Magnetic properties of template-synthesized cobalt/polymer composite ...

JOURNAL OF APPLIED PHYSICS 98, 034318 共2005兲

Magnetic properties of template-synthesized cobalt/polymer composite nanotubes K. Nielsch, F. J. Castaño, C. A. Ross,a兲 and R. Krishnan Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

共Received 7 July 2004; accepted 9 June 2005; published online 15 August 2005兲 An approach to fabricate ferromagnetic/polymer composite nanotubes has been developed. The surfaces of the pores in self-ordered porous alumina membranes are wetted with a polystyrene or poly-l-lactide layer containing a metallo-organic precursor. Decomposition of the precursor leads to the formation of thin-walled magnetic tubes with diameters of 160– 450 nm and wall thicknesses of a few nanometers. The magnetic properties of the tube arrays are interpreted as a result of the tube morphology and microstructure. © 2005 American Institute of Physics. 关DOI: 10.1063/1.2005384兴 I. INTRODUCTION

Magnetic nanowires and nanoparticles exhibit multifunctional properties and are useful in a broad range of applications such as drug or gene delivery,1,2 magnetorheology, as propellers for a biomolecular motor,3 or for magnetic imaging.4 In contrast, there has been very little work reported on magnetic nanotubes, which represent a class of anisotropic multifunctional nanoparticles. By coating the inner or outer nanotube wall with oxides, polymers, biomolecules, or metals, a range of physical and chemical properties may be realized within a single structure. Additionally, multilayer magnetic nanotubes, where the layers are oriented parallel to the tube length, may be expected to have unique magnetoelectronic properties. Towards this goal, we have developed a general method for the fabrication of thin-walled composite polymer/metal tubes and quantified the magnetic properties of these nanostructures. There is an extensive literature on nanotubes made from carbon, which show a range of electrical and mechanical properties depending on their structure.5 In comparison, the literature on noncarbon nanotubes is much more limited.6 Fabrication of metal oxide nanotubes including MoO2,7 Al2O3,7 V2O5,7,8 TiO2,9–11 SiO2,7,9,12 MnO2,9 ZnO,9 WO3,9 and Co3O4 共Ref. 9兲 has been reported. More recently, data on functionalized inorganic nanotubes, such as palladium, gold, and platinum13–15 or tellurium16 nanotubes for catalytic or electronic applications, single-crystal semiconductor nanotubes made from lnGaAs/ GaAs, SiGe, and GaN 共Refs. 17–19兲 for nanoelectronics, and complex oxide nanotubes such as BaTiO3 共Ref. 20兲 with ferroelectric properties have been published. Moreover, organic nanotubes based on a variety of polymers21–25 have also been synthesized, for example, polymethylmethacrylate, polystyrene, Teflon®, and conducting polymers including polypyrole and polyaniline 共PANI兲. In most cases a template-mediated fabrication process is used for the synthesis of nanotubes. Nanotubes can be prepared by templating with sacrificial nanowires,11,15,19 a兲

Author to whom correspondence should be addressed; electronic mail: [email protected]

0021-8979/2005/98共3兲/034318/6/$22.50

viruses,26 or biomolecular microtubules27 in solution, or by using a nanopore template such as porous alumina or a tracketched polymer membrane, which is then coated internally with the material of choice7,9,15,16,21–24 Additionally, nanotubes can be obtained by the roll up of thin films deposited on a sacrificial layer17,18 or by self-assembly.25 However, the preparation of ferromagnetic nanotubes has presented a major synthetic challenge. Cobalt and nickel nanotubes have been grown on biomolecular microtubules by adsorption of a Pd or Pt catalyst, followed by electroless deposition.27 The tobacco mosaic virus was used as a template for the precipitation of iron oxide nanotubes.26 Co and Fe nanotubes were made by pulsed electrodeposition in track-etched membranes.28 Commercially available alumina membranes have been used as templates for Ni and Co nanotube synthesis by dc electrodeposition29,30 with an average tube diameter of 200 nm 共±50% 兲 and a wall thickness of up to 30 nm. In another approach, metal salts are infiltrated into alumina membranes, followed by hydrogen reduction to form FePt and Fe2O3 tubes.31 These papers report the magnetic properties of nanotube arrays formed in disordered porous alumina with a fixed pore diameter of ⬃200 nm. This article describes the fabrication and magnetic properties of high-quality, thin-walled ferromagnetic nanotubes and ferromagnetic/polymer composite nanotubes with diameters of 160 nm and above. For the analysis of the magnetic properties, we have varied the thickness and diameter of our tubes. The composite nature of the tubes may be suitable for applications in biotechnology, where low-density magnetic nanostructures are required. II. NANOTUBE SYNTHESIS

Alumina membranes with 500-nm interpore distance were fabricated in-house by a two-step anodization process of bulk aluminum substrates using 0.15M phosphoric acid 共H3PO3兲 at 195 V for 24 h, followed by a second anodization for 20 h.32–34 Subsequently, the alumina pores were widened from their initial diameter of about 180 nm up to 450 nm by isotropic chemical etching in 5-wt % H3PO3 for up to 3 h at 30 ° C. The thickness of the alumina pore structure was 50– 100 ␮m. Additional porous alumina templates

98, 034318-1

© 2005 American Institute of Physics

Downloaded 22 Aug 2005 to 195.37.184.165. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

034318-2

Nielsch et al.

were made using a process based on imprint lithography. An array of small pits was prepatterned into an evaporated aluminum film on a silicon substrate by imprinting, and subsequently anodized.33 These alumina templates had a thickness of 1.5 ␮m and pores with a diameter of 160 nm, arranged in a regular hexagonal array with an interpore distance of 500 nm. Co2共CO兲8 was used as a metallo-organic precursor, which was dissolved in a solution of dichloromethane 共CH2Cl2兲 containing 2 – 5-wt % polystyrene 共PS兲共Alfa Aesar, M.W. 123,000兲 or poly-l-lactide 共PLLA, Birmingham Polymers, Birmingham, AL, USA兲. The weight ratio of the metallo-organic precursor to the polymer in the dichloromethane was varied between C = 0.4 and 10. In contrast to the synthesis of noble metal nanotubes,13 the preparation of the Co solutions and the infiltration process into the alumina templates must be carried out under an inert gas atmosphere 共N2 or Ar兲. By placing several droplets of the Co2共CO兲8 / polymer solution on the top surface of the alumina membrane using a microsyringe, the cylindrical pores became completely wetted by the polymer solution. During the evaporation of the dichloromethane, the formation of Co2共CO兲8 / polystyrene or Co2共CO兲8 / poly-l-lactide nanotubes takes place at the inner surfaces of the pores by a phase-separation process. The thickness of the polymer/ precursor mixture inside the pores is in the range of 20– 100 nm and decreases at higher precursor concentrations. A polymer layer also forms on top of the alumina membrane, but this was removed using a razor blade. Subsequently, the samples were transferred to a furnace and annealed under vacuum for 24– 72 h at 180 ° C.35–39 This leads to the decomposition of the metallo-organic precursor, so that metallic Co precipitates at the polymeralumina interface to form a Co tube inside each pore. The alumina surface was treated by an ion milling process in order to remove any remaining polymer and cobalt from the top alumina surface. The polymer was finally removed from the pores using a post-annealing step 共艌300 ° C兲. PLLA decomposed completely within 1 h, while PS required a longer heat treatment of 3 – 20 h. Alternatively, the polymer could be removed by soaking in dichloromethane at room temperature. The metallic tubes were released from the alumina matrix by etching in 10-wt % KOH solution. Alternatively, removal of the alumina could be carried out more slowly using dilute phosphoric acid. Electron micrographs were taken in a Zeiss Leo scanning electron microscope after coating the samples with a thin gold layer. For transmission electron microscopy 共Philips CM20T兲 the Co tubes were removed from the KOH solution, washed several times in distilled water and ethanol, and placed on copper grids with holey carbon films. Magnetic measurements were performed in a DMS vibrating sample magnetometer at room temperature. To measure the cobalt concentration, the samples were completely dissolved in 10-ml HCl 共10 wt % 兲 and the cobalt concentration in the HCl solution was determined by atomic absorption spectroscopy.

J. Appl. Phys. 98, 034318 共2005兲

FIG. 1. Cobalt tubes embedded in a self-ordered porous alumina template with an interpore distance of 500 nm and a pore diameter of 300 nm are shown in 共a兲–共c兲. In 共a兲 a top-view image of an array of cobalt/PLLA composite tubes is presented with an outer tube diameter of 300 nm. The precursor concentration of this sample was C = 1.5 and the thickness of the inner PLLA layer is about 70 nm. Thin-walled Co tubes were formed at the polymer/oxide interface during the 24-h vacuum annealing process. 共b兲 An alumina template is shown in cross-section view with two Co/PLLA composite tubes. In this case the precursor concentration was C = 2.5 and the PLLA layer had a thickness of 30– 40 nm. 共c兲 This sample was obtained after heat treatment of the sample shown in 共a兲. The inner PLLA layer was removed by annealing at 300 ° C for 4 h under vacuum conditions. The ultrathin cobalt tubes remain in the pores. 共d兲 A macaroni-like ensemble of cobalt tubes with an average tube diameter of 180 nm is shown. These Co tubes were synthesized from a PS/precursor mixture with a precursor concentration of C = 2.5 at 180 ° C for 24 h in a self-ordered alumina template. For the removal of the polymer layer the same heat treatment as described in 共c兲 was used, and the Co tubes were finally released from the oxide matrix by a KOH etch 共10 wt % 兲 for 30 min.

III. RESULTS AND DISCUSSION

The morphology of two arrays of composite PLLA/ cobalt tubes embedded in an alumina membrane after the annealing process is shown in Figs. 1共a兲 and 1共b兲. The Co tubes themselves are too thin to resolve, but the 40– 70-nm-thick polymer tube is clearly visible. The gap between the polymer and template seen in Fig. 1共a兲 is believed to be a result of the thermal-expansion mismatch between the polymer and the template upon cooling from the precursor decomposition temperature of 180 ° C. Figure 1共c兲 shows a similar sample after a second thermal treatment at 300 ° C, which decomposed the PLLA to leave the Co coating the interior of the pores. A disadvantage of thermal polymer removal is the possible presence of carbon on the interior cobalt tube walls. As an alternative approach the inner polymer


034318-3

Nielsch et al.

FIG. 2. For the synthesis of highly uniform cobalt tubes we use a perfectly ordered alumina pore structure with a thickness of about 1.5 ␮m on a silicon substrate patterned by imprint lithography. The interpore distance of this sample is 500 nm and the pore diameter is 160 nm. Polystyrene was used as a matrix and the precursor concentration was C = 6.66. The cobalt tubes were formed by annealing under vacuum conditions at 180 ° C for 72 h. 共a兲 This pore structure has a perfect hexagonal pore arrangement on a cm2 scale. The inset of 共a兲 shows also the Co/ PS composite tubes inside the pores. Due to the high precursor concentration C the thickness of the tube wall is small 共20– 25 nm兲. 共b兲 and 共c兲 The inner polymer layer has been removed by soaking in dichloromethane. The Co tubes have been partially released from the template by a KOH etch 共10 wt % , 3 min兲. A single Co tube with a length of nearly 1 ␮m is presented in 共b兲. In 共c兲 numerous Co tube segments are lying on top of the alumina pore structure. The pores have been widened from 160 nm up to 300 nm by the KOH etch.

tubes can be dissolved in dichloromethane. In Fig. 1共d兲 the alumina has also been removed to reveal free-standing thinwalled Co tubes. From the dimensions of the nanotubes, it appears that the nucleation of metallic Co occurs at the polymer/oxide interface, which may be a result of catalysis of the Co nucleation by the alumina surface. We have also fabricated perfectly ordered arrays of Co tubes by using an alumina membrane made by anodizing an indented aluminum film.34,40 The alumina membrane shows a highly regular pore arrangement over an area of 1 cm2 and an interpore distance of 500 nm 关Fig. 2共a兲兴. After decomposition of the carbonyl, the polymer was removed by dichloromethane, and the Co tubes were released by a partial KOH etch. As seen in Figs. 2共b兲 and 2共c兲, the etch widened the pores isotropically to about 300-nm diameter and also began to etch secondary pores in intermediate positions between the original pores. Gas evolution during the KOH etch pushed numerous tubes out of the pores, and they can be seen lying on the surface, often broken. Some segments have closed ends, originating from the bases of the pores. Due to their thin walls, the Co tubes are semitransparent in the scanning electron microscope and faint shadows from the pores can be seen through the tube walls.

J. Appl. Phys. 98, 034318 共2005兲

FIG. 3. Transmission electron micrographs of cobalt tubes with a diameter of 180 共a兲 and 350 nm 共b兲 The Co2共CO兲8 concentration in the polystyrene matrix was C = 2.48 and the synthesis was performed at 180 ° C for 72 h. The inset shows the electron-diffraction pattern of a single hexagonal hcp-Co crystallite 共b兲, which is incorporated in one of the tube walls.

For the transmission electron microscopy, a droplet of ethanol solution containing cobalt nanotubes after dissolving the template was placed onto a copper grid with holey carbon film, and the solvent was evaporated prior to microscopy. Figure 3 shows electron micrographs of cobalt/polymer composite tubes with a diameter of around 180 共a兲 and 350 nm 共b兲. The inner polymer tube is nearly transparent to the electron beam. The outer cobalt thin film in Fig. 3共a兲 consists of metallic crystallites with a size of a few nanometers. The electron-diffraction pattern of the tubes is very broad and only a Co-hcp structure could be detected, with no preferential crystallographic orientation. The formation of an oxide film on the surface when the tubes were released from the alumina matrix may also contribute to the broadening of the diffraction pattern. On certain tube areas, particularly in the larger diameter tubes, cobalt crystals with a diameter of up to 100 nm and a hexagonal shape can be found. The hcp electron-diffraction pattern of one large disklike crystallite in Fig. 3共b兲 is shown in the inset. Due to the diffraction pattern and the orientation of the crystals in the image, it appears that these large crystals are oriented with the c axis perpendicular to the tube axis. The major parameters controlling the thickness of the Co tubes are the concentration of precursor in the polymer, C, and the geometry of the template. The expected wall thick-


034318-4

J. Appl. Phys. 98, 034318 共2005兲

Nielsch et al.

ness of the tubes was calculated from the template geometry and the densities of precursor and polymer. For a template with interpore distance Dint, pore diameter D P, and pore length L, if we assume complete filling of the pores by the polymer/precursor mixture, the volume of the mixture in each pore is ␲LD2P / 4. The density of polystyrene and cobalt carbonyl Co2共CO兲8 were taken as ␳PS = 1.2 g / cm3 and ␳Co2共CO兲8 = 1.7 g / cm3, respectively, and the molecular weight of the cobalt carbonyl is M Co2共CO兲8 = 341.9 g / mol. After the carbonyl decomposition, metallic cobalt is formed with density ␳Co = 8.9 g / cm3 and atomic weight M Co = 58.9 g / mol. The final weight of cobalt in each pore was calculated based on the ratio C of the weight of cobalt carbonyl to the weight of polymer: mCo = ␲L关D P⌬ − ⌬2兴

C␳PS␳Co2共CO兲8

2M Co , C␳PS + ␳Co2共CO兲8 M Co2共CO兲8

共1兲

in the case of incomplete pore filling, where ⌬ is the polymer layer thickness. The expected wall thickness of the cobalt nanotube in the pore, DW, is then given by DW =

D P⌬ − ⌬2 C␳PS␳Co2共CO兲8 2M Co . D P␳Co C␳PS + ␳Co2共CO兲8 M Co2共CO兲8

共2兲

The maximum possible wall thickness for the cobalt nanotubes, DW共max兲, is obtained in the limit of C → ⬁, i.e., when the pores are filled only with cobalt carbonyl, and when ⌬ = D P / 2: DW共max兲 =

D PM Co␳Co2共CO兲8 2␳CoM Co2共CO兲8

.

FIG. 4. In-plane 共perpendicular to tube axis兲 and out-of-plane 共parallel to axis兲 hysteresis loops for cobalt tube arrays with an interpore distance of 500 nm and a diameter of 180 nm. The Co2共CO兲8 concentration in the polystyrene matrix was C = 1.67, 2.48, and 3.33. The synthesis was performed at 180 ° C for 72 h. The measured magnetization is normalized to the sample volume in cm3.

共3兲

If a polymer with a higher density is used, such as PLLA 共␳PLLA = 1.27 g / cm3兲, DW共max兲 will increase slightly. For the case that the pores are filled only with cobalt carbonyl before the tube synthesis, the maximum possible wall thickness of the tubes is given by DW共max兲 = 3.3 and 6.6 nm for cobalt tubes with pore diameters of 180 and 400 nm, respectively. For C = 6.6 the corresponding wall thicknesses are 2.7 and 5.3 nm. The wall thickness is small as a result of the large volume reduction upon decomposition of the precursor. In our experiments, for the highest concentrations of the metal precursor 共C = 3.33– 6.66兲, the electron micrographs suggest wall thicknesses of a few nanometers, in agreement with the calculated values of DW共max兲. Magnetic measurements show a systematic change in both saturation magnetization and anisotropy as a function of precursor concentration C. Figure 4 shows that for 180-nm diameter tubes, at low values of C the in-plane and out-ofplane hysteresis loops have similar shapes, but as C increases the samples develop a significant out-of-plane anisotropy. Similar trends were obtained for a pore diameter of 450 nm. Additionally, the saturation magnetization M 共per unit volume of the array of tubes兲 increases with C until it reaches a plateau, which increases with pore diameter as shown in Fig. 5. The increase in magnetization and the development of preferential out-of-plane anisotropy can be understood in terms of the increasing amount of Co produced at higher values of C. Low precursor concentrations lead to the forma-

tion of discontinuous Co nanoparticles with isotropic hysteresis loops, but as C increases the Co precipitates form continuous tubes with a shape-induced easy axis parallel to their length. The saturation magnetization for 180-nm diameter tubes from Fig. 5 is M = 2.2 emu/ cm3, which corresponds to an effective wall thickness of pure Co of 0.3 nm, taking the saturation magnetization of pure cobalt as M s = 1420 emu/ cm3. This is about 10% of the calculated maximum wall thickness DW共max兲 and is thinner than indicated by microscopy 关Fig. 1共d兲兴. To clarify this observation, we have measured the amount of cobalt in each sample by atomic absorption spectroscopy 共AAS兲. This enables the average saturation magnetization per unit volume of cobalt, M eff, to be determined for each sample. At high precursor concentrations, C = 3.3– 6.6, we find M eff = 620– 640 emu/ cm3. In the intermediate range where C = 0.8– 1.6, M eff is ⬃400 emu/ cm3 and for low C 共⬍0.5兲, M eff is less than 200 emu/ cm3.

FIG. 5. The saturation magnetization per unit volume of the tube arrays within the alumina template is plotted as function of the Co2共CO兲8 concentration C in the polystyrene matrix for tube diameters of D P = 180 nm 共from Fig. 3兲 and 450 nm fabricated under identical conditions 共annealed at 180 ° C, 72 h to form metallic Co from the precursor兲.


034318-5

Nielsch et al.

The low effective saturation magnetization of the Co within the samples is attributed to two factors. First, the incomplete filling of the pores by the polymer/precursor mixture limits the amount of precursor available to form each tube. Prior to the decomposition reaction, the polymer/ precursor mixture forms a ⌬ = 20– 100-nm-thick layer on the pore walls, so the maximum amount of Co deposited is proportionately smaller than that predicted by Eq. 共3兲, which assumes complete filling of the pores. As a result, when the polymer thickness is smaller than the pore radius, the Co tube wall thickness should be independent of pore diameter, and the saturation magnetization M per unit volume of the sample would be expected to scale with pore diameter. This is indeed observed for the two pore diameters shown in Fig. 5, where the 450-nm-diameter tubes have a similar wall thickness to that of the 180-nm-diameter tubes. The tendency of the polymer/precursor wetting layer thickness to decrease with increasing C explains why the saturation magnetization and the wall thickness saturate at high C. Second, oxidation of the tube walls occurs during atmospheric exposure after the processing is complete, to produce a thin antiferromagnetic native oxide layer. Such oxidation is commonly observed on the surface of magnetic nanostructures,41 and reduces the effective saturation moment M eff of the Co. For samples with a low C and a very thin tube wall the effects of oxidation are proportionally greater, and these tubes therefore show a very low M eff per unit volume of cobalt. For the samples with highest C, the ratio M eff / M s is up to 0.45, which indicates that just over half the Co is magnetically inert, presumably because it is incorporated into an oxide layer. For example, for a wall thickness of 5.7 nm, the wall could consist of a layer of 2.6 nm of pure Co surrounded on either side by ⬃2 nm of CoO. Most template synthesis methods produce wires, and the magnetic properties of solid nanowires have been extensively studied. The reversal of infinite nanowires is usually described in terms of an idealized curling mechanism,42 in which the magnetization rotates while remaining parallel to the surface of the wire, but experimental measurements and micromagnetic modeling indicate that reversal typically occurs instead via a domain nucleation and propagation mechanism.43 For solid Co nanowires with diameters of the order of 180 nm, switching fields of the order of 200 Oe 共Refs. 44 and 45兲 have been measured. However, the reversal mechanism will be quite different for a nanotube as compared to the reversal of a nanowire. In a thin-walled structure the magnetization remains in the plane of the Co layer during reversal 共i.e., it is circumferential兲 and a curling mechanism is therefore favored, even at large diameters. The switching field for a hollow tube reversing by curling has been modeled,46,47 and the measured coercivities of our samples, 400– 500 Oe for tube diameters of 180– 450 nm, compare well to the predicted nucleation field for curling.46 In the case of our polycrystalline Co tubes, the net magnetocrystalline anisotropy is small and the behavior is primarily a result of the shape. Recent work on 150– 220-nm-diameter Fe3O4 nanotubes31 showed an easy axis along the tube axis and coercive fields of up to 600 Oe, which is also consistent with reversal by curling. However, FePt nanotubes with similar

J. Appl. Phys. 98, 034318 共2005兲

dimensions showed isotropic magnetic behavior, attributed to the high magnetocrystalline anisotropy of the FePt which dominates the shape effects. Unlike arrays of solid nanowires, the composite metal/ polymer tubes have a low magnetic moment, and magnetostatic interaction between the tubes in the array are negligible. The hysteresis loops are therefore characteristic of an ensemble of isolated tubes. The remanence in the out-ofplane direction is approximately half of the saturation magnetization, as a result of the magnetization tilting away from the cylinder axis at remanence. In contrast, an array of closely-spaced tubes showed a preferential magnetization direction perpendicular to the nanotube axis,30 as a result of strong interactions between the tubes. The differences in remanent state between nanotubes and nanowires leads to different magnetic or magnetoresistive behavior, and may be useful in, for example, a tube-shaped high-density recording head.48 The methods that we have demonstrated allow the fabrication of high-quality ultrathin-walled metal tubes and composite polymer/metal tubes. The wall thickness can be accurately controlled by the precursor concentration and the geometry of the template. The cross-sectional shape of the tubes can also be varied by changing the template. The Co tubes are ferromagnetic and their magnetic properties can be understood in terms of their shape and wall thickness. It is noteworthy that in comparison with solid wires, these tubes have very low mass and are therefore less prone to sedimentation in solution. Composite tubes, where the density can be adjusted by varying the internal polymer thickness, may therefore be more suitable than nanowires for applications in ferrofluids or drug delivery. We have prepared ferrofluidic solutions of Co tubes dissolved in water or ethanol which retain their magnetic properties for periods of several weeks or months. Significantly, this synthesis method is not limited to Co. We have also precipitated metallic Fe and Ni from iron pentacarbonyl 关Fe共CO兲5兴 and bis共1,5cyclooctadiene兲nickel 关Ni共COD兲2兴, respectively, and other metal precursors may also be used. Preliminary results for nickel show a nano-particle-like magnetic behavior, suggesting that the tubes consist of agglomerated fine particles, whereas the iron tubes oxidize after preparation. This work suggests the possibility of the formation of complex multilayer nanotubes by sequential precursor decomposition to create layers of different materials, or by combining precursor decomposition with other processes. For example, layers of other materials could also be deposited uniformly inside a thin-walled metallic nanotube by electrodeposition. Although there are many examples of ferromagnetic multilayer nanowires in the literature, the layers are oriented perpendicular to the wire axis, whereas the methods proposed here could generate hollow or filled tubes with layers parallel to the axis. Such materials have not been explored, and offer the possibility of unique physical and chemical properties. For example, bilayer or multilayer nanotubes could be functionalized by adsorption of different kinds of molecules on the interior and exterior surfaces, which could make them ideal for applications such as cataly-


034318-6

sis, drug delivery, or gene therapy,1,2 while concentric multilayer magnetic tubes may be used in magnetoelectronic devices. IV. CONCLUSIONS

Composite Co/polymer nanotubes have been produced by a template-mediated process within the pores of anodic alumina. The interior polymer layer and the template can be removed to leave freestanding polycrystalline Co tubes with wall thicknesses of a few nanometers. The wall thickness can be controlled via the Co precursor concentration. Magnetic measurements indicate an increase in saturation magnetization and anisotropy with increasing wall thickness, and suggest that the outer ⬃2 nm of the Co tube walls has oxidized. The anisotropy of the tubes is governed by shape anisotropy, and the switching field of the tubes is consistent with that expected from curling mechanism in which the magnetization rotates within the plane of the thin Co tube wall. Other materials such as Fe and Ni can be grown using appropriate precursors, suggesting that complex multilayered structures can be synthesized by this method. ACKNOWLEDGMENTS

This work was supported by the National Science Foundation. We thank J. Robinson of MIT for assistance with sample measurement and annealing, H. I. Smith and E. L. Thomas for the use of laboratory facilities, P. Göring from the Max Planck Institute for Microstructure Physics in Halle 共Germany兲 for transmission electron microscopy, and J. Acker from the Leibniz Institute for Solid State and Materials Research in Dresden 共Germany兲 for atomic absorption spectroscopy. A. K. Salem, P. C. Searson, and K. W. Leong, Nat. Mater. 2, 668 共2003兲. C. C. Berry and A. S. G. Curtis, J. Phys. D 36, R198 共2003兲. 3 R. K. Soong, G. D. Bachand, H. P. Neves, A. G. Olkhovets, H. G. Craighead, and C. D. Montemagno, Science 290, 1555 共2000兲. 4 M. Lewin, N. Carlesso, C. H. Tung, X. W. Tang, D. Cory, D. T. Scadden, and R. Weissleder, Nat. Biotechnol. 18, 410 共2000兲. 5 See, for example, articles in Carbon Nanotubes: Synthesis, Structure Properties and Applications, edited by M. Dresselhaus, G. Dresselhaus, and Ph. Avouris 共Springer, Berlin, 2001兲. 6 Y. Xia et al., Adv. Mater. 共Weinheim, Ger.兲 15, 353 共2003兲. 7 B. C. Satishkumar, A. Govindaraj, E. M. Volil, L. Basumallick, and C. N. R. Rao, J. Mater. Res. 12, 604 共1997兲. 8 M. E. Spahr, P. Bitterli, R. Nesper, M. Müller, F. Krumeich, and H. U. Nissen, Angew. Chem. 37, 1263 共1998兲. 9 B. B. Lakshmi, C. J. Patrissi, and C. R. Martin, Chem. Mater. 9, 2544 共1997兲. 10 T. Kasuga, M. Hiramatsu, A. Hoson, T. Sekino, and K. Niihara, Langmuir 14, 3160 共1998兲. 11 K. Subramanya Mayya, D. I. Gittins, A. M. Dibaj, and F. Caruso, Nano Lett. 1, 727 共2001兲. 1 2

J. Appl. Phys. 98, 034318 共2005兲

Nielsch et al.

