Exchange bias and anomalous vertical shift of the hysteresis loops in milled Fe/MnO2 material

Article abstract of DOI:10.1016/j.jmmm.2005.03.014

The present article reports studies on structural and magnetic properties of nanostructured Fe/MnO₂ materials prepared by mechanosynthesis method, with Fe to MnO₂ ratios of 20/80, 50/50 and 60/40. X-ray diffraction patterns indicate that the milled materials have crystalline grain size in the nanoscale region. M?ssbauer spectra of the milled materials suggest the presence of two Fe phases for each sample: a nanocrystalline α-Fe phase with a high degree of disorder/defects and small Fe-oxide particles. The magnetic hysteresis (M(H)) loops, measured at 4.2 K, after the samples were cooled from 300 K in ±10 kOe fields, show unexpected large shifts in both horizontal and vertical directions for the 20/80 sample, while only horizontal shift was detected in the samples with higher Fe concentration. The anomalous vertical shift of the M(H) loop for the 20/80 sample, observed at low cooling field (10 kOe), is being associated with a large contribution from non-collinear magnetic structure of the particles surface. This surface magnetic contribution is strongly influenced by the field cooling magnitude. A simple model is proposed to interpret this result.

Full text of DOI:10.1016/j.jmmm.2005.03.014

ARTICLE IN PRESS
Journal of Magnetism and Magnetic Materials 299 (2006) 11–20
www.elsevier.com/locate/jmmm
Exchange bias and anomalous vertical shift of the hysteresis
loops in milled Fe/MnO material
2
a,
Ã
a
a
a
b
E.C. Passamani , C. Larica , C. Marques , J.R. Proveti , A.Y. Takeuchi ,
c
F.H. Sanchez
a
Departamento de F ı´ sica-CCE, Universidade Federal do Esp ı´ rito Santo, (29060-900), ES, Vit o´ ria, Brazil
b
Centro Brasileiro de Pesquisas F ı´ sicas, Botafogo, (22290-180), Rio de Janeiro, Brazil
Departamento de F ı´ sica-FCE, Universidad Nacional de La Plata, Casilla de Correos 67 (1900), La Plata, Argentina
c
Received 9 January 2005; received in revised form 4 March 2005
Available online 1 April 2005
Abstract
The present article reports studies on structural and magnetic properties of nanostructured Fe/MnO
2
materials
prepared by mechanosynthesis method, with Fe to MnO ratios of 20/80, 50/50 and 60/40. X-ray diffraction patterns
2
¨
indicate that the milled materials have crystalline grain size in the nanoscale region. Mossbauer spectra of the milled
materials suggest the presence of two Fe phases for each sample: a nanocrystalline a-Fe phase with a high degree of
disorder/defects and small Fe-oxide particles. The magnetic hysteresis (MðHÞ) loops, measured at 4.2 K, after the
samples were cooled from 300K in 710 kOe fields, show unexpected large shifts in both horizontal and vertical
directions for the 20/80 sample, while only horizontal shift was detected in the samples with higher Fe concentration. The
anomalous vertical shift of the MðHÞ loop for the 20/80 sample, observed at low cooling field (10 kOe), is being
associated with a large contribution from non-collinear magnetic structure of the particles surface. This surface magnetic
contribution is strongly influenced by the field cooling magnitude. A simple model is proposed to interpret this result.
r 2005 Elsevier B.V. All rights reserved.
PACS: 75.50.Tt; 75.70.Rf; 75.75.+a; 75.70.Cn; 76.80.+y; 75.30.Kz
¨
Keywords: Nanoparticles; Exchange bias effect; Mossbauer effect; Nanomagnetism
1. Introduction
When one or more dimensions of a bulk solid
Ã
are reduced to near or below a certain character-
istic length scale, the associated properties are
modified reflecting the lower dimensionality [1,2].
Corresponding author. Tel.: +55 27 3335 2835;
fax: +55 27 3335 2823.
E-mail address: edson@cce.ufes.br (E.C. Passamani).
