Home
Search
Collections
Journals
About
Contact us
My IOPscience
Enhancement of perpendicular magnetic anisotropy by compressive strain in alternately layered FeNi thin films
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 J. Phys.: Condens. Matter 26 166002 (http://iopscience.iop.org/0953-8984/26/16/166002) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 138.37.211.113 This content was downloaded on 13/06/2014 at 14:36
Please note that terms and conditions apply.
Journal of Physics: Condensed Matter J. Phys.: Condens. Matter 26 (2014) 166002 (6pp)
doi:10.1088/0953-8984/26/16/166002
Enhancement of perpendicular magnetic anisotropy by compressive strain in alternately layered FeNi thin films M Sakamaki and K Amemiya Institute of Materials Structure Science, High Energy Accelerator Research Organization, Tsukuba, Ibaraki 305-0801, Japan E-mail: [email protected] Received 17 December 2013, revised 18 February 2014 Accepted for publication 3 March 2014 Published 3 April 2014 Abstract
The effect of the lattice strain on magnetic anisotropy of alternately layered FeNi ultrathin films grown on a substrate, Cu(tCu = 0–70 ML)/Ni48Cu52(124 ML)/Cu(0 0 1) single crystal, is systematically studied by means of in situ x-ray magnetic circular dichroism (XMCD) and reflection high-energy electron diffraction (RHEED) analyses. To investigate the magnetic anisotropy of the FeNi layer itself, a non-magnetic substrate is adopted. From the RHEED analysis, the in-plane lattice constant, ain, of the substrate is found to shrink by 0.8% and 0.5% at tCu = 0 and 10 ML as compared to that of bulk Cu, respectively. Fe L-edge XMCD analysis is performed for n ML FeNi films grown on various ain, and perpendicular magnetic anisotropy (PMA) is observed at n = 3 and 5, whereas the film with n = 7 shows in-plane magnetic anisotropy. Moreover, it is found that PMA is enhanced with decreasing ain, in the case where a Cu spacer layer is inserted. We suppose that magnetic anisotropy in the FeNi films is mainly carried by Fe, and the delocalization of the in-plane orbitals near the Fermi level increases the perpendicular orbital magnetic moment, which leads to the enhancement of PMA. Keywords: magnetic thin films, x-ray absorption spectroscopy, magnetic anisotropy (Some figures may appear in colour only in the online journal)
1. Introduction
anisotropy is expected to be controlled by the surface/interface effect [5, 6], morphology [7, 8] and strain [9] for instance. Our previous in situ x-ray magnetic circular dichroism (XMCD) study reveals the element specific magnetic anisotropy energies (MAEs) in the FeNi ultrathin films on a Cu(0 0 1) substrate [10], and indicates that PMA is enhanced in the case of Fe-terminated FeNi films, while that of Ni-terminated films is reduced. This can be interpreted to be due to a large positive MAE in the surface Fe, whereas sandwiched-Fe loses MAE. On the other hand, it is reported that anisotropy in the orbital magnetic moment of Fe carries magnetic anisotropy in L10 type FeNi [11]. These results are understood by the fact that the structural strain leads to changes in the orbital magnetic moment of Fe because the buffer/substrate structures and FeNi thicknesses are different between those studies, resulting in different strains. Thus, systematical investigation of the relationship between the strain and magnetic anisotropy is important. Recently, we found from the XMCD study
L10 ordered alloys of noble metals, such as FePt, are known as materials with high uniaxial magnetic anisotropy [1]. On the other hand, the L10 type FeNi multilayer has attracted attention as one of the candidate materials for rare metal-free perpendicularly magnetized films for high-density recording medium. The L10 ordered structure consists of the alternate stacking of two different atomic planes along the fcc [0 0 1] direction, and is expected to exhibit uniaxial magnetic anisotropy in the c axis, [0 0 1]. So far, some efforts have been made to realize perpendicular magnetic anisotropy (PMA) by using alternate monatomic layer deposition on fcc (0 0 1) substrates [2–4]. The magnetic and structural properties of the multilayer are usually investigated after the deposition of the whole film, but in situ observation of the growth process is also important to understand the fundamental magnetic properties, such as the effect on magnetic anisotropy, of each layer. Magnetic 0953-8984/14/166002+6$33.00
1
© 2014 IOP Publishing Ltd Printed in the UK
M Sakamaki and K Amemiya
J. Phys.: Condens. Matter 26 (2014) 166002
(i) (ii)
NiCu (124 ML)
of Ni and Cu, and then a wedge-shaped Cu layer was grown on top. 1 ML Fe and 1 ML Ni layers were alternately grown starting with an Fe layer on the whole area of the substrate, so that the strain effect on the magnetic property of the FeNi layer can be investigated at the same time by using substrates with various lattice constants. The Ni60Cu40/Cu(0 0 1) system has relatively small lattice mismatch, and is known to grow epitaxially on Cu(0 0 1). The in-plane lattice constant of a 20 nm Ni60Cu40 film is reported to be 1.9% smaller than that of bulk fcc Cu [9]. It is also reported that growing a Cu spacer layer on the Ni60Cu40/ Cu(0 0 1) film continuously modulates its in-plane lattice constant. Therefore, the structure of the FeNi film is expected to be controllable when Cu/Ni60Cu40/Cu(0 0 1) is adopted as the substrate, and its influence on the magnetic anisotropy of the FeNi film is investigated. XMCD spectra were measured at room temperature by using circularly polarized x rays, which were obtained by accepting the off-axis radiation from a bending magnet or by using APPLE-type undulators. Polarization switching between opposite circular polarizations was also applied [18, 19]. The sample was mounted with [1 1 0] lying in the horizontal plane, which is the magnetic easy axis of the in-plane magnetized films. A pulsed magnetic field of ∼2 kOe, which is parallel to the incident x-ray beam, was applied before each measurement, and the remanent state was investigated. 2 kOe is strong enough to saturate the magnetization along the easy axis. The normal and grazing x-ray incidence angle geometries were adopted, denoted as NI and GI, respectively. The angle between [1 1 0] and the x-ray beam was 30° at GI. The total electron yield mode was adopted to record the XMCD spectrum.
