Co bilayers: influence of antiferromagnetic grains and mechanisms

Applied Surface Science 405 (2017) 316–320

Contents lists available at ScienceDirect

Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc

Full Length Article

Enhanced exchange bias fields for CoO/Co bilayers: influence of antiferromagnetic grains and mechanisms Cheng-Hsun-Tony Chang a , Shin-Chen Chang a , Jyh-Shen Tsay a,∗ , Yeong-Der Yao b a b

Department of Physics, National Taiwan Normal University, Taipei 116, Taiwan Institute of Physics, Academia Sinica, Nankang, Taipei 11529, Taiwan

a r t i c l e

a b s t r a c t

i n f o

Article history: Received 30 September 2016 Received in revised form 6 January 2017 Accepted 1 February 2017 Available online 10 February 2017 Keywords: Magnetic bilayer Exchange bias field Blocking temperature Magnetic thin films Oxide

The emergence and optimization of devices that can be applied to spintronics have attracted considerable interest, and both experimental and theoretical approaches have been used in studies of exchange bias phenomena. A survey of the literature indicates that great efforts have been devoted to improving exchange bias fields, while only limited attempts have been made to controll the temperature dependence of exchange bias. In this study, the influence of antiferromagnetic grains on exchange bias phenomena in CoO/Co bilayers on a semiconductor surface was investigated. Based on an antiferromagnetic grain model, a correlation between grain size, grain density, blocking temperature, and the exchange bias field was established. For crystallites with a smaller median diameter, the dependence of the thickness of the CoO layer on blocking temperature showed a less pronounced variation. This is due to the larger thermal agitation of the atomic spin moments in the grain, which causes a weaker exchange coupling between atomic spin moments. The enhanced density of antiferromagnetic/ferromagnetic pinning sites resulting from an increased grain density is responsible for the enhancement in the exchange bias fields. The results reported herein provide insights into our knowledge related to controlling the temperature dependence of exchange bias and related mechanisms. © 2017 Elsevier B.V. All rights reserved.

1. Introduction The emergence and optimization of devices that can be applied to spintronics have attracted considerable interest, in both experimental and theoretical approaches in studies of exchange bias (EB) phenomena in recent years [1–26]. Antiferromagnetic (AFM)/ferromagnetic (FM) stacks can function as reference layers that are used in magnetic spin-valve read-heads and magnetic random access memories (MRAMs) [11–13]. EB are sensitive to AFM/FM interfacial conditions via short range exchange coupling so that interfacial roughness, chemical order, grain boundaries, crystallographic order, and micromagnetic effects can all have an influence on performance [14–19]. A survey of the literature indicates that great efforts have been devoted to improving the exchange bias field (HE ) [7,8,18–20]. For instance, after an annealing treatment, microstructural changes such as interface morphology and FeMn chemical composition at the local FeMn/Py interface appear to play a key role in enhancing HE [18]. For CoO on a thicker Co/Ge(100) layer, pinned magnetic moments are increased, result-

∗ Corresponding author. E-mail address: [email protected] (J.-S. Tsay). http://dx.doi.org/10.1016/j.apsusc.2017.02.001 0169-4332/© 2017 Elsevier B.V. All rights reserved.

ing in the formation of a thicker interfacial region, resulting in a larger HE [7]. The nature of the CoO/Fe interface affects EB phenomena, while an over-oxidized interface results in the highest HE [8]. Owing to the areas of larger interface lateral grains, enhanced AFM/FM coupling can cause a large increase in HE with increasing thickness of the Ta buffer layer [19]. In some cases, such as thermally assisted-MRAM (TA-MRAM), well-defined temperature variations in both HE and a coercive force (HC ) are required, in addition to a large HE [21–23]. However, our knowledge related to controlling the temperature dependence of EB is relatively limited [23–26]. CoO exhibits a high magnetocrystalline anisotropy and has an easily accessible Néel temperature (TN = 290 K) [6,24]. Cobalt is a key element and is widely used in magnetic recording media. In our previous reports, we investigated compound formation, microscopic structures, and the initial oxidation that occurs at metal/semiconductor interfaces [27–31]. Interfacial effects cause the axis of magnetization of a Co/Si interface to be easily canted out of plane while the introduction of a silver layer causes the transfer of Si atom interactions and in-plane anisotropy [27,28]. After the exposure of a Co/Ge layer to oxygen, the interplay of oxygen atoms and the interface results in the enhancement of HC [29]. For CoO prepared on Co/Ge(100), the width of the inter-

