Home Search Collections Journals About Contact us My IOPscience

Geometry dependence of coercivity in Ni nanowire arrays

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2008 Nanotechnology 19 075713

(http://iopscience.iop.org/0957-4484/19/7/075713)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 139.184.30.133
The article was downloaded on 07/10/2012 at 19:34

Please note that terms and conditions apply.

IOP PUBLISHING NANOTECHNOLOGY
Nanotechnology 19 (2008) 075713 (6pp) doi:10.1088/0957-4484/19/7/075713

Geometry dependence of coercivity in

Ni nanowire arrays
J Escrig1 , R Lavı́n1 , J L Palma1 , J C Denardin1, D Altbir1 ,
A Cortés2 and H Gómez2
1
Departamento de Fı́sica, Universidad de Santiago de Chile, USACH, Avenida Ecuador 3493,
917-0124 Santiago, Chile
2
Instituto de Quı́mica, Facultad de Ciencias, Universidad Católica de Valparaı́so,
Casilla 4059, Valparaı́so, Chile

E-mail: [email protected]

Received 6 November 2007, in final form 12 December 2007

Published 31 January 2008
Online at stacks.iop.org/Nano/19/075713

Abstract
Magnetic properties of arrays of nanowires produced inside the pores of anodic alumina
membranes have been studied by means of vibrating sample magnetometer techniques. In these
systems the length of the wires strongly influences the coercivity of the array. A simple model
for the coercivity as a function of the geometry is presented which exhibits good agreement
with experimental results. Magnetostatic interactions between the wires are responsible for a
decrease of the coercive field.
(Some figures in this article are in colour only in the electronic version)

1. Introduction the understanding of the magnetization reversal mechanisms

is a permanent challenge for researchers. It has been reported
During recent years magnetic nanoparticles have attracted independently by Forster et al [20] and Hertel [21] that
increasing interest due to their promising applications in soft magnetic nanowires can switch in two different modes
hard disk drives, magnetic random access memory and depending on their thickness. These modes have become
other spintronic devices [1–4]. Also, these elements are known as the transverse reversal mode and the vortex wall
good candidates for biomedical applications. The use of mode. The first refers to a thin ferromagnetic wire (diameter
these particles in devices and architectures is conditioned d < 60 nm). If a wire is thin enough, the exchange interaction
on the possibility of having a fine control of their size forces the magnetization to be homogeneous through any radial
and shape, keeping thermal and chemical stability of the cross section of the cylinder [22]. Similar reversal modes were
nanoparticles [5]. For such applications, different geometries found by Landeros et al [23] in the case of magnetic nanotubes.
have been considered [6–10]. In particular, highly ordered For small particles, the Stoner–Wohlfarth model has
arrays of magnetic nanowires produced inside the pores of been applied to calculate the coercivity when the reversal of
anodic alumina membranes by electrochemical deposition [11] magnetization is driven by a coherent rotation [24]. Also
have been the focus of intense research [12–15]. The high Aharoni [25] calculated the nucleation field in a prolate
ordering, together with the magnetic nature of the wires, gives spheroid when the reversal is driven by curling rotation.
rise to outstanding cooperative properties different from the However, when the transverse reversal mode appears, the
bulk and even from film systems. In these arrays, inter- coercivity has not been modelled yet. Thus, usually
element interactions play an important role and have been experimental results for the coercivity in this case are
the subject of strong investigation. Kazadi Mukenga Bantu interpreted using the Stoner–Wohlfarth model [26].
et al [16] and Rivas et al [17] investigated the structure and In this paper we present experimental results for high
magnetic properties of cobalt nanowires and observed a change aspect ratio Ni nanowire arrays. Based on micromagnetic
of the easy axis depending on the magnetostatic interactions calculations, the coercivity in the presence of a transverse
between them. More recently, Gubbiotti et al [18] and Goolaup reversal mode is modelled, showing good agreement with
et al [19] reported that the switching behaviour of the wires experiments. Magnetostatic interactions are included in order
is affected by magnetostatic interactions. In such particles, to investigate its effect on the magnetic properties of the array.

