Enhancement of perpendicular coercivity in L11 CoPt ...

JOURNAL OF APPLIED PHYSICS 108, 113909 共2010兲

Enhancement of perpendicular coercivity in L11 CoPt thin films by replacement of Co with Cu Fu-Te Yuan,1 An-Cheng Sun,2 Jen-Hwa Hsu,1,a兲 C. S. Tan,3 P. C. Kuo,3 W. M. Liao,4 and H. Y. Lee4 1

Department of Physics and Center for Nanostorage Research, National Taiwan University, Taipei 106, Taiwan 2 Department of Chemical Engineering and Materials Science, Yuan Ze University, Taoyuan 324, Taiwan 3 Department of Materials Science and Engineering, National Taiwan University, Taipei, 106, Taiwan 4 National Synchrotron Radiation Research Center (NSRRC), Hsinchu 300, Taiwan

共Received 5 October 2010; accepted 25 October 2010; published online 7 December 2010兲 Magnetic properties and microstructures of L11 共Co50−xCux兲Pt50 films sputter-deposited at 350 ° C on MgO共111兲 substrates are reported. The addition of Cu significantly improves the alignment of c-axis and chemical ordering. Perpendicular coercivity 共Hc⬜兲 also increases markedly from 0.1 to 1.9 kOe while in-plane coercivity declines from 0.5 to 0.07 kOe for the 20 nm thick films at x = 26. Similar phenomena are observed with larger effects for the 50 nm thick films. The coercive mechanism is attributed to domain-wall pinning produced by the compositional segregation of nanoscaled nonmagnetic Cu-rich and magnetic Co-rich regions within a coherent L11 crystal domain. Therefore, an intermediate value of Hc⬜ can be obtained from this hardening mechanism when further microstructure modifications are enforced, which largely increases the potential for the use in spintronic devices or patterned media. © 2010 American Institute of Physics. 关doi:10.1063/1.3520662兴 I. INTRODUCTION

Rhombohedral CoPt L11 superlattice structure, a metastable phase identified recently, consists of alternatively stacked closed-packed atomic planes of Co and Pt in a sequence of ABCABC with a magnetic easy-axis along 关111兴.1–4 This structure can be formed on MgO共111兲 substrate at temperatures ranging from 200 to 300 ° C by either molecular beam epitaxy or sputtering.1–4 The phase exhibits a very high perpendicular magnetocrystalline anisotropy Ku of 2 – 4 ⫻ 107 erg/ cm3, saturation magnetization M s of about 740 emu/ cm3, anisotropy field Hk of higher than 3 T.2–4 The features like low formation temperature, high Ku, and moderate Hk reveal its agility for advanced magnetic recording media with areal recording density exceeding 1 terabits/ in.2 while possessing high thermal stability and small writing field. However, several properties are required to be further improved. Among them one is the poor alignment of easyaxis; normally found in the currently obtained CoPt films.2–4 Additionally, the small coercivity 共Hc兲 originated from the domain wall motion during magnetic reversal3,4 also limits the applicability of these films for recording media. Therefore, in order to overcome these problems, we propose partial replacement of Co with Cu in Co50Pt50 films to improve the structural and magnetic properties of the L11 film. The use of Cu is based on the following facts: 共1兲 L11-CuPt is thermodynamically stable at room temperature and has a lattice parameter similar to that of L11-CoPt,5 which may preserve the L11 structure of CoPt and even improve the alignment of c-axis; 共2兲 Cu is in-soluble to Co, which is essential a兲

Electronic mail:[email protected].