C.-H.-T. Chang et al. / Applied Surface Science 405 (2017) 316–320

The preparation and the characterization of all specimens were carried out in an ultrahigh vacuum chamber (UHV) with a base pressure of approximately 2×10−10 Torr. The UHV chamber was equipped with facilities for Auger electron spectroscopy (AES), reflection high energy electron diffraction (RHEED), and magnetooptical Kerr effect (MOKE) measurements. By combining argon ion sputtering with AES, it is possible to perform depth-profiling measurements. The equipment components have been described in detail elsewhere [7,27–31]. The Ge(100) substrate was prepared by cycles of Ar+ ion bombardment followed by annealing treatments at 1100 K. The cleanliness of the surface was checked by AES. Co atoms were evaporated from a high purity (99.997%), resistively heated cobalt coil. After exposing the Co/Ge ultrathin films to oxygen, the oxygen was found to be distributed only in the topmost layer and no evidence was found to the formation of an AFM CoO layer [29,30]. To increase the amount of oxygen in the films so as to form CoO overlayers, our strategy was to evaporate Co atoms in an oxygen atmosphere where the gas flow could be accurately controlled by a leak valve [7,31,32]. Experimental evidences from the AES and depth profiling measurements show that the preparation of CoO layers with a concentration ratio of Co and O close to 1:1 can be achieved by evaporating Co atoms in an oxygen atmosphere (see Electronic supplementary information). From Auger signal versus time (AS-t) plot, it is possible to determine the rate of deposition of CoO with thicknesses within the probing depth of AES. Various thicknesses of CoO were obtained by controlling the deposition time. Thicknesses can be further checked by using a thickness monitor and depth profiling measurements. Experimental details for these procedures can be found in References 7, 31, and 32. 3. Results and discussion The objective of this study was to carry out systematic investigations for CoO/Co/Ge(100). In initial experiments, we deposited CoO with a certain thickness on pre-prepared Co/Ge(100). In situ experimental measurements, i.e., AES, RHEED, and MOKE, were

RHEED intensities (arb. units)

FWHM (arb. units)

1.0

0.5

0.0 0

10

20

30

40

CoO thickness (ML) Fig. 1. Intensities and FWHMs of the RHEED spots versus tCoO for CoO/Co/Ge(100).

40 ML CoO/Co/Ge(100) Kerr signals (arb. units)

2. Experimental

CoO/Co/Ge(100) tCo = 25 ML tCo = 20 ML

tCo=25 ML

tCo=20 ML 170 K

300 K

(a) -2.0

0.0

2.0

-2.0

0.0

2.0

Magnetic field (kOe) 0.6

tCo = 25 ML tCo = 20 ML

(b)

0.4

HE (kOe)

facial region depends on the preparation procedure, while the pinned magnetic moments in the interfacial region are responsible for the coincidence for HE and the change of Kerr rotations [7]. For CoO/Co/Ge(100) at different thicknesses of AFM layer, the monotonous increase in the blocking temperature (TB ) is related to the increasing thermal stability of the AFM grains [32]. CoO (AFM)/Co (FM)/Ge has been used as a prototype for investigating EB phenomena. From the literature, EB systems with very thin or polycrystalline AFM layers tend to have TB values smaller than TN because of the reduced AFM anisotropy of smaller grain sizes [5,6]. For monodispersive grains, Monte Carlo simulations indicate that below TB , AFM grains become sufficiently unstable previous to the occurrence of magnetic switching [25]. When the size of bimagnetic core/shell nanoparticles is reduced, the energy barrier associated to the magnetocrystalline anisotropy decreases, and the system rapidly loses its thermal stability [26]. However, the mechanism responsible for the dependence of explaining AFM thickness on both HE and TB is unclear. In this paper, an AFM grain model in which HE and TB are connected with AFM grain sizes and grain densities was successfully established. By reducing the volume of the AFM grains, HE is enhanced due to the fact that the density of the AFM/FM pinning sites is increased, while the dependence of TB on CoO thickness showed a less pronounced variation. These observations provide valuable information for future applications of ultrathin CoO/Co bilayers on semiconductor substrates.

