Delocalized versus localized magnetization reversal in template-grown Ni and Ni80 Fe20 nanowires

Article abstract of DOI:10.1016/j.jmmm.2009.04.001

The study of magnetization reversal processes of highly ordered magnetic nanowire arrays is of particular interest due to their potential application as perpendicular recording media. We electrodeposited high-aspect ratio Ni and Ni₈₀ Fe₂₀ nanowires in nanoporous alumina templates and performed SQUID measurements as a function of temperature (5-350 K) in order to better understand the magnetization reversal mechanisms active in our arrays. A strong coercive field deviation from the expected theoretical value calculated from the Stoner-Wolfarth model assuming delocalized magnetization reversal processes was observed. To explain our results, we then assumed a magnetization reversal triggered by thermal activation over an energy barrier. Taking into account the measured effective anisotropy, obtained experimentally from a low-field susceptibility approximation, we extracted the activation volume and found it to be much smaller than the nanowire physical volumes. This lead us to conclude that domain nucleation was present, mainly due to imperfections that introduce different local anisotropies and thus domain wall nucleation sites.

Full text of DOI:10.1016/j.jmmm.2009.04.001

ARTICLE IN PRESS
Journal of Magnetism and Magnetic Materials 322 (2010) 1319–1322
Contents lists available at ScienceDirect
Journal of Magnetism and Magnetic Materials
journal homepage: www.elsevier.com/locate/jmmm
Delocalized versus localized magnetization reversal in template-grown Ni
and Ni80Fe20 nanowires
Ã
D.C. Leitao a, , C.T. Sousa a, J. Ventura a, K.R. Pirota b, M. Vazquez b, J.B. Sousa a, J.P. Araujo a
a IFIMUP and IN-Institute of Nanoscience and Nanotechnology, Departamento de F´ısica da Faculdade de Ciencias da Universidade do Porto,
Rua do Campo Alegre, 687, 4169-007 Porto, Portugal
b Instituto de Ciencia de Materiales de Madrid, ICMM - CSIC, Campus de Cantoblanco, 28049 Madrid, Spain
a r t i c l e i n f o
a b s t r a c t
Available online 9 April 2009
The study of magnetization reversal processes of highly ordered magnetic nanowire arrays is of
particular interest due to their potential application as perpendicular recording media. We
electrodeposited high-aspect ratio Ni and Ni80Fe20 nanowires in nanoporous alumina templates and
performed SQUID measurements as a function of temperature (5–350 K) in order to better understand
the magnetization reversal mechanisms active in our arrays. A strong coercive field deviation from the
expected theoretical value calculated from the Stoner–Wolfarth model assuming delocalized
Keywords:
Nanoporous alumina
Nanowire
Magnetization reversal
Localization
magnetization reversal processes was observed. To explain our results, we then assumed
a
magnetization reversal triggered by thermal activation over an energy barrier. Taking into account
the measured effective anisotropy, obtained experimentally from a low-field susceptibility approxima-
tion, we extracted the activation volume and found it to be much smaller than the nanowire physical
volumes. This lead us to conclude that domain nucleation was present, mainly due to imperfections that
introduce different local anisotropies and thus domain wall nucleation sites.
& 2009 Elsevier B.V. All rights reserved.
1. Introduction
influencing magnetization reversal processes and the dominant
source of anisotropy in arrays of nanostructures [8,9]. Interest-
ingly, changes in the easy axis orientation may arise from their
interplay. For low porosity and high-aspect ratio arrays (length b
diameter, interwire distance) the intrawire interactions (i.e. shape
anisotropy) prevail, and the magnetization aligns along the axial
axis of the nanowires. However, when the porosity/diameter
increases or the nanowire length decreases, the interwire coupling
overcomes the shape anisotropy, rotating the magnetic easy axis
towards the transverse direction. Such anisotropy dependence on
the nanowire length enables a systematic control and adjustment
of parameters such as coercivity and easy axis direction, as shown
in magnetic/non-magnetic multilayered nanowires when the ratio
of the corresponding thicknesses was changed [10].
