Journal of Magnetism and Magnetic Materials 249 (2002) 241–245
Magnetic properties of ferromagnetic nanowires embedded in nanoporous alumina membranes . a,*, W.J. Blaua, D. Grandjeanb, R.E. Benfieldb, F. Luisc, M. Kroll P.M. Paulusc, L.J. de Jonghc a Physics Department, Trinity College, Dublin, Ireland Centre for Materials Research, University of Kent, Canterbury, UK c Kamerlingh Onnes Laboratorium, Universiteit Leiden, The Netherlands b
Abstract Iron, nickel and cobalt nanowires are prepared within the pores of nanoporous alumina membranes using an electrochemical AC plating procedure. Nanowires produced in this way can be easily varied in diameter (5–250 nm) and length (up to several hundred microns). The magnetisation curves for these nanowire/alumina composites can then be determined not only as a function of the temperature but also as a function of the wire diameter and length. Conclusions regarding the magnetisation reversal processes that take place in the wires can be drawn. For Fe and Ni nanowires, we show that the magnetisation process in wires with a diameter smaller than the domain wall width is independent of the wire length and probably takes place via the formation of a small magnetic domain at the end of the wires and a subsequent propagation of the domain wall along the wire. For Co nanowires a competition between the shape anisotropy and the temperature- and size-dependent magnetocrystalline anisotropy could be observed. r 2002 Published by Elsevier Science B.V. PACS: 75.30.Gw; 75.50.Tt; 75.60. d; 61.46.+w Keywords: Nanowires; Magnetic anisotropy; Magnetisation reversal; Crystallographic structure
1. Introduction One-dimensional ferromagnetic nanostructures are of significant interest because of their possible application as high-density magnetic data storage devices [1,2]. A nanowire can be considered as being one-dimensional if its diameter is in the range of the domain wall width or even smaller and its length is much larger than this [3]. *Corresponding author. Tel.: +353-1-6081-469; fax: +3531-6711-759. . E-mail address:
[email protected] (M. Kroll).
In this paper we present investigations on the magnetic properties of iron, nickel and cobalt nanowires prepared within the pores of an anodically prepared alumina membrane. The average pore diameter can be adjusted easily within the range from 5 to 250 nm by varying the anodising voltage applied. The metals are then plated using an AC plating procedure. Using this approach one gets metal wires with adjustable diameters from 5 to 250 nm. The length of the wires can be controlled by the metal plating time. We prepared wires with average diameters between 6 and 50 nm. Since the domain wall widths for Fe,
0304-8853/02/$ - see front matter r 2002 Published by Elsevier Science B.V. PII: S 0 3 0 4 - 8 8 5 3 ( 0 2 ) 0 0 5 3 7 - 1
242
M. Kroll . et al. / Journal of Magnetism and Magnetic Materials 249 (2002) 241–245
Ni and Co are in the range from 10 to 50 nm, the nanowires discussed here can be regarded as being one-dimensional.
3
parallel perpendicular
1
2
M (Am /kg)
2
0 -1 Fe 12 nm
-2 -3 parallel perpendicular
0.5
2
Nanoporous alumina membranes are prepared by potentiostatic anodic oxidation of aluminium plates or foils in polyprotic acids. Details of the membrane preparation and properties are described elsewhere [4,5]. Metallic nanowires are prepared by plating the metal into the pores of the membrane using an appropriate electrolyte and a constant AC voltage [2,6]. Magnetic measurements are performed within the temperature range from 3 to 300 K and at fields between –5 and +5 T using a SQUID magnetometer. HED investigations were done on beamline ID15 at ESRF in Grenoble/ France.
M (Am /kg)
2. Experimental details
0.0
Ni 12 nm -0.5
3. Magnetic properties Details of the magnetic properties of Fe, Ni, and Co nanowires are described elsewhere [7]. The magnetisation curves for Fe and Ni nanowires within porous alumina membranes with a mean pore diameter of 12 nm are presented in Fig. 1. If the magnetic field is applied parallel to the long axes of the wires one can observe an almost square hysteresis curve as was predicted for infinite long cylinders [8,9]. According to the model there should be no hysteresis if the external field is applied perpendicular to the long axes. This behaviour is clearly seen in Fig. 1 indicating two stable orientations of the magnetic moments, namely pointing parallel and antiparallel to the long axis of the wire. These two orientations are separated by an energy barrier. The slight hysteresis that can be seen for the perpendicular measurement is probably due to a slight misalignment of the wires and due to the influence of the rather weak magnetocrystalline anisotropy. Furthermore, it is obvious that the hysteresis curves for the parallel measurement are slightly sheared. This can be attributed to dipolar interactions between the wires. These interactions are
-1.0
-0.5
0.0
0.5
1.0
Bap (T) Fig. 1. Hysteresis loops for 12 nm Fe and Ni nanowires measured at 4.2 K.
