Structural stability of icosahedral FePt nanoparticles

Abstract

The structural stability of FePt nanoparticles of about 5–6 nm diameter was investigated by dynamic high resolution transmission electron microscopy. The FePt icosahedra were very stable under an electron beam flux of ∼20 A/cm² at 300 kV. Surface sputtering was suppressed due to the large sputtering threshold energy of a Pt-rich shell. Under a flux of ∼50 A/cm², the trapping potential well of the FePt particle on the supporting carbon film was lowered by the magnetic interaction between the electron beam and the particle, which leads to rotational and translational motions of the particle. A large dose of electrons (∼200 A/cm²) initiated melting and recrystallization of the FePt particle. The structure of the FePt nanoparticle, a Pt enriched shell around an Fe/Pt magnetic core, is believed to be responsible for its dynamic behaviour under different beam conditions.

---

Introduction

Multiply-twinned particles (MTPs) are the common structure for small clusters (<10 nm) of face-centered cubic (f.c.c.) materials. The non-crystallographic shape of MTPs can be understood in terms of a modified Curie-Wulff construction.¹ MTPs consist of single-crystal grains of various dimensions, so the particles can exhibit different shapes while the stable morphology of a MTP corresponds to a local minimum of the Gibbs free energy.² The energy barrier separating these local minima and the Boltzmann occupancy factors for the minima have been evaluated theoretically.³ By surmounting the energy barrier, a small particle can fluctuate between various shapes. This unstable state of the particle is called quasimelting. The magnitude of the energy barrier is important to the quasimelting state. Marks and Ajayan found that the energy barrier lowers as the temperature of the particle is raised. Experimentally, a free small particle decoupled from the substrate fluctuates at the lowest possible electron beam flux condition in a microscope (<0.1 A/cm²), which indicates extremely small activation-energy barriers.⁴ However, rather than the quasimelting mechanism, the structural fluctuation of small particles can also be explained as recrystallization from a thermodynamic melting state, which is due to a core excitation or a Coulomb explosion.⁵ For example, the instability of Au particles is affected partly by charge fluctuation of the particles.⁶ In addition, a shear-stress induced layer-by-layer atomic gliding has been proposed for Bi nanoparticle rotation, another example of particle instability, and such a solid-state transition does not result from high energy electron radiation.⁷ To experimentally resolve the mechanism of particle instability, transmission electron microscopy (TEM) is more suitable than other methodologies since it can reveal lattice structures of small particles and the electron beam can excite the particle as well. However, different small particles have diverse complex structures and it is difficult to obtain dynamic information from lattice resolved TEM images for small particles. Therefore, the influence of electron beam irradiation on the structures of small particles in general and MTPs in particular still remains unclear. In this work, we studied the structure stability of FePt nanoparticles under different electron beam intensities. Surface sputtering, rotational and translational motions as well as melting and recrystallization of the FePt particles were observed.

Experimental

The MTPs studied in this paper are FePt nanoparticles with diameters of about 5–6 nm. They were prepared by an inert-gas condensation method based on a DC-sputtering process in a continuous gas flow of Ar and He.⁸ The composition of the nanoparticles is measured to be Fe₅₂Pt₄₈ with an error less than 2%, as determined by energy dispersive X-ray spectroscopy. TEM samples were prepared by dispersing the particles in alcohol by ultrasonic treatment, dropping onto a porous carbon film supported on a copper grid, and then drying in air. The FePt nanoparticles were well separated on the supporting carbon film with inter-particle distances of tens of nanometers. Dynamic high resolution lattice images were recorded by a Philips CM300 FEG/UT electron microscope equipped with a real-time video recording and monitoring system. The microscope was operated under high vacuum of around 1 × 10⁻⁷ Torr. The video-cassette recorder (Mitsubishi HS-U746) has a time resolution of 1/60 s and the point resolution of the system is 0.168 nm at 300 kV.