0957-4484/08/075713+06$30.00 1 © 2008 IOP Publishing Ltd Printed in the UK

Nanotechnology 19 (2008) 075713 J Escrig et al

Figure 1. SEM top view of a highly ordered home-made AAO Figure 2. SEM image of the exposed Ni nanowires with diameter of
template. 50 nm after the alumina template membrane was dissolved away.
Magnification: 17 000×.

2. Experimental methods

The anodic alumina oxide (AAO) porous membranes were

prepared from a 99.99% aluminium foil (0.13 mm thickness,
Sigma-Aldrich) by the two-step anodization technique [27].
The grease on the aluminium surface was removed with
detergent, then successively with acetone and water. The
cleaned aluminium sheets were submitted to an annealing at
350 ◦ C in air for 1 h using a Lindgerg/Blue M model tube
furnace. Then, the aluminium sheets were first etched with
a 5% w/w NaOH solution and afterwards with diluted nitric
acid. Subsequently, the samples were mechanically polished
with alumina (0.3 and 0.05 μm mesh) followed by 1 min
of electropolishing at 15 V in a 40% H2 SO4 , 59% H3 PO4 ,
1% glycerine bath. After this treatment the samples were
submitted to a first anodization at 40 V for 6 h in a 0.3 M oxalic Figure 3. TEM image of free-standing Ni nanowires that have been
grown at −0.9 V.
acid solution at 20 ◦ C. The anodized layer was etched with a
5% H3 PO4 and 1.8% H2 Cr2 O4 solution at room temperature
for 12 h. The ordered pore arrangement was achieved with
a second anodization step that was performed in the same A JEOL 5900 LV apparatus was used for SEM measurements.
conditions as in the first one. A 0.1 M CuCl2 and 20% v/v Figure 2 shows the SEM images of nickel nanowires grown
HCl solution at room temperature was employed to remove the into the 50 nm home-made alumina template supported on the
remaining aluminium from the alumina substrate. To remove substrate. This figure shows that the nanowires have a high
the barrier layer and to open the pores at the bottom, the aspect ratio, are well defined, parallel and close to each other.
membrane was first treated with 5% H3 PO4 and 5% NaOH TEM observations were carried out on a JEOL 2010F
aqueous solution at room temperature. Subsequently, the pores electron microscope operated at 200 kV coupled with an
were widened in a 0.085 M H3 PO4 solution at 37 ◦ C for 15 energy-dispersive spectrometer Princeton Gamma Technology
min. As is observed in figure 1, the membranes present highly system. A 1 nm diameter probe was used for the microanalysis.
ordered pores of diameter d = 2 R = 50 nm and lattice Samples were prepared by crushing glass samples in an agate
constant D = 100 nm. To facilitate the electrical contact, mortar and ultrasonically stirring them in butanol. The smallest
a very thin Au–Pd layer was sputtered on one side of the particles are then collected in the supernatant liquid using a
membrane followed by the electrodeposition [28] of a thicker holey C-coated copper grid. Figure 3 shows a TEM image
nickel layer to achieve the full pore sealing. In this paper of continuous free-standing nanowires. Both analyses account
the electrodeposition was performed at a constant potential for a continuous growth that follows the shape of the pores’
(dc electrodeposition at −0.9 V). The potentiostatic condition template.
allows us to have more precise control of the electrochemical Chemical characterization of nanowires has been realized
reaction, and then a more accurate control of the growth of the by means of energy-dispersive analysis of x-rays (EDAX).
Ni nanowires. Figure 4 shows the spectroscopy data. The peaks correspond
The morphology and the single-crystalline structure to characteristic elemental emissions of Ni.
of individual nanowires were subsequently investigated The magnetic properties of the Ni nanowires were
by scanning electron microscopy (SEM) and transmission measured by a vibrating sample magnetometer (VSM).
electron microscopy (TEM) after dissolution of the template. Figure 5 illustrates the hysteresis cycles for two samples with