0304-8853/$ - see front matter r 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmmm.2005.03.014

ARTICLE IN PRESS
12
E.C. Passamani et al. / Journal of Magnetism and Magnetic Materials 299 (2006) 11–20
In solid materials, the reduction of particle size
may cause an increase of the number of interface
regions, consequently enhancing microscopic in-
terface effects. A suitable technique, which can be
used to reduce crystalline grain sizes as well as to
increase the interface regions in bulk materials, is
the high-energy milling (mechanosynthesis). When
applied to a mixture of powders, this method may
mechanically induce alloying process (mechanical
alloying) at temperatures well below the melting
points of the powder components. For this reason,
the method has largely been employed to produce
metastable phases from systems with positive
cooling (FC) the sample. They have associated
these two effects with a possible interaction
between the FM state of the core with a spin-
glass-like (SGL) state of the g-Fe O shell. The
2
3
vertical shift, measured in core/shell Fe nanopar-
ticles, is also observed in bilayer systems composed
of FM and AFM layers and may reach values of
about 1% of the total magnetization of the FM
contributions [13]. In AFM nanoparticles, Koda-
ma et al. [2] have observed anomalous magnetic
properties, such as large coercivity and loop shifts,
which where attributed to a multisublattice spin
configuration and low coordination at the surface
sites, causing a fundamental change in the
magnetic order throughout the particles. There-
fore, the full comprehension of the loop shifts
(horizontal and vertical) seem to be a complex task
for a systems of magnetic nanoparticles due to
the finite size and surface effects as well as due to
the inter- and intra-particle interactions. Thus, in
the present work, mechanosynthesis technique has
been applied to produce magnetic nanostructure
enthalpy (DH40), like Fe/MnO . In the last
2
decade, it has been shown, by means of scanning
electron microscopy, that the repeated fracture
and cold welding of powder particles, during
milling, generate multilayer-like structures [3].
Recently, it has also been reported that the
mechanosynthesis method can be applied to
produce materials where interfacial phenomena,
for example, the exchange bias (EB) effect, may be
present [4–6].
composites from FM a-Fe (T E1040 K) and
C
The EB phenomenon, which is represented by
the shift of the ferromagnetic (FM) hysteresis loop
away from zero applied field (horizontal shift of
AFM MnO (T E90 K) phases [14]. This adopted
2
N
procedure may lead to an enhancement of the
interfacial magnetic interaction contributions re-
sponsible for the horizontal and vertical shift
effects of the hysteresis loop. X-ray diffractometry
the hysteresis gravity center—H ) due to an
E
unidirectional anisotropy, has been a source of
interest of scientists for over five decades [7–9].
Review articles on EB effect can be found in
references such as [10,11].
In general, the microscopic exchange interaction
at the contact interface between FM and anti-
ferromagnetic (AFM) phases of a material, after it
has been cooled in a magnetic field through the
¨
(XRD), Mossbauer spectroscopy and magnetiza-
tion techniques have been used to investigate the
features of the nanostructured samples and their
interface properties.
2. Experimental
Neel temperature (T ) of the AFM phase, is
´
N
believed to be responsible for the horizontal
macroscopic shift phenomenon observed in
MðHÞ curves. Besides systems with FM/AFM
interfaces, EB effect has also been observed in
samples involving ferrimagnet (FI) (e.g. FI/FM or
FI/AFM) [12] and spin-glass (SG) (e.g. AFM/SG,
FM/SG, FI/SG) [8,9].
Nanostructured (x)Fe:(y)MnO2 (x=y ¼ 20=80,
50/50 and 60/40) materials were prepared by
mechanosynthesis from the mixture of high-purity
powders of Fe and MnO (99.99%). The initial
2
powder mixtures were sealed in a hard steel vial
under high-purity Ar atmosphere (5N5), to pre-
vent Fe oxidation (other than from solid reaction
between the powders). The set was clamped in a
commercial vibrating frame equipment. A massive
cylinder, made of the same material as the vial,
occupying 63% of total internal volume space, was
used as the milling tool. In order to prevent
In core/shell nanoparticles in which the FM Fe
core is covered by a g-Fe O layer, Zheng et al. [8]
2
3
have reported a large EB value (H ) at 2 K and
also a small vertical shift of the hysteresis loop,
both effects having been observed after field-
E

ARTICLE IN PRESS
E.C. Passamani et al. / Journal of Magnetism and Magnetic Materials 299 (2006) 11–20
13
contamination from the milling tools, small
amount of Fe and MnO powder mixture with
defined composition was first milled, since these
powders will partially cover the internal vial
surface as well as the milling tool surfaces. After
that, the excess of powder was removed from the
vial and new portions of Fe and MnO powders, at
2
desired compositions, were added to the vial,
keeping the mass ratio between the solid cylinder
both equipments, being 300 K/min for the VSM
and 10 K/min for the SQUID. From the FC
regime, M(H) loops were obtained at 4.2 K after
cooling the sample from 300 K in the presence of
fields of +10 kOe or ꢀ10 kOe relative to the
measuring field, for the VSM experiments and in
field of 50 kOe for the SQUID ones. For high Fe
contents, the 10 kOe was enough to saturate the Fe
magnetization, while for low Fe content sample
20/80 no saturation was observed in magnetic
fields up to 50 kOe.