(iii) Cu spacer layer (t Cu=0-70 ML)
Cu(001) Figure 1. Cross-sectional view of the substrate. A NiCu(124 ML) layer is grown on an area of Cu(0 0 1), and then a wedge-shaped Cu layer(tCu = 0–70 ML) is grown on top. Three distinct regions, (i) NiCu(124 ML)/Cu(0 0 1), (ii) Cu(10 ML)/NiCu(124 ML)/Cu(0 0 1) and (iii) Cu(0 0 1), are chosen for the detailed analysis.
combining with reflection high-energy electron diffraction (RHEED) analysis that PMA is enhanced when the compressive strain in the in-plane direction was applied to the Fe layer [12]. In that study, Ni(3–21 ML)/Cu(0 0 1) was adopted as the substrate because the magnetic anisotropy and structure of the FeNi layer is controllable by the peculiar magnetic and structural properties of the Ni/Cu(0 0 1) system [13–15], and MAE in Fe and Ni was estimated from the obtained orbital magnetic moment. However, to clarify the magnetic anisotropy of the FeNi layer itself more directly, it is advantageous to exclude the effect of the ferromagnetic Ni substrate. In the present study, we apply in situ XMCD and RHEED analyses to alternately layered FeNi ultrathin films grown on non-magnetic substrates, by using NiCu alloy as a buffer layer. It has been reported that the lattice constant of a Ni thin film is controlled by changing the thickness of the Cu spacer layer on a Ni60Cu40/Cu(0 0 1) substrate [9]. Therefore, we can systematically study the effect of the lattice strain on magnetic anisotropy of the FeNi layer by using the Cu/NiCu/Cu(0 0 1) substrate. Indeed, our result shows that PMA is enhanced when the compressive strain in the in-plane direction is applied to the FeNi films. This is contrary to the previous result on the Ni films, in which the Ni thin films show enhancement of PMA when tensile strain in the in-plane direction is applied [9]. The difference between the Ni and FeNi layers is explained by the fact that magnetic anisotropy in the FeNi films is mainly carried by the Fe layer, and that the delocalization of the inplane orbitals near the Fermi level induced by the compressive strain increases the perpendicular orbital magnetic moment, which leads to enhancement of PMA.
3. Result and discussion Figure 2 displays the RHEED image of NiCu(124 ML)/ Cu(0 0 1) taken at an electron energy of 12.5 kV. A bright round spot at the bottom of the picture is the direct beam from the electron gun, and several streaks were observed along the incident electron beam direction, [1 1 0]. The samples all showed almost the same pattern, which suggests that Fe, Ni, NiCu and Cu grow epitaxially on the Cu(0 0 1) substrate. The in-plane lattice constant of the films was estimated from the comparison of the streak interval of the RHEED pattern, as illustrated in figure 2(b), with that of the Cu(0 0 1) substrate, whose lattice constant is 0.255 nm. In the present study, the nearest in-plane interatomic distance is called the in-plane lattice constant, ain, as indicated in figure 2(a). We first estimate ain of Cu(tCu ML)/NiCu(124 ML)/ Cu(0 0 1) as a function of Cu spacer layer thickness, tCu, as shown in figure 3. The alloy composition in NiCu is estimated to be Ni48Cu52 from the ratio of the deposition rates between Ni and Cu, which were determined from the RHEED oscillation analysis, and is consistent with that obtained by x-ray absorption edge intensity analysis. NiCu film grows as its inplane lattice constant on Cu(0 0 1) decreases, and shrinks by ∼0.8% at a thickness of 124 ML, which is consistent with the
2. Experimental All the experiments were performed in situ in an ultra high vacuum chamber with a base pressure of ∼9 × 10−8 Pa at the beamlines BL-7A [16] and 16A [17] of the Photon Factory, Japan. A Cu(0 0 1) single crystal was cleaned by repeated cycles of Ar+ sputtering at 1.5 keV and subsequent annealing at ∼900 K. NiCu alloy and FeNi layers were grown at room temperature by electron bombardment evaporation. The evaporation rate of Fe, Ni and Cu was separately determined before sample preparation by monitoring the oscillatory intensity of the RHEED spot during the growth of each film. Sample structure is depicted in figure 1. A NiCu(124 ML) layer was grown on an area of Cu(0 0 1) by a simultaneous deposition 2
M Sakamaki and K Amemiya
J. Phys.: Condens. Matter 26 (2014) 166002
(b)
Intensity (arb.unit)
(a)
fcc lattice
ain
d Position
Figure 2. (a) RHEED image of NiCu(124 ML)/Cu(0 0 1) and schematic of the defined in-plane lattice constant, ain. (b) Enlarged image
around the direct-beam spot. Integration over three typical streaks is also shown. ain is estimated from the streak interval, d.
a in (nm)
NiCu(124 ML)/Cu(0 0 1) and (iii) Cu(0 0 1), and compare the magnetic properties of the FeNi films. Figure 4 shows Fe L-edge x-ray absorption and XMCD spectra for 5 ML FeNi films grown on regions, (i), (ii) and (iii). XMCD was measured at the NI configuration, in which the remanent perpendicular magnetic component is detected. The estimated ain from the RHEED analysis is also shown in the figure. In the present case, the FeNi films show perpendicular magnetization when they are grown on (ii) and (iii), while the perpendicular component is not detected at (i). We suppose that the FeNi layers on (i) prefer in-plane magnetization due to the electronic interaction with NiCu alloy, so that it is necessary to insert the Cu spacer layer to break the coupling. The spin magnetic moment, ms, including the magnetic dipole term is estimated from the XMCD sum-rule analysis [20, 21], and is indicated in the figure. The d-hole number, nh, is determined by the white-line intensity after subtracting the step functions, based on the assumption that nh of 5 ML FeNi films on (ii) is nh5ML = 3.4 [22], and that nh is proportional to the white-line intensity. The estimated ms is lower than that of Fe bulk [23] and ∼100 ML L10 FeNi film [11]. We suppose that this is due to systematic errors connected to the procedure of the background and step function subtraction, and to the assumption of nh5ML. Nevertheless, relative comparison of ms among the present samples would be reasonable since the change in nh accompanying the white-line intensity is taken into account, as described above. ms at (ii) is larger than that at (iii), which suggests that the 5 ML FeNi film directly grown on Cu(0 0 1) is not fully magnetized in the perpendicular direction at the remanent state. We suppose that the reduction of PMA causes the domain formation and/or transition to the in-plane magnetization, resulting in the decrease of perpendicular magnetization. In fact,
t Cu (ML) Figure 3. In-plane lattice constant, ain, of Cu(tCu ML)/
NiCu(124 ML)/Cu(0 0 1) as a function of Cu spacer layer thickness, tCu. ain of fcc Cu is shown for comparison. Dotted line is linear fit for tCu = 4–70 ML.
