravova, B. Bittova, M.A.
Gonz´
alez-Fern´
andez, C. J. Serna, M. P. Morales, J. Appl. Phys. 2009,
105, 114309.
[38] Z.Tao,T.-R.T.Han,S.D.Mahanti,P.M.Duxbury,F.Yuan,C.-Y.Ruan,
Phys. Rev. Lett. 2012,109, 166406.
[39] P. E. Bl¨
ochl, Phys.Rev.B1994,50, 17953.
[40] G. Kresse, D. Joubert, Phys.Rev.B1999,59, 1758.
[41] G. Kresse, G. J. Furthm¨
uller, Phys. Rev. B 1996,54, 11169.
[42] J. P. Perdew, K. Burke, M. Ernzerhof, Phys.Rev.Lett.1996,77, 3865.
[43] V. I. Anisimov, J. Zaanen, O. K. Andersen, Phys.Rev.B1991,44, 943.
[44] V. I. Anisimov, I. S. Elfimov, N. Hamada, K. Terakura, Phys. Rev. B
1196,54, 4387.
Ann. Phys. (Berlin)2018, 1700363 1700363 (7 of 7) C2018 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.