2
Nanotechnology 19 (2008) 075713 J Escrig et al

3000 80

Coercive field / Oe

Wall width w / nm
2500
W
70
2000

1500
60
1000

500 50

Radius R / nm
Figure 4. EDAX spectroscopy of a selected area in the Ni nanowire
array. The peaks correspond to characteristic elemental emissions of Figure 6. Coercivity (solid line, equation (1)) and wall width (dotted
nickel. line) as a function of the radius of an isolated Ni nanowire. This
result is independent of the length of the wire.

to the other (M = −M0 ẑ), occurs by means of the propagation

of a transverse domain wall [23]. In this case, a domain wall
appears at one end of the wire and propagates towards the
other. Starting from the equations presented by Landeros et al
[23], we can calculate the width of the domain wall for the
transversal reversal mode as a function of the radius, R .
To calculate the coercive field of an isolated magnetic
µ wire, H0, we used an adapted Stoner–Wohlfarth model [24]
µ in which the length of the coherent rotation is replaced by the
width of the domain wall, w. Following this approach,

H0 2 K (w)
= , (1)
Figure 5. Hysteresis cycles for Ni nanowires (50 nm diameter and
M0 μ0 M02
100 nm interwire distance) as a function of the length. The magnetic
field is applied parallel to the wire axis. where K (l) = 14 μ0 M02 (1 − 3 Nz (l)) and Nz (l) corresponds
to the demagnetizing factor along z [29], given by Nz (l) =
2
1 − F21 [ 4lR2 ] + 38πl
R
, where F21 [x] = F21 [−1/2, 1/2, 2, −x] is
diameter d = 50 nm, lattice parameter D = 100 nm and a hypergeometric function.
lengths L = 4 and 12 μm, measured with the external The dependence of the coercivity H0 (solid line) and width
field along the axis of the wires. From the comparison of of the domain wall (dotted line) is depicted in figure 6 as
these loops, an increase of the coercivity and the remanence a function of the radius of the isolated wire. In this figure
in relation to the length has been observed. Therefore we have considered Ni nanowires defined by M0 = 4.85 ×
we observed that the magnetic behaviour of the arrays 105 A m−1 and the stiffness constant A = 10−11 J m−1 . In
is strongly dependent on the effective magnetic anisotropy equation (1) the length involved is the width of the domain
(mainly determined by shape). wall. However, Landeros et al [23] have shown that the width
of the domain wall, w, is independent of the length of the wire,
3. Model and discussion L , provided the wire is long enough. Then, the coercivity in
figure 6 is independent of the length of the wire. However,
In order to investigate this behaviour, we propose a model in figure 5 we have seen a strong dependence of the coercive
which leads us to calculate the coercivity as a function of the field on the length of the wires. Moreover, the value computed
geometry. We start calculating the coercive field of an isolated from equation (1) is greater than the experimental data. Then
magnetic nanowire assuming that the magnetization reversal is it is clear that the array cannot be considered simply as a set of
driven by a transverse mode. independent wires and interactions have to be included in order
to obtain a better agreement with experiments.
3.1. Isolated magnetic nanowires
3.2. Array of magnetic nanowires
For isolated magnetic Ni nanowires with diameters smaller
than 60 nm, the magnetization reversal, that is, the change of If each individual nanowire is treated as a non-interacting
the magnetization from one of its energy minima (M = M0 ẑ) magnetic dipole, it will contribute to the whole hysteresis loop

