ARTICLE IN PRESS
Journal of Magnetism and Magnetic Materials ] (]]]]) ]]]–]]]
Chemical ordering of FePt nanoparticle self-assemblies by rapid thermal annealing H. Zeng*, Shouheng Sun, R.L. Sandstrom, C.B. Murray Thomas J. Watson Research Center, IBM Research Division, Route 134, P.O. Box 218, Yorktown Heights, NY 10598, USA Received 27 September 2002; received in revised form 27 February 2003
Abstract Self-assembled 4 nm FePt nanoparticle arrays were treated with rapid thermal annealing (RTA) process. The phase transformation from the chemically disordered face-centered-cubic structure of the as-synthesized FePt nanoparticles to the chemically ordered face-centered-tetragonal structure was realized by RTA, with both the annealing temperature and time being greatly reduced, as compared to conventional annealing. The onset of chemical ordering occurred at the annealing condition of around 400 C for only 5 s. Temperature-dependent coercivity measurements revealed strong thermal effects of the nanoparticle assemblies. Curie temperatures (Tc ) of the annealed assemblies were derived from the temperature-dependent hysteresis measurements. Tc of the annealed assemblies increases with increasing annealing temperature, but is about 200–300 K lower than that of bulk FePt (750 K). The drastic reduction of Tc may be attributed to size effect and partial chemical ordering of FePt nanoparticles. r 2003 Elsevier B.V. All rights reserved. Keywords: FePt; Nanoparticles; Rapid thermal annealing
1. Introduction The requirements for ultra-high density magnetic recording have driven the development of new magnetic thin film media with smaller grains, higher coercivity, and minimal exchange coupling between neighboring grains [1]. FePt- and CoPtbased hard magnetic nanoparticle arrays have been viewed as a promising candidate for future recording media applications [2]. Recent solution phase-based synthesis offers a convenient approach to monodisperse FePt nanoparticles with the size being controlled down to only B4 nm [3]. *Corresponding author. Tel.: +1-914-945-1609; fax: +1914-945-4342. E-mail address:
[email protected] (H. Zeng).
However, making a hard magnetic nanoparticle assembly with controlled thickness, mechanical robustness and the desired magnetic properties has been a great challenge. Particular problems associated with the formation of a hard magnetic FePt nanoparticle assembly are the high temperature and long annealing time required for the chemical ordering of FePt, which may lead to nanoparticle aggregation and inter-particle exchange coupling. Severe aggregation will even destroy the selforganized assembly, and lead to domain wall formation [4]. Recent effort has been devoted to lowering the ordering temperature in FePt thin films and nanoparticle arrays by doping of elements such as Cu and Ag [5,6]. Although the doping results in significant reduction of the ordering temperature,
0304-8853/03/$ - see front matter r 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0304-8853(03)00482-7
ARTICLE IN PRESS 2
H. Zeng et al. / Journal of Magnetism and Magnetic Materials ] (]]]]) ]]]–]]]
possible incorporation of impurities in FePt lattices may lead to additional disorder, reducing the achievable anisotropy and thermal stability. Here we report the chemical ordering of FePt with rapid thermal annealing (RTA). A reduction of the ordering temperature in pure FePt comparable to doped assemblies is observed, with the ordering taking place within seconds. Curie temperatures (Tc ) for assemblies annealed at different temperatures are obtained from temperature-dependent hysteresis loop measurements. A dramatic decrease in Tc is observed, as compared to bulk alloy, which may result from size and surface effects and chemical disorder. This may greatly affect room-temperature magnetization, anisotropy and thermal stability.