2
(
milling tool) to sample at about 20:1. The milling
process extended for periods up to 170 h, depend-
ing on the Fe compositions. The samples will
be designed in the text as 20/80 sample for
2
5
0(Fe):80(MnO ) milled for 100 h, as 50/50 for
2
3. Results and discussion
0(Fe):50(MnO ) milled for 40 h and as 60/40
2
for the 60(Fe):40(MnO ) milled for 170 h. More
2
3.1. Structural characterization
details of the milling tools and experimental
procedures have already been reported by Larica
et al. [15].
The crystalline structures of the milled materials
were analyzed by XRD. The history of the milling
process will not be discussed in this article and,
therefore, the investigation will only focus on the
20/80, 50/50 and 60/40 samples, obtained at the
correspondent longest milling time (100, 40 and
170 h, respectively). Fig. 1 shows the XRD
patterns for these samples. Also, the XRD pattern
of a non-milled powder mixture 20/80 was added
for comparison. The non-milled material displays,
as expected for the mixture of powders, the
The influence of the milling process on the
powders has been investigated from materials
collected at pre-defined milling times and will be
published elsewhere. Powder manipulation was
always done inside a glove box under high-purity
argon atmosphere to avoid undesirable Fe oxida-
tion. X-ray diffraction patterns were obtained at
room temperature using a Rigaku diffractometer
with a Cu-K radiation, with the powders being
a
distributed randomly and pressed on a glass
ssbauer spectra (MS) of
tetragonal MnO and the BCC-Fe narrow Bragg
2
sample-holder. The Mo
¨
reflection lines. No Bragg peak shifts have been
observed for milled samples (50/50 and 60/40 in
Fig. 1) if one compares their detectable line
positions with those of the non-milled sample,
despite their relative intensities had changed,
probably due to texture effect produced in the
pressed powder samples, prepared for X-ray
analysis. The latter argumentation may be sup-
ported by the fact that milling time shorter than
170 h in the 60/40 powders does not induced
texture effect, as suggested by their X-ray patterns
(figure not shown). Concerning the milling pro-
cess, the main effect observed in the XRD patterns
of the milled materials is the broadening of the
BCC Bragg peaks, which can be associated with
grain refinement as well as to the increase of lattice
strain. The Fe grain size values were estimated
from the main BCC Bragg peaks (2y ꢁ 44:41)
inverse line width, using the Scherrer formula. The
the milled powders were recorded in the conven-
tional transmission geometry at 300 K, at 11 and
5
7
at 4.2 K, using a 25 mCi Co in Rh matrix
radioactive source, moving in the velocity trian-
¨
gular mode. The Mossbauer measurement at 4.2 K
was done in a He-bath cryostat, while those
experiments at 11 K have been performed in an
APD closed-cycle refrigerator. During the 11 K
¨
measurements, the Mossbauer radioactive source
was kept at 300 K, while the absorber was at 11 K.
The MS are shown without the corrections for the
second-order Doppler-shift. Magnetic properties
were investigated with a vibrating sample magnet-
ometer (VSM) in applied fields (H ) up to 10 kOe
0
and over the temperature range from 4.2 to 300 K
and also using a Quantum design SQUID mag-
netometer at 5 and at 40 K with maximum applied
field of 50 kOe. The cooling rates are different in

ARTICLE IN PRESS
14
E.C. Passamani et al. / Journal of Magnetism and Magnetic Materials 299 (2006) 11–20
the pure magnetic nuclear Zeeman effect plus a
α
small paramagnetic (PM) fraction. Therefore, it
was analyzed using one magnetic component
having the a-Fe hyperfine parameters, with broad
lines due to its nanoscale grain size, and two PM
doublets associated with small Fe clusters and/or
Fe-oxide nanoparticles (FeO) in superparamag-
netic state. Fig. 2(a) displays for the same sample
the spectrum recorded at 11 K, which was fitted
with two magnetic components: one due to the
a-Fe phase, seen in the 300 K spectrum, and a
broad magnetic hyperfine field distribution com-
ponent attributed to small Fe and/or Fe oxide
particles, now in a blocked magnetic state.
Fig. 2(c) and (d) show the MS for the 50/50
sample obtained at 11 and 300 K, respectively.