0021-8979/2010/108共11兲/113909/4/$30.00

for compositional segregation. Our results reveal a significant improvement in the alignment of c-axis of the L11 phase along with a large increase in perpendicular coercivity 共Hc⬜兲, and a substantial reduction in in-plane magnetic 共Hc储兲 component. Moreover, in the present study, microstructure, magnetic domain structure, and magnetic hardening mechanism are studied in detail. II. EXPERIMENTAL

Thin-film samples were fabricated by dc magnetron sputtering. The background pressure was lower than 5 ⫻ 10−9 torr, and the working pressure was fixed at 10 mtorr. MgO共111兲 substrates were heated to 650 ° C for 30 min for surface reconstruction and cleaning. 共Co50−xCux兲Pt50 films with a thickness of 20 and 50 nm were, then, deposited by rotational cosputtering from Co, Cu, and Pt targets after the substrate temperature was stabilized at 350 ° C for 10 min. High-purity targets of Co, Cu, and Pt of 99.999 at. % were used. The chemical composition of 共Co50−xCux兲Pt50 films was controlled by adjusting the sputtering powers of these targets, and was confirmed by calibrated energy dispersion spectroscopy 共EDS兲. X-ray diffractometry 共XRD兲 was used to characterize the structure of the films. The step size of the ␪-2␪ scans was 0.01° and the collection time was 1 s per step. Rocking curves were also measured for the L11共111兲 peaks in the range from ⫺7° to 7° at a scan rate of 2 ° / min. The microstructure was studied by transmission electron microscopy 共TEM兲. Magnetic domain structure was investigated by magnetic force microscopy 共MFM兲. The magnetic properties were measured by using a vibrating sample magnetometer at room temperature.

108, 113909-1

© 2010 American Institute of Physics

Downloaded 07 Dec 2010 to 140.112.101.70. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

113909-2

Yuan et al.

J. Appl. Phys. 108, 113909 共2010兲

FIG. 2. 共Color online兲 Hysteresis loops for the 20 nm thick 共Co50−xCux兲Pt50 films with 共a兲 x = 0 and 共b兲 x = 23, and 50 nm thick films with 共c兲 x = 0 and 共d兲 x = 23.

FIG. 1. 共Color online兲共a兲 XRD diffraction patterns for the 20 nm thick 共Co50−xCux兲Pt50 films with Cu content x ranging from 0 to 32, 共b兲 Dependence of FWHM of rocking curves on x.

III. RESULTS AND DISCUSSION

Figure 1共a兲 shows typical XRD patterns for 20 nm thick 共Co50−xCux兲Pt50 films. The diffraction patterns essentially consist of L11 fundamental 共222兲 and superlattice 共111兲, 共333兲 peaks. The L11共111兲 peak for CoPt binary sample are very broad and weak and its intensity increases significantly when x is increased to 32, revealing improved ordering with an increasing value of x. The order parameter 共Sord兲 elucidates the degree of ordering, as determined from the ratio between integrated intensity of L11共111兲 and L11共222兲 peaks.5 The scattering factors f Co, f Pt, and f Cu used in the calculation of Sord for Co, Pt, and Cu, respectively, are 25.5, 72.5, and 38.4 for 共111兲 peak and 22.3, 64.8, and 34.2 for 共222兲 peak.6 Although the addition of Cu largely enhances the intensity of the superlattice peaks, all the obtained values of Sord are still not exceeding 0.6. The alignment of crystal axes was studied using the rocking curves of L11共111兲 peaks. Figure 1共b兲 plots the full width at half maximum 共FWHM兲 as a function of Cu content. The value of FWHM decreases from 3.2° to a minimum of 2.1° as the value of x is increased from 0 to 23, suggesting improved alignment of the c-axis in the rhombohedral lattice. Experimental results show that substituting Cu for Co can effectively enhance Sord and improve the alignment of 关111兴; that is the magnetic easy-axis. Hysteresis loops for the 20 nm thick and 50 nm thick films with x = 0 and 23 are shown in Figs. 2共a兲–2共d兲. Positive nucleation field together with non-negligible in-plane magnetic component are observed in the films without Cu 关Figs. 2共a兲 and 2共c兲兴. Spontaneous relaxation of magnetization is a typical feature for uniaxial anisotropic films with continuous morphology without magnetic defects.7,8 Significant en-

hancement in Hc⬜ together with a drastic reduction in Hc储 is obtained in the samples with x = 23. Reduced in-plane magnetic component can be explained in terms of improved alignment of c-axis of the L11 phase as shown in Fig. 1共b兲. On the other hand, a large increase in Hc⬜ to about 2 kOe may be related to the reversal mechanism that is dominated by domain wall pinning. Magnetic properties including Hc储, Hc⬜, and saturation magnetization 共M s兲 for Co–Cu–Pt films with thicknesses of 20 and 50 nm are plotted as functions of Cu content in Figs. 3共a兲 and 3共b兲, respectively. For the 20 nm thick films, Hc⬜ increases substantially from around 0.1 to a maximum of 1.9

