Three-dimensional ferromagnetic architectures with multiple metastable states

Article abstract of DOI:10.1063/1.3595339

We demonstrate controllable dual-bath electrodeposition of nickel on architecture-tunable three-dimensional (3D) silver microcrystals. Magnetic hysteresis loops of individual highly faceted Ag-Ni core-shell elements reveal magnetization reversal that comprises multiple sharp steps corresponding to different stable magnetic states. Finite-element micromagnetic simulations on smaller systems show several jumps during magnetization reversal which correspond to transitions between different magnetic vortex states. Structures of this type could be realizations of an advanced magnetic data storage architecture whereby each element represents one multibit, storing a combination of several conventional bits depending on the overall number of possible magnetic states associated with the 3D core-shell shape.

Full text of DOI:10.1063/1.3595339

APPLIED PHYSICS LETTERS 98, 222506 ͑2011͒
Three-dimensional ferromagnetic architectures with multiple
metastable states
F. Nasirpouri,1,2,a͒ M. A. Engbarth,2 S. J. Bending,2 L. M. Peter,3 A. Knittel,4 H. Fangohr,4
and M. V. Milošević5
1Department of Materials Engineering, Sahand University of Technology, Tabriz 51335-1996, Iran
2Department of Physics, University of Bath, Bath BA2 7AY, United Kingdom
3Department of Chemistry, University of Bath, Bath BA2 7AY, United Kingdom
4School of Engineering Sciences, University of Southampton, SO17 1BJ, Southampton, United Kingdom
5Departement Fysica, Universiteit Antwerpen, Groenenborgerlaan 171, B-2020 Antwerpen, Belgium
͑Received 22 March 2011; accepted 5 May 2011; published online 2 June 2011͒
We demonstrate controllable dual-bath electrodeposition of nickel on architecture-tunable
three-dimensional ͑3D͒ silver microcrystals. Magnetic hysteresis loops of individual highly faceted
Ag-Ni core-shell elements reveal magnetization reversal that comprises multiple sharp steps
corresponding to different stable magnetic states. Finite-element micromagnetic simulations on
smaller systems show several jumps during magnetization reversal which correspond to transitions
between different magnetic vortex states. Structures of this type could be realizations of an advanced
magnetic data storage architecture whereby each element represents one multibit, storing a
combination of several conventional bits depending on the overall number of possible magnetic
states associated with the 3D core-shell shape. © 2011 American Institute of Physics.
͓doi:10.1063/1.3595339͔
Recent advances in nanofabrication and synthetic chem-
istry have lead to a dramatic growth in research activity in
the area of magnetic nanostructures,1 with key applications
in data storage, medical diagnosis, and quantum information
processing. Much of the work has focused either on litho-
graphically patterned thin films or nanoscale particles and
clusters. In contrast, very little work has been attempted on
three-dimensional ͑3D͒ mesostructures, whose sizes are
comparable with the relevant characteristic physical length
scales ͑e.g., ferromagnetic domain size and domain wall
width͒. It has been recently ͑Ref. 2͒ demonstrated that