They have the similar features as observed for the
60/40 sample, except for the nanocrystalline a-Fe
phase fraction that is 70% for 60/40 sample and
2
3
2% for the 50/50 sample. MS obtained at 11 and
00 K for the milled 20/80 sample are displayed in
θ
Fig. 2(e) and (f), respectively. The spectra are
mainly composed of a broad doublet associated
with the (Mn,Fe) O phase, which is also con-
Fig. 1. Room temperature X-ray diffraction patterns of the
samples: 60(Fe):40(MnO ) milled for 170 h (a), 50(Fe):50(MnO
milled for 40 h (b), 20(Fe):80(MnO ) milled for 100 h (c) and the
non-milled 20(Fe):80(MnO ) (d).
2
2
)
2
2
3
2
firmed by the corresponding XRD pattern. A
weak magnetic component, with hyperfine para-
meter values compatible with that of a-Fe phase,
observed at 300 K (indicated by arrows in
Fig. 2(f)), should also be seen in the low-
temperature spectrum, but it was not visually
observed possibly due to the noisy background
(Fig. 2(e)). Important feature, that should be noted
in the spectra displayed in Fig. 2(e) and (f), is that
obtained grain size values were not corrected for
instrumental broadening, which means that they
may be underestimated. It is important to say that
there is a considerable reduction of the BCC-Fe
grain sizes with increase of milling time, i.e., from
2
1
00 nm (non-milled) to as low as 9 nm (milled for
70 h). In addition, the XRD patterns of the 60/40
¨
the Mossbauer line shape becomes asymmetric at
and 50/50 samples have peaks related to an Fe-
oxide (FeO) overlapped with tetragonal MnO and
11 K (Fig. 2(e)). In addition, it is observed that the
quadrupole splitting value reduces and the lines of
the spectrum become narrower at 11 K. These
effects may be an indication of a partial magnetic
ordering of the sample at that temperature,
associated with a fraction of the (Mn,Fe) O
2
BCC-Fe phase lines. For the 20/80 sample, the
XRD pattern shows a set of peaks associated with
the (Mn,Fe) O phase and some asymmetry
2
3
around the angular position of 44.41, which may
indicate the presence of a nanocrystalline BCC-Fe
phase, as also observed in MS discussed below.
MS for the 60/40, 50/50 and 20/80 samples are
presented in Fig. 2(a)–(f), at temperatures of 11
and 300 K. A 4.2 K, spectrum for the 20/80 sample
is also displayed in this figure as (g). The 300 K
spectrum for the 60/40 sample is displayed in
Fig. 2(b). It has basically six symmetric lines due to
2
3
superparamagnetic particles undergoing a transi-
tion to a magnetically blocked state, since this
oxide phase has an expected PM to FI transition at
the range between 35 and 60 K, depending on Fe
content [14]. The 4.2 K spectrum of the 20/80
sample presents a well-defined six absorption lines,
indicating that the (Mn,Fe) O nanoparticles are
2
3
now at a blocked magnetic state. The fit of this

ARTICLE IN PRESS
E.C. Passamani et al. / Journal of Magnetism and Magnetic Materials 299 (2006) 11–20
15
5
7
Fig. 2. 11 and 300 K Fe Mo
¨
ssbauer spectra of the samples: 60(Fe):40(MnO
0 h (c and d) and 20(Fe):80(MnO ) milled for 100 h (e and f) samples. The 4.2 K spectrum of the sample 20(Fe):80(MnO
00 h is also shown in (g). The full lines through the experimental points are the fit results of the spectra. The subspectra are shown as
2
) milled for 170 h (a and b), 50(Fe):50(MnO
2
) milled for
4
1
2
2
) milled for
dot-lines. The sextet associated with a-Fe is only shown for the spectrum (a), but it was also used in the fits of the spectra displayed in
b, c, d and f.
spectrum was done using two sextets, which can
be attributed to the two Fe sites often found in
Fe-oxides.
¨
magnetic state at 11 K, as indicated by Mossbauer
results. Further Mossbauer investigation under
¨
external field on this sample is in progress.
