M. Beerman, A. Pakhomov, Y. Bao, and K. Krishnan

1
M. Beerman, A. Pakhomov, Y. Bao, and K.
Krishnan Department of Materials Science and
Engineering University of Washington, Seattle, WA
98195, USA Work supported by National Science
Foundation NSF/DMR 0203069 Campbell Endowment at
UW An anamolous peak has been observed in zero
field cooled measurements on systems composed of
e-phase cobalt nanoparticles. The samples were
synthesized via a thermal decomposition process
resulting in spherical nanoparticles surrounded
by an organic surfactant 1,2. In addition to
the blocking temperature peak, which is dependant
on particle volume 3 and displays interesting
interaction effects, an additional peak appears
at 8 K regardless of particle diameter. Zero
field cooled and field cooled magnetization
versus temperature measurements on spherical
particles with mean diameters in the range of
5-20 nm were performed with a SQUID magnetometer,
including careful background signal checks.
Measurements after heat treatment, to promote
oxidation, show that the blocking temperature
peak can be effectively removed in fully oxidized
powders. The 8 K peak, however, remains after
heat treatment, but with a reduced magnitude, and
M(H) dependance is hysteretic below this
temperature. Additional measurements were
performed on commercial micrometer-sized cobalt
oxide powder, where this feature is also present,
but gives a much weaker contribution to
susceptibility. Unreacted synthesis precursors as
well as samples with different surfactant species
all showed at least some indication of a peak at
8 K. This feature, universal for the system under
investigation, has not been observed in Co
nanoparticle systems prepared by co-sputtering or
co-evaporation of Co with non-magnetic materials.
Furthermore, hysteresis measurements, both before
and after heat treatment, show a bias when cooled
in field from 30 K to below the peak temperature.
Analyses of the data leads us to conclude that
freezing of frustrated uncompensated spins on the
surface of cobalt oxide is the most reasonable
explanation of this anomalous low temperature
behavior. 1 V. Puntes et al. Science 291
(2001) 2115 2 Y. Bao, et. al., JMMM, in press.
3 L. Néel, Annales de Géophysique 5 (1949) 99
The low temperature peak consistently appears at
8 K, regardless of particle diameter. Peak
susceptibility, however, decreases as particle
diameter increases suggesting that the peak is
due to a surface effect. One possible
explanation is surface oxidation. In order to
explore this possibility, ZFC measurements were
performed before and after heating the sample in
air at 120ºC for 24 hours. The temperature has
been plotted on a log scale to reveal the peak at
8 K in the ZFC plot to the right. The 8 K peak
may be associated with a layer of cobalt oxide at
the surface as suggested by the figure below.
A uniaxial magnetic particle has two stable
magnetization orientations as seen in the figure
to the left. If an external field is applied
along the easy axis then one of the two
orientations has a lower energy as seen to the
right. As the temperature increases, thermal
fluctuations provide the activation energy for
the system to relax over the energy barrier into
the field direction. This behavior is well known
and follows
27kBTB
Epsilon cobalt nanospheres surrounded by a
mixture of organic surfactants (TOPO and OA) are
synthesized via thermal decomposition of cobalt
carbonyl in di-chlorobenzene solvent. The
diameter D may be controlled between 5 and 20 nm,
and the surfactant length is approximately 1.7 nm.
Surfactants
ZFC curve before heating. 6 nm spheres with TOPO
and OA surfactants.
?-cobalt
D
No External Field
?-cobalt core Co3O4 shell
Applied External Field
Co3O4
where ?m is the characteristic measurement time
20s, ?0 is the microscopic fluctuation attempt
time 10-10s, KA is the anisotropy constant 0.21
J/m3 for ?-Co, and V is the magnetic particle
volume. The barrier height may be approximated
by 27 kBTB where kB is Boltzmanns constant and
TB is the blocking temperature, which corresponds
to the peak temperature in moment versus
temperature measurements (as identified to the
left).
ZFC curve of the same sample after heating.
The particles are precipitated from solution onto
a thin carbon film for transmission electron
microscopy (TEM).This TEM image shows a self
assembled monolayer of 10 nm spheres.(Phillips
400 T at 100 keV)
Before heating
After heating
After heat treatment, the ?-cobalt core is
completely replaced with the spinel phase of
cobalt oxide, Co3O4. The blocking peak,
originally at 60 K, is absent in the ZFC plot to
the middle right. If the 8 K peak is due to a
surface effect, then one would expect that the
magnitude of the peak to remain the same before
and after heat treatment. The magnitude of the
peak susceptibility, however, has dropped by
nearly a factor of ten, but this is accounted for
by the shoulder of the blocking peak. The phases
were identified by x-ray diffraction using a
Rigaku 12.5 kW rotating anode x-ray source and
goniometer. The diffraction peaks in scan a
(lower right figure) correspond to ?cobalt, and
the peaks in scan b correspond to Co3O4.
ZFC
TB
Two curves are plotted here zero field cooled
(ZFC) and field cooled (FC). ZFC measurements
start at low temperature after cooling the sample
in zero external field. Magnetization is then
measured as the temperature slowly increases. FC
measurements start at high temperature where a
field is applied and the magnetization is
measured as the temperature slowly decreases.
The rate of cooling is constant in both cases.
The interesting observation is the second peak at
low temperature on the ZFC curve and its
associated sharp uptake in magnetization as seen
in the FC curve. The focus of this poster
presentation is to further characterize and
describe this feature. Magnetic measurements were
performed on a Quantum Design MPMS 5
superconducting quantum interference device.
For more information concerning particle
synthesis and self assembly please
see Controlled Self-assembly of Colloidal
Cobalt Nanocrystals Mediated by Magnetic
Interactions Poster presentation 2-ypm52
Location H
XRD a. before and b. after heating.