“architecture-tunable” mesostructures can be grown by elec-
trodeposition onto highly oriented pyrolitic graphite ͑HOPG͒
substrates. These structures are perfectly 3D in stark contrast
to lithographically patterned structures, which invariably
have rough edges and surfaces.
The main objective of this letter is to point out the cor-
relation between highly faceted structures and their magnetic
properties. For example, it is already known that the shape of
magnetic micro- or nanostructures has pronounced effects on
magnetic anisotropy.3 We use electrodeposition, a procedure
by which Pb, Ag, Bi, and Sn can be readily deposited in the
form of highly faceted microcrystals ͑Refs. 4–6͒ while
nickel is found to form wires or particles without any recog-
nizable facets. In what follows, we report the fabrication and
magnetic characterization of highly faceted 3D Ni mesos-
tructures, using a dual-bath method.7 We argue that the 3D
shape of the sample hosts many more stable magnetic states
than two-dimensional analogs, and hence can be used for
multibit data storage.
counter electrode and a standard Ag/AgCl electrode and
high-purity silver wire as reference electrodes. To establish
appropriate deposition potentials, cyclic voltammograms
͑CV͒ were captured in the prepared electrolytes.
A two-step dual-bath potentiostatic method was used to
electrodeposit Ag-Ni core-shell structures. In the first step,
highly faceted silver mesocrystals were electrodeposited
from nitrate electrolytes at a pH=2–2.5. The sample was
then gently rinsed in deionized water and blow dried with
argon. In the second step nickel was electrodeposited from a
Watts bath containing 2.3 mol/L H2O Ni sulfate, 0.6 mol/L
H2O Ni chloride, and 0.5 mol/L H2O boric acid, with a pH
=3–4. The thickness of the nickel shell was determined from
Faraday’s law with a cathodic current efficiency of 100%
using the computer-controlled deposition software
͑AUTOLAB-GPES͒. For the given surface area of the working
electrode and the effective surface density of silver islands,
the nickel thickness was estimated to be 100 nm from the
known total charge passing through the cell ͑For instance, for
the pyramid structure shown in Fig. 1͑d͒, a total charge of 16
mC was conducted to obtain a 100 nm thick Ni film.
The scanning electron micrographs of the electrodepos-
ited highly faceted silver mesocrystals with different sizes
and shapes are shown in Fig. 1. A Ni shell was subsequently
electrodeposited on silver mesocrystals from a Watts bath at
Electrodeposition was carried out under potentiostatic
mode in a conventional three-electrode cell connected to a
potentiostat ͑microAutolab III͒ using HOPG from SPI Sup-
plies ͑grade 1͒ as working electrode, a platinum plate as
FIG. 1. Different shapes of silver mesocrystals electrodeposited at ͑a͒ –60
mV for 10 s from 250 mM Ag nitrate ͑b͒ Ϫ60 mV for 30 s from 100 mM Ag
nitrate, ͑c͒ Ϫ70 mV for 20 s from 100 mM Ag nitrate, and ͑d͒ 1 V for 60 s
followed by 10 s at open circuit potential and finally Ϫ10 mV for 30 s from
100mM Ag nitrate.
a͒
Electronic mail: nasirpouri@sut.ac.ir.
0003-6951/2011/98͑22͒/222506/3/$30.00
98, 222506-1
© 2011 American Institute of Physics