FIG. 3. 共Color online兲 Dependence of 共a兲 in-plane 共H储兲 as well as out-ofplane 共H⬜兲 coercivities, and 共b兲 saturation magnetization 共M s兲 on Cu content for the 20 nm thick and 50 nm thick films.


113909-3

Yuan et al.

kOe, while Hc储 drops from 0.5 to less than 0.07 kOe at x = 26. The change is even more striking in films with 50 nm thickness, in which Hc⬜ increases from 0.2 to 2.2 kOe and Hc储 decreases from 1.5 to 0.05 kOe at x = 23. Additionally, M s declines almost linearly from 740 to 300 emu/ cm3 when x increases from 0 to 37 in both 20 and 50 nm thick films as indicated in Fig. 3共b兲. This result reveals a simple magnetic dilution effect with the replacement of Cu in this composition range. Although the reduced M s leads to decrease in Ku, it is worthy to note that the Cu-substitution does not obviously degrade the room-temperature anisotropy field 共HkRT兲, i.e., the field at the joint of in-plane and out-of-plane magnetization curves. A similar HkRT of about 2 T is retained in the films when x is varied from 0 to 26. Detailed TEM analysis were undertaken to study the microstructure and clarify the mechanism of Hc⬜ enhancement and the role of Cu in reducing the c-axis distribution. Figures 4共a兲–4共c兲 display a plane-view image, selected area electron diffraction 共SAD兲 pattern, and cross-sectional image for the 20 nm thick film with x = 23. The plane-view image 关Fig. 4共a兲兴 indicates the continuous morphology of the film. No clear grain boundaries or second phase 共precipitates兲 are found. The coherent lattice structure reveals a perfect texture. The SAD pattern in Fig. 4共b兲 with zone axis of 关001兴 points out that the satellite spots of L11 surround the MgO兵200其 spots, confirming epitaxial growth. The elongated and blurred satellite spots also suggest the existence of defects and internal strain. The defects are clearly identified in the cross-sectional image in Fig. 4共c兲, which are mainly stacking faults parallel to the film plane. Comparing to the Co–Cu–Pt film, planar defects are much denser and complex in CoPt binary sample as shown in Fig. 4共d兲. Twins, stacking faults, and high-angle boundaries confine the coherent length to about 2 nm. The “size” effect causes the broadened L11共111兲 diffraction peak with Lorentzian shape,9,10 and reduces the intensity of the peak, resulting in underestimation of Sord. The addition of Cu drastically decreases the planar defects, leading to markedly narrowed c-axis distribution. Furthermore, the local analysis of composition by EDS reveals that the Co–Cu–Pt film is uniform and the composition is close to Co27Cu23Pt50. The results of TEM study confirm that the film has a single L11 phase and the domain wall pinning is not originated from the second phase. To further investigate the composition distribution, the element mapping of Co and Cu was carried out in the selected area as marked in Fig. 4共c兲, which was determined from the scanning TEM 共STEM兲. The resolution was fixed at 128⫻ 128 pixel and the duration of mapping was 20 s, an optimum compromise between the resolution and image stability. The quantitative distribution obtained from the Co and Cu mapping results is shown in Fig. 4共e兲. The Co regions are represented by red color while the Cu regions by green color. The subnanometer-scaled compositional precipitation is detected. Our STEM results elucidate the coercive mechanism that is discussed below. Spinodal decomposition occurs in the epitaxially grown Co– Cu–Pt films, resulting in randomly distributed nanometerscaled Cu-rich and Co-rich L11 regions. Partial isolation of magnetic Co-rich phase by the nonmagnetic Cu-rich regions provides extra impedance for domain wall motion, leading to