2. Synthesis and assembly of FePt nanoparticles Monodisperse FePt nanoparticles were synthesized by high temperature (297 C) decomposition of Fe(CO)5 and reduction of Pt(acac)2 in the presence of oleic acid and oleyl amine [3]. The Fe and Pt composition is tuned by varying the molar ratio of Fe(CO)5 and Pt(acac)2. The syntheses show that 0.5 mmol of Fe(CO)5 and 0.5 mmol of Pt(acac)2 yield Fe38Pt62, while 1.1 mmol of Fe(CO)5 and 0.5 mmol of Pt(acac)2 lead to Fe56Pt44 nanoparticle materials [7]. Alternatively, the FePt nanoparticles were made by solution phase reduction of FeCl2 and Pt(acac)2 [8]. The advantage of this reduction approach is that the final FePt composition can be readily controlled by the actual molar ratio of FeCl2 and Pt(acac)2. For example, a 0.6/0.4 ratio of FeCl2/Pt(acac)2 gives Fe60Pt40 nanoparticles, while a 0.5/0.5 ratio yields Fe50Pt50 nanoparticles. Under suitable conditions, the monodisperse nanoparticles suspended in solution tend to selforganize into ordered arrays after solvent evaporation. This self-organization is influenced by the nature of the interactions exhibited among the particles, and between the particles and the substrate. The synthetic strategy to nanoparticle superlattices relies on a large number of weak and non-directional interactions, such as ionic bonds, hydrogen bonds and van der Waals interactions to
Fig. 1. SEM image of a highly ordered 4 nm FePt nanoparticle self-assembly.
organize the particles’ self-assembly. The uniformity in the size of the FePt nanoparticles is very critical in forming 2-D and 3-D superlattices. Fig. 1 illustrates the SEM image of a highly ordered 4 nm FePt assembly on Si substrate, which shows cubic assembly structure.
3. RTA and phase transformation In this study, an AG Associates HeatPulse 410 Rapid Thermal Processing unit was used to anneal the FePt nanoparticle assemblies on glass substrates, with the annealing temperatures (Ta ) ranging from 300 C to 600 C, and the annealing time (referring to time elapsed at the set temperature) from 1 to 30 s. The heating rate is 200 C/s and the cooling profile can be roughly described by Newton’s Law of Cooling, with the initial cooling rate greater than 50 C/s for the first second. At the annealing temperature of 600 C, the cooling time from 600 C to 400 C is B7 s; at 500 C, the cooling time from 500 C to 400 C is B5 s. RTA was performed under clean nitrogen atmosphere, with a flow rate of 40 cm3/s. All samples were predried in vacuum to evaporate the solvent. As-synthesized FePt nanoparticles possess disordered face-centered-cubic (FCC) structure, as can be seen from the X-ray diffraction (XRD) pattern shown in Fig. 2(a), and have very low magnetocrystalline anisotropy, showing
ARTICLE IN PRESS H. Zeng et al. / Journal of Magnetism and Magnetic Materials ] (]]]]) ]]]–]]] (111) (110)
(200)
580 C (002) (201) (112)
(d)
550 C
(c)
500 C
(b) 460 C
(a)
20
as-syn
30
40
50
60
70
2θ ( Co Ka radiation) pattern of 4 nm Fig. 2. XRD (l ¼ 1:79 A, FePt nanoparticle self-assembly: (a) as-synthesized and annealed by RTA for 5 s at annealing temperatures of (b) 460 C, (c) 500 C, (d) 550 C and (e) 580 C.
superparamagnetic behavior at room temperature. Upon annealing, FePt undergoes a phase transformation from disordered FCC to chemically ordered face-centered-tetragonal (FCT), which yields high magnetocrystalline anisotropy. Typically, annealing temperatures above 500 C are required for the ordering of FePt nanoparticles. Figs. 2(b)–(e) show the XRD patterns of samples after RTA at from 460 C to 550 C for 5 s. It can be seen that the superlattice (0 0 1), (1 1 0) and (2 0 1) peaks of the FCT phase emerges at Ta X460 C, indicating the formation of partially ordered FCT structure. As the annealing temperature further increases, the intensity of these peaks becomes stronger, and the original (2 0 0) FCC peak become split into (2 0 0) and (0 0 2) tetragonal peaks, suggesting the improvement in the degree of ordering. Magnetic measurements show that annealed FePt nanoparticle assemblies are ferromagnetic. Figs. 3(a) and (b) compare Hc as a function of Ta for RTA samples and samples annealed in a box oven for 30 min. With the annealing time fixed to be 5 s, the room-temperature coercivity (Hc ) increases with increasing annealing temperatures, from 300 Oe at Ta ¼ 360 C to 17,000 Oe at Ta ¼ 580 C, as seen in Fig. 3(a), which is consistent with the improvement of chemical ordering. Hc ðTa Þ curve shows two distinct regions:
Hc (kOe)
Intensity (arb. unit)
(001)
(e)
26 24 22 20 18 16 14 12 10 8 6 4 2 0
3
a (300 K ) b (300 K) c (10 K)
350
400
450
500
550
600
Ta (C)
Fig. 3. Hc as a function of the annealing temperature Ta for (a) RTA for 5 s, measured at 300 K; (b) box oven annealing for 30 min, measured at 300 K; and (c) RTA for 5 s, measured at 10 K.
