Sun Co Nano Particles Jap 1999

JOURNAL OF APPLIED PHYSICS

VOLUME 85, NUMBER 8

15 APRIL 1999

Synthesis of monodisperse cobalt nanocrystals and their assembly into magnetic superlattices „invited… Shouheng Suna) and C. B. Murray IBM T. J. Watson Research Center, Yorktown Heights, New York 10598

High temperature, solution phase reduction of cobalt chloride in the presence of stabilizing agents was employed to produce magnetic colloids ~ferrofluids! of cobalt nanocrystals. We systematically synthesized and isolated nearly monodisperse nanocrystal samples ranging in size from 2 to 11 nm while maintaining better than a 7% std. dev. in diameter. As synthesized cobalt particles are each a single crystal with a complex cubic structure related to the beta phase of elemental manganese ~e-Co!. Annealing the nanocrystals at 300 °C converts them quantitatively to the more common hexagonal-close-packed crystal form. Deposition of these uniform cobalt particles on solid substrates by evaporation of the carrier solvent results in the spontaneous assembly of two-dimensional and three-dimensional magnetic superlattices ~colloidal crystals!. A combination of x-ray powder diffraction, transmission electron microscopy, and superconducting quantum interference device magnetometry were used to characterize both the dispersed nanocrystals and the assembled superlattices. © 1999 American Institute of Physics. @S0021-8979~99!50908-0#

properties depend strongly on NC ~grain! size. At the smallest sizes the NCs become superparamagnetic. The orientation of magnetic polarization in each NC begins fluctuating randomly with ambient thermal energy and the NCs can no longer be used in magnetic storage media. Development of detailed understanding of the properties of magnetic NCs is essential to the development of future magnetic recording technology. It is expected that the organized magnetic media might contribute to an effort to reach magnetic recording densities between 100 Gb/in.2 and 1 Tb/in.2 Numerous physical and chemical methods have been employed to produce metal NCs, including sputtering, metal evaporation, grinding, electrodeposition, solution phase metal salt reduction, and neutral organometallic precursor decomposition. Significant progress has been made in preparing nearly monodisperse noble metal NCs ~Au, Ag, and Pd, Pt!.24–29 Strong magnetic interactions in Ni, Fe, and Co particle systems make it difficult to form stable colloids. Uncontrolled agglomeration of the magnetic particles often makes it impossible to employ separation procedures which could narrow the distribution, and prevents the ready formation of the smooth, thin films required in magnetic recording applications. The magnetic properties of the NCs depend strongly not only on particle size but on the precise crystal structure and the presence of faults. The air sensitivity of the magnetic transition metal NCs has complicated their handling and has prompted use of the metal oxide nanoparticles in many applications despite the weaker magnetic properties.30,31 In this article, we describe the synthesis of monodisperse cobalt NCs with precise control over NC size ~2–11 nm! and size distribution ~s,7% std. dev.!. We further demonstrate the preparation of ordered thin films ~superlattices, colloidal crystals! of these monodisperse cobalt NCs. Each cobalt NC is sheathed in a robust organic ligand shell to limit oxidation and prevent irreversible aggregation of the particles. Only two stable forms @hexagonal-close-packed ~hcp! and face-