¨
In summary, the XRD and Mossbauer data can
be organized within two groups. The first one is
related to the 60/40 and 50/50 samples that are
composed by nanocrystalline AFM MnO2 and
FM a-Fe phases. In addition to these phases, the
3.2. Magnetic properties
3.2.1. Group one: 50/50 and 60/40 samples
Magnetic hysteresis (MðHÞ) loops for the 60/40
and 50/50 samples, measured at 4.2 K after FC
6
0/40 and 50/50 samples also have a contribution
from an Fe-oxide phase (with an ordering tem-
perature lower than 300 K). The second group
is formed by nanocrystalline FI (Mn,Fe) O and
from 300 K in a 10 kOe external field (H ), are
0
shown in Fig. 3(a) and (b), respectively. The
interesting features of these MðHÞ curves are: (i)
the hysteresis loops are conventional-like shape
(symmetric with respect to their gravity centers),
and tends toward the saturation regime for fields
large than 10 kOe; (ii) no clear vertical shift
is observed in the MðHÞ loops for the 60/40
sample, while a small vertical shift is apparent
in the loop for the 50/50 sample, i.e.
2
3
a-Fe phases for the 20/80 sample. The former
occupied nearly 96% of total Fe volume fraction
of the 20/80 sample. Besides that, the milled 20/80
sample results may suggest a broad distribution of
magnetic particles, where the small particles are in
superparamagnetic state at 11 K and the large
particles can be related to a-Fe (magnetic lines
indicated by arrows in Fig. 2(f)) and/or due to the
max
0
max
0
jMðþH ÞjajMðꢀH Þj (see in Fig. 3(b) the
(
Mn,Fe) O particles undergoing to a blocked
2
DM value, which corresponds to twice the vertical
3

ARTICLE IN PRESS
16
E.C. Passamani et al. / Journal of Magnetism and Magnetic Materials 299 (2006) 11–20
shift of magnetization value); (iii) there is a
significant horizontal shift of the hysteresis loops
samples (see also values in Table 1): the H value is
E
larger for the sample with lower Fe composition
(50/50), while HC is larger for the sample with
higher Fe content (60/40). However, in both
(see Fig. 3(a) and (b)), which is associated with the
EB effect (H ). In particular, from the analysis of
E
Fig. 3, the EB field (H ) and the coercive field (H )
E
samples the H values are larger in the FC mode,
C
C
values, defined in Fig. 3(a) and (b), assume
opposite Fe dependence for the 60/40 and 50/50
relative to the correspondent ZFC one. Particu-
larly, the horizontal loop shift (EB effect) of the
60/40 sample can basically be attributed to an
uniaxial magnetic anisotropy originated from
magnetic interface interaction between AFM
MnO and FM a-Fe. For the 50/50 sample, in
2
addition to the above EB contribution, the EB
effect may be also related to a contribution from
the large Fe-oxide fraction, i.e. it can be magnetic
coupled with FM a-Fe phase and/or the Fe-oxide
phase may yield its own EB contribution due to
the uncompensated spins at the particle surfaces
(EB effect from AFM/SG). The apparent absence
of the vertical shift in the 60/40 sample may be
explained taking into account the large contribu-
tion of the FM a-Fe phase, which may mask the
interface contribution to the net magnetization.
The vertical shift of 50/50 sample is about 3% of
the magnetization value, similar to that reported
by Zheng et al. [8].
(
a)
3
.2.2. Group two: 20/80 sample
Special features were observed in the milled
0/80 sample. For this sample, the temperature-
2
dependent magnetization curves (MðTÞ) measured
in ZFC and FC processes, from 4.2 to 300 K in a
0
.1 kOe probe field, are displayed in Fig. 4. The
ZFC curve presents a very broad peak at about
0 K, while a relative sharp peak is observed at
(
b)
4
about 30 K for the FC mode. Moreover, the FC
curve has systematically higher magnetization
values than that of ZFC for temperature lower
Fig. 3. Ten kilo-oersted FC magnetic hysteresis loops obtained
at 4.2 K for the samples: 60(Fe):40(MnO ) milled for 170 h (a)
and 50(Fe):50(MnO
2
2
) milled for 40 h (b).
Table 1
Summary of magnetic properties obtained in field-cooled curves for the milled Fe
x
(MnO
2
)
y
samples obtained at +10 kOe
C
(kOe) H (kOe)
max
max
Samples x=y
MðþH0 Þ (emu/g)
MðꢀH0 Þ (emu/g)
M
R
(emu/g)
H
E
6
5
2
0/40
0/50
0/80
95.1
15.4
11.2
95.0
14.2
6.6
41.1
5.2
0.36
0.47
0.96
0.82
0.77
1.96
1.3
The M
MðHÞ loops. The H
maximum applied field values (positive and negative).