3
Nanotechnology 19 (2008) 075713 J Escrig et al

of the array with a small square-shaped loop. Consequently, in

the case of an array of identical non-interacting wires, a square
macroscopic hysteresis loop would be observed with a single
Barkhausen jump. Nevertheless, in the array, the distances
between the wires are smaller or comparable to their diameter
and/or length. Then the wires in the array are subject to
important magnetostatic interactions between them that must
be considered. The interaction of each wire with the stray
fields produced by the array—an effective antiferromagnetic
coupling between neighbouring wires—strongly influences the
coercivity [30–33]. This stray field depends on the length of the
wires, explaining in this way our results in figure 5. In these
interacting systems, the process of magnetization reversal can
be viewed as the overcoming of a single energy barrier, E .
In an array with all the wires initially magnetized in the same Figure 7. Calculated coercivity obtained from equation (2) using
direction, the magnetostatic interaction between neighbouring ε = 20 (black solid line corresponds to R = 15 nm and grey solid
wires favours the magnetization reversal of some of them. A line to R = 25 nm) and stray fields (dotted line corresponds to
reversing field aligned opposite to the magnetization direction R = 15 nm and dashed line to R = 25 nm) as a function of the
length of the Ni nanowire arrays. Red dots correspond to the coercive
lowers the energy barrier, thereby increasing the probability of fields obtained from figure 5. Interwire distance for all cases is
switching. The dependence of the applied field on the energy D = 100 nm.
barrier is often described [34] by the expression
 
H 2
E = U 1 − , by Escrig et al [14], to avoid size-dependent results, we have
H0 considered a sample which contains a minimum of N ≈ 70 000
where H is the applied field and H0 denotes the intrinsic wires. The calculated coercivity Hc and stray field per wire as
coercivity of an isolated wire. For single-domain particles a function of the length of the wires for two different radii are
having a uniaxial shape anisotropy, the energy barrier at zero depicted in figure 7, where ε = 20 has been used. A strong
applied field, U , is just the energy required to switch by influence of the length is observed which can be ascribed to
coherent rotation, K (L). If we assume that the switching field the strength of the magnetostatic interaction among nanowires,
Hs is equal to Hc , then which in turn depends on the length. At small lengths an almost
linear dependence of the coercivity is observed, which was
Hc = H0 − Hint, (2) previously measured by Vázquez et al [37]. However, in dot
arrays, d > L , a different behaviour is observed in which
where Hint corresponds to the stray field induced within the
case the coercivity decreases ( Hint increases with L ) [38].
array given by
In figure 7, dots illustrate the coercive fields obtained from
 1/2 the hysteresis loops shown in figure 5. A good agreement
2 K (L) ε| Ẽ int (D)|
Hint = . (3) between the coercivities obtained with VSM and our analytical
μ0 M02 K (L) calculations is shown.
We now investigate the dependence of the coercivity on
In the previous equation we have assumed that the reversal of the diameter of the wires, d , and the interwire distance, D ,
individual nanowires produces a decrease of the magnetostatic by defining the ratio d/D . When the wires are in contact,
energy E int that equals the magnetic anisotropy barrier E . d/D = 1; when they are infinitely separated, d/D = 0. Our
Besides, ε is an adjustable parameter that depends on the results are shown in figure 8 for two different radii and four
distribution of magnetic wires in space and on the long- different lengths, where ε = 20 has been used. In the range of
distance correlation among the wires. The value of ε cannot be parameters considered, we observe that an increase of the ratio
obtained from first principles, although values between unity d/D results in a decrease of the coercivity. As also shown
and some tens are reasonable [35]. Besides, Ẽ int (D) is the in figure 7, an increase in the length of the wires produces an
magnetostatic interaction between two nanowires separated a
increase of the coercivity.
distance D . Such an interaction has been calculated by Laroze
Labelled dots 1, 2, 3 and 4 in figure 8 correspond to
et al [36] and is given by
the cases described in table 1. Note the agreement between
⎛ ⎞ experimental and calculated values through equation (2).
E int (D) μ0 M02 R 2 ⎝ 1
⎠. Deviations between experimental measurements and analytical
Ẽ int (D) ≡ = 1− 
V 2L D 1+ L2 results can originate from the dispersion of the lengths and
D2
positions of each wire in the array and a reduction in the
In order to understand the role of the length of the wires in homogeneity of the diameter of nanopores [37]. Also it is
the array we calculate the normalized magnetostatic interaction important to point out that in our model we have not considered
energy per wire  (no dimensional stray field) for arrays of
N −1  N
any kind of crystalline anisotropy, which can be present,
different lengths, i= 1 j =i+1 ( Ẽ int (D)/μ0 M0 N). As stated
2
depending on the different synthesis procedures. The addition