317

0.2

0.0 160

180

200

220

240

Temperature (K) Fig. 2. (a) Kerr signals versus temperature for CoO/Co/Ge(100) at 300 K and 170 K. (b) Exchange bias fields versus temperature for CoO/Co/Ge(100).

then carried out. Both TB and HE can be determined from lowtemperature MOKE data. Using different thicknesses of CoO, we repeated the measurements to obtain the CoO thickness dependent AES (Fig. S1), RHEED (Fig. 1), TB (Fig. 3a), and HE (Fig. 4a). To explain the trend for TB (Fig. 3a), we established an AFM grain model by considering the AFM grain size (Fig. 3b). To explain the trend in HE (Fig. 4a), we further considered the AFM grain density in the AFM grain model (Fig. 4b). After preparing CoO overlayers on Co/Ge(100), the surface structures were investigated using a RHEED technique. Fig. 1 shows the RHEED intensities and full widths at half maximum (FWHM) versus tCoO for CoO/Co/Ge(100). Under grazing incidence conditions, both the RHEED intensities and the FWHM are sensitive to the ordered/disordered structure of the top MLs of CoO. For CoO/Co/Ge(100) with tCo = 25 ML, the RHEED intensity increases to a maximum value for tCoO between 20 and 25 ML as tCoO increases. The initial enhancement in RHEED intensity indicates that ordered CoO overlayers are formed. RHEED intensity was reduced for CoO layers that are thicker than 25 ML, which is due to defect forma-

318

C.-H.-T. Chang et al. / Applied Surface Science 405 (2017) 316–320

crystallite size for producing the diffraction spot can be expressed by d = K/Bcos B

Fig. 3. (a) Blocking temperatures versus tCoO for CoO/Co/Ge(100). (b) A model of the reduced TB for smaller CoO grains in a CoO/Co EB system.

where d is the average crystallite size; K is the Scherrer constant with a value of 0.9;  is the wavelength; B is the FWHM of the Bragg peak; and  B is the Bragg angle [35,37]. From Eq. (1), the average crystallite size d is inversely proportional to the FWHM and this implies that a larger FWHM corresponds to smaller grains. In Fig. 1 for CoO on 25 ML Co/Ge(100), as tCoO increases, the FWHM of the RHEED spot decreases to a minimum value between 20 and 25 ML CoO, followed by an increase. This indicates the CoO is nucleated to form larger grains in cases of a CoO layer thinner than 20 ML while for a thicker CoO layer, defect formation in the films causes the formation of smaller grains in the surface layers. For the case of CoO/Co/Ge(100) with tCo = 20 ML, the trends for both the RHEED intensity and FWHM were similar to those for tCo = 25 ML. However, the RHEED intensity was slightly larger while the FWHM was smaller for CoO/Co/Ge(100) for tCo = 20 ML. A larger RHEED intensity is indicative of a more ordered state for the CoO overlayers, while a smaller FWHM is indicative of larger grain sizes for CoO/Co/Ge(100) for tCo = 20 ML, as compared to that for tCo = 25 ML. The finding of a larger grain size for thinner Co films is the cornerstone of the proposed model for explaining the relation between TB and grain sizes. The magnetic properties of the CoO/Co/Ge(100) were systematically investigated using a MOKE technique. Fig. 2(a) shows Kerr signals versus magnetic field at different temperatures. At 300 K, the hysteresis loops for CoO/Co/Ge(100) are symmetrical. No EB occurs in the case of a T > TN . Under conditions of cooling in a magnetic field of +2 kOe to low temperatures, the shifted hysteresis loop and the enhanced HC characterize the EB phenomena of the CoO/Co/Ge(100) system. Fig. 2(b) shows the values for HE versus temperature for 40 ML CoO/Co/Ge(100). For the tCo = 25 ML in Fig. 2(b), HE is zero at temperatures above 213 K and then increases from 0.04 to 0.38 kOe as the sample temperature decreases from 206 to 170 K. The TB for 40 ML CoO/25 ML Co/Ge(100) was determined to be 213 K. For tCo = 20 ML in Fig. 2(b), the evolution for HE is similar to that for tCo = 25 ML while the TB is larger. In the following paragraphs, we focus on the establishment of a model that connects TB and HE with the grain size and density. Fig. 3(a) shows the TB versus the thickness of the CoO layer for CoO/Co/Ge(100). As tCoO increases from 15 to 40 ML for CoO/Co/Ge(100) with tCo = 25 ML, TB increases from 198 to 213 K. For very thin or polycrystalline AFM layers, TB is smaller than TN [5,6]. To reach TB close to TN , previous investigations on EB show that the CoO thickness should be at least about 10 nm [9,38]. This effect is related to grain size and the thickness of the AFM layer through finite size effects [6,10]. The relation between TB and the magnetic anisotropy energy of the AFM layer KAFM can be expressed by K AFM V m = kB ln(f 0 )