previous result [9]. Thus, ain of NiCu(124 ML)/Cu(0 0 1) is estimated to be 0.253 nm. When the Cu spacer layer is added to NiCu(124 ML)/Cu(0 0 1), ain shows a rapid increase up to tCu = 4 ML, and then shows monotonous increase. Although there is no data between 10 and 70 ML, the data between 4 and 10 ML show a linear increase, and the extrapolated line reaches the data at 70 ML. Thus, we apply a linear fit between 4 and 70 ML. A linear change in ain of the Cu spacer layer is also reported in the previous study [9]. The lattice constant of fcc Cu(0 0 1) is reached if we extrapolate the fitting result up to ∼120 ML. The FeNi films were then grown on Cu(tCu ML)/ NiCu(124 ML)/Cu(0 0 1) and bare Cu(0 0 1) substrates, and the x-ray beam was irradiated at different positions in order to obtain the signal originated from different film configurations with various lattice constants. We here focus on the three distinct regions, (i) NiCu(124 ML)/Cu(0 0 1), (ii) Cu(10 ML)/ 3
M Sakamaki and K Amemiya
J. Phys.: Condens. Matter 26 (2014) 166002
(b)
(a) (i) a in =0.253
m s =1.51
B
(ii) a in =0.254
(iii) a in =0.255
m s =0.96
B
Figure 4. Fe L-edge circularly polarized x-ray absorption spectra (XAS) (a) and XMCD spectra (b) for 5 ML FeNi films grown on three different substrates, (i) NiCu(124 ML)/Cu(0 0 1), (ii) Cu(10 ML)/NiCu(124 ML)/Cu(0 0 1) and (iii) Cu(0 0 1). ain of each substrate indicated in the figure is estimated from the data in figure 3. XAS and XMCD were measured at normal incidence angle configurations. The remanent magnetization state was investigated after applying the magnetic field parallel to the incident x-ray beam. Estimated spin magnetic moment, ms, is indicated in the figure.
several fine domain structures are observed for FeNi bilayers when PMA is reduced [24]. Because we measure the XMCD spectra at the remanent state, there might exist some domain structures. To confirm that the reduction of the perpendicular magnetization indicates the transition toward the in-plane magnetization, Fe L-edge XMCD spectra for n ML FeNi films grown on (ii) are shown in figure 5, in which XMCD was taken at NI and GI configurations. In the case of 3 and 5 ML FeNi films, the XMCD intensity at NI is twice as large as that at GI. This means that the films show perpendicular magnetization because XMCD probes the magnetic moment projected onto the direction of the incident x ray. On the other hand, the 7 ML film shows no XMCD signal at NI, while a clear XMCD signal is observed at GI, which corresponds to the inplane magnetization. These results indicate that the XMCD intensity taken at NI directly reflects the magnitude of PMA. From the previous study, it is found that the magnetic direction of Ni is the same as that of Fe in the present FeNi thickness region [12]. Therefore, only the Fe L-edge measurement was performed in order to avoid the time-dependent changes caused by surface effects such as oxidation. To investigate the relation between the structure and magnetic anisotropy systematically, we plot the XMCD intensity at the Fe L3 peak top for n ML FeNi films as a function of the in-plane lattice constant, ain, in figure 6. XMCD was measured at the NI configuration. The 3 ML FeNi films show perpendicular magnetization, and XMCD intensity is almost the same independent of ain. On the other hand, the 5 ML FeNi film also shows perpendicular magnetization in a wide ain region including (ii) and (iii), but larger XMCD intensity is observed at smaller ain. As observed in figure 4, XMCD was not detected at (i), which indicates in-plane magnetization. In
(ii) a in =0.254
Figure 5. Fe L-edge XMCD spectra for n ML FeNi films grown on Cu(10 ML)/NiCu(124 ML)/Cu(0 0 1), where ain = 0.254. The spectra were taken at normal (NI) and grazing incidence (GI) angle configurations. 3 and 5 ML FeNi films show perpendicular magnetization, whereas 7 ML FeNi film shows in-plane magnetization.
the case of n = 7, FeNi films show almost no XMCD signal regardless of ain, which also indicates in-plane magnetization. Thus, it is concluded that the compressive strain in the inplane direction enhances PMA in the FeNi films. 4
M Sakamaki and K Amemiya
J. Phys.: Condens. Matter 26 (2014) 166002
(i) a in =0.253 (ii) a in =0.254 (iii) a in =0.255
m s =1.26 B m l /m s =0.13
m s =1.25 B m l /m s =0.23
m s =1.30 B m l /m s =0.15
a in (nm) Figure 6. XMCD intensity at Fe L3 peak top for n ML FeNi films as a function of in-plane lattice constant, ain. ain is obtained by RHEED analysis. Three different substrates, (i) NiCu(124 ML)/ Cu(0 0 1), (ii) Cu(10 ML)/NiCu(124 ML)/Cu(0 0 1) and (iii) Cu(0 0 1) are indicated in the figure. XMCD was measured at normal incidence angle configuration. The remanent magnetization state was investigated after applying the magnetic field parallel to the incident x-ray beam.
Then, we estimate the uniaxial magnetic anisotropy constant, Ku, with the help of a theoretical calculation. If we assume that the volume of the unit cell is the same as that of bulk L10 FeNi, the ratio of the lattice parameters, c/a, is estimated to be 0.99 and 0.97 for the FeNi films on (ii) and (iii), respectively, whereas that of bulk L10 FeNi is reported to be 1.0036 [25]. Ku at c/a = 0.99 and 0.97 are estimated to be 0.50 and 0.42 MJ m−3, respectively, from the first-principle calculations for L10 type FeNi [26]. The difference in the estimated Ku between (ii) and (iii) can explain the observed change in PMA. Finally, we discuss the detailed spectral structure and obtained magnetic moment. Figure 7 shows Fe L-edge XMCD for 3 ML FeNi films at three different regions, (i), (ii) and (iii). It is reasonable to compare the 3 ML FeNi films on various substrates because all these films are fully magnetized in the perpendicular direction, as shown in figure 6. The estimated spin magnetic moment is almost the same among three samples, whereas the ratio of orbital and spin magnetic moments, ml/ms, shows a distinct difference, and a larger value is obtained at (ii). In addition, the spectral shape between the L3 and L2 edges at (ii) seems different from that at (i) and (iii), i.e., the XMCD curve in this region shows a positive signal at (i) and (iii), while it shows almost zero at (ii). In the literature, a small negative XMCD signal between the L3 and L2 edges is observed for bcc Fe, fcc Ni and fcc Co thin films, and is explained as being due to the diffuse magnetic moment, which is originated from the polarized 4s electrons [27, 28]. Normally, bulk Fe shows a positive shoulder above the L3 edge, and the signal approaches zero before the L2 peak, as observed in the present case of (i) and (iii). Thus, we suppose that the observed spectral change at (ii) is due to the diffuse magnetic moment, and the compressive strain induces the diffuse magnetism. Indeed, it is reported from the first-principles calculations for L10 type FeNi that only the Fe components of the orbital density of states exist at the Fermi level [29], and the Fe d x 2 − y2 and dyz, zx orbitals delocalize near the Fermi level by the
Figure 7. Fe L-edge XMCD spectra for 3 ML FeNi films grown on (i) NiCu(124 ML)/Cu(0 0 1), (ii) Cu(10 ML)/NiCu(124 ML)/ Cu(0 0 1) and (iii) Cu(0 0 1), taken at normal incidence angle configurations. The remanent magnetization state was investigated. Estimated spin and orbital magnetic moments, ms and ml, are indicated in the figure. Inset shows enlarged view of the region between the L3 and L2 edges.