In the last years there has been growing interest on
nanostructures as they provide several versatile applications and
new physical phenomena. Particularly, arrays of highly ordered
nanostructures have their potential application in a wide range of
areas ranging from semiconductors or magneto-optics to biome-
dical or sensor devices mostly due to their novel properties caused
by the constrained dimensions [1–4]. Moreover, nano-ordered
arrays of magnetic nanowires offer the prospect of application as
magnetic perpendicular recording media. However, a complete
understanding of the magnetization reversal processes towards
the ability to control it as desired is still an unresolved issue.
In theory, single ideal infinite long nanowires, with dominating
To fabricate arrays of nanowires, one usually requires expen-
sive and time consuming lithographic techniques, like e-beam or
focused ion beam lithographies. An alternative route to obtain
such arrays is the use of easily fabricated anodized nanoporous
alumina as templates for the subsequent growth of magnetic
materials [11]. These templates present enormous advantages
such as the possibility to build a net of aligned and ordered
nanostructures and the ability to change their dimensions as
desired. These geometrical properties, such as diameter and
distance between nanowires (given by the anodization potential)
and length (given by the anodization time) can lead to out-
standing magnetic behaviors [8,9].
shape anisotropy, undergo magnetization reversal by coherent
pffiffiffi
rotation if their diameter DoDc ¼
pdW, where Dc is the critical
diameter and dW is the domain wall (DW) width, or by curling if
D4Dc, being both delocalized modes. However, in real systems,
the finite nanowire lengths together with different local aniso-
tropies give rise to potential domain wall nucleation sites and
thus volume localized magnetization reversal [5–7]. In fact,
intrawire and interwire magnetostatic effects are the key feature
Ã
Corresponding author.
E-mail address: dleitao@fc.up.pt (D.C. Leitao).
0304-8853/$ - see front matter & 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmmm.2009.04.001

ARTICLE IN PRESS
1320
D.C. Leitao et al. / Journal of Magnetism and Magnetic Materials 322 (2010) 1319–1322
In this work we used nanoporous alumina substrates with an
average pore diameter of ꢀ35 nm, interpore distances of ꢀ100 nm
(porosity ꢀ 10%) and pore lengths of ꢀ1:3 m, as templates for the
m
electrodeposition of Ni and N80Fe20 nanowires. Due to their high-
aspect ratio (length/diameter ’ 37) and low porosity these
nanowires fit into the category where the shape anisotropy is
dominant and consequently should hold an easy axis along the
nanowire axial axis [9]. We performed SQUID (Quantum Design)
magnetization measurements as a function of temperature (from
5 to 350 K) and extracted the coercive field (Hc) temperature
dependance for both samples. Although the diameters of our pore
templates place the Ni nanowires (Dc ’ 45 nm) in the range of
magnetization reversal by coherent rotation and the Ni80Fe20
nanowires (Dc ’ 30 nm) should endure nucleation of reverse
domains by curling, we were able to conclude that the
magnetization reversal is indeed nano-localized in a volume ꢀ 2
orders of magnitude smaller than the wire volume, as a result of
lower local anisotropies induced by imperfections along the wires
and at the ends, that reduce the energy necessary to locally
reverse the magnetization.
2. Experimental details
For the growth of nanowires, nanoporous alumina templates,
obtained by a standard two-step anodization method [12,13] of
high-purity (99.997%) aluminum foils, were employed. Prior to
anodization, samples were properly cleaned and degreased,
followed by electropolishing for surface improvement, setting
the necessary conditions for a well organized porous structure to
be obtained. First anodizations were performed at a constant
voltage of 40 V for 15 h, with a 0.3 M oxalic acid solution at a
temperature between 2 and 6 ꢁC. These anodization conditions
resulted in nanoporous alumina substrates with an average pore
diameter of ꢀ35 nm and separation ꢀ100 nm. The second
anodization was performed only for 30 min, producing pore
Fig. 1. Magnetization dependence on applied field at 300 K for (a) Ni80Fe20 and (b)
Ni nanowires samples.