I. INTRODUCTION

Nanometer size crystals, nanocrystals ~NCs!, display many properties which are both quantitatively and qualitatively different from their respective bulk materials and from the discrete atomic or molecular species from which they are derived. Novel NC properties arise from the large fraction of atoms which reside on the surface of the particles and from the finite number of atoms in each crystalline core. It was in magnetic systems that the first finite size effects were recognized1 with more subtle effects noted later in nonmagnetic metals.2 The increased surface area of NCs has long been exploited to optimize the chemical reactivity of catalysts.3,4 Recently the study of finite size effects of metal NCs has intensified with the promise of uncovering the evolution of material properties with particle size and harnessing novel properties in new materials and devices.5–12 The potential to stabilize unusual phases of materials in NC form has also generated great interest.13 Semiconductor NCs showing size-tunable optical properties have been integrated into exploratory optical and electronic devices.14–16 These early nanoelectronic devices are often a major departure from present technology and represent important efforts to identify potential alternatives to conventional silicon device fabrication, which is expected to approach fundamental limits with the next 10 years. Magnetic storage technology is advancing even more rapidly toward its scaling limits. Thin granular films of ferromagnetic NCs formed by sputter deposition are already the basis of conventional rigid magnetic storage media ~hard drives!. Progress in magnetic recording density are due in part to the development of media with finer and finer grain magnetic films.17–23 The study of nanoscale magnetic domains are of both fundamental and pressing technical interest as the grain size of advanced recording media is rapidly shrinking to dimensions where magnetic a!

Electronic mail: [email protected]

0021-8979/99/85(8)/4325/6/$15.00

4325

© 1999 American Institute of Physics

Downloaded 12 Feb 2004 to 129.107.14.178. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

4326

J. Appl. Phys., Vol. 85, No. 8, 15 April 1999

centered-cubic ~fcc!# of bulk elemental cobalt are known. In practice, many bulk cobalt samples display mixed structures in which low energy stacking faults introduce a combination of hcp and fcc character. Solution phase chemical synthesis is generally not thermodynamically controlled and thus can allow preparation of crystal phases which are metastable. Here, we describe the reduction of cobalt chloride with lithium triethyl borohydride ~superhydride! to produce cobalt particles with an unusual structure distinct from the fcc and hcp forms, having the cubic symmetry of the b phase of manganese. The similarity of this novel form of cobalt to the b phase of Mn was recently recognized by Dinega and Bawendi tentatively designated as the epsilon phase of cobalt e-Co.32 It is a soft magnetic material which displays a magnetocrystalline anisotropy that is lower than uniaxial hcp type. However, the e-Co NCs can be converted quantitatively to the common hcp phase by annealing at 300 °C. The fact that e-Co NCs are soft magnetic materials ~low coercivity! provides several practical benefits. The reduced magnetic dipole interaction between particles facilitates stabilizing the particles during the synthesis, assists size selective processing and the formation of ordered films. II. EXPERIMENT

The synthetic experiments were carried out using standard airless procedures and commercially available reagents. Dioctylether solution of superhydride (LiBEt3H) is prepared by mixing the tetrahydrofuren ~THF! superhydride solution ~Aldrich! with dioctylether and evaporating THF under vacuum. The particle synthesis began with the injection of 2 M dioctylether superhydride solution into a hot ~200 °C! CoCl2 dioctylether solution in the presence of oleic acid ~octadec-9-ene-1-carboxylic acid, CH3~CH2!7CH vCH~CH2!7COOH) and trialkylphosphine (PR3 , R 5n-C4H9 , or n-C8H17!. Reduction occurred instantly upon injection, leading to the simultaneous formation of many small metal clusters ~nuclei!. Continued heating at 200 °C allowed steady growth of these clusters into nanometer sized, single crystals of cobalt. The steric bulk of the alkylphosphine controlled the rate of particle growth. Short chain alkylphosphines ~e.g., tributylphosphine! allowed faster growth, favoring production of large particles while bulkier species ~e.g., trioctylphosphine! reduced particle growth and favored production of smaller NCs. Growth of the particles was quenched by cooling the dispersion. Particle size could not be readily increased by extended heating as side reactions, potentially catalyzed by the cobalt particles, led to the progressive decomposition of organic components in the dispersion. These organically stabilized cobalt NCs were readily dispersed in aliphatic, aromatic, and chlorinated solvents and precipitated by short chain alcohols, facilitating size selective precipitation which isolated nearly monodisperse samples.33 In a typical experiment, cobalt chloride ~anhydrous! ~1 mmol!, oleic acid ~1 mmol!, and dioctylether ~20 ml! were mixed under nitrogen and heated to 100 °C. Tributylphosphine ~3 mmol! was added and the mixture was then heated to 200 °C. Dioctylether superhydride solution ~1 mL, 2 mmol