R C
and H values are determined from the hysteresis gravity center taking into account the horizontal and vertical shifts of the
max
0
E
field corresponds to the horizontal hysteresis gravity center shift. MðꢂH Þ values are determined for the

ARTICLE IN PRESS
E.C. Passamani et al. / Journal of Magnetism and Magnetic Materials 299 (2006) 11–20
17
Fig. 4. ZFC and FC magnetization curves for the sample
0(Fe):80(MnO ) milled for 100 h, measured with a 0.1 kOe
Fig. 5. Fifty kilo-oersted FC magnetic hysteresis loops
obtained at 5 and 40 K for the sample 20(Fe):80(MnO ) milled
for 100 h. The close lozenges correspond to the loop obtained at
2
2
2
field in a 4.2–300 K temperature range. The arrows indicate the
thermal cycle direction.
5
4
K, while open squares correspond to the loop obtained at
0 K.
than 300 K (Fig. 4), which may be an indication
for the presence of a magnetically ordered phase,
consistent with the weak magnetic component
observed in the MS background recorded at
reported in the literature [2,16]. Thus, combining
the MðTÞ and MðHÞ results for the 20/80 sample,
it can be inferred that there are evidences for the
presence of interacting magnetic particles and/or
of uncompensated spin-frozen state with magnetic
behavior similar to that of a SGL state assumed to
occur at the particle surfaces.
3
00 K. Therefore, the MðTÞ results seem to
indicate a behavior typical of a particle ensemble
having a broad distribution of sizes. This assump-
tion is supported by the MS of this sample that
shows basically two components (PM and mag-
netic) at 300 K. Moreover, the changes in the peak
shapes of the DC magnetization measurements in
ZFC and FC modes and the roughly conventional
MðTÞ shape of Fig. 4 may give indication for inter-
particle interactions with a blocking temperature
at about 30 K, as also suggested by Zheng et al. [8],
for nanoparticles system.
Magnetic hysteresis loops for the 20/80 sample,
measured at 4.2 K after ZFC and FC in a +10 kOe
applied field, cooled from 300 K, are shown in
Fig. 6. Now, not only the FC but also the ZFC
MðHÞ loop shows that the magnetization value
remains unsaturated in fields up to 10 kOe, an
effect already discussed above. In the present case,
a SGL state may be caused by the inter-particle
interactions (clusters-glass-like state) and/or un-
compensated magnetic moments at the particle
surfaces. The latter is responsible for the EB effect,
also observed in Fe-core and g-Fe O -shell system
Magnetic hysteresis loops for the 20/80 sample,
measured at 5 and at 40 K after FC in a +50 kOe
field, from 300 K, are shown in Fig. 5. The 40 K
MðHÞ loop (opened square) displays no hysteresis,
i.e. it has a nearly zero coercive field and reversible
magnetization values in whole field range (see the
figure inset, in the 710 kOe range). On the other
hand, the 5 K MðHÞ loop shows unsaturated
behavior and magnetization irreversibility even
for field values near 50 kOe (full lozenges), an
effect characteristic of non-collinear magnetic
structures found in low-dimensional systems, as
2
3
[8]. There are other important features in Fig. 6
that deserve to be stressed: (i) the HC field value
has significantly increased from 0.7 kOe for the
ZFC loop to 1.96 kOe for the FC one; (ii) the FC
MðHÞ loops obtained with opposite cooling field
directions (+ and ꢀ relative to the measuring
field) are rather like mirror images of each other
relative to the ZFC loop; (iii) the FC loops are
asymmetrical (the areas in the first and third

ARTICLE IN PRESS
18
E.C. Passamani et al. / Journal of Magnetism and Magnetic Materials 299 (2006) 11–20
In general, the vertical shift of the hysteresis
loop is often observed in bilayer of FM/AFM [6,9]
as well as in (FM core)/(SGL shell) magnetic
nanoparticles [8]. However, in both systems the
magnitude of the vertical shift is only few
percentage of the material total magnetization
[
8,13], while it is about 20% in the present 20/80
case. Moreover, for FM/AFM bilayer systems the
horizontal and vertical shifts were attributed to be
related to each other. However, not too large
vertical shift was reported even for cases were very
high EB values occurred [8,13]. In the case of core/
shell nanoparticles, the reported high EB value of
6.3 kOe was associated with magnetic exchange
interaction among large interface areas of rever-
sible and frozen spins, the latter in a SGL state.
Therefore, the anomalous vertical shift observed in
the 20/80 sample may not be simply explained
under the same assumption [8,13]. On the other
hand, anomalous magnetic properties, such as
large coercivity and loop shift, have been reported
for AFM nanoparticles systems of NiO [2]. To
explain their results, the authors have assumed a
broad distribution of reversal fields as well as
different surface and bulk anisotropies. It is
important to point out that Kodama et al. [2] do
not account for the vertical loop shift, although
their experimental data (shown in Fig. 1 of their
paper) seems to display the vertical shift effect.