for Ta o500 C, Hc increases with Ta slowly at a rate of about 30 Oe/ C; for 500oTa o550 C, however, this rate jumps to 250 Oe/ C. Further increases in Ta only increases Hc slightly, suggesting that a high degree of ordering is already achieved at Ta ¼ 550 C. We also noticed that RTA samples show significantly higher Hc than box furnace annealed specimens at roughly the same annealing temperatures. Detailed studies on such differences are still underway. When FePt is annealed at high temperatures, ordering and nanoparticle aggregation can occur at the same time. Room-temperature coercivity alone may not be a suitable indication of the degree of ordering. This is because for FePt nanoparticles as small as 4 nm, thermal fluctuations to the energy barrier Ku V ; where Ku is the anisotropy constant and V the particle volume, would lead to a decrease in room-temperature coercivity. The smaller the individual volume of magnetization reversal, the stronger this thermal effect. For well-isolated FePt assemblies, it is reasonable to assume that the volume of magnetization reversal is equal to the particle volume, as the reversal is likely cooperative within the whole particle. When aggregation occurs, the effective volume of magnetization reversal is increased. This may lead to an increased energy barrier Ku V ; and thus enhanced thermal stability, yielding relatively
ARTICLE IN PRESS H. Zeng et al. / Journal of Magnetism and Magnetic Materials ] (]]]]) ]]]–]]]
4
high room-temperature coercivity. To study the effects of ordering and aggregation, the temperature dependence of coercivity was measured. Fig. 3(c) shows Hc measured at 10 K for RTA samples annealed at different temperatures. At such a low measuring temperature, the contribution from thermal effects is negligible, and the coercivity can serve as an indication of the degree of ordering. It can be seen that at Ta ¼ 400 C, Hc at 10 K reaches 6 kOe, suggesting the onset of chemical ordering, although the XRD results indicate that the superlattice (0 0 1) peak only becomes visible at Ta ¼ 460 C. TEM observations confirm that the nanoparticles are well isolated. As Ta increases, Hc increases, suggesting improved ordering. For Ta ¼ 580 C, Hc at 10 K is 24,000 Oe. This corresponds to Ku B2:5 107 erg/ cm3, which is consistent with values of highly ordered FePt reported by others [9]. Fig. 4 shows the measured temperature-dependent coercivity, normalized by Hc at 10 K, for RTA samples with Ta ¼ 360 C, 460 C, 520 C and 580 C. For particle ensembles with a single energy barrier, Hc ðTÞ behavior can be described by Sharrock’s formula [10], provided that the reversal mechanism is coherent rotation. Nevertheless, in reality, the Hc ðTÞ behavior is rather complicated. It can be seen that Hc of all samples decreases dramatically with increasing measuring temperatures, due to thermal fluctuations. However, Hc ðTÞ
1.0
Hc /Hc(10 K)
0.8 0.6 0.4 580 C 520 C 460 C 360 C
0.2 0.0 0
50
100 150 200 250 300 T (K)
Fig. 4. Normalized coercivity Hc =Hc (10 K) as a function of measuring temperature for Ta =(a) 360 C, (b) 460 C, (c) 520 C and (d) 580 C. The dashed lines are guides to the eye.