4
Nanotechnology 19 (2008) 075713 J Escrig et al

References
µ
µ
µ [1] Koch R H, Deak J G, Abraham D W, Trouilloud P L,
µ Altman R A, Lu Yu, Gallagher W J, Scheuerlein R E,
Roche K P and Parkin S S P 1998 Phys. Rev. Lett.
81 4512–5
[2] Cowburn R P, Koltsov D K, Adeyeye A O, Welland M E and
Tricker D M 1999 Phys. Rev. Lett. 83 1042–5
[3] Wolf S A, Awschalom D D, Buhrman R A, Daughton J M,
von Molnar S, Roukes M L, Chtchelkanova A Y and
Treger M 2001 Science 294 1488–95
[4] Gerrits Th, van der Berg H A M, Hohlfeld J, Bar L and
Rasing Th 2002 Nature 418 509–12
[5] Puntes V F, Krishnan K M and Alivisatos A P 2001 Science
291 2115
[6] Landeros P, Escrig J, Altbir D, Laroze D,
Figure 8. Dependence of calculated coercivity with the ratio of the d’Albuquerque e Castro J and Vargas P 2005 Phys. Rev. B
diameter, d = 15 nm (black lines) and 25 nm (grey lines), to lattice 71 094435
parameter, D . We have used ε = 20. Experimental points 1, 2, 3 and [7] Wachowiak A, Wiebe J, Bode M, Pietzsch O,
4 are discussed in the text. Morgenstern M and Wiesendanger R 2002 Science
298 577–580
[8] Landeros P, Escrig J, Altbir D, Bahiana M and
Table 1. Parameters for different Ni nanowire arrays. Geometrical d’Albuquerque e Castro J 2006 J. Appl. Phys. 100 044311
parameters and Hc have been measured in this paper (superscript 1) [9] Bachmann J, Jing J, Knez M, Barth S, Shen H, Mathur S,
or taken from [39] (superscript 2) and [40] (superscript 3). Gosele U and Nielsch K 2007 J. Am. Chem. Soc.
129 9554–5
Dot d (nm) D (nm) L (μm) Hc (Oe) [10] Escrig J, Altbir D and Nielsch K 2007 Nanotechnology
(2) 18 225704
1 30 100 ≈1 1200 [11] Masuda H and Fukuda K 1995 Science 268 1466
2( 1 ) 50 100 ≈12 624 [12] Nielsch K, Wehrspohn R B, Barthel J, Kirschner J, Gosele U,
3( 1 ) 50 100 ≈4 366 Fischer S F and Kronmuller H 2001 Appl. Phys. Lett.
4( 3 ) 50 65 ≈3 420 79 1360
[13] Wang Z K, Lim H S, Zhang V L, Goh J L, Ng S C, Kuok M H,
Su H L and Tang S L 2006 Nano Lett. 6 1083–6
of such anisotropy does not modify qualitatively the results, [14] Escrig J, Altbir D, Jaafar M, Navas D, Asenjo A and
Vázquez M 2007 Phys. Rev. B 75 184429
and its effect is equivalent to changing the aspect ratio of the [15] Liu M, Lagdani J, Imrane H, Pettiford C, Lou J, Yoon S,
wire. Harris V G, Vittoria C and Sun N X 2007 Appl. Phys. Lett.
90 103105
[16] Kazadi Mukenga Bantu A, Rivas J, Zaragoza G,
4. Conclusions Lopez-Quintela M A and Blanco M C 2001 J. Appl. Phys.
89 3393
[17] Rivas J, Kazadi Mukenga Bantu A, Zaragoza G, Blanco M C
In conclusion, by means of analytical calculations and
and Lopez-Quintela M A 2002 J. Magn. Magn. Mater.
experimental measurements we have investigated the important 249 220–7
role of magnetostatic interaction in the magnetic properties of [18] Gubbiotti G, Tacchi S, Carlotti G, Vavassori P, Singh N,
nanowire arrays. We have derived an analytical expression Goolaup S, Adeyeye A O, Stashkevich A and
that allows us to obtain the coercivity as a function of the Kostylev M 2005 Phys. Rev. B 72 224413
[19] Goolaup S, Adeyeye A O and Singh N 2006 J. Appl. Phys.
magnetostatic interactions present in the array. The effect of 100 114301
the stray field originating from the magnetostatic interactions [20] Forster H, Schrefl T, Scholz W, Suess D, Tsiantos V and
between the wires of the array must be included to obtain a Fidler J 2002 J. Magn. Magn. Mater. 249 181
quantitative agreement between experimental and theoretical [21] Hertel R 2002 J. Magn. Magn. Mater. 249 251
[22] Hertel Riccardo and Kirschner Jurgen 2004 Physica B
results. Our results lead us to conclude that the length
343 206–10
of the wires strongly influences the coercivity of the array. [23] Landeros P, Allende S, Escrig J, Salcedo E, Altbir D and
Good agreement between VSM measurements and theoretical Vogel E E 2007 Appl. Phys. Lett. 90 102501
calculations is obtained. [24] Stoner E C and Wohlfarth E P 1948 Phil. Trans. R. Soc. A
240 599
Stoner E C and Wohlfarth E P 1991 IEEE Trans. Magn.
Acknowledgments 27 3475 (Reprinted)
[25] Aharoni A 1997 J. Appl. Phys. 82 1281
[26] Wegrowe J-E, Kelly D, Franck A, Gilbert S E and
This work was supported by the Millennium Science Ansermet J-Ph 1999 Phys. Rev. Lett. 82 3681
Nucleus Basic and Applied Magnetism P06-022F, Fondecyt [27] da Fontoura Costa L, Riveros G, Gómez H, Cortés A, Gilles M,
Dalchiele E A and Marotti R E 2005 http://arxiv.org/
no. 11070010 and AFOSR no. FA9550-07-1-0040.
cond-mat/0504573
The CONICYT PhD program and Graduate Direction of [28] Cortés A, Palma J L, Denardin J C, Marotti R E, Dalchiele E A
Universidad de Santiago de Chile are also acknowledged. and Gomez H 2007 unpublished