S. Sun and C. B. Murray

superhydride!! was injected during vigorously stirring. A color change from dark blue to black was observed, indicating reduction of the blue Co21 complex to form cobalt particles. The black solution was stirred at 200 °C for 20 min and then cooled to room temperature. Particles were precipitated by adding ethanol to the dispersion. The supernatant was discarded either by decanting or by centrifugation. The waxy magnetic precipitate was redispersed in 10 mL hexane with 100–500 ml of oleic acid to ensure stability. Size selective precipitation and centrifugation were employed to fractionate the colloids. Average particle size was coarsely controlled by selecting the type of alkylphosphine used in combination with oleic acid during the growth. Bulky P~C8H17) 3 limited growth to produce particles ~2–6 nm! and less bulky P~C4H9 ) 3 led to larger ~7–11 nm! particles. Fine tuning of the particle size was done by fractionating the coarsely adjusted samples and selecting the desired size fraction. In this way a series of nearly monodisperse NC samples were produced. NC superlattice ~colloidal crystals! was formed by spreading a thin layer of cobalt dispersion on a solid substrate and allowing it to dry slowly. Evaporation of a dilute dispersion of 9 nm cobalt particles ~;2% particles! over a period of 30 min at temperatures between 20 and 80 °C produced thin films containing close-packed monolayer and multilayer regions. The e-Co NCs were annealed under a mixture of argon ~95%! and hydrogen ~5%! or under vacuum ~0.2 mm Hg! at 300 °C to convert the particles into hcp NCs. During this process the protective organic shell was removed and subsequent exposure to air resulted in the rapid formation of cobalt oxide nanoparticles. Oxidation of the annealed samples was prevented by treating the annealed samples with acetone before exposing them to air. Cobalt and phosphorus elemental analyses of cobalt particles was performed on an inductively coupled plasma– atomic emission spectrometer at Galbreith Lab, Tennessee. Transmission electron microscopy ~TEM! was used to determine particle size and size distribution. Commercial 30 nm Au particles ~BB International, United Kingdom! were used as an internal standard for low resolution TEM measurements on a Philips EM 420 ~120 kV!. High resolution TEM images showing the lattice structure of the individual particles were obtained on JEOL 4000 FX ~400 kV!. X-ray powder diffraction was performed on an Inel CPS 590 diffractometer under Co K a radiation ~l51.788 965 Å!. The samples for TEM analysis were prepared by drying alkane dispersion of cobalt particles on amorphous carbon coated copper grid, silicon nitride membranes, and silicon ~100! wafers while x-ray samples employed silicon ~100! wafers as substrates. Magnetic studies were carried out using MPMS2 Quantum Design superconducting quantum interference device magnetometer. Measurements of isolated particle properties were carried out on as synthesized particles or by diluting the cobalt NCs dispersions with octadecylamine to 20% cobalt. The ordered thin films ~superlattices of NCs! were prepared by evaporation of octane solution of NCs on silicon nitride

Downloaded 12 Feb 2004 to 129.107.14.178. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

J. Appl. Phys., Vol. 85, No. 8, 15 April 1999

S. Sun and C. B. Murray

4327

substrates. The films were then loaded with the substrate parallel to the magnetic field used in the measurement. III. RESULTS AND DISCUSSION