Fig. 6. ZFC and 10 kOe FC magnetic hysteresis loops obtained
at 4.2 K for the sample 20(Fe):80(MnO ) milled for 100 h. The
2
open circles correspond to the +10 kOe FC experiment, while
open triangles correspond to the ꢀ10 kOe FC experiment. ZFC
experiment is represented by the closed squares.
quadrant are different), an effect that suggests the
presence of a system made of interacting magnetic
layers; (iv) the FC loops present a high EB effect
(
the H field value is smaller than the values
H ꢃ1 kOe) for this 20/80 sample case. However,
E
E
reported for core/shell Fe nanoparticles [8]; finally,
(
v) there is an anomalous vertical shift of the FC
MðHÞ loop observed for the 20/80 sample, i.e. the
magnitude of the magnetization in positive field
max
direction (MðþH Þ), relative to the cooling field,
0
is larger than that for measuring field in opposite
Also, their H (HE ¼ loopshift) fields seem not to
C
max
0
low (Fig. 6—minor loop) and high (Fig. 5—close
direction (MðꢀH Þ). For the experiments, with
depend on the cooling field magnitude. In con-
trast, in our case, the vertical shift and H field are
E
to major loop) cooling fields, the H field and the
E
drastically reduced at 50 kOe (DM near zero).
Thus, in principle, the vertical shift effect seems to
have different mechanism for cases where the spins
interaction is FM/AFM or FM/SGL or in AFM
(multisublattices) nanoparticles and in the FI/SGL
nanoparticles (present case—anomalous vertical
loop shift).
Two models (A and B) may explain the present
case (20/80 sample). In both models, however, the
magnetization of the particles surface at 4.2 K are
in an uncompensated magnetic state (low coordi-
nation at the surface sites), as indicated by our
MðHÞ results and in accordance with those
reported for nanoparticle systems [2,8,9].
max
max
vertical shifts (DM ¼ jMðþH0 Þj ꢀ jMðꢀH0 Þj)
are different, with DM value smaller for the higher
cooling field (DM near null at 50 kOe). Anyhow,
the loop vertical shift indicates that part of the
magnetic moments frozen in the FC direction
cannot be reversed by the applied field. In
addition, the HE field is also a FC dependent,
similar to those results reported by Bianco et al. [9]
for a system composed of Fe nanoparticles
embedded in an Fe oxide matrix. Thus, the
reduction of DM and HE field with increasing
the FC magnitude may be associated with an
increase of the FI domains at a cost of the SGL
state (the non-collinear state reduces with increas-
ing FC magnitude).
In (A), it may assume the model proposed by
Kodama et al. [2] for AFM nanoparticles, which

ARTICLE IN PRESS
E.C. Passamani et al. / Journal of Magnetism and Magnetic Materials 299 (2006) 11–20
19
seems to be also valid for our FI (Mn,Fe) O3
2
surface roughness due to the grain refinement. In
model (B), the surface spins configuration of the
FI (Mn,Fe) O nanoparticles is substantially
nanoparticles. There, the large loop shift is
associated with a multisublattice spin configura-
tion (distribution of reversal fields), instead of two
sublattices often found in bulk AFM (FI) materi-
als, and with low spin coordination at the surface
sites. Also, they suggested that a relatively weak
magnetic coupling between the sublattices (dis-
tribution of reversal fields throughout the parti-
cles) allows a variety of reversal paths for the spins
upon cycling the applied field, resulting in large
coercivity and loop shifts. However, for AFM NiO
nanoparticles, the loop vertical shift seems to be
smaller than our case. In summary, model (A)
suggests that FI or AFM materials may account
for the anomalous loop shifts.
2
3
changed with increasing cooling field magnitude
as suggested by the DM and H field behaviors.
E
4. Conclusion
In summary, mechanosynthesis technique has
been applied to produce exchange biased
¨
Fe/MnO samples. X-ray diffraction, Mossbauer
2
spectroscopy and bulk magnetization measure-
ments were employed to study the structural and
magnetic properties of milled Fe:MnO materials.
2
X-ray diffraction experiments indicate that the
grain sizes of the milled powder reach nanoscale
¨
region. Mossbauer spectroscopy suggests that,
¨
Based on the Mossbauer and magnetization
MðTÞ results, that indicate a broad distribution of
particle sizes, an alternative model (B) may be
proposed. In this case, small particles are assumed
to be pinned to the surface of the larger particles,
at low temperatures, when an uniaxial anisotropy
is induced by the applied field in the FC process.