5
Nanotechnology 19 (2008) 075713 J Escrig et al

[29] Beleggia M, Tandon S, Zhu Y and De Graef M 2004 J. Magn. [35] Allia P, Coisson M, Tiberto P, Vinai F, Knobel M,
Magn. Mater. 272–276 e1197 Novak M A and Nunes W C 2001 Phys. Rev. B 64 144420
[30] Adeyeye A O, Bland J A C, Daboo C and Hasko D G 1997 [36] Laroze D, Escrig J, Landeros P, Altbir D, Vazquez M and
Phys. Rev. B 56 3265 Vargas P 2007 Nanotechnology 18 415708
[31] Hertel R 2001 J. Appl. Phys. 90 5752 [37] Vázquez M, Pirota K, Torrejón J, Navas D and
Hernández-Vélez M 2005 J. Magn. Magn. Mater. 294 174
[32] Vázquez M, Pirota K, Hernández-Vélez M, Prida V M,
[38] Grimsditch M, Jaccard Y and Schuller I K 1998 Phys. Rev. B
Navas D, Sanz R, Batallán F and Velázquez J 2004 J. Appl. 58 11539
Phys. 95 6642 [39] Nielsch K, Wehrspohn R B, Barthel J, Kirschner J, Fischer S F,
[33] Bahiana M, Amaral F S, Allende S and Altbir D 2006 Phys. Kronmuller H, Schweinbock T, Weiss D and Gosele U 2002
Rev. B 74 174412 J. Magn. Magn. Mater. 249 234–40
[34] Sharrock M P 1994 J. Appl. Phys. 76 6413–8 [40] Navas D 2006 PhD Thesis Universidad Autónoma