Inert atmosphere solution phase metal salt reduction is a well known technique for the preparation of metal particles. A variety of ligands, including polymers and surfactants, have been used to control particle growth, stabilize metal dispersions, and limit oxidation of the particles. The colloids are stable when a repulsive force of sufficient strength and range exists to counteract the combined attractive forces. Metal particles experience strong Van der Waals attractions which combined with the magnetic dipole interactions make stabilizing these systems very challenging. We have employed a combination of surfactants—trialkylphosphine and oleic acid—to control particle growth, stabilize the particles, and prevent oxidation. Trialkylphosphine reversibly coordinates neutral metal surface sites, slowing but not stopping particle’s growth, i.e., it cannot prevent the particles from eventually growing to undispersable aggregates at high temperature when used alone. Oleic acid, when employed alone, is an excellent stabilizing agent, but it binds so tightly to the particle surface during synthesis that greatly impedes the particle to growth. The combination of trialkylphosphine/oleic acid produced a tight ligand shell which allowed the particles to grow steadily while protecting them from aggregation and oxidation. It is broadly accepted that a temporal separation of the nucleation and growth stages is desirable for the production of a monodisperse colloid. Ideally, a large number of critical nuclei should be formed in a short interval of time followed by the simultaneous and steady growth of those nuclei. In our synthesis, we injected a reducing agent ~superhydride! to induce rapid reduction of cobalt ~II! chloride in hot ~200 °C! media. This procedure provided the temporally discrete homogeneous nucleation desired. Growth of the nuclei was continued by addition of cobalt containing species to the surface of the particles and was halted by cooling to below 100 °C. Cobalt NC samples of extremely narrow size distributions ~s,7%! in diameter were isolated after the synthesis by size selective precipitation. The particles dispersed readily in alkane solvent and were precipitated by alcohols. Since the attractive forces between particles are strongly dependent on the particle size, the gradual addition of a flocculant ~e.g., ethanol! induces aggregation of the largest particles in the dispersion before the smallest became unstable. Sequential precipitation of fractions of the dispersion and collection by centrifugation were exploited to narrow the size distribution. The results of these synthesis and separation procedures are displayed in Fig. 1. The two sets of cobalt particles shown with mean diameter of 6 and 9 nm are nearly monodisperse. Each particle is separated from its neighbor by the organic ligand shell. There are only two stable crystal phases known for elemental cobalt at ambient pressures. The bulk hcp form is stable at temperatures below 425 °C, while the bulk fcc form is the stable structure at higher temperature. The cobalt NCs

FIG. 1. TEM micrographs of cobalt nanocrystals with sizes of ~a! 6 and ~b! 9 nm from superhydride reduction. The particles are deposited on amorphous carbon from hexane dispersion.

produced here are neither fcc nor hcp. Figure 2 shows the diffraction pattern of 11 nm of NCs which has been indexed to a crystal structure with the symmetry found in the b phase of elemental Mn ~e-Co phase!.32 In bulk form the b phase of elemental manganese was obtained by quenching manganese from between ; 800 and 1100 °C.34 This type of manganese has a complicated primitive cubic cell with 20 atoms present in a cube 6.30 Å on a side.34 Table I shows a comparison of the d spacings and relative intensities of reflections observed in our sample ~e-Co! and in a structure with b-Mn type symmetry. The assignment of even the smallest features in the cobalt diffraction pattern in Fig. 2 is remarkable. No distinct

FIG. 2. XRD pattern of 11 nm e-Co nanocrystals. Inset: Size-dependent XRD. The patterns are collected from particle thin films on Si~100! substrate.

Downloaded 12 Feb 2004 to 129.107.14.178. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

4328

J. Appl. Phys., Vol. 85, No. 8, 15 April 1999

S. Sun and C. B. Murray

TABLE I. Diffraction comparison of 11 nm cobalt NCs and b phase of manganese. d Co~Å!

Relative int.

d Mn~Å!