Results and discussion

Previous studies indicate that FePt nanoparticles have an icosahedral structure with a Pt enriched shell around an Fe/Pt core.^8,9 The focal series of 20 lattice images were taken at a current density of 20 A/cm² at 300 kV for ∼100 s. It is remarkable to find that under such conditions the FePt nanoparticle was very stable. Images obtained from the exit wave reconstruction acquisition process show no differences in particle structure between the first ten images and the second ten images. Both reconstructed images can well resolve (1) the lattice fringes corresponding to {111} and {220} planes with spacings of 0.220 nm and 0.135 nm, and (2) the same atomic occupancy at edge columns of the FePt nanoparticle. With such irradiation, it was found that the structure of the FePt MTP remained an icosahedron for 30 min. Fig. 1 shows reconstructed exit wave phase images of one representative FePt nanoparticle under an electron beam flux of ∼20 A/cm² at 300 kV for ∼100 s, 20min, and 30 min.⁸ The particle sizes varied slightly during the irradiation. The number of shells of the FePt nanoparticle changed from 14 to 13 and 12.5 after being exposed to intense electron beam irradiation for 20 and 30 min, respectively. From the HRTEM image in Fig. 1c, it can be found that the outermost shell at the bottom of the nanoparticle is removed and the number of shells is marked as 12.5 consequently. The sputtering of surface atoms resulted from inelastic scattering between the incident beam and the surface atoms as a certain amount of energy was transferred to the surface atoms which is larger than the surface binding energy. At 300 kV, the maximum transferable kinetic energy is 15.25 eV for Fe and 4.37 eV for Pt,¹⁰ and the sputtering threshold energies of Fe and Pt are in the ranges 4–8 eV and 8–16 eV, respectively. Since the particle has a Pt rich shell and a Fe-rich core,⁹ and the maximum energy transferred to Pt (4.37 eV) is far lower than its sputtering threshold energy (8–16 eV), the particle is stable with an extremely low and relatively stable sputtering rate (∼20 min/shell) at such flux conditions. Although the energetic data may not justify the occurrence of sputtering erosion, it indicates that slight sputtering erosion may contribute to the removal of the particle surface layer.

Fig. 1 Reconstructed exit wave phase images of one representative FePt nanoparticle under an electron beam flux of ∼20 A/cm² at 300 kV.⁸ (a) Primary exit wave phase image of the icosahedral FePt nanoparticle with {111} and {220} planes resolved. (b) Exit wave phase image exposed to electron beam for 20 min with one layer peeled off. (c) Exit wave phase image exposed to electron beam for 30 min with one and a half layers peeled off.

With increased electron beam flux, the FePt nanoparticle is found to be less stable on the supporting carbon film. Rotational and translational motions of the FePt particle were observed as the electron-beam flux increased to ∼50 A/cm². The images in Fig. 2 were selected from a videotape recording over a period of 140 min. The FePt particle began to rotate on the supporting carbon film after 30 min irradiation under the beam flux. Real-time lattice fringe changes can be observed with the orientation change of the particle. The particle rotated from a [112] (Fig. 2a) to a [211] orientation (Fig. 2b) in 20 min. It then rotated to [111] (Fig. 2c) and [112] (Fig. 2d) orientations consecutively. The time interval of orientation was ∼20 min. Particles usually rotated with different speeds, which is most likely due to the cohesion variations of the particles and the supporting carbon film. After ∼60 min of the electron irradiation, the particle appeared to detach from the supporting carbon film and float in vacuum. Fig. 2c demonstrates such a case where the particle and the supporting film cannot be brought into focus simultaneously, indicating that they are at different heights in terms of the beam optics. Since the depth of field of the imaging system is about 10–20 nm which is much larger than the diameter of the FePt nanoparticle, the height of the particle above the supporting film is more than 20 nm. Despite the fact that the particle was floating and heat transferred to the film was greatly decreased, the particle retained the icosahedral shape and it was not in the quasimelting state. Ajayan and Marks⁴ proposed that the energy well to trap an Au particle on an MgO substrate is much deeper than the morphological potential energy barriers of the Au particles. Therefore, once the Au particle was detached from the substrate, a quasimelting state initiated and the particle fluctuated between different morphologies. In our experiment, since the detached FePt particle was not in the quasimelting state, the magnitude of the potential well that traps the FePt particle on the supporting carbon film is relatively lower than those of the morphological barriers of the FePt particles. This may be due to the magnetic interaction between the electron beam and the particle with an Fe/Pt magnetic core which generates a mechanical torque and lowers the trapping well.