Fig. 4. (a) Exchange bias fields versus tCoO for CoO/Co/Ge(100) measured at 170 K. (b) An AFM grain model of the enhanced HE and smaller variance of TB for smaller CoO grains in the CoO/Co EB system.

tion at the surface layers. From the kinematic theory of diffraction, the FWHM of a diffraction spot is approximately inversely proportional to the mean diameter of the periodic structure [33,34]. The theory of the x-ray scattering maxima broadening by small crystallites can be applied in the case of electron scattering [35,36]. From Scherrer’s formula, the relation between the FWHM and average

(1)

(2)

where Vm is the median grain volume; kB is Boltzmann constant;  is the relaxation time; f0 is taken to be 109 s−1 [10,39]. Dm is the median diameter and is proportional to the average crystallite size d in Eq. (1). By substituting the median grain volume Vm with ␲Dm 2 ·tAFM /4 in Eq. (2), the relation between TB and the thickness of the AFM layer tAFM can be expressed by = [K AFM ␲Dm 2 /4kB ln(f 0 )]t AFM

(3)

The hypothesis that grain volume is proportional to AF film thickness is valid up to a threshold thickness of the AFM layer. By increasing tAFM , the blocking temperature increases linearly due to the increase in the volume of the AFM grains (␲Dm 2 ·tAFM /4) [10,39].

C.-H.-T. Chang et al. / Applied Surface Science 405 (2017) 316–320 Table 1 The exchange bias field (HE ), the slope ˇ of the TB versus tCoO plot and the resultant ˇ·HE values for different tCo .

319

tCo (ML)

HE (kOe)

ˇ (slope)

HE ·tCo ·ˇ3/2

FM layer [5,6]. From the modified equation concerning the HE and pinned interfacial magnetizations [7,38,40], the relation between HE and the nominal thickness  of the pinned uncompensated layer in fractions of a ML is

25 20

0.36 0.14

0.58 ± 0.03 1.3 ± 0.1

3.98 ± 0.21 4.15 ± 0.32

HE =  · J · SAFM SFM /a2AFM MFM tFM

The plot of TB versus tCoO in Fig. 3(a) is consistent with this inference. For CoO/Co/Ge(100) with tCo = 25 ML, TB increases much more slowly as compared to that with tCo = 20 ML. Although the KAFM often increases with increasing tAFM due to changes in crystallization at the grain boundaries [10,39], crystallite ordering is similar for CoO on 25 and 20 ML Co/Ge(100) as inferred from the similar RHEED intensities in Fig. 1. However, a larger FWHM was found for CoO on 25 ML Co/Ge(100) as compared to that for CoO on 20 ML Co/Ge(100). This implies grain sizes as well as the median diameters Dm of the crystallites are smaller in the specimen for CoO on 25 ML Co/Ge(100). According to Equation (3), the slope ˇ of a plot of TB versus tCoO is proportional to Dm 2 , i.e., ˇ∝Dm 2

(4)