compressive strain [26]. We suppose that the delocalization of the in-plane orbitals enhances both the diffuse magnetism and the perpendicular orbital magnetic moment, resulting in the enhancement of PMA in the FeNi films. 4. Summary We have performed a systematic study on the effect of the lattice strain on magnetic anisotropy of alternately layered FeNi ultrathin films grown on the substrate Cu(tCu = 0–70 ML)/ Ni48Cu52(124 ML)/Cu(0 0 1) single crystal, by means of in situ XMCD and RHEED analyses. From the RHEED analysis, the in-plane lattice constant, ain, of the substrate was found to shrink by 0.8% and 0.5% at tCu = 0 and 10 ML as compared to that of bulk Cu, respectively, so that continuous modulation of ain was confirmed. Fe L-edge XMCD was measured for n ML FeNi films grown on various ain, and PMA is observed at n = 3 and 5, whereas the film with n = 7 shows in-plane magnetic anisotropy. Moreover, we revealed that PMA is enhanced with decreasing ain, in the case where a Cu spacer layer is inserted. At the same time, the enhancement of the perpendicular magnetic moment and the diffuse magnetic moment are observed with decreasing ain. We suppose that magnetic anisotropy in the FeNi films is mainly carried by Fe, and the delocalization of the in-plane orbitals near the Fermi level increases the 5
M Sakamaki and K Amemiya
J. Phys.: Condens. Matter 26 (2014) 166002
[9] Lauhoff G, Vaz C A F, Bland J A C, Lee J and Suzuki T 2000 Phys. Rev. B 61 6805 [10] Sakamaki M and Amemiya K 2011 Appl. Phys. Express 4 073002 [11] Kotsugi M et al 2013 J. Magn. Magn. Mater. 326 235 [12] Sakamaki M and Amemiya K 2013 Phys. Rev. B 87 014428 [13] O'Brien W L, Droubay T and Tonner B P 1996 Phys. Rev. B 54 9297 [14] Matthews J W 1970 Thin Solid Films 5 187 [15] Schulz B and Baberschke K 1994 Phys. Rev. B 50 13467 [16] Amemiya K, Kondoh H, Yokoyama T and Ohta T 2002 J. Electron Spectrosc. Relat. Phenom. 124 151 [17] Amemiya K, Toyoshima A, Kikuchi T, Kosuge T, Nigorikawa K, Sumii R and Ito K 2010 AIP Conf. Proc. 1234 295 [18] Amemiya K et al 2013 J. Phys.: Conf. Ser. 425 152015 [19] Tsuchiya K, Shioya T, Aoto T, Harada K, Obina T, Sakamaki M and Amemiya K 2013 J. Phys.: Conf. Ser. 425 132017 [20] Thole B T, Carra P, Sette F and van der Laan G 1992 Phys. Rev. Lett. 68 1943 [21] Carra P, Thole B T, Altarelli M and Wang X 1993 Phys. Rev. Lett. 70 694 [22] Eguchi K, Takagi Y, Nakagawa T and Yokoyama T 2013 J. Phys.: Conf. Ser. 430 012129 [23] Cullity B D 1972 Introduction to Magnetic Materials (Reading, MA: Addison-Wesley) [24] Ramchal R, Schmid A K, Farle M and Poppa H 2004 Phys. Rev. B 69 214401 [25] Albertsen J F 1981 Phys. Scr. 23 301 [26] Miura Y, Ozaki S, Kuwahara Y, Tsujikawa M, Abe K and Shirai M 2013 J. Phys.: Condens. Matter 25 106005 [27] O'Brien W L and Tonner B P 1994 Phys. Rev. B 50 12672 [28] Le Cann X, Boeglin C, Carriére B and Hricovini K 1996 Phys. Rev. B 54 373 [29] Ravindran P, Kjekshus A, Fjellvåg H, James P, Nordström L, Johansson B and Eriksson O 2001 Phys. Rev. B 63 144409
perpendicular orbital magnetic moment, which results in the enhancement of PMA. Acknowledgments The present work has been performed under the approval of the Photon Factory Program Advisory Committee (no. 2010S2-001). The authors are grateful for the financial support of the Quantum Beam Technology Program from the Ministry of Education, Culture, Sports, Science and Technology (MEXT). References [1] Ovanov O A, Solina L V and Demshina V A 1973 Phys. Met. Metallogr. 35 81 [2] Shima T, Okamura M, Mitani S and Takanashi K 2007 J. Magn. Magn. Mater. 310 2213 [3] Mizuguchi M, Sekiya S and Takanashi K 2010 J. Appl. Phys. 107 09A716 [4] Kojima T, Mizuguchi M and Takanashi K 2011 J. Phys.: Conf. Ser. 266 012119 [5] Amemiya K, Sakai E, Matsumura D, Abe H, Ohta T and Yokoyama T 2005 Phys. Rev. B 71 214420 [6] Abe H, Amemiya K, Matsumura D, Kitagawa S, Watanabe H, Yokoyama T and Ohta T 2006 J. Magn. Magn. Mater. 302 86 [7] Sakamaki M and Amemiya K 2011 J. Phys.: Conf. Ser. 266 012020 [8] Cherifi S, Stanescu S, Mocuta C, Deville J-P, Boeglin C, Ohresser P and Brookes N B 2001 Surf. Sci. 482–85 1056
6
Search
Collections
Journals
About
Contact us
My IOPscience
Enhancement of perpendicular magnetic anisotropy by compressive strain in alternately layered FeNi thin films
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 J. Phys.: Condens. Matter 26 166002 (http://iopscience.iop.org/0953-8984/26/16/166002) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 138.37.211.113 This content was downloaded on 13/06/2014 at 14:36
Please note that terms and conditions apply.