behaviors of samples annealed at low and high temperatures show different trends: for low Ta samples (p460 C), the rate of Hc reduction is initially quite large, while decreases gradually as the measuring temperature approaches 300 K; for high Ta samples, this trend tends to be reversed. According to Sharrock’s formula, the temperature dependence of Hc may originate from two factors: (1) the temperature dependence of intrinsic properties such as Ms and Ku and (2) the thermal fluctuation term kB T=ðKu V Þ: One of the possible reasons for the difference in Hc ðTÞ behaviors we observed may be that the temperature dependence of intrinsic properties is competing with the thermal fluctuation term. This temperature dependence of Ms and Ku may originate from various factors, including the size, surface effects and disorder of nanoparticles. Different Hc ðTÞ trends seem to suggest that for the sample annealed at low Ta ; the thermal fluctuation term kB T=ðKu V Þ is dominating; while for the sample annealed at high Ta ; the dominating effect is the temperature dependence of intrinsic properties. The reason lies in that for high Ta samples, strong inter-particle exchange coupling and/or possible aggregation makes the thermal fluctuation term less important. For isolated, randomly oriented particles with uniaxial anisotropy, the remanence ratio S ¼ Mr =Ms is 0.5 [11]. S greater than 0.5 suggests that there may exist inter-particle exchange coupling. Therefore, S can be used to study inter-particle exchange interactions. Again, due to the thermal effects, room-temperature S cannot be used for such purposes. Instead, S at 10 K is measured for these samples and is plotted as a function of Ta in Fig. 5. It can be seen that S decreases gradually and monotonically with decreasing Ta : However, for samples annealed at Ta p400 C, S is still greater than 0.5, even though our TEM analysis suggests that the particles are well isolated. This may be attributed to the fact that our low Ta samples contain a mixture of both a uniaxial and cubic anisotropy, which is consistent with partial ordering observed from XRD results. It is known that for isolated, randomly oriented particles with cubic anisotropy, S is 0.83 for K1 > 0 and 0.87 for K1 o0 [12].
ARTICLE IN PRESS H. Zeng et al. / Journal of Magnetism and Magnetic Materials ] (]]]]) ]]]–]]]
0.74
1.0
0.72
0.8 Ms/M0
S = Mr /Ms
0.70 0.68
J=1/2
0.6 0.4
Ta = 360 C, Tc = 410 K
0.66
Ta = 460 C, Tc = 450 K
0.2
0.64
5
Ta = 520 C, Tc = 580 K
0.0
0.62
0.0
0.60 350
400
450
500
550
600
The intrinsic magnetic properties of Ms ; Ku and the curie temperature Tc for the bulk FePt are affected by size and surface effects, as well as chemical disorder of the FePt nanoparticles. Ms as a function of temperature was obtained from temperature-dependent hysteresis loop measurements for several FePt nanoparticle samples with different Ta : Fig. 6 shows Ms ðTÞ=Ms ð0Þ vs. T=Tc curves for Ta ¼ 360 C, 460 C and 520 C; the theoretical curve calculated from Weiss’s theory using the Brillouin function with quantum number J ¼ 1=2 is also plotted. Tc for these samples can be determined by fitting the experimental data with the theoretical curve. As can be seen, Tc decreases from 580 to 410 K as Ta decreases from 520 C to 360 C. Compared to Tc of 750 K for bulk FePt, the reduction of Tc in FePt nanoparticle assemblies for Ta ¼ 520 C is 170 K and for Ta ¼ 360 C, 340 K. Okamoto et al. [13] reported a weak dependence of Tc on ordering parameters in epitaxial FePt thin films, from 750 K for the ordering parameter of 0.79–700 K for the ordering parameter of 0.5. Partial ordering in nanoparticle assemblies due to insufficient heat treatment therefore may account for part of the Tc reduction. Another significant factor that contribute to the reduction of Tc is the particle size and surface effect. For 4 nm FePt nanoparticles, nearly 30% of the atoms are at the surface if the thickness of the ( The surface surface layer is assumed to be 2 A. atoms have reduced exchange coupling, which may contribute to the size dependent Tc change. The size and surface effects on Ms of nanoparticles
0.4
0.6
0.8
1.0
T/Tc
Ta
Fig. 5. Remanence ratio S ¼ Mr =Ms measured at 10 K as a function of Ta :
0.2
Fig. 6. Ms ðTÞ=Ms ð0Þ vs. T=Tc for Ta =60 C, 460 C and 520 C.