plane) magnetization versus applied field behaviors for both
samples at 300 K. For Ni80Fe20 [Fig. 1(a)] the expected anisotropic
MðHÞ cycle, typical of single domain behavior, is observed. An easy
axis with Hc ꢃ 950 Oe and a perpendicular hard axis with Hc ꢃ
170 Oe were identified. On the other hand, in the case of Ni
[Fig. 1(b)], we observe an almost isotropic behavior. Nevertheless,
an easy axis along the longitudinal nanowire direction
with coercive field Hc ꢃ 350 Oe and a hard axis along nanowire
axial direction with Hc ꢃ 210 Oe can still be identified. Notice
lengths of ꢀ1:3
mm. The thickness of the existent barrier-layer at
the pores bottom can be estimated to be ꢀ60 nm [14]. To
electrodeposit the metallic nanowires in the nanoporous alumina
templates, without having to manipulate the thin and fragile
membranes, the deposition current has to flow through this oxide
barrier, which therefore must be thinned down. The reduction
procedure [11] gives origin to dendrites (small channels in the
alumina barrier-layer) enabling the current to tunnel through the
resulting ꢀ2:5 nm thick barrier. Afterwards, a pulsed electrode-
position method [11] was followed to grow Ni and Ni80Fe20
nanowires. Different electrolytes were employed: a standard
Watts bath (NiSO4 ꢂ 6H2O, NiCl2 ꢂ 6H2O, H3BO3) was used for Ni
deposition, while for Ni80Fe20 the solution was composed of
NiSO4 ꢂ 7H2O, FeSO4 ꢂ 7H2O, H3BO3, sacarine and sodium-lauryl
sulfate. We confirmed the ratio of Ni:Fe in the nanowires by
Energy-dispersive X-ray spectroscopy (EDS) to be 80:20. Deposi-
tion pulses with a current density of 38 mA=cm2 were applied for
both Ni and Ni80Fe20. The electrodepositions were carried out at
30 ꢁC. Different regimes can be observed in the UðtÞ electro-
deposition curves enabling a high control over the deposition
process. To obtain a good filling uniformity the nanowires were
left to overflow the surface membrane for a considerable time.
that, according to the literature,
a much more anisotropic
behavior would be expected for this system [8,15]. EDS analysis
revealed the presence of about 13% of Cu within the Ni nanowires.
The consequent decrease of the corresponding saturation
magnetization (MSat) by ꢀ15% (MNSait ’ 480 emu=cc [8], while
Ni87Cu13
Sat
to
M
’ 400 emu=cc) leads to the decrease of the
magnetostatic interactions and to the reduction of the easy
axis uniaxiality and Hc value. The observed almost-isotropic
behavior for the Ni nanowires sample can therefore be explained
by this Cu addition [10].
3.2. Temperature dependence of coercive field
We then performed MðHÞ cycles in the 5–350 K range, allowing
us to obtain the temperature dependence of Hc for both samples
(Fig. 2). We can observe that Hc increases as the temperature
decreases, being the variation more accentuated at low
temperatures. This behavior can be explained either by the
different thermal expansion coefficients of the aluminum and
nanowires, leading to an increase of effective anisotropy (Keff
with temperature or by thermal fluctuations [5]. The first
explanation can be discarded since major influence of
)
3. Experimental results and discussion
a
magnetoelastic anisotropy is only observed when the nanowire
array is embedded in an aluminum matrix [16], which is not our
case [17]. This fact is corroborated by the constant behavior of Keff
with temperature (not shown). So, thermal activation over an
energy barrier processes must be taken into account. A detailed
3.1. Hysteresis cycles
Figs. 1(a) and (b) show easy axis (along nanowires longitudinal
axis) and hard axis (transverse to nanowires, i.e. in the sample

ARTICLE IN PRESS
D.C. Leitao et al. / Journal of Magnetism and Magnetic Materials 322 (2010) 1319–1322
1321
U0 ’ 83 000 K, while for Ni, U0 ’ 25 000 K, both with acceptable
correlation factors. Considering the calculated effective anisotro-
3
3
py, and from Eq. (3) we obtain Veff ’ ð20 nmÞ and Veff ’ ð26 nmÞ
for Ni80Fe20 and Ni, respectively. We thus conclude that the
experimental Veff is smaller than the expected volume for a Neel
ꢁ
1
ꢁ
180 DW nucleation (ꢀ2 V180 DW), but is ꢀ100 times smaller than
the wire volume. These differences are attributed to imperfections
and inhomogeneities along the nanowire, but mainly to geome-
trical features at the ends that lead to spin canting, reducing the
magnetostatic energy needed to nucleate a domain wall. Micro-
magnetic calculations in finite length wires have also demon-
strated the influence of the ends in lowering the energy for
domain nucleation [20]. In fact, previous works showed that the
magnetization reversal can start in small regions of the wires,
typically ꢀ 2 orders of magnitude smaller than the wire length
[5,7,18,21], which is agreement with this work.