Journal of Physics: Condensed Matter J. Phys.: Condens. Matter 26 (2014) 166002 (6pp)
doi:10.1088/0953-8984/26/16/166002
Enhancement of perpendicular magnetic anisotropy by compressive strain in alternately layered FeNi thin films M Sakamaki and K Amemiya Institute of Materials Structure Science, High Energy Accelerator Research Organization, Tsukuba, Ibaraki 305-0801, Japan E-mail: [email protected] Received 17 December 2013, revised 18 February 2014 Accepted for publication 3 March 2014 Published 3 April 2014 Abstract
The effect of the lattice strain on magnetic anisotropy of alternately layered FeNi ultrathin films grown on a substrate, Cu(tCu = 0–70 ML)/Ni48Cu52(124 ML)/Cu(0 0 1) single crystal, is systematically studied by means of in situ x-ray magnetic circular dichroism (XMCD) and reflection high-energy electron diffraction (RHEED) analyses. To investigate the magnetic anisotropy of the FeNi layer itself, a non-magnetic substrate is adopted. From the RHEED analysis, the in-plane lattice constant, ain, of the substrate is found to shrink by 0.8% and 0.5% at tCu = 0 and 10 ML as compared to that of bulk Cu, respectively. Fe L-edge XMCD analysis is performed for n ML FeNi films grown on various ain, and perpendicular magnetic anisotropy (PMA) is observed at n = 3 and 5, whereas the film with n = 7 shows in-plane magnetic anisotropy. Moreover, it is found that PMA is enhanced with decreasing ain, in the case where a Cu spacer layer is inserted. We suppose that magnetic anisotropy in the FeNi films is mainly carried by Fe, and the delocalization of the in-plane orbitals near the Fermi level increases the perpendicular orbital magnetic moment, which leads to the enhancement of PMA. Keywords: magnetic thin films, x-ray absorption spectroscopy, magnetic anisotropy (Some figures may appear in colour only in the online journal)
1. Introduction
anisotropy is expected to be controlled by the surface/interface effect [5, 6], morphology [7, 8] and strain [9] for instance. Our previous in situ x-ray magnetic circular dichroism (XMCD) study reveals the element specific magnetic anisotropy energies (MAEs) in the FeNi ultrathin films on a Cu(0 0 1) substrate [10], and indicates that PMA is enhanced in the case of Fe-terminated FeNi films, while that of Ni-terminated films is reduced. This can be interpreted to be due to a large positive MAE in the surface Fe, whereas sandwiched-Fe loses MAE. On the other hand, it is reported that anisotropy in the orbital magnetic moment of Fe carries magnetic anisotropy in L10 type FeNi [11]. These results are understood by the fact that the structural strain leads to changes in the orbital magnetic moment of Fe because the buffer/substrate structures and FeNi thicknesses are different between those studies, resulting in different strains. Thus, systematical investigation of the relationship between the strain and magnetic anisotropy is important. Recently, we found from the XMCD study
L10 ordered alloys of noble metals, such as FePt, are known as materials with high uniaxial magnetic anisotropy [1]. On the other hand, the L10 type FeNi multilayer has attracted attention as one of the candidate materials for rare metal-free perpendicularly magnetized films for high-density recording medium. The L10 ordered structure consists of the alternate stacking of two different atomic planes along the fcc [0 0 1] direction, and is expected to exhibit uniaxial magnetic anisotropy in the c axis, [0 0 1]. So far, some efforts have been made to realize perpendicular magnetic anisotropy (PMA) by using alternate monatomic layer deposition on fcc (0 0 1) substrates [2–4]. The magnetic and structural properties of the multilayer are usually investigated after the deposition of the whole film, but in situ observation of the growth process is also important to understand the fundamental magnetic properties, such as the effect on magnetic anisotropy, of each layer. Magnetic 0953-8984/14/166002+6$33.00
1
© 2014 IOP Publishing Ltd Printed in the UK
M Sakamaki and K Amemiya
J. Phys.: Condens. Matter 26 (2014) 166002
(i) (ii)
NiCu (124 ML)
of Ni and Cu, and then a wedge-shaped Cu layer was grown on top. 1 ML Fe and 1 ML Ni layers were alternately grown starting with an Fe layer on the whole area of the substrate, so that the strain effect on the magnetic property of the FeNi layer can be investigated at the same time by using substrates with various lattice constants. The Ni60Cu40/Cu(0 0 1) system has relatively small lattice mismatch, and is known to grow epitaxially on Cu(0 0 1). The in-plane lattice constant of a 20 nm Ni60Cu40 film is reported to be 1.9% smaller than that of bulk fcc Cu [9]. It is also reported that growing a Cu spacer layer on the Ni60Cu40/ Cu(0 0 1) film continuously modulates its in-plane lattice constant. Therefore, the structure of the FeNi film is expected to be controllable when Cu/Ni60Cu40/Cu(0 0 1) is adopted as the substrate, and its influence on the magnetic anisotropy of the FeNi film is investigated. XMCD spectra were measured at room temperature by using circularly polarized x rays, which were obtained by accepting the off-axis radiation from a bending magnet or by using APPLE-type undulators. Polarization switching between opposite circular polarizations was also applied [18, 19]. The sample was mounted with [1 1 0] lying in the horizontal plane, which is the magnetic easy axis of the in-plane magnetized films. A pulsed magnetic field of ∼2 kOe, which is parallel to the incident x-ray beam, was applied before each measurement, and the remanent state was investigated. 2 kOe is strong enough to saturate the magnetization along the easy axis. The normal and grazing x-ray incidence angle geometries were adopted, denoted as NI and GI, respectively. The angle between [1 1 0] and the x-ray beam was 30° at GI. The total electron yield mode was adopted to record the XMCD spectrum.