lie in two-folds: in addition to the lose of moment for the surface atoms, which leads to lower Ms for smaller particles, the reduction of Tc also decreases room-temperature Ms : For example, for bulk FePt with Tc at about 750 K, room-temperature Ms is only 1.5% less than Ms at 0 K; however, for the sample annealed at Ta ¼ 360 C with Tc B410 K, the reduction in Ms amounts to B20%. Finally, as the magnetocrystalline anisotropy Ku decreases with increasing temperature, and vanishes at around Tc ; a reduction in Tc of about 200–300 K for nanoparticle assemblies means that Ku decreases with increasing temperature faster than that of bulk.
4. Conclusions Chemical ordering for 4 nm FePt nanoparticle self-assembly can be improved by rapid thermal annealing, with significantly lowered annealing temperature and shortened time. Partially ordered assemblies show a remanence ratio greater than 0.5, which is attributed to a mixture of a cubic and uniaxial anisotropy. Significant reduction in the curie temperature is due largely to the particle size and surface effects, as well as partial ordering. To use these FePt nanoparticle assemblies for future magnetic recording applications, further optimizing of annealing conditions leading to fully ordered, well-isolated nanoparticles is required. It may be advantageous to utilize particles larger than 4 nm to alleviate problems associated with nanoparticle size and surface effects.
ARTICLE IN PRESS 6
H. Zeng et al. / Journal of Magnetism and Magnetic Materials ] (]]]]) ]]]–]]]
Acknowledgements This work was supported in part by US DoD/ DARPA under grant DAAD 19-01-1-0546.
References [1] D. Weller, A. Moser, IEEE Trans. Magn. 35 (1999) 4423. [2] D. Weller, et al., IEEE Trans. Magn. 36 (2000) 10. [3] S. Sun, C.B. Murray, D. Weller, L. Folks, A. Moser, Science 287 (2000) 1989. [4] H. Zeng, S. Sun, T.S. Vedantam, P. Liu, Z. Dai, Z.L. Wang, Appl. Phys. Lett. 80 (2002) 2583. [5] T. Maeda, T. Kai, A. Kikitsu, T. Nagase, J. Akiyama, Appl. Phys. Lett. 80 (2002) 2147.
[6] J.W. Harrel, D.E. Nikles, S.S. Kang, S. Wang, Fourth International Conference on Fine Particle Magnetism, Pittsburg, 2002 (Abstract BA4I). [7] S. Sun, E.E. Fullerton, D. Weller, C.B. Murray, IEEE Trans. Magn. 37 (2001) 1239. [8] S. Sun, et al., J. Phys. Chem. B 107 (2003) 5419. [9] A. Cebollada, R.F.C. Farrow, M.F. Toney, Structure and magnetic properties of chemically ordered magnetic binary alloys in thin film form, in: H.S. Nalwa (Ed.), Magnetic Nanostructures, American Scientific Publishers, Los Angeles, 2002. [10] M.P. Sharrock, J. Appl. Phys. 76 (1994) 6413. [11] B.D. Cullity, Introduction to Magnetic Materials, Addison-Wesley, Reading, MA, 1972. [12] J. Garcia-Otero, M. Porto, J. Rivas, A. Bunde, J. Appl. Phys. 85 (1999) 2287. [13] S. Okamoto, N. Kikuchi, O. Kitakami, T. Miyazaki, Y. Shimada, Phys. Rev. B 66 (2002) 024413.