Monodisperse FePt Nanoparticles and Ferromagnetic FePt Nanocrystal Superlattices

See allHide authors and affiliations

Science  17 Mar 2000:
Vol. 287, Issue 5460, pp. 1989-1992
DOI: 10.1126/science.287.5460.1989


Synthesis of monodisperse iron-platinum (FePt) nanoparticles by reduction of platinum acetylacetonate and decomposition of iron pentacarbonyl in the presence of oleic acid and oleyl amine stabilizers is reported. The FePt particle composition is readily controlled, and the size is tunable from 3- to 10-nanometer diameter with a standard deviation of less than 5%. These nanoparticles self-assemble into three-dimensional superlattices. Thermal annealing converts the internal particle structure from a chemically disordered face-centered cubic phase to the chemically ordered face-centered tetragonal phase and transforms the nanoparticle superlattices into ferromagnetic nanocrystal assemblies. These assemblies are chemically and mechanically robust and can support high-density magnetization reversal transitions.

The synthesis of nanoparticles with controlled size and composition is of fundamental and technological interest. The effort to understand the physics of ever smaller structures has been paralleled by attempts to exploit their beneficial properties. The increased surface area and tailored surface chemistry of metal nanoparticles have long been used to optimize the activity and specificity of catalysts (1). Semiconductor nanocrystals show size-tunable optical properties and have been integrated into exploratory optical and electronic devices (2, 3). Engineering the interparticle spacing of metal particle arrays has revealed evidence of tunable metal-insulator transitions (4). Small metal particle arrays have been used to build single-electron devices (5,6). Progress in ultrahigh-density magnetic recording is due in part to the development of metal thin film media with smaller particles, tighter size distributions, and optimized compositions (7, 8). Recently, the study of magnetization processes on nanometer length scales has intensified (9,10).

We report on the synthesis of monodisperse FePt nanoparticles with controlled size and composition and the fabrication of ferromagnetic FePt nanocrystal superlattice assemblies with tunable interparticle spacings. FePt alloys are an important class of materials in permanent magnetic applications because of their large uniaxial magnetocrystalline anisotropy [Ku ≅ 7 × 106 J/m3 (11)] and good chemical stability. As the magnetic stability of individual particles scales with the anisotropy constant, Ku , and the particle volume, V, small FePt particles may be suitable for future ultrahigh-density magnetic recording media applications (12). Until now, the synthesis of FePt particle thin films has mainly relied on vacuum deposition techniques (13–15). Postdeposition annealing has proven essential to transform the as-deposited chemically disordered face-centered cubic (fcc) structure into the chemically ordered face-centered tetragonal (fct) phase, which has high Ku . Random nucleation in the initial stages of growth, however, typically results in broad distributions of particle sizes, which may be further aggravated by agglomeration during annealing. Solution phase chemical synthesis, in contrast, has been successfully used to prepare monodisperse metal particles (16–19). However, synthetic procedures have not yet been developed to prepare monodisperse hard magnetic FePt nanoparticles.

To prepare FePt nanoparticles, we used a combination of oleic acid and oleyl amine to stabilize the monodisperse FePt colloids and prevent oxidation. The synthesis is based on the reduction of Pt(acac)2 (acac = acetylacetonate, CH3COCHCOCH3) by a diol and the decomposition of Fe(CO)5 in high-temperature solutions. The use of diol or polyalcohol (for example, ethylene glycol or glycerol) to reduce metal salts to metal particles is referred to as the “polyol process” (20). We modified this procedure by using a long-chain 1,2-hexadecanediol to reduce the Pt(acac)2to Pt metal. Thermal decomposition of Fe(CO)5 has been used to produce Fe particles (17). Both chemical reactions were initiated together in the presence of oleic acid and oleyl amine, providing a convenient route to monodisperse FePt nanoparticles (21).

The size and composition of these FePt nanoparticles can be readily controlled. Their composition is adjusted by controlling the molar ratio of iron carbonyl to the platinum salt. For example, with dioctylether as solvent, a 3:2 molar ratio of Fe(CO)5 to Pt(acac)2 gave Fe48Pt52 particles, a 2:1 molar ratio yielded Fe52Pt48, and a 4:1 molar ratio produced Fe70Pt30 (22). The FePt particle size can be tuned from 3 to 10 nm by first growing 3-nm monodisperse seed particles in situ and then adding more reagents to enlarge the existing seeds to the desired size. These particles are isolated and purified by centrifugation after the addition of a flocculent (for example, ethanol) and can be redispersed in nonpolar solvents in a variety of concentrations (23).