taken into account using a mean field approximation (for details cf. Ref. [7]). After correcting for the demagnetising field we get squareness ratios of 0.9 to 0.95 for all Fe and Ni wires even at room temperature. For infinite long cylinders there are basically two theoretical predictions for the magnetisation reversal. Both processes include the simultaneous reversal of all magnetic moments alongp the ffiffiffi wire. For wires with diameters DP larger than plW (lW : domain wall width) the magnetisation reversal should takepplace via a curling ffiffiffi mode. For wires with DP o plW the magnetic moment should be reversed in a homogeneous rotation. Both modes can be distinguished by investigating the coercive field BC0 at low temperatures. For the homogeneous rotation or rotation in unison BC0 is predicted to be independent of DP ; whereas for the curling process BC0
M. Kroll . et al. / Journal of Magnetism and Magnetic Materials 249 (2002) 241–245
1
uniform rotation
2Bc0 /µ0 Msb
curling
0.1
0.01 0.1
1
10
Dp /(π 1/2λ w) Fig. 2. Dependence of the coercive field BC0 for Fe (K) and Ni (J) nanowires on the wire diameter. Data for larger wire diameters than those investigated here ( ) are taken from Huysmans et al. [10].
Co nanowires, Dp=25 nm
12
parallel perpendicular
4
2
M (Am /kg)
8
0 T = 314 K || Bc = 0.112 T
-4
⊥
-8
Bc = 0.036 T
-12 12 8
parallel perpendicular
4
2
M (Am /kg)
should be dependent on BC0 : Ferre! et al. [10], however, already found a deviation from the predicted curling modes for Ni nanowires with a diameter of approximately 35 nm in polycarbonate membranes. Fig. 2 shows the values received for the coercive fields at a low temperature for Fe and Ni wires. The data for large wire diameters are taken from Huysmans et al. [11]. For large diameters the experimental data fit very well to the predicted curve for the curling process. For small wire diameters the coercive field is nearly independent of the wire diameter as was predicted for rotation in unison. The values, however, are too low by a factor of three. For a given wire diameter the coercive field is independent of the wire length. Investigations on the activation energy and the magnetic viscosity, not shown here [7], prevail that magnetisation reversal in these nanowires takes place via a nucleation of a small domain probably at the end of the wires or at imperfections and a subsequent propagation of the domain wall along the wire. Cobalt nanowires not only show the strong shape anisotropy that can be observed for Fe and Ni wires, but also a temperature- and sizedependent magnetocrystalline anisotropy along the hexagonal c-axis of HCP-Co. This c-axis is known to be perpendicular to the long axis of the wire [12]. Therefore at low temperatures a
243
0 T=5K || Bc = 0.100 T
-4 -8
⊥
Bc = 0.076 T
-12 -1.0
-0.5
0.0 Bap (T)
0.5
1.0
Fig. 3. Magnetisation curves for Co nanowires at 314 and 5 K.
competition between shape anisotropy and magnetocrystalline anisotropy can be observed (cf. Fig. 3). At room temperature, however, the magnetic properties are predominated by the shape anisotropy. This can be explained by the strong temperature dependence of the magnetocrystalline anisotropy constant. The competition between the two kinds of anisotropies is also controlled by the diameter of the wires. The smaller the pore diameter the less important the magnetocrystalline anisotropy becomes even at low temperatures. High-energy X-ray diffraction (HED) and EXAFS measurements show that the Co nanowires consist of two different Co phases, the hexagonal cobalt phase which shows a strong magnetocrystalline anisotropy, and a cubic Co phase which in the bulk is only known as a high temperature modification (stable above 661 K) (cf. Fig. 4) [13]. This coexistence of the two Co phases was also shown by other groups (see, e.g. Refs. [12,14]). The ferromagnetic FCC modification, however, does not show a significant magnetocrystalline anisotropy. Since the amount of
244
M. Kroll . et al. / Journal of Magnetism and Magnetic Materials 249 (2002) 241–245
the pores is much more affected by the current density. Experiments to control the Co phase using different current densities are currently under progress.