Relative int.

hkl

2.73 2.49 2.16 2.04 1.94 1.85 1.63 1.44 1.37 1.20 1.13

5 4 8 100 56 31 9 7 7 23 18

2.823 2.577 2.231 2.104 1.997 1.904 1.6872 1.4874 1.4115 1.2377 1.1721

5 5 7 100 60 25 9 6 5 25 20

210 211 220 221 310 311 321 330 420 510 520

peaks corresponding to CoO and CoP phase were detected. Inset in Fig. 2 shows a series of x-ray patterns for different nanocrystal sizes. The x-ray reflections broaden as the crystallite size decreases. The measurement of x-ray line widths yields a crystal coherence length which matches TEM particle size measurements, indicating that each particle is a single crystal. The e-Co particles were synthesized at 200 °C. This structure was not stable at temperatures above 300 °C. Heating the particles at 300 °C under Ar1H2 ~5%! or under vacuum converted the material to the hcp phase. Figure 3~a! shows the x-ray pattern of a film of 9 nm e-Co NCs prior to annealing. Figure 3~b! shows the pattern of the same film after annealing at 300 °C under 0.2 mm Hg vacuum for 3 h. The particles have converted to the hcp phase without a change in crystal size ~x-ray and TEM measurements!. Heating a similar sample at 500 °C produced the expected transition of the hcp phase to a predominantly fcc structure which retained a significant number of hcp like stacking faults. At these higher temperatures both growth and some sintering of neighboring particles increased the NC size and degraded the size distribution. Thus the e-Co NCs served as an intermedi-

FIG. 3. XRD patterns of the e-Co particles before and after annealing: ~a! as-synthesized 9 nm particles and ~b! after annealed at 300 °C under vacuum for 3 h.

FIG. 4. TEM micrographs of ~a! hcp cobalt particles derived from annealed e-Co particles at 300 °C and ~b! Cobalt oxide particles from oxidation of the annealed cobalt particles.

ate in the formation of high quality hcp Co NC films. It should be mentioned that since the protecting ligands were removed during the annealing, the sample became very air sensitive. Exposing the annealed sample to ambient environment would result in immediate formation of a mixture cobalt oxide phases. This has been confirmed by x-ray diffraction ~XRD! and TEM experiments. These oxidized samples resemble the high quality cobalt oxide nanocrystals reported by Wang et al.35 It was possible to avoid excessive oxidation if the annealed samples were immersed in undried acetone prior to exposure of the sample to air. It is believed that trace amounts of oxygen and moisture in the acetone gently react with the particle surface, forming a thin oxide layer which hinders further oxidation. A similar controlled oxidation technique has been used in protecting Pd nanoparticles.36 Figure 4 displays two TEM images from annealed NC samples. Figure 4~a! is an image of hcp cobalt particles obtained by annealing e-Co particles under vacuum at 300 °C, and Fig. 4~b! is an image of oxidized Co particles produced when the surface of the particles was not treated before exposure to air. The particles in monolayer regions of the film were largely undisturbed during the transformations but particles in multilayer regions show signs of collapse and sintering. This same trend in the relative thermal stability of NC monolayers and multilayers has also been noted previously in cobalt oxide nanocrystal systems.35 Preparation of nearly monodisperse e-Co NCs allows the size evolution of properties in a new magnetic material to be observed. Their facile conversion to hcp nanoparticles also permits the magnetic properties of NC arrays to be modified after assembly. Figure 5 shows 5 K hysteresis measurements from a e-Co superlattice sample on silicon nitride before and after annealing. The as synthesized 9 nm e-Co particles showed a coercive field (H c ) of 500 Oe @Fig. 5~a!#. After annealing at 300 °C under argon and hydrogen ~5%! for 3 h, the sample showed an increase in coercivity to 1450 Oe @Fig. 5~b!#. Further annealing at 500 °C for 3 h did not give higher H c even though TEM and x-ray analysis showed that grain growth occurred along with the conversion to a largely fcc structure. Instead, H c decreased to 800 Oe. This coercivity change confirms that it is the phase transformation of the NCs and not a change in particle size that is responsible for the change in coercivity at 300 °C.