When the FePt colloids are spread on a substrate and the carrier solvent is allowed to slowly evaporate, FePt nanoparticle superlattices are produced (24). A drop (∼0.5 μl) of dilute FePt dispersion (∼1 mg/ml) was deposited on a SiO-coated copper grid for transmission electron microscopy (TEM) studies. The results reveal that the particles are monodisperse with σ ≤ 5% in diameter and readily self-assemble into three-dimensional (3D) superlattices. A TEM image (Fig. 1A) shows a thin section of a hexagonal close-packed 3D array of 6-nm Fe50Pt50 particles with a nearest neighbor spacing of ∼4 nm maintained by the oleic acid and oleyl amine capping groups. Room temperature ligand exchange of these long-chain capping groups for shorter RCOOH/RNH2 (R = dodecyl down to hexyl chains) allows the interparticle distance to be adjusted. Ligand exchange with hexanoic acid/hexylamine yields a cubic packed multilayer of 6-nm Fe50Pt50 particles with ∼1-nm spacings (Fig. 1B). Such a transition from hexagonal to cubic packing has been observed in monodisperse cobalt nanoparticle assemblies (19). The symmetry of the observed superlattices is influenced by several experimental parameters including the relative dimensions of the metal core and the organic capping, as well as the annealing history of the sample. The FePt particle assemblies show no obvious aggregation upon annealing at temperatures up to 600°C under static N2 atmosphere (1 atm). High-resolution scanning electron microscopy (HRSEM) was used to image the annealed FePt particle assembly on a thermally oxidized Si substrate. For a 180-nm-thick, 4-nm Fe52Pt48 particle assembly annealed at 560°C for 30 min, the HRSEM images of both surface (Fig. 1C) and cross section (not shown in figure) of the assembly show that the particles are well separated with no agglomeration occurring. Interparticle spacings, however, are reduced from ∼4 to ∼2 nm, as indicated by TEM, HRSEM, and small angle x-ray scattering experiments. Some coherent strain is observed in the superlattices because of this shrinkage. Rutherford backscattering measurements on these annealed 4-nm Fe52Pt48 particle assemblies indicate 40 to 50% (atomic %) carbon content. This shows that annealing at high temperature does not result in the loss of stabilizing ligands; rather, they are converted to a carbonaceous coating around each particle. Energy dispersive x-ray spectroscopy measurements carried out during the SEM and scanning TEM imaging of the films confirm that the average nanocrystals are slightly iron-rich and that the carbon is intimately associated with the particles.

Figure 1

(A) TEM micrograph of a 3D assembly of 6-nm as-synthesized Fe50Pt50particles deposited from a hexane/octane (v/v 1/1) dispersion onto a SiO-coated copper grid. (B) TEM micrograph of a 3D assembly of 6-nm Fe50Pt50 sample after replacing oleic acid/oleyl amine with hexanoic acid/hexylamine. (C) HRSEM image of a ∼180-nm-thick, 4-nm Fe52Pt48 nanocrystal assembly annealed at 560°C for 30 min under 1 atm of N2 gas. (D) High-resolution TEM image of 4-nm Fe52Pt48nanocrystals annealed at 560°C for 30 min on a SiO-coated copper grid.

Electron diffraction of the as-synthesized Fe52Pt48 particles reveals a typical fcc pattern whereas Fe52Pt48 particles annealed at 560°C show an fct internal structure. Figure 1D is a high-resolution TEM micrograph of individual 4-nm Fe52Pt48particles annealed at 560°C for 30 min. The particles are single crystals with a lattice spacing of 2.20 Å, characteristic of the (111) planes in the chemically ordered fct FePt phase (25). The change of the internal particle crystal structure upon annealing is seen in wide-angle x-ray diffraction (XRD) studies.Figure 2 shows a series of such XRD patterns for ∼1-μm-thick Fe52Pt48assemblies as a function of annealing temperature at a constant annealing time of 30 min. As-synthesized particles exhibit the chemically disordered fcc structure (Fig. 2A). Annealing induces the Fe and Pt atoms to rearrange into the long-range chemically ordered fct structure, as indicated by the (111) peak shifts and evolution of the (001) and (110) peaks (Fig. 2, B to E). At annealing temperatures below 500°C, only partial chemical ordering is observed (Fig. 2, B and C). The chemical ordering can be increased by annealing at higher temperatures (Fig. 2, D and E) or by increasing the annealing time (not shown in figure). Diffraction patterns of the samples annealed above 540°C match those of fct FePt (11, 25). Fitting of the XRD line shapes of the sample annealed at 550°C confirms that the particles are individual single crystals with dimensions matching the average particle size of 4 nm (σ = 5% in diameter) determined by statistical analysis of the TEM images. Annealing 4-nm Fe52Pt48 particles at higher temperature, for example, ≥600°C, however, results in an increase of the average particle size and a broadening of the size distribution based on TEM and XRD analyses.

Figure 2

XRD patterns (A) of as-synthesized 4-nm Fe52Pt48 particle assemblies and a series of similar assemblies annealed under atmospheric N2 gas for 30 min at temperatures of (B) 450°C, (C) 500°C, (D) 550°C, and (E) 600°C. The indexing is based on tabulated fct FePt reflections (25). The diffraction patterns were collected with a Siemens D-500 diffractometer with Cu Kα radiation (wavelength λ = 1.54056 Å).

