Journal of Nanoparticle Research

, Volume 13, Issue 12, pp 7005–7012

Synthesis of single-crystal Sm-Co nanoparticles by cluster beam deposition

Authors

    • Department of Physics and AstronomyUniversity of DE
  • W. Li
    • Department of Physics and AstronomyUniversity of DE
  • G. C. Hadjipanayis
    • Department of Physics and AstronomyUniversity of DE
  • D. J. Sellmyer
    • Department of Physics and AstronomyUniversity of Nebraska
Research Paper

DOI: 10.1007/s11051-011-0612-8

Cite this article as:
Akdogan, O., Li, W., Hadjipanayis, G.C. et al. J Nanopart Res (2011) 13: 7005. doi:10.1007/s11051-011-0612-8

Abstract

Single-crystal Sm-Co nanoparticles have been successfully produced by a cluster beam deposition technique. Particles have been deposited by DC magnetron sputtering using high Ar pressures on both single-crystal Si substrates and Au grids for the magnetic and structural/microstructural properties, respectively. Oxidation of the particles is prevented by using carbon buffer and cover layers. Nanoparticles have a uniform size distribution with an average size of 4.2, 6 and 7 nm at 1, 1.5 and 2 Torr of Ar pressure, respectively. At 1 Torr, the particles have the disordered 1:7 structure and a high coercivity of 19 kOe at 10 K. These particles show a superparamagnetic behavior with a blocking temperature of TB = 145 K. From this value of TB and the particle volume, the value of anisotropy constant K is estimated to be around 2.2 × 10ergs/cc. Heat is introduced to the particles during their flight to the substrate to increase the particle size. Nanoparticles of SmCo5 with an average size of 15 nm and high room temperature coercivity have been produced. No change in magnetic and structural properties of the samples has been observed even after 10 months. Cluster beam deposition could play a key role for the production of rare earth nanoparticles for many applications.

Keywords

SputteringCoercivityNanoparticlesSm-CoRare earth metals

Introduction

Synthesis of high-anisotropy magnetic nanoparticles with desired structure and size is of great fundamental and technological interest (McHenry and Laughlin 2000). Rare earth intermetallic compounds (RE-TM) are known to have very high anisotropy; however, their high reactivity makes it harder to produce nanoparticles of these materials with the desired properties. Direct production of these nanoparticles with good magnetic properties without post-annealing could be the key to use these particles for potential applications (Hadjipanayis 1999). Previously, good magnetic properties were achieved in thin films made by physical vapor deposition, but these required a subsequent post-annealing (Sayama et al. 2004; Zhang et al. 2011). Very few studies exist on the preparation of magnetically hard nanoparticles of RE-TM compounds including those via ball milling (Kirkpatrick et al. 1996; Akdogan et al. 2009a, Akdogan et al. 2009b, Akdogan et al. 2010), chemical synthesis (Gu et al. 2003; Hou et al. 2007; Matsushita et al. 2010) and cluster gun (CG) (Stoyanov et al. 2003a; Tuaillon-Combes et al. 2003). The latter technique was successfully used to fabricate FePt (CoPt) (Stoyanov et al. 2003b; Qiu and Wang 2006; Liu et al. 2011) and Co/CoO(Skumryev et al. 2003) core–shell nanoparticles with excellent magnetic properties. These results are promising since the same technique could be used to produce RE-TM nanoparticles with good magnetic properties. Recently, single-crystal YCo5 nanoparticles with excellent magnetic properties have been produced with this technique (Balasubramanian et al. 2011a), which opened the way for the fabrication of Sm-Co nanoparticles; however, the latter are expected to be harder to synthesize due to the amorphization and oxidation problems of Sm-based nanoparticles. One of the main advantages of this technique is that it can be easily scaled up to produce larger amounts of nanoparticles with excellent material purity and without the presence of chemical by-products (Binns et al. 2005; Wegner et al. 2006).

Sm-Co compounds are very attractive for nanofabrication (Gutfleisch et al. 2011), because they have a large anisotropy in bulk and it is easier to control their composition through cluster beam deposition. Especially, fabrication of well-separated, single-crystal nanoparticles with moderate coercivity plays a key role in the development of high-density recording media (Frey and Sun 2010). The SmCo7 alloys have properties between those of SmCo5 (1:5) and Sm2Co17 (2:17). SmCo7 (1:7), with TbCu7 structure, is a metastable phase, which could only be formed during mechanical alloying and sputtering (Liu et al. 2005).