J. Appl. Phys. 108, 113909 共2010兲

FIG. 4. 共Color online兲共a兲 Plane-view TEM images, 共b兲 SAD patterns, 共c兲 cross-section microstructure for the Co27Cu23Pt50 and 共d兲 Co50Pt50 film with 20 nm in thickness, and 共e兲 two-dimensional quantitative distribution of Co and Cu obtained from element mapping in the square area marked in 共c兲. The red color denotes Co region 共Co K␣1兲 and green color denotes Cu regions 共Cu K␣1兲.

increased Hc⬜ but unchanged HkRT. Additionally, due to the Cu–Pt and Co–Pt regions have identical lattice structures and similar lattice parameters, the mixing of them improves the alignment of c-axis, which effectively reduces in-plane magnetic component. However, the size of the Co-rich and Curich regions below 1 nm as shown in Fig. 4共e兲 is believed to be underestimated due to the competing effects of image drift as well as overlapping of regions with different compositions. The image drift will be fast at the locations of the sample with very thin thickness but the drift diminishes at the thick parts. On the other hand, at the film locations with larger thickness the overlap of composition segregation regions along the normal direction also leads to underestimate of the size. Therefore, a compromise between the resolution and image stability during the observation has been undertaken as indicated previously.


113909-4

J. Appl. Phys. 108, 113909 共2010兲

Yuan et al.

IV. CONCLUSION

FIG. 5. 共Color online兲 Magnetic domain patterns for the Co50Pt50 films with 共a兲 20 nm and 共b兲 50 nm thick; magnetic domain patterns for the Co27Cu23Pt50 films with 共c兲 20 nm and 共d兲 50 nm thick, respectively.

Figures 5共a兲–5共d兲 present the MFM images for the 20 and 50 nm thick samples with x = 0 and 23. Large mazelike domain pattern with average size ranging between 200 and 250 nm is observed in the films without Cu 关Figs. 5共a兲 and 5共b兲兴. This domain structure is typically observed in the epitaxially grown perpendicular anisotropic films such as L10 FePt and CoPt,7,8,11,12 indicative of the absence of domain wall pinning. Driven by lowering the total magnetostatic energy of the film, the mazelike patterns form spontaneously before the external field is reduced to zero, which largely reduces Hc⬜. However, the Cu-substitution significantly alters the domain morphology from maze-like to particulatelike and reduces the domain size to approximately 100 nm as shown in Figs. 5共c兲 and 5共d兲. The changed domain pattern and increased Hc⬜ suggest strengthened domain-wall pinning effect, which confirms the domain wall pinning mechanism induced by the compositional segregation. The effect of addition of Cu in L10 FePt and CoPt films has been reported previously by several groups. It was found that Cu could facilitate the order-disorder transformation by forming Fe–Cu–Pt or Co–Cu–Pt ternary alloy, resulting in increased coercivity and reduced ordering temperature.13–16 Additionally, Cu addition could improve the 共001兲 texture of L10 FePt films annealed by rapid thermal process.17,18 Results of calorimetric studies have confirmed that the alloying of Cu decreases the activating energy of order transformation.19 Nevertheless, Cu plays a totally different role in L11 Co–Cu–Pt phase. The addition of Cu does not significantly enhance the chemical ordering 共Sord ⬍ 0.5兲 and have no effect on ordering temperature. Instead, it improves the quality of the epitaxial growth by suppressing the formation of various planar defects and increases Hc⬜ arising from nanoscaled compositional segregation.