Fig. 2 Dynamic HRTEM images of one FePt nanoparticle taken under an electron beam flux of ∼50 A/cm² at 300 kV showing continuous rotation, accumulation and alignment of the particles.

One interesting phenomenon is that the particles began to accumulate and line up, as shown in Fig. 2e and f. Each FePt nanoparticle can be regarded as a magnetic entity with random magnetic momentum direction. The moment of force between the magnetic momentum and magnetic field induced by the incident electron beam makes the FePt nanoparticle rotate towards the direction where the magnetic momentum is perpendicular to the incident electron beam. The magnetic dipole interaction between the nanoparticles resulted in the translational motion of the nanoparticles towards each other. The secondary electron emission charged the detached FePt nanoparticles positively.¹¹ The FePt nanoparticles nearby were polarized as electric dipoles with polarization directions parallel to the electron beam. The Coulomb force between the parallel electric dipoles prevents the nanoparticles gathering too close. Then the nanoparticles gathered in a line with separations of several nanometers (Fig. 2e and f).

With an electron beam flux of ∼200 A/cm², the nanoparticle showed typical behaviors of the quasimelting state as well as the process of melting and recrystallization. Rotational motion can be observed via changes in lattice fringes in the real-time HRTEM images after the particle was irradiated for 30 min. The period required to rotate the particle is shorter than that under a flux of ∼50 A/cm². The results suggest that the energy deposited in the particle by each inelastic scattering event accumulates and a dynamic heat balance between the particle and the supporting carbon film cannot be established at these flux conditions without involving a morphological change of the particle. Different from the rotation of the particle discussed previously, the structural fluctuation of the particle at the flux of ∼200 A/cm² involved a melting state, as shown in Fig. 3 and 4. Fig. 3 shows a representative case with processes of floating, fluctuation, dissolution and recrystallization. Fig. 4 shows a case in which the states of the nanoparticles cycle between quasimelting and recrystallization of the truncated icosahedron, both twin and single crystal. HRTEM images (Fig. 3c, 4b and g) reproduced from the video recorder and CCD camera show that there exist no lattice fringes throughout the nanoparticle. The melting state was however not stable and lasted for several minutes. It recrystallized starting from the edge (Fig. 4c) which was on the supporting carbon film. After ∼150 s, a truncated icosahedral structure formed on the supporting carbon film with the (111) plane in contact with the film. Due to its high thermal conductivity, the supporting film was a virtual heat sink for the temperature gradient which was the driving force for the crystallization of the truncated icosahedral structure. The truncated icosahedral structure was almost as stable as the initial icosahedral one. It existed for ∼30 minutes before it was melted (Fig. 4e) under the same electron beam conditions. It then became an unstable twin structure and was melted again after ∼1 minute. At the end of our recording, the particle adopted a single crystal structure as shown in Fig. 4h. The initial particle (Fig. 4a) consisted of 13 shells with ∼8000 atoms and ∼5.8 nm in size. The recrystallized truncated icosahedron shown in Fig. 4c consisted of 14 shells with ∼7000 atoms. Compared with the truncated icosahedron, the recrystallized twin structure was smaller. The final single crystal structure was much smaller than the initial one. As shown in Fig. 4h, there were two sets of lattice fringes which consisted of 22 and 16 lattice fringes corresponding to d₂₀₀ = 0.19 nm. Supposing the thickness of the single crystal to be the average of the length and width of the crystal, then the crystal dimensions were about 4.2 × 3.0 × 3.6 nm³, which are similar to the previously reported single crystal nature of 4 nm Fe₅₂Pt₄₈ particles.¹² Such an ideal structure theoretically contained ∼3000 atoms. Some FePt atoms in the melting state can also be observed at the bottom right of the crystallite in Fig. 4h. The cohesion energy for surface reconstructed decahedra using the embedded atom method (EAM) on a palladium cluster indicated that the single crystalline structure had the lowest cohesion energy when the cluster size was smaller than 15³ = 3375 atoms.¹³ Our observation on the FePt particle was consistent with the calculation predictions.