For many applications, control of the particle size is one of the main problems. Most often, post-annealing is necessary to increase the particle size. However, control on the agglomeration of the particles is very limited, which makes the end product non-uniform with much bigger agglomerates than desired. In the case of FePt nanoparticles, Colak and Hadjipanayis (2009) used silica coating to prevent agglomeration of chemically synthesized nanoparticles. In cluster beam deposition samples, the agglomeration problem is solved by using a more confined plasma (through a special iron pole piece) and applying high sputtering powers (Qiu et al. 2006, Qiu and Wang 2006; Liu et al. 2011). However, these techniques have not yet been applied to RE-Co nanoparticles. Another problem with post-annealing is the oxidation, which is especially important for RE-TM nanoparticles. In order to control the particle size and oxidation, we introduce in-flight annealing of the particles (which is explained in next section in detail).

Previous attempts to produce high anisotropy Sm-Co nanoparticles with CBD failed; Stoyanov et al. (2003a) produced Sm-Co nanoparticles that lacked crystallinity and gave 5 mT coercivities even at 5 K, compared to 19 kOe we got at 10 K. Tuaillon-Combes et al. (2003) also produced Sm-Co nanoparticles with CBD technique. However, X-ray photoelectron spectroscopy of the deposited clusters showed that the Sm atoms were segregated on the cluster surface. They achieved RT coercivity of 700 Oe only after ex situ annealing at 570 °C in Nb matrix. In this study, we have produced Sm-Co nanoparticles with good magnetic properties from 1:5 and 2:17 targets using inert gas condensation in a CG.

Experimental procedure

Schematic illustration of the sputtering chamber can be seen in Fig. 1. In addition to the two magnetron guns for film sputtering, the system has a heater for in-flight annealing (Fig. 1i) and a CG attached to it for nanoparticle production (Fig. 1ii). CG is a magnetron gun in a high-pressure chamber (aggregation chamber). Aggregation chamber consists of two sections namely zone 1 and zone 2. Zone 1 is the region very close to the target where the extraction of the atoms from the alloy target (in our case 1:5 and 2:17 alloy) occurs. Since a high Ar pressure of 0.5–5 Torr is used rather than the usual 5 mTorr in classical sputtering, hot atoms are cooled and condensed in a cold inert gas to create the clusters (zone 2). The particles then leave the CG through a small orifice on top of the gun toward the silicon substrate, which is mounted on a rotating platform. As soon as the particles leave the aggregation chamber, particle size stops increasing. In order to increase the size further, a heater is introduced right after the CG with adjustable temperatures between 200 and 900 °C (Fig. 1i).
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-011-0612-8/MediaObjects/11051_2011_612_Fig1_HTML.gif
Fig. 1

Schematic illustration of the magnetron sputtering chamber: heater (i) and the CG (ii) can be seen in close-up

The base pressure in the sputtering chamber was 2 × 10−7 Torr, and high-purity Ar (99.9999%) was used for the deposition with a pressure of 5 mTorr inside the main chamber and 1–2 Torr inside the CG. A DC power of 25 Watts was applied to both the 1:5 and 2:17 alloy targets. Samples were sputtered on 500-μm-thick Si (100) wafers for magnetic measurements and on Au grids for transmission electron microscopy (TEM). The samples were coated with carbon using a DC power of 24 Watts (sputtering rate = 2.6 Å/sec) in order to prevent oxidation. Microstructure characterization and composition analyses of the samples were performed with JEOL JEM-3010 TEM and JSM 6330F SEM. Magnetic measurements at room temperature and below were taken with a Quantum Design Versalab vibrating sample magnetometer (VSM) with a maximum field of 3 T and with a quantum design squid (superconducting quantum interference device) with a maximum field of 5.5 T. X-ray diffraction (XRD) measurements were taken with Rigaku Ultima IV.

Results and discussion

Particles from the SmCo5 Target

Figure 2 shows the SEM (a) and a TEM bright-field (BF) (b) image of Sm-Co nanoparticles deposited at 1 Torr. SEM image was taken from a sample of nanoparticles deposited for 500 s. The Sm-Co nanoparticles on the TEM grid were deposited for 10 s. SEM image indicates that longer deposition times only increase the amount of the particles. Therefore, particle formation stops after the particle beam comes out of the gun, and the morphology is independent of the substrate and the deposition time. Binns et al. (2005) showed similar results on Fe nanoparticles deposited with the CBD technique. Nucleation and growth take place in the aggregation chamber where the particles spent most of their time because of the high pressure. Particle travel time after the aggregation chamber is too fast to affect their size. The inset of Fig. 1b indicates a narrow size distribution with an average size of 4.2 nm (the ImageJ program (Rasband and ImageJ 1997) has been used to determine the average particle size). Energy-dispersive X-ray spectroscopy (EDS) of the sample shows a composition of 15.5 at % Sm and 84.5 at % Co, which is very close to that of SmCo5. The slight difference between the target and the end product composition is due to the different sputtering rates of Sm and Co. High-resolution TEM (HRTEM) image clearly shows single-crystal SmCo7 nanoparticles (Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-011-0612-8/MediaObjects/11051_2011_612_Fig2_HTML.gif
Fig. 2

a SEM image of Sm-Co nanoparticles on Si substrate, b BF image of Sm-Co nanoparticles deposited on TEM grid at 1 Torr from 1:5 target; inset shows the size distribution of the nanoparticles

https://static-content.springer.com/image/art%3A10.1007%2Fs11051-011-0612-8/MediaObjects/11051_2011_612_Fig3_HTML.jpg
Fig. 3