4. Conclusions
Fig. 2. Temperature dependence of Hc for Ni80Fe20 (triangles) and Ni (squares)
nanowires. Full lines show the best fit of Eq. (2) with
data.
a ¼ 1:5 to the experimental
In this work we performed magnetization measurements as a
function of temperature for Ni and Ni80Fe20 electrodeposited
nanowire arrays. The Ni80Fe20 nanowire sample showed the
expected longitudinal anisotropic behavior characteristic of a
single domain nanowire. On the other hand, Ni nanowires exhibit
an almost isotropic behavior with easy and hard axis Hc smaller
than those found in the literature. EDS characterization revealed
an inclusion of Cu, which leads to a decrease of the magnetostatic
interactions, thus lowering the effective anisotropy present in this
sample. The temperature dependence of Hc was also studied in
order to gain some insight into the magnetization reversal
mechanisms of these arrays of nanowires. We were able to
conclude that it deviates from the simple prediction for infinite
cylinders of coherent rotation and curling. Instead, the reversal
description of the application of the model can be found in Ref.
[5]. However, in a simple explanation, the thermally excited
magnetization reversal follows an Arrhenius law, setting the
critical barrier height U0 as
U0 ¼ kBT lnðG0tcÞ.
U0 may depend on the temperature by magnetoelastic effects, but
as we have stated before this is not significant in the present work.
We thus obtain for the temperature dependence of Hc [18]:
(1)
ꢀ
ꢁ
1=a
Hc
kBT lnðG0tcÞ
¼ 1 ꢄ
,
(2)
Hc0
U0
where Hc0 is the coercive field at 0 K, kB is the Boltzmann constant,
T is the temperature, G0 is the attempt frequency (G0ꢀ109 Hz), tc
takes place via domain nucleation triggered by
a thermal
activated process. We also conclude that due to the presence of
imperfections along and at the ends of the nanowires the volume
for reversal nucleation is smaller than the expected for a 180ꢁ
domain wall nucleation and much smaller (ꢀ 2 orders of
magnitude) than the wire volume.
is the measurement time (tcꢀ100 s) and the exponent
a
depends
¼ 2 for coherent
¼ 1:5 for domain wall
on the reversal processes presented
magnetization reversal processes;
(a
a
nucleation). This concept of magnetization reversal triggered by
a thermal activated process gives rise to a volume (Veff ) where the
reversed domain starts to nucleate [18]:
Veff ¼ U0kB=Keff
.
(3)
Acknowledgments
Next we argue about the applicability of this model to our
Work supported in part by projects FEDER/POCTI/n2-155/94.
D.C. Leitao and C.T. Sousa are thankful to FCT for doctoral Grants
SFRH/BD/25536/2005 and SFRH/BD/38290/2007. J.P. Araujo also
thanks the Fundacao Gulbenkian for its financial support within
the ‘‘Programa Gulbenkian de Estimulo a Investigacao Cientifica’’.
experimental results of HcðTÞ. First, we performed the fitting of
Eq. (2) with parameter
coherent rotation processes. Since this is a delocalized reversal
mode, meaning that the reversal takes place all over the sample,
a ¼ 2 (not shown) which holds for
the effective volume ðVeff
Þ
is equal to the wire volume
3
[Vwire ꢃ ð108 nmÞ ]. However, the Veff values obtained from the
fits were much smaller than the wire volumes. In fact, the
theoretically predicted delocalized coherent rotation or curling
modes [19] fail in the case of real nanowires mainly due to their
finite dimensions, existence of imperfections and interactions
among nanowires [7].
References
[1] F. Casanova, C.E. Chiang, C.P. Li, I.V. Roshchin, A.M. Ruminski, M.J. Sailor,
I.K. Schuller, Nanotechnology 19 (2008) 31.
[2] Z.L. Xiao, C.Y. Han, U. Welp, H.H. Wang, V.K. Vlasko-Vlasov, W.K. Kwok, D.J.