Superconducting quantum interference device magnetometry measurements of 4-nm FePt particles show that the as-synthesized particle assemblies are superparamagnetic (coercivityH c = 0 Oe) at room temperature. The temperature-dependent magnetization was measured in a 10-Oe field between 5 and 400 K with the standard zero–field-cooling and field-cooling procedures (26). These studies indicate that superparamagnetic behavior is blocked at 20 to 30 K. This low blocking temperature is consistent with low magnetocrystalline anisotropy of the fcc structure. Annealing converts the particles to the high-anisotropy fct phase and transforms them into room temperature nanoscale ferromagnets. The coercivity of these ferromagnetic assemblies is tunable by controlling annealing temperature and time, as well as the Fe:Pt ratio and particle size. Curve A of Fig. 3 shows representative in-plane coercivity data of a series of ∼140-nm-thick, 4-nm Fe52Pt48 samples as a function of annealing temperature. There is little difference between in-plane and out-of-plane coercivities and hysteresis behavior, indicating random orientation of the easy axes of individual FePt nanocrystals. FePt stoichiometry-dependent coercivity data (curve B of Fig. 3) demonstrate that Fe-rich FexPt1−x(x ∼ 0.52 to 0.60) nanocrystal assemblies have the largest coercivity, consistent with earlier reports on vacuum-deposited FePt thin films (12, 27).

Figure 3

Curve A shows the in-plane coercivity of a series of ∼140-nm-thick, 4-nm Fe52Pt48assemblies as a function of annealing temperature. Each sample is annealed for 30 min under 1 atm of N2 gas. Curve B indicates the composition-dependent coercivity of ∼140-nm-thick FePt assemblies annealed at 560°C for 30 min.

The annealed FePt nanocrystal assemblies are smooth ferromagnetic films that can support high-density magnetization reversal transitions (bits). A ∼120-nm-thick assembly of 4-nm Fe48Pt52 nanocrystals with an in-plane coercivity of H c = 1800 Oe was selected for initial recording experiments. Atomic force microscopy studies of this sample indicate a 1-nm root-mean-square variation in height over areas of 3 μm by 3 μm. A static write/read tester was used for the recording experiments (28). The read-back sensor voltage signals (Fig. 4A) from written data tracks correspond to linear densities of 500, 1040, 2140, and 5000 flux changes per millimeter (fc/mm) (curves a to d, respectively). These write/read experiments demonstrate that this 4-nm Fe48Pt52 ferromagnetic nanocrystal assembly supports magnetization reversal transitions at moderate linear densities that can be read back nondestructively. Much higher recording densities, beyond the highest currently achievable linear densities of ∼20,000 fc/mm, can be expected if the thickness of these ferromagnetic assemblies can be reduced to ∼4 nm. On the basis of conventional grain size scaling arguments, such media should support recording densities about 10 times larger than the CoCr-based alloy media presently used industry wide (12). A critical factor in achieving such a goal is the thermal stability of the magnetic transitions, which can be assessed with a dynamic coercivity method. This method relies on remanent coercivity measurements as a function of the applied magnetic field pulse width (29).Figure 4B shows such data, from which the ratio of the energy barrier for magnetization reversal (KuV) to the thermal energy (k B T, wherek B is the Boltzmann constant and T is the temperature) is extracted as an inverse slope parameter (29). In a simple uniform rotation picture, which is justified for the present small, isolated particles, the inverse slope becomes 〈Ku 〉〈V〉/k B T= 48. An average anisotropy constant of 〈Ku 〉 = 5.9 × 106 J/m3 can be extracted, assuming spherical 4-nm-diameter particles. These preliminary results raise the expectation that thinner and higher coercivity assemblies will indeed permit drastically reduced bit cell sizes.

Figure 4

(A) Magneto-resistive (MR) read-back signals from written bit transitions in a 120-nm-thick assembly of 4-nm-diameter Fe48Pt52nanocrystals. The individual line scans reveal magnetization reversal transitions at linear densities of (a) 500, (b) 1040, (c) 2140, and (d) 5000 fc/mm. (B) Dynamic coercivity measurements (H CR) of the sample in (A) at 300 K over a range from 5 ns to 65 s. The measured data (•) are fit to a dynamic coercivity law for pulse width > 10−6 s (solid curve) (28).

The present synthesis provides a simple procedure for the preparation of monodisperse FePt nanoparticles and ferromagnetic FePt nanocrystal superlattice assemblies. The reported control of particle size, composition, crystal structure, and interparticle spacings will enable rigorous testing of current theoretical descriptions of nanometer-scale magnetics. Initial recording studies demonstrate that an assembly of magnetic nanocrystals as small as 4 nm can support stable magnetization reversal transitions (bits) at room temperature. It is expected that optimized ferromagnetic nanocrystal superlattices may contribute to future magnetic recording at areal densities in the terabits per square inch regime.

  • * To whom correspondence should be addressed. E-mail: ssun{at}


View Abstract

Navigate This Article