From linear fitting of the experimental data in Fig. 3(a), the slopes ˇ were determined to be 0.58 ± 0.03 and 1.3 ± 0.1 for tCo = 25 and 20 ML, respectively. The values of ˇ are summarized in Table 1. Based on the reduced variance of the TB for smaller CoO grains in the CoO/Co EB system, a model that takes the sizes of the CoO grains into consideration was established. This model is schematically illustrated in Fig. 3(b). For smaller CoO grains, the thermal agitation of the atomic spin moments in the AFM grain is larger and results in a weaker exchange coupling between atomic spin moments at the AFM/FM interface. This effect serves as the physical reason for the less pronounced variance of the TB for CoO/Co/Ge(100) with tCo = 25 ML. Under conditions of cooling in a magnetic field of +2 kOe to 170 K, HE was recorded for CoO/Co/Ge(100) layers with different CoO thicknesses. For comparative purposes, Fig. 4(a) shows the evolution of HE versus tCoO with both the tCo = 25 ML and tCo = 20 ML. As tCoO increases for CoO/Co/Ge(100) with tCo = 25 ML, HE is zero when CoO is thinner than 5 ML and reaches a nonzero value at 10 ML. This shows that the onset of the EB phenomenon is between 5 and 10 ML. The onset of the EB phenomenon has been reported to be influenced by finite size effects, crystallinity, and interfacial conditions [5,8,31]. By increasing the tCoO to 25 ML, HE was increased to a maximum value of 0.46 kOe. The increase in HE can be attributed to the increasing anisotropy energy of the AFM layer KAFM and the increasing number of pinned uncompensated spins at the interface [7,31]. For the case of a thicker CoO, KAFM is somewhat stable [10,39]. The reduction in HE for a CoO thicker than 25 ML can be attributed to the increase in the size of the AFM domain, which is often observed in continuous AFM/FM layers [6,31]. For thinner CoO layers, the interfacial conditions are not as stable as compared to that for thicker CoO. HE versus tCoO plots for CoO/Co/Ge(100) with different tCo values are compared in Fig. 4(a), showing that the HE values are stable for tCoO values that are larger than approximately 30 ML. In the following discussion for the exchange bias field, we focus on thicker CoO that is in a more stable state. From Fig. 4(a), the HE for tCo = 25 ML is larger than that for tCo = 20 ML. By assuming a coherent rotation of the magnetization of an EB system and not taking into account the lateral magnetic structure of the layers, the AFM layer related HE can be expressed as HE ∝



KAFM AAFM /MFM tFM

(5)

where AAFM is the exchange stiffness of the AFM; MFM is the saturation magnetization of the FM layer; and tFM is the thickness of the

(6)

where J is the interface exchange energy; SAFM and SFM are the magnetic moments of the AFM and FM, respectively; aAFM is the size of the unit cell of the AFM. According to a previous report [7] and the data shown in Fig. 1, the nominal thickness , increases while the median diameter Dm of the crystallites in the specimen become reduced with increasing CoO thickness. We assume that the same interfacial area ·Dm 2 is influenced in the process of increasing CoO thickness. The relationship between the nominal thickness  and grain size can be given by ∝Dm −2

(7)

and shows that the density of the AFM/FM pinning sites is enhanced by smaller grains. Considering the variation in grain density from Eq. (7) and the uncompensated spin correlation of the ferromagnetic layer from a previous report [41], Equation (6) can be rewritten as H E ∝/(t FM ·Dm )

(8)

In additional, by increasing tAFM , the blocking temperature increases linearly as the result of an increase in the volume of the AFM grains (␲Dm 2 ·tAFM /4) [10,39]. Combining Eqs. (4) and (7), the exchange bias field HE in Eq. (8) can be rewritten as H E ∝t FM −1 ·ˇ−3/2

(9)

By substituting the values for ˇ, tFM (=tCo ) and HE , the resultant HE ·tCo ·ˇ3/2 values for CoO/Co/Ge(100) with tCo = 25 and 20 ML are nearly identical, as shown in Table 1. This result shows that by combining the effects of the grain density and grain size, the effect is suitable for explaining why the HE for tCo = 25 ML is larger than that of tCo = 20 ML. Based on Eqs. (3) and (9), an AFM grain model that takes grain size as well as grain density into consideration is proposed to explain the reduced variance of TB and the enhanced HE for CoO/Co EB system with smaller CoO grains as schematically illustrated in Fig. 4(b). For smaller CoO grains, the thermal agitation of the atomic spin moments in the AFM grain is larger, showing a weaker exchange coupling. This is responsible for the reduced variance in TB in Fig. 3(a). However, the density of the AFM/FM pinning sites is enhanced for smaller grains, based on the assumption that the influenced interfacial volume remains constant. The enhanced density of AFM/FM pinning sites is responsible for the enhancements in HE in Fig. 4(a) for a CoO/Co EB system. From a technological point of view, the optimization of AFM grain sizes and grain densities is of importance for controlling variance of TB and the enhancements of HE . The results provide the insights into our knowledge related to controlling the temperature dependence of exchange bias and related mechanisms. 6. Conclusion As tCoO increases for CoO/Co/Ge(100), the enhancement in magnetic anisotropy KAFM causes the value for HE to increase for CoO layers thinner than 25 ML while the increase in the AFM domain size results in a reduction of HE for thicker CoO layers. By increasing the value of tAFM , the value for TB also increases owing to the increasing volume of the AFM grains and their enhanced thermal agitation of the atomic spin moments. For CoO on thicker Co/Ge(100) layers, a larger FWHM is evident in the case of smaller grain sizes. By considering grain size as well as grain density, an AFM grain model