Fig. 3 Dynamic HRTEM images of one FePt nanoparticle taken under an electron beam flux of ∼200 A/cm² showing four processes: floating, fluctuation, dissolution and recrystallization.

Fig. 4 Dynamic HRTEM images of one FePt nanoparticle taken under an electron beam flux of ∼200 A/cm² showing cycles between the quasimelting state and recrystallization of the truncated icosahedron, both twin and single crystal.

Conclusions

The experimental results show that FePt magnetic nanoparticles can be very stable under an electron beam by each inelastic scattering event TEM operation conditions (∼20 A/cm²). With a higher beam flux of ∼50 A/cm², a quasimelting phase was still not observed but the rotational motion and displacement of the nanoparticle on the supporting carbon film were the results of the magnetic and electrical interactions between the particles and the beam. However, as the beam flux increased to ∼200 A/cm², thermodynamic melting and recrystallization occurred repeatedly and the final morphology of the particle was determined by the stable shape of an atom cluster with a specific size.

Acknowledgements

This work was supported by the Berkeley Scholar Program, the National Natural Science Foundation of China (No. 50671003), the Program for New Century Excellent Talents in University (NCET-06-0175), the Director, Office of Science, Office of Basic Energy Science, of the U. S. Department of Energy under contract No. DE-AC02-05CH11231, and the Deutsche Forschungsgemeinschaft SFB 445.

References

1. L. D. Marks, Rep. Prog. Phys., 1994, 57, 603–649.
2. L. D. Marks, P. M. Ajayan and J. Dundurs, Ultramicroscopy, 1986, 20, 77–82.
3. P. M. Ajayan and L. D. Marks, Phys. Rev. Lett., 1988, 60, 585–587.
4. P. M. Ajayan and L. D. Marks, Phys. Rev. Lett., 1989, 63, 279–282.
5. P. Williams, Appl. Phys. Lett., 1987, 50, 1760–1762.
6. S. Iijima and T. Ichihashi, Phys. Rev. Lett., 1986, 56, 616–619.
7. A. Be'er, R. Kofman, F. Phillipp and Y. Lereah, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 74, 224111.
8. R. M. Wang, O. Dmitrieva, M. Farle, G. Dumpich, H. Q. Ye, H. Poppa, R. Kilaas and C. Kisielowski, Phys. Rev. Lett., 2008, 100, 017205.
9. R. M. Wang, O. Dmitrieva, M. Farle, G. Dumpich, M. Acet, S. Mejia-Rosales, E. Perez-Tijerina, M. J. Yacaman and C. Kisielowski, J. Phys. Chem. C, 2009, 113, 4395–4400.
10. N. J. Zaluzec and J. F. Mansfield, Intermediate Voltage Microscopy and its Application to Materials Science, ed. K. Rajan, Philips Electron Optics Publishing Group, Mahwah, New Jersey, 1987.
11. R. F. Egerton, P. Li and M. Malac, Micron, 2004, 35, 399–409.
12. S. H. Sun, C. B. Murray, D. Weller, L. Folks and A. Moser, Science, 2000, 287, 1989–1992.
13. J. L. Rodriguez-Lopez, J. M. Montejano-Carrizales, U. Pal, J. F. Sanchez-Ramirez, H. E. Troiani, D. Garcia, M. Miki-Yoshida and M. Jose-Yacaman, Phys. Rev. Lett., 2004, 92, 196102.

---