HRTEM image of SmCo7 nanoparticles deposited from the 1:5 target

XRD data shown in Fig. 4 consist of two very broad peaks due to the small particle size. Rietveld refinement of the XRD data (Fig. 4) has been done with PCW (powdercell) program (Kraus and Nolze 1996). 2θ data between 32.3o and 37o are excluded from the XRD pattern to remove the Si substrate peaks from the analysis. Phase analysis of the refinement shows 14.7% of Sm2O3 and 85.3% of SmCo7. Scherer sizes are 3.5 nm and 5.4 nm for the SmCo7 and Sm2O3, respectively. The 4.2 nm size found from the size distribution (Fig. 2b inset) is the average of these two types of nanoparticles.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-011-0612-8/MediaObjects/11051_2011_612_Fig4_HTML.gif
Fig. 4

XRD data and Rietveld refinement of nanoparticles deposited at 1 Torr from the 1:5 target

The hysteresis loop at room temperature (Fig. 5a inset) shows no magnetic hardness with high-field magnetization slope characteristic of superparamagnetic particles. Zero field–cooled and field-cooled (150 Oe) magnetization versus temperature data (Fig. 5b) indicate that the 1:7 nanoparticles are superparamagnetic with a blocking temperature of 145 K. Anisotropy constant K can be calculated from the value of the blocking temperature TB by using the following equation, which is valid for superparamagnetic particles (Cullity 1972);
$$ K = \frac{25kT}{V} $$
(1)
where k, TB, K and V are the Boltzmann constant, temperature, anisotropy constant and the particle volume, respectively. From this value of TB and the particle volume, the value of anisotropy constant K is estimated to be around 2.2 × 10erg/cc (Eq. 1). This value is significantly lower than the bulk value for this structure, which is 5 × 10erg/cc. This behavior is expected, since from bulk to nanoparticles, the surface to volume ratio increases very rapidly that a large percentage of the nanoparticle atoms are on the surface. Larger surface area creates more defects and/or changes the electronic structure, which may cause a reduction in the anisotropy of the particles as compared to bulk (Giri et al. 1999). Recently, a lower K value of 3.5 × 106 erg/cc has been observed in Y-Co nanoparticles produced with cluster beam deposition compared to the bulk K value of 5.5 × 107 erg/cc (Balasubramanian et al. 2011b). Our result is much closer to the bulk value due to the high crystallinity of the nanoparticles. Hysteresis loops at 10 and 40 K were also measured with SQUID (Fig. 5a). The observed coercivities are 7.6 and 19 kOe for 40 and 10 K, respectively. When the same sample was measured after 10 months, no change has been observed in the magnetic properties. This result points out the fact that these nanoparticles are stable after they are formed and their oxidation occurs only during the formation of the particles (Fig. 6). Stable Sm-Co nanoparticles are attractive for use in high-density magnetic recording media (Frey and Sun 2010).
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-011-0612-8/MediaObjects/11051_2011_612_Fig5_HTML.gif
Fig. 5

a Hysteresis loops at 10 and 40 K, and in the inset at 300 K, b Magnetization versus temperature data of nanoparticles deposited at 1 Torr from the 1:5 target

https://static-content.springer.com/image/art%3A10.1007%2Fs11051-011-0612-8/MediaObjects/11051_2011_612_Fig6_HTML.gif
Fig. 6

First- and second-quadrant magnetization curves of the 1:7 nanoparticles (from 1:5 target) when they are freshly made and after 10 months

Particles from the Sm2Co17 Target

In the second part of this study, the nanoparticles were deposited from a 2:17 alloy target. Superparamagnetic 3.5 nm (after extracting Sm2O3 from Rietveld) (Fig. 7a) 1:7 nanoparticles have been produced with a blocking temperature of TB = 70 K at 1 Torr (Fig. 7b inset). The corresponding magnetocrystalline anisotropy constant K is estimated to be around 1 × 10ergs/cc (Eq. 1), and the observed coercivities are 6 and 11 kOe at 40 and 10 K, respectively (Fig. 7b). Rietveld analysis was used in both samples (from the 1:5 and 2:17 targets) to determine the lattice constants a and c. Nanoparticles from the 2:17 target have a and c values close to the 2:17 structure, while nanoparticles from the 1:5 target have a and c close to the 1:5 (Fig. 8). This is consistent with the observation that the coercivity and the anisotropy constant of the 1:7 nanoparticles made from the 1:5 target are higher than those produced from the 2:17 target.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-011-0612-8/MediaObjects/11051_2011_612_Fig7_HTML.gif
Fig. 7

a BF image and b hysteresis loops at 10 and 40 K of the nanoparticles made from the 2:17 target at 1 Torr; inset shows corresponding magnetization versus temperature graph

https://static-content.springer.com/image/art%3A10.1007%2Fs11051-011-0612-8/MediaObjects/11051_2011_612_Fig8_HTML.gif
Fig. 8