320

C.-H.-T. Chang et al. / Applied Surface Science 405 (2017) 316–320

could be established that explains the reduced variance of TB and the enhanced HE for a CoO/Co EB system with smaller CoO grains. When the median diameter Dm of the crystallites in the specimen is smaller, the dependence of CoO thickness on TB shows a less pronounced variation due to the larger thermal agitation of the atomic spin moments in the AFM grain. The enhanced density of AFM/FM pinning sites resulting from an increased grain density is responsible for the enhancement in the HE . These observations provide valuable information for future applications in spintronics for ultrathin CoO/Co bilayers on semiconductor substrates. Acknowledgments The authors wish to acknowledge support from Ministry of Science and Technology of Taiwan under Contract Nos. MOST 1032112-M-003-006 and MOST 104-2112-M-003-004. We appreciate Dr. Milton S. Feather for the English editing of the manuscript. This article was subsidized by the National Taiwan Normal University (NTNU), Taiwan. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.apsusc.2017.02. 001. References [1] Y. Fan, K.J. Smith, G. Luepke, A.T. Hanbicki, R. Goswami, C.H. Li, H.B. Zhao, B.T. Jonker, Nature Nanotechnol. 8 (2013) 438. [2] S. Ouazi, S. Vlaic, S. Rusponi, G. Moulas, P. Buluschek, K. Halleux, S. Bornemann, S. Mankovsky, J. Minar, J.B. Staunton, H. Ebert, H. Brune, Nature Commun. 3 (2012) 1313. [3] P.K. Manna, S.M. Yusuf, Phys. Rep. 535 (2014) 61. [4] F. Matsukura, Y. Tokura, H. Ohno, Nature Nanotechnol. 10 (2015) 209. [5] J. Nogués, J. Sort, V. Langlais, V. Skumryev, S. Surinach, J.S. Munoz, M.D. Baró, Phys. Rep. 422 (2005) 65. [6] J. Nogués, I.K. Schuller, J. Magn. Magn. Mater. 192 (1999) 203. [7] S.C. Chang, J.S. Tsay, C.H.T. Chang, Y.D. Yao, Appl. Surf. Sci. 354 (2015) 95. ´ ´ ´ [8] E. Młynczak, J. Gurgul, J. Przewoznik, D. Wilgocka-Slezak, K. Freindl, N. Spiridis, J. Korecki, Appl. Surf. Sci. 304 (2014) 86. [9] A.D. Lamirand, M.M. Soares, A.Y. Ramos, H.C.N. Tolentino, M.D. Santis, J.C. Cezar, A.D. Siervo, M. Jamet, Phys. Rev. B 88 (2013) 140401(R). [10] K. O’Grady, L.E. Fernandez-Outon, G. Vallejo-Fernandez, J. Magn. Magn. Mater. 322 (2010) 883. [11] H.D. Gan, R. Malmhall, Z.H. Wang, B.K. Yen, J. Zhang, X.B. Wang, Y.C. Zhou, X.J. Hao, D.H. Jung, K. Satoh, Y.M. Huai, Appl. Phys. Lett. 105 (2014) 192403.