In conclusion, this study investigates how Cusubstitution affects 共Co50−xCux兲Pt50 films with 20 and 50 nm in thicknesses. Experimental results indicate that adding Cu facilitates L11 ordering, improves the alignment of 关111兴 of L11 phase, and significantly enhances perpendicular coercivity. The optimum compositions for the 350 ° C-deposited 20 nm and 50 nm thick films are x = 26 and 23, respectively. In those films, Hc⬜ is increased from 0.1 to 1.9 and 2.2 kOe. Meanwhile, Hc储 decreases from 0.5 and 1.5 kOe to 0.07 kOe. Moreover, a high HkRT of 2 T is retained, even for the films in which Co is replaced by Cu 26 at. %. The coercive mechanism is linked to nanometer-scaled domain wall pinning. Based on it, a medium Hc⬜ can be attained when the film is grown with microstructural modifications such as forming islandlike, granular, or patterned structures, which solves the typical problem of high coercivity arising from magnetic isolation in the high-Ku films. The enhanced perpendicular coercivity and unique coercive mechanism substantially promote the use of Co–Cu–Pt L11 phase in the applications of magnetic layers with a large perpendicular magnetic anisotropy such as advanced magnetic recording patterned media with a small writing field.

ACKNOWLEDGMENTS

This work was supported by the National Science Council of Taiwan under Grant No. NSC 99-2112-M-002-020MY3 and partially supported by the Ministry of Economic Affairs of Taiwan under Grant No. 99-EC-17-A-08-S1-006. 1

S. Iwata, S. Yamashita, and S. Tsunashima, IEEE Trans. Magn. 33, 3670 共1997兲. H. Sato, T. Shimatsu, Y. Okazaki, H. Muraoka, H. Aoi, S. Okamoto, and O. Kitakami, J. Appl. Phys. 103, 07E114 共2008兲. 3 A. C. Sun, F. T. Yuan, J. H. Hsu, and H. Y. Lee, Scr. Mater. 61, 713 共2009兲. 4 F. T. Yuan, A. C. Sun, and J. H. Hsu, Scr. Mater. 62, 762 共2010兲. 5 H. Lang, T. Mohri, and W. Pfeiler, Intermetallics 7, 1373 共1999兲. 6 B. D. Cullity and S. R. Stock, Elements of X-Ray Diffraction, 3rd edition 共Prentice-Hall, Saddle River, New Jersey, 2001兲, p. 634. 7 J.-U. Thiele, L. Folks, M. F. Toney, and D. K. Weller, J. Appl. Phys. 84, 5686 共1998兲. 8 K. Barmak, J. Kim, L. H. Lewis, K. R. Coffey, M. F. Toney, A. J. Kellock, and J.-U. Thiele, J. Appl. Phys. 98, 033904 共2005兲. 9 D. Balzar, J. Appl. Crystallogr. 25, 559 共1992兲. 10 D. Balzar and H. Ledbetter, J. Appl. Crystallogr. 26, 97 共1993兲. 11 T. Suzuki, N. Honda, and K. Ouchi, J. Appl. Phys. 85, 4301 共1999兲. 12 G. Q. Li, H. Takahoshi, H. Ito, H. Saito, S. Ishio, T. Shima, and K. Takanashi, J. Appl. Phys. 94, 5672 共2003兲. 13 T. Maeda, T. Kai, A. Kikitsu, T. Nagase, and J. Akiyama, Appl. Phys. Lett. 80, 2147 共2002兲. 14 C. L. Platt, K. W. Wierman, E. B. Svedberg, R. van de Veerdonk, J. K. Howard, A. G. Roy, and D. E. Laughlin, J. Appl. Phys. 92, 6104 共2002兲. 15 J. H. Min, B. H. An, J. U. Cho, H. M. Ji, S. J. Noh, Y. K. Kim, H. L. Liu, Y. D. Ko, and J. S. Chung, J. Appl. Phys. 101, 09K513 共2007兲. 16 T. Huang, F. Wang, J. Kuo, and X. Xu, Rare Met. Mater. Eng. 28, 14 共2009兲. 17 C. L. Platt and K. W. Wierman, J. Magn. Magn. Mater. 295, 241 共2005兲. 18 M. L. Yan, Y. F. Xu, and D. J. Sellmyer, J. Appl. Phys. 99, 08G903 共2006兲. 19 K. Barmak, J. Kim, D. C. Barry, K. W. Wierman, E. B. Svedberg, and J. K. Howard, J. Appl. Phys. 95, 7486 共2004兲. 2