Lattice constants versus Sm concentration. Dottedlines represent the samples from the 2:17 and 1:5 targets

In-flight annealing of the particles

In order to increase the particle size, post-annealing has been applied at 400 °C for 30 min on the sample deposited at 1 Torr from the 1:5 target. Uncontrollable agglomeration has occurred, which resulted in a non-uniform microstructure (Fig. 9), and the whole sample was heavily oxidized (Fig. 9 inset). In another attempt, the pressure in the chamber was increased to 1.5 and 2 Torr to decrease the mean free path of the atoms that will lead to more collisions and larger particle sizes (Wegner et al. 2006). An average particle size of 6 and 7 nm has been observed for 1.5 and 2 Torr, respectively. Nevertheless, the increase in the size is far below the desired values.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-011-0612-8/MediaObjects/11051_2011_612_Fig9_HTML.gif
Fig. 9

BF image of the post-annealed (400 °C for 30 min.) sample; inset shows the corresponding SAED of the sample

Since none of the known methods for increasing the particle size worked, a novel method was introduced, which involved heating the nanoparticles during their flight to the substrate. Heater temperature of 750 °C and 2 Torr chamber pressure lead to a mixture of small particles with an average size of ~7 nm which are not fully crystalline and big nanoparticles of SmCo5 with an average size of 15 nm (Fig. 10a). This could be due to the increased pressure in the aggregation chamber from 1 to 2 Torr which in turn increases the flow rate. As a result of increased flow rate, some particles will not have enough time to crystallize or agglomerate and fly unchanged to the substrate while others agglomerate because of the introduced heat and create much bigger and crystalline particles. The larger nanoparticles are single crystalline with the SmCo5 structure and are covered with 5-nm amorphous Sm-O layer (forms during the in-flight annealing) (Fig. 10b and inset). At temperatures of 900 °C and above, it is found that the nanoparticles are completely consumed by the Sm-O.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-011-0612-8/MediaObjects/11051_2011_612_Fig10_HTML.jpg
Fig. 10

a BF and b HRTEM image of the nanoparticles deposited at a 750 °C heater temperature and with a 2 Torr aggregation chamber pressure; inset shows the SAED of the particle in HRTEM image

Hysteresis loops at 300 and 50 K are shown Fig. 11a. Sample coercivity is 0.7 and 3 kOe for 300 and 50 K, respectively. Much higher switching fields are also observed in the sample consisting of bigger particles (third quadrant of the loops); however, the volume ratio of the small nanoparticles is much higher than that of the big ones, which drops the coercivity considerably. ZFC curve of the M versus T graph has two slopes indicating a bimodal size distribution (Fig. 11b). The apparent high switching field of more than 10 kOe at RT (observed on the demagnetization curve) is related to the larger particles and makes them attractive for high-density recording media provided they can be separated from the smaller ones.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-011-0612-8/MediaObjects/11051_2011_612_Fig11_HTML.gif
Fig. 11

a Hysteresis loops at 300 and 50 K, b magnetization versus temperature graph of the nanoparticles deposited at 2 Torr aggregation chamber pressure and with 750 °C heater temperature

Conclusions

Sm-Co nanoparticles with different sizes have been successfully produced with the cluster beam deposition technique. Particle size has been controlled by introducing in-flight heat to the particles. During RT synthesis, 3.5-nm superparamagnetic 1:7 nanoparticles have been successfully produced with different lattice constants a and c from 1:5 and 2:17 targets. With the heater addition to the system (at 750 °C), the particle size distribution becomes bimodal with the small particles having an average size of 7 nm and the big nanoparticles having an average size of 15 nm covered with 5 nm Sm-O shell. The sample has 0.7 and 3 kOe at 300 and 50 K, respectively. Our current efforts are focused on increasing the amount of the big nanoparticles in the sample. Cluster beam deposition could be the key technique to produce high-coercivity RE-TM nanoparticles without post-annealing.

Acknowledgment

The authors would like to thank A. M. Gabay for helpful discussions. The authors also thank Dr. Melania Marinescu and EEC for providing the targets. Work supported by DOE DE-FG02-04ER4612.

Copyright information

© Springer Science+Business Media B.V. 2011