Thus, the magnetization of these small particles
would be harder to reverse by the external field
depending on the Fe concentration, one may have
basically particles with a-Fe features (60/40 and
50/50), while the major fraction of the spectrum,
observed in the low Fe content (20/80) sample,
originates from a broad distribution of particle
sizes for the (Mn,Fe) O phase in a PM state at
2
3
300 K. In all samples, horizontal shifts of the field
cooled MðHÞ loops were observed, indicating that
the milled materials present EB effect. Further-
more, for the 20/80 sample, a large vertical shift of
the hysteresis loop was observed. This latter effect,
clearly observed in low cooling field (10 kOe)
process, is being associated with a large contribu-
tion of non-collinear magnetic structure at the
particle surfaces (thick layer). This surface mag-
netic contribution is strongly influenced by the FC
magnitude.
(
due to its large SGL character) than those of the
larger particles core, which are easily reversible.
Furthermore, the small particles are practically in
a SGL state, while in the large particles the FI core
contributes predominantly to the magnetization.
Under the random arrangement of the magnetiza-
tion in the ZFC experiments, the MðHÞ loop
shows a conventional feature, as expected for
systems of interacting magnetic particles without
any uniaxial anisotropy. The FC experiment
would produce a preferential orientation of the
uncompensated surface magnetization in the
direction of cooling field. Under this situation,
the net magnetization of the hard component
Acknowledgments
(
small particles and larger particles surface) would
We acknowledge the financial support of UFES
and CNPq for this work.
not be zero and its magnetic contribution would
increase the sample net magnetization when the
measuring field is parallel to the cooling field and it
would decrease when the measuring field is anti-
parallel, yielding an enhanced vertical shift of the
magnetization loops. Yet, in model (B), instead of
small particles pinned to the large one, it could be
thought basically as particles with relative large
References
[
[
1] T. Ambrose, C.L. Chien, Phys. Rev. Lett. 76 (1996) 1743.
2] R.H. Kodama, S.A. Makhlouf, A.E. Berkowitz, Phys.
Rev. Lett. 79 (1997) 1393.
[3] E. Hellstern, L. Schultz, Appl. Phys. Lett. 48 (1986) 114.

ARTICLE IN PRESS
20
E.C. Passamani et al. / Journal of Magnetism and Magnetic Materials 299 (2006) 11–20
[
[
[
[
[
[
4] J. Sort, X. Amilis, S. Surin
Appl. Phys. Lett. 75 (1999) 3177.
5] T.A. Anhoj, C.S. Jacobsen, S. Morup, J. Appl. Phys. 95
2004) 3649.
˜
ach, J.S. Mun
˜
oz, M.D. Baro
´
,
[12] P.J. van der Zaag, R.M. Wolf, A.R. Ball, C. Bordel, L.F.
Feiner, R. Jungblut, J. Magn. Magn. Mater. 148 (1995)
346.
(
´
[13] J. Nogues, C. Leighton, I. Schuller, Phys. Rev. B 61 (2000)
1315.
6] S. Mishra, J. Losby, S. Roy, N. Ali, K. Marasinghe, IEEE
Trans. Magn. 40 (2004) 2716.
7] W.H. Meiklejohn, C.P. Bean, Phys. Rev. B 102 (1) (1956)
[14] T.F. Connolly, E.D. Copenhaver, Solid State Physics
Literature Guides: Bibliography of Magnetic Materials
and Tabulation of Magnetic Transition Temperatures, vol.
5, IFI/Plenum, London, 1972.
1413.
8] R.K. Zheng, G.H. Wen, K.K. Fung, X.X. Zhang, J. Appl.
Phys. 95 (2004) 5244.
[15] C. Larica, K.M.B. Alves, E. Baggio-Saitovitch, A.P.
˜
Guimaraes, J. Magn. Magn. Mater. 145 (1995) 306.
9] L.D. Bianco, D. Fiorani, A.M. Testa, E. Bonetti, L.
Signorini, Phys. Rev. B 70 (2004) 052401.
[16] R.H. Kodama, E. Berkowitz, Phys. Rev. B 59 (1999)
6321;
[
[
10] J. Nogues, I.K. Schuller, J. Magn. Magn. Mater. 192
1999) 203.
11] R.L. Stamps, J. Phys. D 33 (2000) R247.
(
R.H. Kodama, S.A. Makhlouf, E.J. McNiff Jr., S. Foner,
Phys. Rev. Lett. 77 (1996) 394.