Exchange bias was originally observed in samples
consisting of micron sized cobalt particles
surrounded by a cobalt oxide (CoO) shell. Bias
is due to exchange interaction between the
ferromagnetic core and the anti-ferromagnetic
shell. The nanoparticle samples that we are
discussing here have an ?cobalt core and a Co3O4
shell. The Neel temperature for Co3O4 is 40 K
(as opposed to 270 K for CoO). A powder sample
was prepared with 6 nm spherical particles. The
sample was cooled in an applied field of 1000 G
to the measurement temperature. Hysteresis plots
between /- 2T at various temperatures appear in
the figure below. The applied field axis is
reduced so that the exchange bias shift in the
hysteresis loop is noticeable.
Spin-glass-like freezing is one possible physical
mechanism that can lead to a particle
size-independent maximum in the ZFC curves. Bulk
spin glasses have been well studied and show a
shift in the glass transition to lower
temperatures as the applied field increases. We
observed similar shifts in our measurements,
which might be an indication of spin-glass
freezing. The variation in saturation
magnetization at low temperatures observed in
?Fe2O3 and Pd-Fe nanoparticle systems has been
attributed to surface spin-glass like freezing.
The anisotropy is not bound to the crystal
lattice due to the irregularities at the region
of broken bonds near the surface. The spin glass
transition temperature Tg is approximately 8K in
this case. At temperatures above Tg, the surface
spins are strongly paramagnetic and do not couple
even in high fields as suggested by the figure
(below left). In the presence of an applied
field below Tg the spins will freeze into the
field direction (below right).
The evidence as outlined in the Heat Treatment
section of this presentation is compelling for
the surface oxide explanation of the 8 K
susceptibility peak. However, a thorough
investigation requires background measurements of
the synthesis precursors as well as a
commercially obtained Co3O4 powder. The cobalt
particle reaction precursor is Co2(CO)8. A
sample was dried and measured on the SQUID
magnetometer. The ZFC FC curve below does show
some indication of the peak at 8 K. One possible
explanation of the peak is that some clustering
of the precursor results in small particle
aggregates with an associated blocking peak at 8
K. Another explanation is direct oxidation of
the precursor to form the uncompensated surface
spin state, which is consistent with the heat
treatment measurements.
Co3O4
?-cobalt
T gt Tg
T lt Tg
The figure to the left is the family of field
cooled susceptibility curves from which the
linear susceptibilty ?0, non linear
susceptibility ?2 and the non linear exponent ?
were obtained and plotted in the figure to the
right.
The ZFC plot below was obtained from the
commercial Co3O4 powder. At first look, the 8 K
peak is absent, however, a closer inspection
reveals some indication of a peak. Again this is
consistent with the surface oxide interpretation.
We have previously observed that the magnitude
of the 8 K peak is inversely proportional to the
particle diameter, and this sample consists of a
wide size distribution of particles ranging from
hundreds of nm to tens of microns. The surfaces
of the smallest particles in this distribution
are likely contributing to the 8 K peak.
Applied field
In order to test the applicability of this model,
we estimated the linear ?0 and non-linear ?2
contributions to magnetic susceptibility, and the
critical exponent ?, as a function of temperature
in the range 2.5-10K. These data were extracted
from a family of field cooled (FC) curves
measured at different fields with a low ramp rate
on a sample containing 6-nm particles, then
compared to the following expansion for the
susceptibility
Magnetic property measurements of the
surfactant-coated core-shell Co/Co3O4
nanoparticles reveal contributions from three
interacting and correlated entities single
domain metallic cobalt, which is primarily
responsible for the blocking phenomena
antiferromagnetic Co3O4 which is revealed in
exchange bias and the surface layer of the oxide
where spins are coupled below a critical
temperature, which may be defined as either spin
glass freezing or ferromagnetic Curie transition.
Since the blocking effect can be controlled by
annealing, the low temperature behavior can be
isolated and studied separately from blocking,
while the interaction between surface and core
may be examined by studying the system in the
pre-oxidized state.
The exchange field H-ex is the extent of the
shift of the hysteresis loop from the origin.
The figure below is a plot of H-ex as a function
of temperature. The blocking temperature of 6 nm
particles is approximately 80 K, and the greatest
exchange field occurs just below this temperature
at about 50 K.
Isothermal magnetization curves were extracted
from the set of FC measurements. Then linear
susceptibility ?0 was found from the initial
slope of the isothermal M(H) curve. The
inflection point of this curve gives the
transition temperature, which is about 7.5 K. A
power law fit was employed to obtain ?2 and ?.
The non-linear susceptibility is well behaved
below the transition temperature and is two
orders of magnitude smaller than ?0. Above the
transition temperature, however, ?2 and ? are not
well determined due to the limitations of the
measurements. The nonlinear exponent ? is
constant below Tg, with a value of 3/2 for low
fields. These trends do not compare well to bulk
spin glasses, where the nonlinear susceptibility
has a maximum at the freezing temperature, and
the exponent ? has a minimum.
Contact Information Please visit the Integrated
Nanosciences Mesoscale Engineering Group
website at http//depts.washington.edu/kkgroup O
r you may email the authors at Kannan M.
Krishnan kannanmk_at_u.washington.edu Michael
Beerman mbeerman_at_u.washington.edu
K. H. Fischer, J. A. Hertz, Spin Glasses,
Camridge University Press (1991). B. Martinez,
et. al., Phys. Rev. Lett., 80, 181 (1997).
W.H. Meiklejohn, C.P. Bean, Phys. Rev., 102,
1413 (1956).