222506-2
Nasirpouri et al.
Appl. Phys. Lett. 98, 222506 ͑2011͒
hibit modest signal to noise levels because of the inherent
noise background in EDS. However, the accumulation of a
high density of pixels in the vicinity of the core-shell is good
evidence of highly local electrodeposition.
We carried out the magnetic Mz-Hz measurements at 5 K
using a linear array of 2ϫ2 ␮m2 GaAs/AlGaAs heterostruc-
ture Hall probes, where z is the direction perpendicular to the
plane of the Hall probe. Individual Ag-Ni core-shell struc-
tures were placed on top of one of the active Hall elements
using a piezoelectric nanomanipulator such that the base of
pyramidal-shaped structures was in direct contact with the
sensor with the apex pointing upwards. The Hall probes were
operated with a 20 µa 32 Hz ac current and the Hall voltage
detected with a lock-in amplifier. An external magnetic field
was applied perpendicular to the Hall array. Figure 3 shows
the magnetization curves obtained for one particular pyramid
͓Fig. 1͑d͔͒ and a truncated icosahedron ͓Fig. 1͑b͔͒ at the
same applied field sweep rate. Surprisingly, the magnetiza-
tion reversal occurs in a stepwise fashion in both hysteresis
loops. Each step corresponds to a different, stable magnetic
state, as verified by performing minor loops ͑back-tracking͒
for several magnetization steps. Minor loops were recorded
by sweeping the magnetic field to different levels after satu-
rating the sample in the opposite direction. Depending on the
magnetic history, we find six different stable states at zero
magnetic field for the pyramidal structure, and ten for the
more complex icosahedral shaped magnetic element. In Fig.
3͑b͒, the magnetic states decay relatively fast upon reversal
of the magnetic field. This is a consequence of a rather thick
Ni shell on the sample. Namely, the metastability of mag-
netic states in 3D ferromagnets is mainly due to different
demagnetization scenarios in the vicinity of 3D edges. As
thinner shells will have the more pronounced ͑sharp͒ edges,
more possible states and their enhanced stability can be ex-
pected.
FIG. 2. ͑Color online͒ ͑a͒ 3D AFM image of a typical Ag pyramidal me-
socrystal with the height from the base to the apex of 5.2 and 10 ␮m base
side length, ͑b͒ Ag mesocrystal after electrodeposition of a 100 nm Ni shell
on top. EDS elemental maps of silver ͑c͒ and nickel ͑d͒ taken on the actual
Ag-Ni mesocrystal as shown in insets.
a potential of −800 mVAg/AgCl. CVs taken in the Watts bath
on bare HOPG and with silver electrodeposited on HOPG,
showed a considerable difference in the equilibrium deposi-
tion potential of nickel. When nickel is electrodeposited on
the bare HOPG, it starts to reduce at potentials more negative
than −840 mVAg/AgCl. In the presence of silver metallic cen-
ters on HOPG, nickel starts to grow at potentials more nega-
tive than −670 mVAg/AgCl. At this potential, the nickel shell
is solely electrodeposited on the silver centers. Ex-situ
atomic force micrographs ͑AFM͒ taken from a pyramidal
silver mesocrystal before and after electrodeposition of
nickel are shown in Figs. 2͑a͒ and 2͑b͒. Surface roughening
of the polycrystalline nickel shell is clearly seen as a result of
the well-established 3D nucleation and growth mechanism of
electrodeposition of nickel ͑Ref. 8͒ or due to the presence of
hydrogen evolution ͑Ref. 9͒ on the crystalline Ag core. The
energy dispersive X-ray spectroscopy ͑EDS͒ elemental maps
show that nickel is primarily deposited on top of the silver
mesocrystals, as seen in Figs. 2͑c͒ and 2͑d͒. Both maps ex-
Figure 3 demonstrates not only the existence of different
magnetic states for our 3D mesostructures, but also the
controllable, sharp switching between different magnetiza-
tion states. These two properties are essential for a data stor-
age element like conventional magnetic binary bits or qua-
ternary ͑four-state͒ bits based on the vortex chirality and its
polarization.10 However, although an electronic method for
reversing vortex polarity has already been established,11 re-
FIG. 3. ͑Color online͒ Hysteresis
loops and back-tracking curves of a ͑a͒
truncated icosahedral and ͑b͒ pyrami-
dal Ag-Ni structures with a typical 100
nm nickel shell ͑shown in upper left
insets͒ measured in an external mag-
netic field applied perpendicular to the
base of the element at 5 K. Arrows
show the directions of application of
magnetic field and backward sweep to
obtain minor loops. Dashed lines indi-
cate the positions of magnetization
steps. Note: in panel ͑b͒ the full hys-
teresis loop is shifted horizontally by
Ϫ20 Oe for clarity.