(iii) Cu spacer layer (t Cu=0-70 ML)
Cu(001) Figure 1. Cross-sectional view of the substrate. A NiCu(124 ML) layer is grown on an area of Cu(0 0 1), and then a wedge-shaped Cu layer(tCu = 0–70 ML) is grown on top. Three distinct regions, (i) NiCu(124 ML)/Cu(0 0 1), (ii) Cu(10 ML)/NiCu(124 ML)/Cu(0 0 1) and (iii) Cu(0 0 1), are chosen for the detailed analysis.
combining with reflection high-energy electron diffraction (RHEED) analysis that PMA is enhanced when the compressive strain in the in-plane direction was applied to the Fe layer [12]. In that study, Ni(3–21 ML)/Cu(0 0 1) was adopted as the substrate because the magnetic anisotropy and structure of the FeNi layer is controllable by the peculiar magnetic and structural properties of the Ni/Cu(0 0 1) system [13–15], and MAE in Fe and Ni was estimated from the obtained orbital magnetic moment. However, to clarify the magnetic anisotropy of the FeNi layer itself more directly, it is advantageous to exclude the effect of the ferromagnetic Ni substrate. In the present study, we apply in situ XMCD and RHEED analyses to alternately layered FeNi ultrathin films grown on non-magnetic substrates, by using NiCu alloy as a buffer layer. It has been reported that the lattice constant of a Ni thin film is controlled by changing the thickness of the Cu spacer layer on a Ni60Cu40/Cu(0 0 1) substrate [9]. Therefore, we can systematically study the effect of the lattice strain on magnetic anisotropy of the FeNi layer by using the Cu/NiCu/Cu(0 0 1) substrate. Indeed, our result shows that PMA is enhanced when the compressive strain in the in-plane direction is applied to the FeNi films. This is contrary to the previous result on the Ni films, in which the Ni thin films show enhancement of PMA when tensile strain in the in-plane direction is applied [9]. The difference between the Ni and FeNi layers is explained by the fact that magnetic anisotropy in the FeNi films is mainly carried by the Fe layer, and that the delocalization of the inplane orbitals near the Fermi level induced by the compressive strain increases the perpendicular orbital magnetic moment, which leads to enhancement of PMA.
3. Result and discussion Figure 2 displays the RHEED image of NiCu(124 ML)/ Cu(0 0 1) taken at an electron energy of 12.5 kV. A bright round spot at the bottom of the picture is the direct beam from the electron gun, and several streaks were observed along the incident electron beam direction, [1 1 0]. The samples all showed almost the same pattern, which suggests that Fe, Ni, NiCu and Cu grow epitaxially on the Cu(0 0 1) substrate. The in-plane lattice constant of the films was estimated from the comparison of the streak interval of the RHEED pattern, as illustrated in figure 2(b), with that of the Cu(0 0 1) substrate, whose lattice constant is 0.255 nm. In the present study, the nearest in-plane interatomic distance is called the in-plane lattice constant, ain, as indicated in figure 2(a). We first estimate ain of Cu(tCu ML)/NiCu(124 ML)/ Cu(0 0 1) as a function of Cu spacer layer thickness, tCu, as shown in figure 3. The alloy composition in NiCu is estimated to be Ni48Cu52 from the ratio of the deposition rates between Ni and Cu, which were determined from the RHEED oscillation analysis, and is consistent with that obtained by x-ray absorption edge intensity analysis. NiCu film grows as its inplane lattice constant on Cu(0 0 1) decreases, and shrinks by ∼0.8% at a thickness of 124 ML, which is consistent with the
2. Experimental All the experiments were performed in situ in an ultra high vacuum chamber with a base pressure of ∼9 × 10−8 Pa at the beamlines BL-7A [16] and 16A [17] of the Photon Factory, Japan. A Cu(0 0 1) single crystal was cleaned by repeated cycles of Ar+ sputtering at 1.5 keV and subsequent annealing at ∼900 K. NiCu alloy and FeNi layers were grown at room temperature by electron bombardment evaporation. The evaporation rate of Fe, Ni and Cu was separately determined before sample preparation by monitoring the oscillatory intensity of the RHEED spot during the growth of each film. Sample structure is depicted in figure 1. A NiCu(124 ML) layer was grown on an area of Cu(0 0 1) by a simultaneous deposition 2
M Sakamaki and K Amemiya
J. Phys.: Condens. Matter 26 (2014) 166002
(b)
Intensity (arb.unit)
(a)
fcc lattice
ain
d Position
Figure 2. (a) RHEED image of NiCu(124 ML)/Cu(0 0 1) and schematic of the defined in-plane lattice constant, ain. (b) Enlarged image
around the direct-beam spot. Integration over three typical streaks is also shown. ain is estimated from the streak interval, d.
a in (nm)
NiCu(124 ML)/Cu(0 0 1) and (iii) Cu(0 0 1), and compare the magnetic properties of the FeNi films. Figure 4 shows Fe L-edge x-ray absorption and XMCD spectra for 5 ML FeNi films grown on regions, (i), (ii) and (iii). XMCD was measured at the NI configuration, in which the remanent perpendicular magnetic component is detected. The estimated ain from the RHEED analysis is also shown in the figure. In the present case, the FeNi films show perpendicular magnetization when they are grown on (ii) and (iii), while the perpendicular component is not detected at (i). We suppose that the FeNi layers on (i) prefer in-plane magnetization due to the electronic interaction with NiCu alloy, so that it is necessary to insert the Cu spacer layer to break the coupling. The spin magnetic moment, ms, including the magnetic dipole term is estimated from the XMCD sum-rule analysis [20, 21], and is indicated in the figure. The d-hole number, nh, is determined by the white-line intensity after subtracting the step functions, based on the assumption that nh of 5 ML FeNi films on (ii) is nh5ML = 3.4 [22], and that nh is proportional to the white-line intensity. The estimated ms is lower than that of Fe bulk [23] and ∼100 ML L10 FeNi film [11]. We suppose that this is due to systematic errors connected to the procedure of the background and step function subtraction, and to the assumption of nh5ML. Nevertheless, relative comparison of ms among the present samples would be reasonable since the change in nh accompanying the white-line intensity is taken into account, as described above. ms at (ii) is larger than that at (iii), which suggests that the 5 ML FeNi film directly grown on Cu(0 0 1) is not fully magnetized in the perpendicular direction at the remanent state. We suppose that the reduction of PMA causes the domain formation and/or transition to the in-plane magnetization, resulting in the decrease of perpendicular magnetization. In fact,
t Cu (ML) Figure 3. In-plane lattice constant, ain, of Cu(tCu ML)/
NiCu(124 ML)/Cu(0 0 1) as a function of Cu spacer layer thickness, tCu. ain of fcc Cu is shown for comparison. Dotted line is linear fit for tCu = 4–70 ML.