Miller, J.M. Hiller, R.E. Cook, G.A. Willing, G.W. Crabtree, Appl. Phys. Lett. 81
(2002) 15.
[3] H. Masuda, H. Asoh, M. Watanabe, K. Nishio, M. Nakao, T. Tamamura, Adv.
Mater. 13 (2001) 3.
[4] D. Navas, M. Hernandez-Velez, M. Vazquez, W. Lee, K. Nielsch, Appl. Phys. Lett.
90 (2007) 19.
[5] P.M. Paulus, F. Luis, M. Kroll, G. Schmid, L.J. de Jongh, J. Magn. Magn. Mater.
224 (2001) 2.
[6] R. Skomski, H. Zeng, M. Zheng, D.J. Sellmyer, Phys. Rev. B 62 (2000) 6.
[7] D.J. Sellmyer, M. Zheng, R. Skomski, J. Phys. Condens. Matter 13 (2001) 25.
[8] K. Nielsch, R.B. Wehrspohn, J. Barthel, J. Kirschner, U. Gosele, S.F. Fischer, H.
Kronmuller, Appl. Phys. Lett. 79 (2001) 9.
Instead, we conclude that inhomogeneous localized modes
occur. Although usually energetically unfavorable in perfect
nanowires, structural inhomogeneities and geometrical features
at the ends lead to a local energy decrease, favoring these reversal
processes [7,6,19]. In Fig. 2 the full lines represent the fit of HcðTÞ
to Eq. (2) using as a fitting parameter U0 and
a ¼ 1:5 for DW
nucleation processes. For this case, the expected effective volume
of nucleation should correspond to the nucleation of a 180ꢁ DW
3
that accordingly to Ref. [5] is about ð35 nmÞ for Ni80Fe20 and
[9] M. Vazquez, K. Pirota, M. Hernandez-Velez, V.M. Prida, D. Navas, R. Sanz,
F. Batallan, J. Velazquez, J. Appl. Phys. 95 (2004) 11.
3
ð40 nmÞ for Ni. For Ni80Fe20 we have obtained an energy barrier of

ARTICLE IN PRESS
1322
D.C. Leitao et al. / Journal of Magnetism and Magnetic Materials 322 (2010) 1319–1322
´
[10] L. Carignan, C. Lacroix, A. Ouimet, M. Ciureanu, A. Yelon, D. Menard, J. Appl.
[17] The samples are only supported underneath by the remaining aluminum foil.
The aluminum that many times is left on the sides of the alumina matrix and
that causes the high effects of the magnetoelastic anisotropy, specially on Ni,
was removed.
[18] W. Wernsdorfer, B. Doudin, D. Mailly, K. Hasselbach, A. Benoit, J. Meier,
J.P. Ansermet, B. Barbara, Phys. Rev. Lett. 77 (1996) 9.
[19] A. Aharoni, Introduction to the Theory of Ferromagnetism, second ed., Oxford
Science Publications, 2000.
Phys. 102 (2007) 023905.
[11] K. Nielsch, F. Muller, A.P. Li, U. Gosele, Adv. Mater. 12 (2000) 8.
[12] H. Masuda, K. Fukuda, Science 268 (1995) 5216.
[13] D.C. Leitao, C.T. Sousa, J. Ventura, F. Carpinteiro, J.G. Correia, M.M. Amado, J.B.
Sousa, J.P. Araujo, Phys. Status Solidi C 5 (2008) 3488–3491.
[14] K. Nielsch, J. Choi, K. Schwirn, R.B. Wehrspohn, U. Gosele, Nano Lett. 2 (2002)
7.
[15] M. Vazquez, K. Pirota, J. Torrejon, D. Navas, M. Hernandez-Velez, J. Magn.
Magn. Mater. 294 (2005) 2.
[20] R. Ferre, K. Ounadjela, J.M. George, L. Piraux, S. Dubois, Phys. Rev. B 56 (1997)
21.
[16] D. Navas, K.R. Pirota, P.M. Zelis, D. Velazquez, C.A. Ross, M. Vazquez, J. Appl.
Phys. 103 (2008) 7.
[21] H. Zeng, R. Skomski, L. Menon, Y. Liu, S. Bandyopadhyay, D.J. Sellmyer, Phys.
Rev. B 65 (2002) 13.