[12] G. Gubbiotti, S. Tacchi, L.D. Bianco, E. Bonfiglioli, L. Giovannini, M. Tamisari, F. Spizzo, R. Zivieri, J. Appl. Phys. 117 (2015) 17D150. [13] K. Akmaldinov, C. Ducruet, C. Portemont, I. Joumard, I.L. Prejbeanu, B. Dieny, V. Baltz, J. Appl. Phys. 115 (2014) 17B718. [14] A. Kohn, A. Kovacs, R. Fan, G.J. McIntyre, R.C.C. Ward, J.P. Goff, Sci. Rep. 3 (2013) 2412. [15] D.L. Cortie, A.G. Biternas, R.W. Chantrel, X.L. Wang, F. Klose, Appl. Phys. Lett. 105 (2014) 032402. ´ [16] E. Młynczak, B. Matlak, A. Kozioł-Rachwał, J. Gurgul, N. Spiridis, J. Korecki, Phys. Rev. B 88 (2013) 085442. [17] T. Saerbeck, H. Zhu, D. Lott, H. Lee, P.R. LeClair, G.J. Mankey, A.P.J. Stampfl, F. Klose, J. Appl. Phys. 114 (2013) 013901. [18] K.Y. Kim, H.C. Choi, S.Y. Jo, C.Y. You, J. Appl. Phys. 114 (2013) 073908. [19] J.H. Hsu, A.C. Sun, P. Sharma, Thin Solid Films 542 (2013) 87. [20] C.J. Chen, R.K. Chiang, S. Kamali, S.L. Wang, Nanoscale 7 (2015) 14332. [21] I.L. Prejbeanu, S. Bandiera, J. Alvarez-Herault, R.C. Sousa, B. Dieny, J.P. Nozieres, J. Phys. D: Appl. Phys. 46 (2013) 074002. [22] I.L. Prejbeanu, M. Kerekes, R.C. Sousa, H. Sibuet, O. Redon, B. Dieny, J.P. Nozieres, J. Phys. Conden. Matter 19 (2007) 165218. [23] G. Vinai, J. Moritz, S. Bandiera, I.L. Prejbeanu, B. Dieny, Appl. Phys. Lett. 104 (2014) 162401. ˜ V. Skumryev, J. Sort, D. Givord, J. Nogués, [24] J.A.D. Toro, D.P. Marques, P. Muniz, Phys. Rev. Lett. 115 (2015) 057201. [25] D. Ledue, A. Maitre, F. Barbe, L. Lechevallier, J. Magn. Magn. Mater. 372 (2014) 134. [26] G.C. Lavorato, E. Lima Jr., D. Tobia, D. Fiorani, H.E. Troiani, R.D. Zysler, E.L. Winkler, Nanotechnology 25 (2014) 355704. [27] J.S. Tsay, T.Y. Fu, M.H. Lin, C.S. Yang, Y.D. Yao, Appl. Phys. Lett. 88 (2006) 102506. [28] C.H.T. Chang, T.Y. Fu, J.S. Tsay, J. Appl. Phys. 117 (2015) 17B733. [29] H.W. Chang, J.S. Tsay, Y.L. Chiou, K.T. Huang, W.Y. Chan, Y.D. Yao, J. Magn. Magn. Mater. 310 (2007) E741. [30] H.W. Chang, J.S. Tsay, Y.L. Chiou, K.T. Huang, W.Y. Chan, Y.D. Yao, J. Appl. Phys. 99 (2006) 08J705. [31] H.W. Chang, J.S. Tsay, S.C. Chang, P.C. Chuang, Y.D. Yao, J. Alloys Compd. 562 (2013) 69. [32] C.H.T. Chang, S.C. Chang, J.S. Tsay, Y.D. Yao, AIP Adv. 6 (2016) 056101. [33] G. Ertl, J. Küppers, Low Energy Electrons and Surface Chemistry, second ed., VCH, Weiheim, 1985. [34] A.S. Arrott, in: J.A.C. Bland, B. Heinrich (Eds.), Ultrathin Magnetic Structures I, Springer, Berlin, 1994. [35] N.N. Berchenko, M.S. Fruginskii, I.V. Kurilo, I.O. Rudyj, I.S. Virt, Surf. Interface Anal. 40 (2008) 641. [36] Z.G. Pinsker, Electron Diffraction, Butterworths Scientific Publications, London, 1953. [37] J.M. Raitano, S. Khalid, N. Marinkovic, S.W. Chan, J. Alloys Compd. 644 (2015) 996. [38] A.E. Berkowitz, K. Takano, J. Magn. Magn. Mater. 200 (1999) 552. [39] G. Vallejo-Fernandez, L.E. Fernandez-Outon, K. O’Grady, Appl. Phys. Lett. 91 (2007) 212503. [40] H. Ohldag, A. Scholl, F. Nolting, E. Arenholz, S. Maat, A.T. Young, M. Carey, J. Stöhr, Phys. Rev. Lett. 91 (2003) 017203. [41] K. Takano, R.H. Kodama, A.E. Berkowitz, Phys. Rev. Lett. 79 (1997) 1130.