4. Conclusions
Fig. 4. High-energy X-ray diffraction pattern for Co nanowires in a nanoporous alumina membrane with 48 nm mean pore diameter.
FCC-Co increases if the wire diameter is decreased, the shape anisotropy becomes predominant in small Co wires even at low temperatures. In contrast to other groups [10] we found that the position of the hexagonal c-axis does not change with the diameter, but remains more or less constant at about 801 relative to the long axis of the wires. Therefore only the varying amounts of FCC- and HCP-Co can be made responsible for the results obtained. It is still unclear what causes the amount of FCC-Co to increase in small wires. Recent attempts to control the Co phase and the orientation of certain axes like the hexagonal caxis by applying an external magnetic field (approx. 0.35 T) during the metal plating have failed. The increasing ratio of surface to volume atoms in small wires is likely to influence the thermal properties like the melting point or the transition temperature between two different Co phases. Quantitative analysis of this correlation is, however, still to be done. Furthermore, the local pH during electroplating is known to be responsible for the Co phase. In our experiments the pH of the solution is kept constant at 3.5 (composition of the electrolyte: 50 g/l CoSO4 7H2O, 25 g/l H3BO3, 20 g/l glycerine, pH adjustment by H2SO4). However, the local pH, i.e. the pH inside
The magnetic properties of Fe and Ni wires are dominated by a strong shape anisotropy with the easy axis parallel to the long axis of the wires. Low temperature investigations on the coercive field show that magnetisation reversal in nanowires with a diameter significantly larger than the domain wall width takes place via a curling mode. For small nanowires theoretical predictions for infinite long wires need to be modified. Here magnetisation reversal probably includes the formation of a small domain at the end of the wires and the propagation of the domain wall along the wire. The magnetic properties of Co wires are dominated by a temperature- and size-dependent competition between shape anisotropy and magnetocrystalline anisotropy. HED measurements show varying amounts of FCC- and HCP-Co in wires with different diameters. Different mixtures of these two Co phases cause different magnetic properties. Attempts to control the Co phase and the orientation of certain axes are in progress.
Acknowledgements M.K. acknowledges a grant funded by the Deutsche Forschungsgesellschaft (DFG).
References [1] N. Tsuya, Y. Saito, H. Nakamura, S. Hayano, A. Furugohri, K. Ohta, Y. Wakui, T. Tokushima, J. Magn. Magn. Mater. 54–57 (1986) 1681. [2] S. Kawai, I. Ishiguro, J. Electrochem. Soc. 123 (7) (1976) 1047. [3] L.J. de Jongh (Ed.), Physics and Chemistry of Metal Cluster Compounds, Kluwer Academic Publishers, Dordrecht, 1994.
M. Kroll . et al. / Journal of Magnetism and Magnetic Materials 249 (2002) 241–245 [4] G.E. Thompson, R.C. Furneaux, G.C. Wood, J.A. Richardson, J.S. Goode, Nature 272 (1978) 433. [5] R.C. Furneaux, W.R. Rigby, A.P. Davidson, Nature 337 (1989) 147. . H.-P. Kormann, [6] T.-A. Hanaoka, A. Heilmann, M. Kroll, T. Sawitowski, G. Schmid, P. Jutzi, A. Klipp, U. Kreibig, R. Neuendorf, Appl. Organomet. Chem. 12 (1998) 367. . G. Schmid, L.J. de Jongh, [7] P.M. Paulus, F. Luis, M. Kroll, J. Magn. Magn. Mater. 224 (2001) 180. [8] E.H. Frei, S. Shtrikman, D. Treves, Phys. Rev. 106 (3) (1957) 446. [9] A. Aharoni, S. Shtrikman, Phys. Rev. 109 (5) (1958) 1522.
245
[10] R. Ferr!e, K. Ounadjela, J.M. George, L. Piraux, S. Dubois, Phys. Rev. B 56 (21) (1997) 14066. [11] G.T.A. Huysmans, J.C. Lodder, J. Wakui, J. Appl. Phys. 64 (4) (1988) 2016. [12] G.J. Strijkers, J.H.J. Dalderop, M.A.A. Broeksteeg, H.J.M. Swagten, W.J.M. de Jonge, J. Appl. Phys. 86 (9) (1999) 5141. . [13] R.E. Benfield, D. Grandjean, M. Kroll, R. Pugin, T. Sawitowski, G. Schmid, J. Phys. Chem. B 105 (10) (2001) 1961. [14] A. Fert, L. Piraux, J. Magn. Magn. Mater. 200 (1999) 338.