Downloaded 12 Feb 2004 to 129.107.14.178. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

J. Appl. Phys., Vol. 85, No. 8, 15 April 1999

S. Sun and C. B. Murray

4329

FIG. 7. TEM image of a 2D assembly of 9 nm cobalt nanocrystals. Inset: High resolution TEM image of a single particle. FIG. 5. Hysteresis loops of cobalt particles at 5 K from ~a! as-synthesized 9 nm e-Co particles, ~b! hcp cobalt particles from e-Co NCs annealed at 300 °C, and ~c! fcc1hcp cobalt particles from e-Co NCs annealed at 500 °C.

The temperature dependence of magnetization was measured in a 10 Oe field between 5 and 300 K using zero-fieldcooling ~ZFC! procedure.37 The results from 9 nm cobalt particles shown in Fig. 6~a! are typical for crystals in the nanoparticle regime. All particles studied are well below the critical size at which particle becomes a single domain magnet,38 and are small enough to display superparamagnetism. Thermal energy at room temperature is sufficient to overcome the coercive field of the particles and thus the orientation of magnetization of isolated particles fluctuates in time. Cooling at zero magnetic field freezes the moments of individual NCs into random orientations. When external field ~10 Oe! is applied, it energetically favors the moments of the individual particles to reorient with the field at low tempera-

FIG. 6. ZFC magnetization vs temperature of e-Co particles: ~a! M – T curve from as-synthesized 9 nm particles and ~b! M – T curve from same cobalt particles but diluted with octadecylamine.

ture, but there is not enough thermal energy to overcome the coercive field pinning the moments. As the sample is allowed to warm, the particles gain some thermal energy and begin to reorient their magnetic polarization with the 10 Oe external field and the total magnetization increases. At the blocking temperature the magnetization reaches a maximum as thermal energy becomes comparable to the energy gained by aligning the particle magnetic vectors in the weak field and the magnetization drops with further heating. The broad transition from superparamagnetic to ferromagnetic behavior shown in Fig. 6~a! at around 165 K is probably due largely to the magneto-static particle interactions in the close-packed arrays. As shown in Fig. 6~b!, diluting particles with ;80% 1-octadecylamine reduced the magneto-static coupling of the particles and gave a sharper transition at about 105 K. Development of a detailed understanding inter-particle of coupling requires the study of ordered NC arrays in which the size and spacing of particles can be precisely controlled. The e-Co NCs are very uniform in both size and shape, allowing them to self-organize readily into two-dimensional ~2D! and three-dimensional ~3D! superlattices. Slow evaporation of a carrier solvent from a cobalt dispersion spreaded on a flat substrate allows well-organized superlattice structures to be formed. A TEM image of a monolayer region in such a structure is shown in Fig. 7. The crystal structure of the particles is confirmed by electron diffraction and is also revealed by a high resolution TEM image shown in the inset of Fig. 7. The high-resolution transmission electron microscopy shows clear cobalt lattice planes with a spacing of 1.99 Å, close to d 31051.997 Å spacing expected for e-Co. The use of high boiling solvent like dodecane as the carrier solvent for the cobalt NCs allows slow evaporation at higher temperature. This added thermal energy permits the particles to diffuse to their lowest energy superlattice sites during evaporation, producing well-defined 3D hcp superlattice structure, as shown in Fig. 8. The hcp superlattice formation is driven by attractive van der Waals forces and magnetic interactions between the particles. This assembly process is reversible simply by immersing the sample into alkanes. Oleic acid on the particle surface can be replaced by

Downloaded 12 Feb 2004 to 129.107.14.178. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

4330

J. Appl. Phys., Vol. 85, No. 8, 15 April 1999

S. Sun and C. B. Murray 1

FIG. 8. TEM image of a 3D superlattice of 9 nm cobalt nanocrystals assembled on amorphous carbon film at 70 °C.