previous result [9]. Thus, ain of NiCu(124 ML)/Cu(0 0 1) is estimated to be 0.253 nm. When the Cu spacer layer is added to NiCu(124 ML)/Cu(0 0 1), ain shows a rapid increase up to tCu = 4 ML, and then shows monotonous increase. Although there is no data between 10 and 70 ML, the data between 4 and 10 ML show a linear increase, and the extrapolated line reaches the data at 70 ML. Thus, we apply a linear fit between 4 and 70 ML. A linear change in ain of the Cu spacer layer is also reported in the previous study [9]. The lattice constant of fcc Cu(0 0 1) is reached if we extrapolate the fitting result up to ∼120 ML. The FeNi films were then grown on Cu(tCu ML)/ NiCu(124 ML)/Cu(0 0 1) and bare Cu(0 0 1) substrates, and the x-ray beam was irradiated at different positions in order to obtain the signal originated from different film configurations with various lattice constants. We here focus on the three distinct regions, (i) NiCu(124 ML)/Cu(0 0 1), (ii) Cu(10 ML)/ 3
M Sakamaki and K Amemiya
J. Phys.: Condens. Matter 26 (2014) 166002
(b)
(a) (i) a in =0.253
m s =1.51
B
(ii) a in =0.254
(iii) a in =0.255
m s =0.96
B
Figure 4. Fe L-edge circularly polarized x-ray absorption spectra (XAS) (a) and XMCD spectra (b) for 5 ML FeNi films grown on three different substrates, (i) NiCu(124 ML)/Cu(0 0 1), (ii) Cu(10 ML)/NiCu(124 ML)/Cu(0 0 1) and (iii) Cu(0 0 1). ain of each substrate indicated in the figure is estimated from the data in figure 3. XAS and XMCD were measured at normal incidence angle configurations. The remanent magnetization state was investigated after applying the magnetic field parallel to the incident x-ray beam. Estimated spin magnetic moment, ms, is indicated in the figure.
several fine domain structures are observed for FeNi bilayers when PMA is reduced [24]. Because we measure the XMCD spectra at the remanent state, there might exist some domain structures. To confirm that the reduction of the perpendicular magnetization indicates the transition toward the in-plane magnetization, Fe L-edge XMCD spectra for n ML FeNi films grown on (ii) are shown in figure 5, in which XMCD was taken at NI and GI configurations. In the case of 3 and 5 ML FeNi films, the XMCD intensity at NI is twice as large as that at GI. This means that the films show perpendicular magnetization because XMCD probes the magnetic moment projected onto the direction of the incident x ray. On the other hand, the 7 ML film shows no XMCD signal at NI, while a clear XMCD signal is observed at GI, which corresponds to the inplane magnetization. These results indicate that the XMCD intensity taken at NI directly reflects the magnitude of PMA. From the previous study, it is found that the magnetic direction of Ni is the same as that of Fe in the present FeNi thickness region [12]. Therefore, only the Fe L-edge measurement was performed in order to avoid the time-dependent changes caused by surface effects such as oxidation. To investigate the relation between the structure and magnetic anisotropy systematically, we plot the XMCD intensity at the Fe L3 peak top for n ML FeNi films as a function of the in-plane lattice constant, ain, in figure 6. XMCD was measured at the NI configuration. The 3 ML FeNi films show perpendicular magnetization, and XMCD intensity is almost the same independent of ain. On the other hand, the 5 ML FeNi film also shows perpendicular magnetization in a wide ain region including (ii) and (iii), but larger XMCD intensity is observed at smaller ain. As observed in figure 4, XMCD was not detected at (i), which indicates in-plane magnetization. In
(ii) a in =0.254
Figure 5. Fe L-edge XMCD spectra for n ML FeNi films grown on Cu(10 ML)/NiCu(124 ML)/Cu(0 0 1), where ain = 0.254. The spectra were taken at normal (NI) and grazing incidence (GI) angle configurations. 3 and 5 ML FeNi films show perpendicular magnetization, whereas 7 ML FeNi film shows in-plane magnetization.
the case of n = 7, FeNi films show almost no XMCD signal regardless of ain, which also indicates in-plane magnetization. Thus, it is concluded that the compressive strain in the inplane direction enhances PMA in the FeNi films. 4
M Sakamaki and K Amemiya
J. Phys.: Condens. Matter 26 (2014) 166002
(i) a in =0.253 (ii) a in =0.254 (iii) a in =0.255
m s =1.26 B m l /m s =0.13
m s =1.25 B m l /m s =0.23
m s =1.30 B m l /m s =0.15
a in (nm) Figure 6. XMCD intensity at Fe L3 peak top for n ML FeNi films as a function of in-plane lattice constant, ain. ain is obtained by RHEED analysis. Three different substrates, (i) NiCu(124 ML)/ Cu(0 0 1), (ii) Cu(10 ML)/NiCu(124 ML)/Cu(0 0 1) and (iii) Cu(0 0 1) are indicated in the figure. XMCD was measured at normal incidence angle configuration. The remanent magnetization state was investigated after applying the magnetic field parallel to the incident x-ray beam.
Then, we estimate the uniaxial magnetic anisotropy constant, Ku, with the help of a theoretical calculation. If we assume that the volume of the unit cell is the same as that of bulk L10 FeNi, the ratio of the lattice parameters, c/a, is estimated to be 0.99 and 0.97 for the FeNi films on (ii) and (iii), respectively, whereas that of bulk L10 FeNi is reported to be 1.0036 [25]. Ku at c/a = 0.99 and 0.97 are estimated to be 0.50 and 0.42 MJ m−3, respectively, from the first-principle calculations for L10 type FeNi [26]. The difference in the estimated Ku between (ii) and (iii) can explain the observed change in PMA. Finally, we discuss the detailed spectral structure and obtained magnetic moment. Figure 7 shows Fe L-edge XMCD for 3 ML FeNi films at three different regions, (i), (ii) and (iii). It is reasonable to compare the 3 ML FeNi films on various substrates because all these films are fully magnetized in the perpendicular direction, as shown in figure 6. The estimated spin magnetic moment is almost the same among three samples, whereas the ratio of orbital and spin magnetic moments, ml/ms, shows a distinct difference, and a larger value is obtained at (ii). In addition, the spectral shape between the L3 and L2 edges at (ii) seems different from that at (i) and (iii), i.e., the XMCD curve in this region shows a positive signal at (i) and (iii), while it shows almost zero at (ii). In the literature, a small negative XMCD signal between the L3 and L2 edges is observed for bcc Fe, fcc Ni and fcc Co thin films, and is explained as being due to the diffuse magnetic moment, which is originated from the polarized 4s electrons [27, 28]. Normally, bulk Fe shows a positive shoulder above the L3 edge, and the signal approaches zero before the L2 peak, as observed in the present case of (i) and (iii). Thus, we suppose that the observed spectral change at (ii) is due to the diffuse magnetic moment, and the compressive strain induces the diffuse magnetism. Indeed, it is reported from the first-principles calculations for L10 type FeNi that only the Fe components of the orbital density of states exist at the Fermi level [29], and the Fe d x 2 − y2 and dyz, zx orbitals delocalize near the Fermi level by the
Figure 7. Fe L-edge XMCD spectra for 3 ML FeNi films grown on (i) NiCu(124 ML)/Cu(0 0 1), (ii) Cu(10 ML)/NiCu(124 ML)/ Cu(0 0 1) and (iii) Cu(0 0 1), taken at normal incidence angle configurations. The remanent magnetization state was investigated. Estimated spin and orbital magnetic moments, ms and ml, are indicated in the figure. Inset shows enlarged view of the region between the L3 and L2 edges.