222506-3
Nasirpouri et al.
Appl. Phys. Lett. 98, 222506 ͑2011͒
scaling is necessary in order to perform computationally fea-
sible simulations, but is also more suited for the data storage
concept. We indeed find that the spatially averaged magneti-
zation undergoes several jumps during the reversal process,
via different 3D magnetic vortex states. Our simulations are
still very computationally challenging, as the mesh reso-
lution has to be significantly smaller than the shell thickness.
Mirroring our experiments, we have backtracked the ob-
served magnetic states revealing the stability of four vortex
states at Hext=0 Oe with ͑i͒ the core of the vortex on the top
of the pyramidal structure ͓symmetric state, inset ͑a͒ in Fig.
4͔, or ͑ii͒ the core on one of the side faces ͓asymmetric
vortex state, insets ͑b͒ and ͑c͒ in Fig. 4͔. The states of Figs.
4͑b͒ and 4͑c͒ do not only differ in the vortex core position
but also in the way the magnetization curls around the core.
A closer comparison between these states reveals that, on
average, the z-component of the state in Fig. 4͑c͒ is more
negative. High demagnetization fields that occur in the vicin-
ity of the sharp edges, which separate the triangular faces,
lead to the energetic differences between the observed vortex
states. Our simulations are done at T=0 K, but the calcu-
lated energy differences between the micromagnetic states
are in the electron volt range and thus can be thermally stable
at higher temperatures ͑for energy barrier EM as the mean
value of the activation energy E, to other states exceeding
50 KBT͒.14,15
FIG. 4. ͑Color online͒ The magnetization loops and micromagnetic states
from a top-down perspective with z-component magnetization color bar
obtained by micromagnetic simulations, for a pyramidal Nickel shell with a
base side length of 500 nm, a height of 250 nm, and thickness of 5 nm. The
ជ
calculated, spatially averaged magnetization M is normalized with respect to
the saturation magnetization ͑MS =510000 A/m͒. The dashed line is the
mirror image of the reversal curve. Inset ͑a͒ shows a symmetric vortex with
its core sitting on the top of the shell. Insets ͓͑b͒ and ͑c͔͒ show the asym-
metric vortex state, where the core ͑indicated by black dots͒ sits on one of
the side faces. Point ͑d͒ corresponds to the reversed symmetric vortex state
͑not shown due to a lack of space͒. The states have the following micro-
magnetic energies: ͓͑a͒ and ͑d͔͒ 26:6 eV, ͑b͒ 22:0 eV, and ͑c͒ 24:0 eV.
This work was supported by EPSRC in the U.K. under
Grant Nos. EP/E039944/1 and EP/E040063/1, DYNAMAG
project ͑EU FP7/2007-2013 Grant No. 233552͒, and FWO-
Vlaanderen.
versing the chirality in a reproducible way is still a formi-
dable task. The 3D architecture of our magnetic elements
offers a simple alternative either for multibit storage, or stor-
age of a combination of conventional bits ͑2n magnetic states
needed to store all combinations of n bits͒. The scalability of
our mesostructures for this application is possible by making
them far smaller as they become faceted practically immedi-
ately upon electrochemical nucleation. The only reason that
we report data here on micron-sized crystals is because the
manipulation and Hall probe magnetometry is greatly facili-
tated in larger samples.
To verify the above scenario we have used the finite-
element, micromagnetic solver Nmag ͑Ref. 12͒ with
hierarchical matrices ͑Ref. 13͒ in order to simulate the
magnetization process of pyramidal shell structures. The
magnetocrystalline anisotropy of nickel is neglected in our
model which is reasonable at room temperature when the
cubic anisotropy of Nickel is small. Stronger deviations may
be expected between simulation and experiment as we per-
formed experiments at low temperatures in order to reduce
the noise of the Hall sensor. Figure 4 shows the obtained
simulation results for the reversal behavior of a Ni pyramidal
shell of Fig. 2 but scaled down by a factor of 20. This down-
1F. Nasirpouri and A. Nogaret, Nanomagnetism and Spintronics ͑World
Scientific, Singapore, 2010͒.
2Z. L. Xiao, C. Y. Han, W.-K. Kwok, H. Wang, U. Welp, J. Wang, and G.
W. Crabtree, J. Am. Chem. Soc. 126, 2316 ͑2004͒.
3C. A. Ross, Annu. Rev. Mater. Res. 31, 203 ͑2001͒.
4M. P. Zach, K. H. Ng, and R. M. Penner, Science 290, 2120 ͑2000͒.
5A. Müller, S. E. C. Dale, M. A. Engbarth, S. J. Bending, and L. M. Peter,
Cryst. Eng. Comm. 12, 2135 ͑2010͒.
6S. E. C. Dale, S. J. Bending, and L. M. Peter, Langmuir 25, 11228 ͑2009͒.
7A. Blondel, B. Doudin, and J.-Ph. Ansermet, J. Magn. Magn. Mater. 165,
34 ͑1997͒.
8S. Morin, A. Lachenwitzer, O. M. Magnussen, and R. J. Behm, Phys. Rev.
Lett. 83, 5066 ͑1999͒.
9W. Schwarzacher, J. Phys.: Condens. Matter 16, R859 ͑2004͒.
10C. Phatak, M. Tanase, A. K. Petford-Long, and M. De Graef, Ultramicros-
copy 109, 264 ͑2009͒.
11K. Yamada, S. Kasai, Y. Nakatani, K. Kobayashi, H. Kohno, A. Thiaville,
and T. Ono, Nature Mater. 6, 270 ͑2007͒.
12T. Fischbacher, M. Franchin, G. Bordignon, and H. Fangohr, IEEE Trans.
Magn. 43, 2896 ͑2007͒.
13A. Knittel, M. Franchin, G. Bordignon, T. Fischbacher, S. J. Bending, and
H. Fangohr, J. Appl. Phys. 105, 07D542 ͑2009͒.
14H. N. Bertram and M. Williams, IEEE Trans. Magn. 36, 1 ͑2000͒.
15R. P. Cowburn, J. Appl. Phys. 93, 9310 ͑2003͒.

Applied Physics Letters is copyrighted by the American Institute of Physics (AIP). Redistribution of journal
material is subject to the AIP online journal license and/or AIP copyright. For more information, see
http://ojps.aip.org/aplo/aplcr.jsp