a variety of other acids such as octanoic acid (CH3~CH2!6COOH) and 5-dodecenoic acid (CH3~CH2!5CHvCH~CH2!3COOH) through simple ligand exchange. Engineering the lattice spacing of the magnetic NC superlattice is now possible by simply varying the length and bulkiness of the alkyl groups of the carboxylic acid. However, octanoic acid and shorter chains were insufficient to sterically stabilize the colloid. One idealized model for future ultra-high-density recording media has been proposed in which uniform particles with an average diameter of 8–10 nm or less and a high H c of 2500 Oe are employed to encode data in individual magnetic grains.17 Although the intrinsic coercivity of elemental cobalt is too low to be a candidate in advanced media, our synthetic approach is being extended to prepare ordered arrays of higher coercivity intermetallic nanoparticles ~e.g., CoPt!. The potential utility of these ordered magnetic superlattice systems is not limited to magnetic media but may reveal interesting magneto transport effects as well. Granular materials consisting of nanometer size magnetic particles in an dielectric or nonmagnetic metal matrix have been under intensive study recently because of their novel giant magnetoresistive properties.39–43 IV. CONCLUSIONS

General synthetic routes to monodisperse cobalt NCs with b-Mn ~e-Co! form have been presented. Controlled growth and steric stabilization of the NCs were provided by the combination of oleic acid and trialkylphosphine. Annealing the particles at 300 °C resulted in the quantitative phase transformation of the e-Co structure to the hcp structure. Procedures to induce the self-organization of the magnetic NCs into magnetic superlattices were identified. ACKNOWLEDGMENTS

The authors wish to thank B. A. Scott, D. Mitzi, M. T. Prikas ~IBM!, M. G. Bawendi ~MIT! for helpful discussions, and IBM Corporation for financial support.