H. Nakamura and Y. Matsui, J. Am. Chem. Soc. 117, 2651 共1995兲. M. Steinhart, Z. H. Jia, A. K. Schaper, R. B. Wehrspohn, U. Gösele, and J. H. Wendorff, Adv. Mater. 共Weinheim, Ger.兲 15, 706 共2003兲. 14 M. Nishizawa, V. P. Menon, and C. R. Martin, Science 268, 700 共1995兲. 15 Y. Sun, B. Mayers, and Y. Xia, Adv. Mater. 共Weinheim, Ger.兲 15, 641 共2003兲. 16 B. Mayers and Y. Xia, Adv. Mater. 共Weinheim, Ger.兲 14, 279 共2002兲. 17 V. Ya. Prinz, V. A. Seleznev, A. K. Gutakovsky, A. V. Chehovskiy, V. V. Preobrazhenskii, M. A. Putyato, and T. A. Gavrilova, Physica E 共Amsterdam兲 6, 828 共2000兲. 18 O. G. Schmidt and K. Eberl, Nature 共London兲 410, 168 共2001兲. 19 J. Goldberger, R. He, Y. Zhang, S. Lee, H. Yan, H. J. Choi, and P. Yang, Nature 共London兲 422, 599 共2003兲. 20 Y. Luo et al., Appl. Phys. Lett. 83, 440 共2003兲. 21 C. R. Martin, Science 266, 1961 共1994兲. 22 M. Steinhart, J. H. Wendorff, A. Greiner, R. B. Wehrspohn, K. Nielsch, J. Schilling, J. Choi, and U. Göselle, Science 296, 1997 共2002兲. 23 S. I. Moon and T. J. McCarthy, Macromolecules 36, 4253 共2003兲. 24 M. Steinhart, S. Senz, R. B. Wehrspohn, U. Gösele, and J. H. Wendorff, Macromolecules 36, 3646 共2003兲. 25 Y. Long, Z. Chen, N. Wang, Y. Ma, Z. Zhang, L. Zhang, and M. Wan, Appl. Phys. Lett. 83, 1863 共2003兲. 26 W. Shenton, T. Douglas, M. Young, G. Stubbs, and S. Mann, Adv. Mater. 共Weinheim, Ger.兲 11, 253 共1999兲. 27 M. Mertig, R. Kirsch, and W. Pompe, Appl. Phys. A: Mater. Sci. Process. 66, S723 共1998兲. 28 G. Tourillon, L. Pontonnier, J. P. Levy, and V. Langlais, Electrochem. Solid-State Lett. 3, 20 共2000兲. 29 J. Bao, C. Tie, Z. Xu, Q. Zhou, D. Shen, and Q. Ma, Adv. Mater. 共Weinheim, Ger.兲 13, 1631 共2001兲. 30 J. Bao, Z. Xu, J. Hong, X. Ma, and Z. Lu, Scr. Mater. 50, 19 共2004兲. 31 Y. C. Sui, R. Skomski, K. D. Sorge, and D. J. Sellmyer, Appl. Phys. Lett. 84, 1525 共2004兲. 32 H. Masuda, K. Yada, and A. Osaka, Jpn. J. Appl. Phys., Part 2 37, L1340 共1998兲. 33 K. Nielsch, J. Choi, K. Schwirn, R. B. Wehrspohn, and U. Gösele, Nano Lett. 2, 677 共2002兲. 34 K. Nielsch, F. J. Castano, S. Matthias, W. Lee, and C. A. Ross, Adv. Eng. Mater. 7, 217 共2005兲. 35 J. R. Thomas, J. Appl. Phys. 37, 2914 共1966兲. 36 D. P. Dinega and M. G. Bawendi, Angew. Chem., Int. Ed. 38, 1788 共1999兲. 37 V. F. Puntes, K. M. Krishnan, and A. P. Alivisatos, Science 291, 2115 共2001兲. 38 C. B. Murray, S. H. Sun, H. Doyle, and T. Betley, MRS Bull. 26, 985 共2001兲. 39 N. Cordente, M. Respaud, F. Senocq, M. J. Casanove, C. Amiens, and B. Chaudret, Nano Lett. 1, 565 共2001兲. 40 J. S. Choi, G. Sauer, P. Göring, K. Nielsch, R. B. Wehrspohn, and U. Gösele, J. Mater. Chem. 13, 1100 共2003兲. 41 C. A. Ross et al., Phys. Rev. B 62, 14252 共2000兲. 42 E. H. Frei, S. Shtrikman, and D. Treves, Phys. Rev. 106, 446 共1957兲. 43 R. Ferre, K. Ounadjela, J. M. George, L. Pirauz, and S. Dubois, Phys. Rev. B 56, 14066 共1997兲. 44 T. M. Whitney, J. S. Jiang, P. C. Searson, and C. L. Chien, Science 261, 1316 共1993兲. 45 J. M. Garcia, A. Asenjo, J. Velazquez, D. Garcia, M. Vazquez, P. Aranda, and E. Ruiz-Hitzky, J. Appl. Phys. 85, 5480 共1999兲. 46 C. R. Chang, C. M. Lee, and J. S. Yang, Phys. Rev. B 50, 6461 共1994兲. 47 C. M. Lee and C. R. Chang, Mater. Chem. Phys. 43, 183 共1996兲. 48 S. Khizroev, M. H. Kryder, D. Litvinov, and D. A. Thompson, Appl. Phys. Lett. 81, 2256 共2003兲. 12 13