compressive strain [26]. We suppose that the delocalization of the in-plane orbitals enhances both the diffuse magnetism and the perpendicular orbital magnetic moment, resulting in the enhancement of PMA in the FeNi films. 4. Summary We have performed a systematic study on the effect of the lattice strain on magnetic anisotropy of alternately layered FeNi ultrathin films grown on the substrate Cu(tCu = 0–70 ML)/ Ni48Cu52(124 ML)/Cu(0 0 1) single crystal, by means of in situ XMCD and RHEED analyses. From the RHEED analysis, the in-plane lattice constant, ain, of the substrate was found to shrink by 0.8% and 0.5% at tCu = 0 and 10 ML as compared to that of bulk Cu, respectively, so that continuous modulation of ain was confirmed. Fe L-edge XMCD was measured for n ML FeNi films grown on various ain, and PMA is observed at n = 3 and 5, whereas the film with n = 7 shows in-plane magnetic anisotropy. Moreover, we revealed that PMA is enhanced with decreasing ain, in the case where a Cu spacer layer is inserted. At the same time, the enhancement of the perpendicular magnetic moment and the diffuse magnetic moment are observed with decreasing ain. We suppose that magnetic anisotropy in the FeNi films is mainly carried by Fe, and the delocalization of the in-plane orbitals near the Fermi level increases the 5
M Sakamaki and K Amemiya
J. Phys.: Condens. Matter 26 (2014) 166002
[9] Lauhoff G, Vaz C A F, Bland J A C, Lee J and Suzuki T 2000 Phys. Rev. B 61 6805 [10] Sakamaki M and Amemiya K 2011 Appl. Phys. Express 4 073002 [11] Kotsugi M et al 2013 J. Magn. Magn. Mater. 326 235 [12] Sakamaki M and Amemiya K 2013 Phys. Rev. B 87 014428 [13] O'Brien W L, Droubay T and Tonner B P 1996 Phys. Rev. B 54 9297 [14] Matthews J W 1970 Thin Solid Films 5 187 [15] Schulz B and Baberschke K 1994 Phys. Rev. B 50 13467 [16] Amemiya K, Kondoh H, Yokoyama T and Ohta T 2002 J. Electron Spectrosc. Relat. Phenom. 124 151 [17] Amemiya K, Toyoshima A, Kikuchi T, Kosuge T, Nigorikawa K, Sumii R and Ito K 2010 AIP Conf. Proc. 1234 295 [18] Amemiya K et al 2013 J. Phys.: Conf. Ser. 425 152015 [19] Tsuchiya K, Shioya T, Aoto T, Harada K, Obina T, Sakamaki M and Amemiya K 2013 J. Phys.: Conf. Ser. 425 132017 [20] Thole B T, Carra P, Sette F and van der Laan G 1992 Phys. Rev. Lett. 68 1943 [21] Carra P, Thole B T, Altarelli M and Wang X 1993 Phys. Rev. Lett. 70 694 [22] Eguchi K, Takagi Y, Nakagawa T and Yokoyama T 2013 J. Phys.: Conf. Ser. 430 012129 [23] Cullity B D 1972 Introduction to Magnetic Materials (Reading, MA: Addison-Wesley) [24] Ramchal R, Schmid A K, Farle M and Poppa H 2004 Phys. Rev. B 69 214401 [25] Albertsen J F 1981 Phys. Scr. 23 301 [26] Miura Y, Ozaki S, Kuwahara Y, Tsujikawa M, Abe K and Shirai M 2013 J. Phys.: Condens. Matter 25 106005 [27] O'Brien W L and Tonner B P 1994 Phys. Rev. B 50 12672 [28] Le Cann X, Boeglin C, Carriére B and Hricovini K 1996 Phys. Rev. B 54 373 [29] Ravindran P, Kjekshus A, Fjellvåg H, James P, Nordström L, Johansson B and Eriksson O 2001 Phys. Rev. B 63 144409
perpendicular orbital magnetic moment, which results in the enhancement of PMA. Acknowledgments The present work has been performed under the approval of the Photon Factory Program Advisory Committee (no. 2010S2-001). The authors are grateful for the financial support of the Quantum Beam Technology Program from the Ministry of Education, Culture, Sports, Science and Technology (MEXT). References [1] Ovanov O A, Solina L V and Demshina V A 1973 Phys. Met. Metallogr. 35 81 [2] Shima T, Okamura M, Mitani S and Takanashi K 2007 J. Magn. Magn. Mater. 310 2213 [3] Mizuguchi M, Sekiya S and Takanashi K 2010 J. Appl. Phys. 107 09A716 [4] Kojima T, Mizuguchi M and Takanashi K 2011 J. Phys.: Conf. Ser. 266 012119 [5] Amemiya K, Sakai E, Matsumura D, Abe H, Ohta T and Yokoyama T 2005 Phys. Rev. B 71 214420 [6] Abe H, Amemiya K, Matsumura D, Kitagawa S, Watanabe H, Yokoyama T and Ohta T 2006 J. Magn. Magn. Mater. 302 86 [7] Sakamaki M and Amemiya K 2011 J. Phys.: Conf. Ser. 266 012020 [8] Cherifi S, Stanescu S, Mocuta C, Deville J-P, Boeglin C, Ohresser P and Brookes N B 2001 Surf. Sci. 482–85 1056
6