Magnetic Properties of Fine Particles, edited by J. L. Dormann and D. Fiorani ~North-Holland, Amsterdam, 1992!. 2 J. A. A. J. Perenboom, P. Wyder, and F. Meier, Phys. Rep. 78, 173 ~1981!; V. M. Shalaev, ibid. 272, 61 ~1996!. 3 J. S. Bradley, in Cluster and Colloids, edited by G. Schmid ~VCH, 1994!, p. 523. 4 S. Klingelho¨fer, W. Heitz, A. Greiner, S. Oestreich, S. Fo¨rster, and M. Antonietti, J. Am. Chem. Soc. 119, 10116 ~1997!. 5 E. Bar-Sadeh, Y. Goldstein, C. Zhang, H. Deng, B. Abeles, and O. Millo, Phys. Rev. B 50, 8961 ~1994!. 6 D. B. Janes, V. R. Kolagunta, R. G. Osifchin, J. D. Bielefeld, R. P. Andres, J. I. Henderson, and C. P. Kubiak, Superlattices Microstruct. 18, 275 ~1995!. 7 A. N. Korotkov, R. H. Chen, and K. Likharev, J. Appl. Phys. 78, 2520 ~1995!. 8 R. P. Andres et al., Science 272, 1323 ~1996!. 9 D. L. Klein, R. Roth, A. K. L. Lim, A. P. Alivisatos, and P. L. McEuen, Nature ~London! 389, 699 ~1997!. 10 S. Tiwari, F. Rana, H. Hanafi, A. Hartstein, E. F. Crabbe, and K. Chan, Appl. Phys. Lett. 68, 1377 ~1996!. 11 J. J. Welser, S. Tiwari, S. Riston, K. Y. Lee, and Y. Lee, IEEE Electron Device Lett. 18, 278 ~1997!. 12 Y. Shi, K. Saito, H. Ishikura, and T. Hiramoto, J. Appl. Phys. 84, 2358 ~1998!. 13 S. H. Tolbert, A. B. Herhold, L. E. Brus, and A. P. Alivisatos, Phys. Rev. Lett. 76, 4384 ~1996!. 14 V. Colvin et al., Nature ~London! 370, 354 ~1994!. 15 B. O. Dabbousi et al., Appl. Phys. Lett. 66, 1316 ~1995!. 16 C. B. Murray, C. R. Kagan, and M. G. Bawendi, Science 270, 1335 ~1995!. 17 E. S. Murdock, R. F. Simmons, and R. Davison, IEEE Trans. Magn. 28, 3078 ~1992!. 18 T. Yogi and T. A. Nguyen, IEEE Trans. Magn. 29, 307 ~1993!. 19 P.-L. Lu and S. H. Charap, IEEE Trans. Magn. 30, 4230 ~1994!. 20 E. Grochowski and D. A. Thompson, IEEE Trans. Magn. 30, 3797 ~1994!. 21 R. M. H. New, R. F. W. Pease, and R. L. White, J. Vac. Sci. Technol. B 12, 3196 ~1994!. 22 D. N. Lambeth, E. M. T. Velu, G. H. Bellesis, L. Lee, and D. E. Laughlin, J. Appl. Phys. 79, 4496 ~1996!. 23 S. H. Charap, P.-L. Lu, and Y. He, IEEE Trans. Magn. 33, 978 ~1997!. 24 G. Schmid, Chem. Rev. 92, 1709 ~1992!. 25 M. T. Reetz and W. Helbig, J. Am. Chem. Soc. 116, 7401 ~1994!. 26 M. T. Reetz and S. A. Quaiser, Angew. Chem. Int. Ed. Engl. 34, 2240 ~1995!. 27 R. G. Osifchin, W. J. Mahoney, J. D. Bielefeld, R. P. Andres, J. I. Henderson, and C. P. Kubiak, Superlattices Microstruct. 18, 283 ~1995!. 28 S. A. Harfenist, Z. L. Wang, M. M. Alvarez, I. Vezmar, and R. L. Whetten, J. Phys. Chem. 100, 13904 ~1996!. 29 T. Teranishi and M. Miyake, Chem. Mater. 10, 594 ~1998!. 30 M. D. Bentzon, J. van Wonterghem, S. Morup, A. Tholen, and C. J. Koch, Philos. Mag. B 60, 169 ~1989!. 31 M. D. Bentzon and A. Tholen, Ultramicroscopy 38, 105 ~1990!. 32 D. Dinega and M. G. Bawendi ~unpublished!. The e-Co designation has been tentatively chosen to distinguish this structure (P4 3 32) from other structures, proposed a third polymorph of elemental cobalt. 33 C. B. Murray, D. J. Norris, and M. G. Bawendi, J. Am. Chem. Soc. 115, 8706 ~1993!. 34 R. W. G. Wychoff, Crystal Structure ~Wiley, New York, 1963!, Vol. 1, p. 51. 35 J. S. Yin and Z. L. Wang, Phys. Rev. Lett. 79, 2570 ~1997!; J. Phys. Chem. B 101, 8979 ~1997!. 36 M. N. Vargaftik et al., J. Chem. Soc. Chem. Commun., 937 ~1985!. 37 M. Hanson, C. Johansson, M. S. Pedersen, and S. Mo” rup, J. Phys.: Condens. Matter 7, 9269 ~1995!. 38 D. L. Leslie-Pelecky and R. D. Rieke, Chem. Mater. 8, 1770 ~1996!. 39 C. L. Chien, J. Appl. Phys. 69, 5267 ~1991!. 40 A. E. Berkowitz et al., Phys. Rev. Lett. 68, 3745 ~1992!. 41 J. Q. Xiao, J. S. Jiang, and C. L. Chien, Phys. Rev. Lett. 68, 3749 ~1992!. 42 Y. G. Pogorelov, M. M. P. de Azevedo, and J. B. Sousa, Phys. Rev. B 58, 425 ~1998!. 43 S. Takahashi and S. Maekawa, Phys. Rev. Lett. 80, 1758 ~1998!.

Downloaded 12 Feb 2004 to 129.107.14.178. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp