Đề tài Study of perpendicular exchange bias mechanism in mnpd/co multilayers

MINISTRY OF ND TRAINING HANOI UNIVERSITY OF TECHNOLOGY INTERNATIONAL TRAINING INSTITUTE FOR MATERIALS SCIENCE (ITIMS) MASTER THESIS OF MATERIALS SCIENCE STUDY OF PERPENDICULAR EXCHANGE BIAS MECHANISM IN MnPd/Co MULTILAYERS NGUYEN HUU DZUNG Supervisor: Prof. D.Sc. Nguyen Phu Thuy Hanoi – 2007 EDUCATION A ii HANOI UNIVERSITY OF TECHNOLOGY INTERNATIONAL TRAINING INSTITUTE FOR MATERIALS SCIENCE (ITIMS) Batch ITIMS – 2005 Title of MSc Thesis: Study of perpendicular exchange bias mechanism in MnPd/Co multilayers Author: Nguyen Huu Dzung Supervisor: Prof. D.Sc. Nguyen Phu Thuy Referees: 1. Dr. Nguyen Thang Long 2. Dr. Nguyen Phuc Duong Abstract The multilayers of [MnPd/Co]10 have been investigated for the first time. The results indicate that large perpendicular exchange bias field and magnetic anisotropy were found in these samples below the blocking temperature TB ~ 240 K. The dependence of exchange bias on the layer thickness has also been studied. The easy axis direction strongly depends on both the Co and MnPd thicknesses. The origin of the perpendicular anisotropy was attributed to the magneto-elastic effect due to the strained CoPd interfacial alloy forming at the interface between the Co and MnPd layers. In order to explain the perpendicular exchange bias mechanism, a phenomenological picture was put forward in which the fluctuations of the MnPd spins at the interface play an important role. Besides, the results show the anomalous effect related to field- induced anisotropy, i.e. the parallel field cooling enhanced the perpendicular anisotropy property instead of the perpendicular one. Keywords: Perpendicular exchange bias, perpendicular magnetic anisotropy, magnetic thin films, multilayers. iii TRƯỜNG ĐẠI HỌC BÁCH KHOA HÀ NỘI VIỆN ĐÀO TẠO QUỐC TẾ VỀ KHOA HỌC VẬT LIỆU (ITIMS) Khĩa ITIMS – 2005 Tiêu đề của luận văn: Nghiên cứu cơ chế trao đổi dịch vuơng gĩc trong hệ màng mỏng đa lớp MnPd/Co Tác giả: Nguyễn Hữu Dũng Người hướng dẫn: GS. TSKH. Nguyễn Phú Thùy Người phản biện: 1. TS. Nguyễn Thăng Long 2. TS. Nguyễn Phúc Dương Tĩm tắt Lần đầu tiên, hệ màng mỏng đa lớp [MnPd/Co]10 đã được nghiên cứu. Kết quả cho thấy độ lớn trường trao đổi dịch và năng lượng dị hướng từ vuơng gĩc lớn đã thu được ở dưới nhiệt độ blocking TB ~ 240 K. Sự phụ thuộc của hiện tượng trao đổi dịch vào chiều dày các lớp cũng đã được xem xét. Hướng của trục dễ phụ thuộc mạnh vào chiều dày của cả hai lớp Co và MnPd. Nguồn gốc của dị hướng từ vuơng gĩc được gán cho hiệu ứng từ đàn hồi do sự hình thành của hợp kim CoPd ở mặt tiếp xúc giữa lớp Co và MnPd. Để giải thích cơ chế của hiện tượng trao đổi dịch vuơng gĩc, một mơ hình hiện tượng luận đã được đề xuất trong đĩ sự thăng giáng của các spin lớp MnPd ở mặt tiếp xúc đĩng một vai trị quan trọng. Ngồi ra, hệ màng đa lớp cịn thể hiện hiệu ứng dị thường liên quan tới dị hướng cảm ứng từ trường, tức là, quá trình làm nguội trong từ trường song song với bề mặt màng làm tăng cường tính dị hướng vuơng gĩc thay vì từ trường làm nguội vuơng gĩc. Từ khĩa: Hiện tượng trao đổi dịch vuơng gĩc, dị hướng từ vuơng gĩc, hệ màng mỏng đa lớp MnPd/Co. iv ACKNOWLEDGEMENTS First and foremost, I thank my supervisor Prof. D.Sc. Nguyen Phu Thuy for the guidance and inspiration over the last one year at the ITIMS. I would like to thank him for his invaluable advice, comments and suggestions. I would like to express most sincerely my gratitude to Dr. Nguyen Anh Tuan as my co-supervisor at the ITIMS. I would like to thank him for his guidance and valuable discussions. I also wish to extend my warmest thanks to Dr. Nguyen Thang Long for his useful discussions and also for MFM and AFM measurements at the College of Technology, Vietnam National University, Hanoi; to Dr. Nguyen Phuc Duong for reading my thesis and his feedback; to Dr. Nguyen Nguyen Phuoc for many discussions and frank advice; to M.Sc. Do Hung Manh for cross-section images and composition analysis at the Institute of Materials Science, Vietnamese Academy of Science and Technology. Besides, I also wish to extend my thank to Prof. D.Sc. Than Duc Hien for the encouragement and the financial support from State Program on Fundamental Research. Thanks are further extended to all members at the ITIMS for their encouragement and kind supports throughout the present thesis. Especially, I thank M.Sc. Le Thanh Hung for his useful help in experiments. Finally, I would like to thank my family and my friends for their love and encouragement during this study. October 2007 _________________ Nguyen Huu Dzung v LIST OF NOTATIONS θ Angle between incident X-ray and crystal plane (hkl) AF Antiferromagnet(s)/ Antiferromagnetic AFM Atomic force microscope at.% Atomic percent EDS Energy dispersive spectrometer FC Field cooling fct Face centered tetragonal structure FESEM Field emission scanning electron microscope FM Ferromagnet(s)/ Ferromagnetic hcp Hexagonally close packed structure H External magnetic field HC Coercitive force (Coercitivity) HE Exchange bias field HFC Cooling field JK Unidirectional anisotropy (exchange bias coupling) energy Keff Effective magnetic anisotropy KS Surface/interfacial anisotropy KU Uniaxial magnetic anisotropy energy KV Volume anisotropy M Magnetization MFM Magnetic force microscope MS Saturation magnetization of ferromagnetic layer RF Radio frequency SEM Scanning electron microscope vi T Measurement temperature TB Blocking temperature TC Curie temperature tCo Ferromagnetic layer thickness tMnPd Antiferromagnetic layer thickness TN Néel temperature VSM Vibrating sample magnetometer WDS Wavelength dispersive spectrometer XRD X-ray diffraction ZFC Zero field cooling vii LIST OF FIGURES Fig. 1-1. Schematic diagram of the spin configuration of an FM/AF bilayer at different states (After [20]). 5 Fig. 1-2. Schematic diagram of the spin structures assumed in some of the proposed models within each category. 10 Fig. 1-3. Schematic view of spin configuration of FePt/FeMn multilayer based on modified Malozemoff model (After N.N. Phuoc et al. [59]). 14 Fig. 2-1. Schematic view of the MnPd target used in the present thesis. 15 Fig. 2-2. Schematic view of [MnPd/Co]N multilayer structure used in the present thesis. 17 Fig. 2-3. Schematic diagram of glancing incident θ/2θ scan X- ray diffraction configuration. 18 Fig. 3-1. X-ray diffraction spectra of [MnPd(10 nm)/Co(x nm)]10 multilayers, (a) x = 2.5 nm, (b) x = 3.5 nm, (c) x = 4.5 nm. 24 Fig. 3-2. Cross-sectional view of [MnPd(10 nm)/Co(7.5 nm)]10 as-deposited multilayer. 25 Fig. 3-3. MFM image of [MnPd(10 nm)/Co(3.5 nm)]10 as- deposited multilayer. 26 Fig. 3-4. Schematic diagram of measurement configurations for samples at 120K. Here, the measurement field direction (H) is the same as the cooling field (HFC). 27 viii Fig. 3-5. Parallel and perpendicular hysteresis loops measured at T = 120 K for [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) multilayers. 28 Fig. 3-6. Parallel and perpendicular hysteresis loops measured at T = 120 K for [MnPd(y nm)/Co(3.5 nm)]10 (y = 3.5, 5.5, 7.5, 10, 15.5, 30 nm) multilayers. 29 Fig. 3-7. Schematic diagram of measurement configurations at room temperature. Here, HFC denotes the cooling field direction and H denotes measurement field directions. Note that all samples were measured in two different directions. 31 Fig. 3-8. Parallel and perpendicular hysteresis loops measured at room temperature for [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) multilayers cooled in the field perpendicular to the plane. 32 Fig. 3-9. Parallel and perpendicular hysteresis loops measured at room temperature for [MnPd(10 nm)/Co (x nm)]10 (x = 2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) multilayers cooled in the field parallel to the plane. 33 Fig. 3-10. Parallel and perpendicular hysteresis loops measured at room temperature for [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) multilayers cooled in the zero field. 34 Fig. 3-11. Parallel and perpendicular hysteresis loops measured at room temperature for [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) as-deposited multilayers. 35 ix Fig. 3-12. Magnetization – temperature curve of [MnPd(10 nm)/Co(3.5 nm)]10 multilayer in the presence of a field of 2500 Oe. 36 Fig. 4-1. The Co thickness dependence of perpendicular and parallel exchange bias fields (HE), coercitivity (HC), unidirectional anisotropy constant (JK). 40 Fig. 4-2. The MnPd thickness dependence of perpendicular and parallel exchange bias fields (HE), coercitivity (HC). 42 Fig. 4-3. (a) The plot of the product of Keff and tCo versus tCo and (b) the plot of KU versus tCo of [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) multilayers at 120K. 45 Fig. 4-4. Anisotropy energies of [MnPd/Co]10 multilayers which were treated at different conditions. (a) Plot of the product of Keff and tCo versus tCo and (b) plot of KU versus tCo at room temperature. 47 Fig. 4-5. Schematic diagram of multilayer structure after annealing. 49 Fig. 4-6. Schematic view of spin configurations of MnPd/Co multilayer: (a) perpendicular-to-the-plane easy axis and (b) parallel-to-the-plane easy axis. 54 x CONTENTS Preface 1 Chapter 1 Introduction 1.1 Background 3 1.2 Overview on exchange bias 6 1.3 Previous studies on perpendicular exchange bias 12 Chapter 2 Experimental 2.1 Introduction 15 2.2 Sample preparation 15 2.3 Experimental techniques 18 2.3.1 Glancing incident X-ray diffraction 18 2.3.2 Field emission scanning electron microscope 18 2.3.3 Stylus-method profilemetry 19 2.3.4 Energy dispersive X-ray spectrometer 19 2.3.5 Wavelength dispersive X-ray spectrometer 20 2.3.6 Magnetization hysteresis loops 21 2.3.7 Magnetization – temperature curve 22 2.3.8 Magnetic force microscope & atomic force microscope 22 Chapter 3 Experimental results 3.1 Introduction 23 3.2 Crystallographic structure 23 3.2.1 Glancing incident X-ray diffraction 23 3.2.2 Cross-section observation 25 3.3 Magnetic properties 25 xi 3.3.1 Domain observation 26 3.3.2 Magnetization hysteresis loops at low temperature 26 3.3.3 Magnetization hysteresis loops at room temperature 30 3.3.4 Temperature dependence of magnetization in MnPd/Co multilayers 36 Chapter 4 Discussions 4.1 Introduction 37 4.2 Crystallographic structure 37 4.2.1 Glancing incident X-ray diffraction 37 4.2.2 Cross-section observation 38 4.3 Magnetic properties 38 4.3.1 Domain observation 39 4.3.2 Thickness dependence of exchange bias 39 4.3.2.1 Co thickness dependence of exchange bias 39 4.3.2.2 MnPd thickness dependence of exchange bias 41 4.3.3 Perpendicular magnetic anisotropy in MnPd/Co multilayers 43 4.3.3.1. Perpendicular anisotropy at low temperature 44 4.3.3.2. Perpendicular anisotropy at room temperature 46 4.3.3.3. Effect of annealing on perpendicular anisotropy 46 4.3.3.4. Anomalous field induced anisotropy 50 4.3.4 Temperature dependence of magnetization in MnPd/Co multilayers 51 4.4 Explanation of exchange bias coupling mechanism 52 Conclusions and further direction 56 References 58 - 1 - PREFACE Exchange bias has been studied extensively for over half of a century but most of the research has been carried out in the configuration called parallel exchange bias. In this configuration, the cooling field and the measurement field are applied in the plane. Beside parallel exchange bias, there has been very little work carried out in the perpendicular configuration with the cooling field and the measurement field along the film normal. Perpendicular exchange bias is recently of renewed interest because it is relevant in the quest for a better understanding of the microscopic origin of the exchange bias phenomenon and it might lead to wide applications in magnetic sensors, perpendicular recording media, perpendicular magnetic read heads and also magnetic random access memories (MRAMs). In this thesis, the studies on perpendicular exchange bias in [MnPd/Co]10 multilayers are reported for the first time. Since the objective of the present thesis is to study the perpendicular exchange bias mechanism, the approach is to investigate both the parallel and perpendicular exchange biases. Besides, perpendicular anisotropy of the samples at low and room temperatures is also investigated due to its important contribution to the effect. The present thesis consists of 4 chapters. Chapter 1 is to give an overview on exchange bias in both theoretical and experimental research; and also previous studies on perpendicular exchange bias. Chapter 2 focuses on the sample preparation and experimental techniques. Some descriptions on the apparatuses and measurements that were used in the present thesis are introduced. - 2 - Chapter 3 represents the experimental results. The aim and configurations of measurements and also sample processing procedures are given. Chapter 4 is to discuss the results of crystallographic and magnetic properties of [MnPd/Co]10 multilayers. The behavior of exchange bias in both the parallel and perpendicular directions will be summarized. After that, based on that result and the magnetic anisotropy behavior of the samples, we try to give a phenomenological picture to explain the perpendicular exchange bias coupling mechanism. Finally, conclusions and further direction as well as the list of references are given at the end of the thesis. - 3 - Chapter 1 1. INTRODUCTION 1.1 Background Nowadays, magnetic materials play an important role in the information technology oriented social. There are various applications using magnetic materials such as magnetic recordings, magnetic sensors, magnetic heads, and electronic motors. It is of particular interest to note that through rapid technological developments in recent years, thin films and multilayers have received much attention. Among studies on magnetic materials, the exchange bias coupling between ferromagnetic (FM) and (AF) materials is of great interest. Since discovered in 1956 by Meiklejohn and Bean [1], there have been many studies published in the literature on this effect because of various applications such as spin valves, magnetic read heads, magnetic random access memories (MRAMs). Although it has been studied extensively, physical origin of this effect is still in controversy. Exchange bias effect is a phenomenon observed in a system consisting of antiferromagnetic and ferromagnetic materials, in which the magnetization hysteresis loop is shifted along the field axis after the sample undergoing the so-called field cooling process through the Néel temperature of the antiferromagnetic material. In other words, its characteristic signature is the shift of the center of the hysteresis loop from its normal position at H = 0 to HE. However, in order to compare different types of exchange bias systems often rather than using the loop shift itself, the so-called unidirectional anisotropy energy or exchange bias coupling energy JK = HEMStFM (where MS - 4 - is the saturation magnetization and tFM is the thickness of the FM layer) is evaluated instead. The exchange bias effect is only observed below a certain temperature. The temperature at which the exchange bias field becomes zero, HE = 0, is usually denoted as blocking temperature (TB). Exchange bias can be qualitatively understood by assuming an exchange interaction at the AF-FM interface (Fig 1-1). When a field is applied in the temperature range TN < T < TC, the FM spins line up with the field, while the AF spins remain random (see Fig 1-1-(a)). When cooling to T < TN, in the presence of the field (so-called cooling field which is denoted as HFC in present thesis), due to the interaction at the interface, the AF spins next to the FM align ferromagnetically to those of the FM (assuming that the interaction is ferromagnetic). The other spin planes in the AF follow the AF order so as to produce zero net magnetization (see Fig 1-1-(b)). When the field is reversed, the FM spins start to rotate. However, the AF spins remain unchanged due to its large anisotropy. Therefore, the interfacial interaction between the AF-FM spins try to align parallel the FM spins. In other words, the AF spins exert a microscopic torque on the FM spins, to keep them to their original position (see Fig 1-1-(c)). The field needed to reverse completely the FM spins is larger if it is in contact with the AF because an extra field is to overcome a microscopic torque. As the field is back to its original direction, the FM spins will start to rotate back at a smaller field because it now exerts a torque with the same direction as the applied field (see Fig. 1-1-(d) and Fig 1-1-(e)). The material behaves as if there is an extra biased field; the hysteresis loop is therefore shifted along the field axis (see the hysteresis loop in Fig 1-1). If the AF anisotropy is large, one should only observe a shift of the hysteresis loop, while for small AF anisotropies, the only observed effect should be a coercivity enhancement (without any loop - 5 - FM AF FM AF (d) (c) (b) (a) HFC Field cooling H M O HE Fig. 1-1. Schematic diagram of the spin configuration of an FM/AF bilayer at different states. (After [20]) FM AF FM AF (e) FM AF - 6 - shift). Nevertheless, in general, both the effects can be observed simultaneously, due to, for example, structural defects or grain size distribution, which bring about local variations of the AF anisotropy. Although this simple phenomenological model gives an intuitive picture, it fails to quantitatively understand of these phenomena. In particular, the theoretically predicted exchange bias field is much larger than the experimental value. In an attempt to reduce this discrepancy, many models such as planar domain wall model [2], random-field model [3-5], spin flop model [6] put forward. However, there have not been experimental confirmations of these models and they are therefore in controversy. It is due to the fact that the role of the many different parameters involved in exchange bias, such as anisotropy, interface roughness, spin configuration or magnetic domain is far from being understood. A clear understanding of exchange bias at the microscopic level is still lacking. Therefore, from the fundamental point of view, the subject of exchange bias is still a hot topic for the years to come and it is of great interest to study this phenomenon together with its associated effects for a better understanding of physical origin. 1.2 Overview on exchange bias So far, exchange bias has been investigated extensively both experimentally and theoretically. Regarding experimental research, from a view point of material form, studies on exchange bias can be relatively divided into 3 categories: exchange bias in particles, exchange bias in nanostructures and exchange bias in (continuous) thin films. Fine particles were the first type of system where exchange bias was reported. Since its discovery, exchange bias in particles has been concentrated on a number of materials, mainly ferromagnetic metals covered by their - 7 - antiferromagnetic oxides, such as Co/CoO [1, 7, 8], Ni/NiO [9], Fe/FeO [10], Fe/Fe2O3 [11], Fe/Fe3O4 [12]. Recently, the number of studies on exchange bias in small particles has been reduced because most of the applications using this effect are in the form of thin films. Moreover, these systems are not suitable for studies of fundamental aspects of exchange bias due to uncontrolled distribution of the particle size and shape, difficulty to identify the nature of the interface, stoichiometry and crystallinity of the AF material. However, studies of FM-AF exchange interactions in fine particle systems has still found interest in applications to improve the performance of permanent magnetic materials (by means of an enhancement of the coercivity which typically accompanies the hysteresis loop shift) [13-15] or to increase in the superparamagnetic limit in magnetic recording media [16, 17]. Hence, in fine particle systems, exchange bias studies may be particularly interesting not only for the loop shift itself, but also for other exchange bias related phenomena. Today, the industrial demand to systematically reduce the size of spin- valve and other exchange bias based devices is also fueling new research in lithographically fabricated exchange biased nanostructures [15]. Different kinds of nanostructured systems where exchange bias has been studied, including artificial nanostructures (e.g., lithographically fabricated nanostructures), chemical surface modification (e.g., oxidation, nitration or sulfation), FM nanoparticles embedded in an AF matrix, controlled core-shell nanoparticles, surface effects (e.g., ferromagnetic, ferrimagnetic or antiferromagnetic particles with magnetically disordered surfaces) [15, 18- 23]. The recent advances in magnetic fine particle production and the fabrication of magnetic nanostructures by lithographic methods have propelled a renewed interest in nanostructures in general and exchange biased - 8 - ones in particular. However, exchange bias theories for nanostructures are still lacking [15]. Although there has been some research on exchange bias in nanoparticles in the last decades, the bulk of exchange bias research has focused mainly on thin film systems. This is firstly due to the possibility of an increased number of FM/AF combinations in thin films. Secondly, the greater control of the FM/AF interface that thin films allow, in which the microstructure of both the AF and FM layers (e.g., grain size, orientation, crystalline quality) and, to some extent, the interface (e.g., roughness, spin structure or interface layers) can be controlled. Finally, the fundamental role of exchange bias in spin valve and tunneling devices has triggered the explosive increase of research in FM/AF thin film systems. In the point view of the AF material form, studies on exchange bias in thin films can be divided into 2 categories: exchange bias with insulating AF films and with metallic AF films. Almost all the reported investigations of exchange bias with insulating AF films involve oxides CoO, NiO, NixCo1-xO [24-26] except FeF2, MnF2 [27, 28]. Oxidized film systems give usually large exchange bias, e.g., the largest interfacial energy ever found is in Fe3O4/CoO bilayers (JK = 2.2 erg/cm2) [29]. However, since most of these oxidized film systems exhibit exchange bias at low temperature, the applications based on this type are uncommon and it has received less attention than before. Apart from oxides, the most popular materials are FeF2 and MnF2, exhibiting interesting phenomena such as positive exchange bias, double-shifted loops (depending on temperature and the cooling field) [28, 30]. Meanwhile, studies on exchange bias with metallic AF films focus on alloys of Mn with transition metals such as Pd, Pt, Ir [24, 31, 32] or - 9 - ferromagnetic metals as Fe, Ni [33-35]. As for interfacial energy aspect, its value in the published reports is usually in the range from 0.1 to 0.5 erg/cm2 (lower than oxidized film systems). Recently, Imakita et al. [36] obtained the largest exchange bias energy at room temperature in CoFe/MnIr with the JK value up to 1.3 erg/cm2 capable of using for the future read heads in hard disk drivers. Jiao et al. showed that exchange bias might exist in the Gd/Cr bilayers and Cr/Gd/Cr trilayers regardless of the condition of TC > TN and the anomalous dependence of the exchange bias field which increased with temperature until TC [37]. As for theoretical research, many models have been proposed to understand its mechanism and have been achieved different results with experimental observations. The models may be classified as either macroscopic, mesoscopic, or microscopic (see Fig. 1-2). Most of the works have been concentrated on the discrepancy between the theoretically predicted and experimental exchange bias field. Mauri et al. [2] proposed a model based on the formation of a planar domain wall. The interfacial exchange energy is thus due to the wall energy in the AF layer giving the same order of the experimental exchange bias field in some cases. Malozemoff [3-5] put forward a random-field arising from the random defects at the interfaces, which are argued to be more likely in the real systems. Due to the random-field, the AF is broken into domains. In this case, domain walls perpendicular to the interface are energetically favorable. Therefore, the interfacial exchange energy is also of the same order of the wall energy in the AF. Takano, Berkowitz and coworkers proposed a model for the exchange anisotropy of AF/FM bilayers in which the AF layer consists of (essentially uncoupled) grains [6]. The exchange bias field is due to uncompensated surface spins of antiferromagnetic grains. As for exchange bias in the systems - 10 - a) Macroscopic Meiklejohn-Bean Malozemoff Mauri Takano-Berkowitz Koon, Schulthess-Butler b) Mesoscopic Fig. 1-2. Schematic diagram of the spin structures assumed in some of the proposed models within each category. c) Microscopic FM AF FM AF FM AF FM AF FM AF - 11 - with compensated interfacial spin, Koon [38] put forward perpendicular coupling between the AF and the FM spins. Schulthess and Butler [39] have shown that Koon’s perpendicular coupling, together with uncompensated spins (similar to Malozemoff or Takano et al. suggestions [3-5]) can explain simultaneously the loop shift and coercivity enhancement encountered in FM–AF bilayers. Meanwhile, Miltényi, Nowak, Misra, Beckmann et al. used Monte Carlo at finite temperature to study a FM–AF couple with defects in the bulk of the AF, i.e., not necessarily at the interface. They found the formation of domains in the bulk of the AF, perpendicular to the FM–AF interface, which gave rise to uncompensated spins at the interface, which were responsible for the hysteresis loop shift. They also found that increasing the number of defects, within certain limits, increases the number of AF domains, leading to larger exchange bias [40-46]. Suess et al. [47-49] have developed a model based on perpendicular coupling and randomly distributed exchange coupled AF grains. Interestingly, the origin of exchange bias is found to be in the energy stored in the domain walls between AF grains with different orientations. Lederman et al. have recently reported that if the FM layer couples differently to each of the two AF sublattices. It could give rise to exchange bias. Actually, using this simple concept many of the experimentally observe defects in FM/FeF2 bilayers can be explained [50]. It is should be noted that all these theories are applied for the case of parallel exchange bias phenomena. In which, the cooling field and the measurement field are applied in the plane. Beside parallel exchange bias, there has been very little work carried out in the perpendicular configuration - 12 - with the cooling field and the measurement field along the film normal, namely perpendicular exchange bias. 1.3 Previous studies on perpendicular exchange bias Parallel exchange bias has been studied for a long time, but perpendicular exchange bias has been observed recently in the FM/AF systems with perpendicular magnetic anisotropy. Perpendicular exchange bias is of renewed interest because it is relevant in the quest for a better understanding of the microscopic origin of the exchange bias phenomenon and it might lead to wide applications in magnetic sensors, perpendicular recording media, perpendicular magnetic read heads and also magnetic random access memories (MRAMs). The exchange bias effect was measured for the first time in FeF2-CoPt hetero-systems with perpendicular anisotropy by Kagerer et al. [51, 52] .The exchange bias field exhibits a strong dependence on the cooling field and temperature. Maat et al. [53] studied perpendicular exchange bias in the system of [Co/Pt]/CoO multilayers and found that the perpendicular exchange bias field is larger than the parallel one, which can be attributed to the anisotropy in the CoO induced by the CoO (111)-textured growth of the films thus producing the difference between the spin projections on the parallel and perpendicular directions. Conversely, Marrows [54], who carried out research on perpendicular exchange bias in [Co/Pd]/FeMn multilayers, found that the difference of parallel and perpendicular exchange bias might not due to the texture of the film because the discrepancy between parallel and perpendicular exchange biases were clearly observed in the weak textured film systems. This difference can be attributed to the fluctuations of the AF spin at the interface, which naturally played a key role in determining any exchange bias [54]. Garcia et al. [55] found a large anomalous enhancement - 13 - of perpendicular exchange bias in [Co/Pt]/FeMn by introduction of a nonmagnetic spacer between the ferromagnetic and the antiferromagnetic layers, which was presumably interpreted as the enhancement of perpendicular magnetic anisotropy. Sort et al. [56] found that a temperature range of square loops behavior on Co/Pt multilayers with perpendicular anisotropy could be extended by using exchange bias with either FeMn or IrMn layers. This was attributed to additional anisotropy induced to the multilayers by exchange bias coupling. Recently, a 1/cosθ dependence of exchange bias field on the angle θ between the applied field and the perpendicular-plane cooling field was observed by Kim et al. [57] in [Pt/Co]4/MnIr multilayers and Sun et al. [58] in FeMn/[FeNi/FeMn]15 multilayers. This 1/cosθ dependence was ascribed to the strong out-of-plane anisotropy. They also found that the hysteresis loops became asymmetric at intermediate angle with a shift not only along the field axis but also along the magnetization axis [57, 58]. It is only lately that N.N. Phuoc et al. has used the modified Malozemoff model with the assumption of spin canting at the interface of FM and AF layers to explain the perpendicular exchange bias effect in [FePt/FeMn]10 multilayers and found that the canting spins at the interface play an important role in the effect (see Fig. 1-3) [59]. Among these studies on perpendicular exchange bias, very few materials have been investigated, mainly Co/Pt multilayers with CoO [53], Co/Pt, CoFe/Pt, Co/Pd multilayers with FeMn [54-56], Co/Pt multilayers with MnIr [56, 57], Co/Pt multilayers with FeF2 [51, 52]. In which, the multilayers are ferromagnetic and have perpendicular anisotropy. The exchange bias effect in MnPd/Co bilayers has received much attention by N.N. Phuoc et al. [60], N.T. Nam et al. [61] and N.P. Thuy et al. [62]. All these works have concentrated on parallel exchange bias in the bilayers, - 14 - KAFparallel KAFperpendicular KAF α AF FM AF FM Substrate Fig. 1-3. Schematic view of spin configuration of FePt/FeMn multilayer based on modified Malozemoff model (After N.N. Phuoc et al. [59]). showing that the exchange bias coupling between the Co and MnPd layers is of huge values. However, perpendicular exchange bias in this kind of material has never been investigated before. Therefore, a study on perpendicular exchange bias in these systems is necessary for a better understanding of physical origin of exchange bias and also related phenomena. We will show in this thesis that perpendicular exchange bias can be indeed observed in the samples produced by the multilayer thin film technique from the same materials. - 15 - Chapter 2 2. EXPERIMENTAL 2.1 Introduction Samples used in the present thesis were prepared by a RF sputtering system at ambient temperature. A vibrating sample magnetometer (VSM) and a magnetic force microscope (MFM) were used to characterize the magnetic properties of samples. Structural analyses were performed by X-ray diffraction (XRD). Cross-section images were carried out by a field emission scanning electron microscope (FESEM) and surface image were taken from an atomic force microscope. The film compositions were analyzed by an energy dispersive X-ray spectrometer (EDS) and a wavelength dispersive X- ray spectrometer (WDS). Deposition rates for materials were calculated from the thickness of the corresponding single layers determined by an Alpha-step model. 2.2 Sample preparation Thin films of MnPd/Co multilayers were deposited onto single crystal Si(111) substrates at ambient temperature by an Alcatel SCM 400 RF sputtering system at the ITIMS. Antiferromagnetic layers were prepared from a composite target (see Fig. 2-1). The target is a circular Pd Mn Pd Fig. 2-1. Schematic view of the MnPd target used in the present thesis. - 16 - target with sectorial Mn pieces glued conductively to it. In the present thesis, the area compositions of Mn and Pd on the target were about 60:40, respectively. Meanwhile, a circular Co target was used to prepared ferromagnetic layers. The RF sputtering system has two power sources used for two targets. The targets were placed in its positions in the deposition chamber and after that, the deposition chamber was pumped out until the pressure inside was less than 5 × 10-6 mbar. Samples were fabricated in Ar gas. The gas flow was regulated by a mass flow controller and kept at a constant rate during the deposition. Sputtering process was carried out in the condition of the Ar pressure kept at about 5 × 10-3 mbar. No external magnetic field was applied in the deposition and substrates were at ambient temperature. The samples used in the present thesis are Si/[MnPd/Co]10 multilayer thin films. The MnPd and Co layers were deposited alternately onto single crystal Si(111) substrates (see Fig. 2-2) at the power of 150 W for the MnPd target and 300 W for the Co target. The corresponding deposition rates for MnPd and Co layers are 2.3 × 10-2 nm/s and 2.8 × 10-2 nm/s. The compositions of MnPd layer were determined using a wavelength dispersive X-ray spectrometer (WDS) and an energy dispersive X-ray spectrometer (EDS). The results showed the Mn and Pd compositions are 11: 89, respectively. - 17 - N bilayers x = 2.5 – 10 nm y = 3.5 – 30 nm N = 10 bilayers MnPd y (nm) Co x (nm) Co x (nm) Co x (nm) Co x (nm) MnPd y (nm) MnPd y (nm) MnPd y (nm) Si substrate Fig. 2-2. Schematic view of [MnPd/Co]N multilayer structure used in the present thesis. - 18 - 2.3 Experimental techniques 2.3.1 Glancing incident X-ray diffraction In order to analysis the structure of sample, θ/2θ scan X-ray diffraction was carried out using a PANalytical-Philips X’pert Pro system at Hanoi University of Technology. A Cu target is used as the X-ray source. A double- crystal monochromator is used to obtain monochromatic and collimated Cu Kα1 radiation (λ=0.154056). The incident X-ray and the sample were fixed. The incident angle of the X-ray beam was of 1 degree with respect to the sample surface. Meanwhile, the detector rotated so that the θ/2θ scan configuration was preserved during the measurements. In the present thesis, the angle 2θ was from 25 to 70 degrees. Diffracted beam Sample Incident beam Fig. 2-3. Schematic diagram of glancing incident θ/2θ scan X-ray diffraction configuration. 2.3.2 Field emission scanning electron microscope Cross-section images were observed by a field emission scanning electron microscope (FESEM). The best resolution of the system is up to 2 nm (standard mode), 3 to 5 times better than conventional SEM. Because a field emission source provides narrower probing electron beams at low temperature - 19 - and high energy with acceleration voltage from 0.5 to 30kV (variable at 0.1 kV/step). The magnification of the system is in the range of X 20 - X 800000. In present thesis, observations of cross-section were carried out by a Hitachi FE-SEM S4800 microscope system at the Institute of Materials Science, Vietnamese Academy of Science and Technology. After the sample was broken into half, they were immediately used to view. 2.3.3 Stylus-method profilemetry The stylus method consists of measuring the mechanical movement of a stylus as it is made to trace the topography of a film-substrate step. The film thickness is directly read out as the height of the resulting step-contour trace. The profilemeter used in this thesis is called Alpha-step model with the vertical resolution of about 1 Å. The Alpha-Step IQ is guaranteed step height repeatability which makes it easier to precisely determine the thickness of thin films, roughness, etch depth in a wide extending below 8 nm and tall step height up to 2 mm. The performance is due to modern ultra-low noise electronics and precision mechanical components. The stylus scanning motion provides exceptional stability for extremely repeatable measurements. To determine deposition rate for a material, a single layer with a film- substrate step was prepared in a specific time. The single layer was measured for three times in order to receive the mean thickness. Hence, one can calculate the deposition rate for the material. In this thesis, the deposition rates for MnPd and Co are respectively 2.3 × 10-2 nm/s and 2.8 × 10-2 nm/s. These thickness measurements were carried out at the Institute of Materials Science, Vietnamese Academy of Science and Technology. 2.3.4 Energy dispersive X-ray spectrometer (EDS) - 20 - X-rays emitted from a sample under electron bombardment are collected with a liquid nitrogen-cooled solid state detector and analyzed via computer according to their energy. Typically, the computer programs used in EDS will display a real time histogram of number of X-rays detected per channel (variable, but usually 10 electron volts/channel) versus energy expressed in KeV. Using EDS, all of the energies of the characteristic X-rays incident on the detector are measured simultaneously and data acquisition is therefore very rapid across the entire spectrum. However, the resolution of an EDS detector is considerably worse than that of a WDS spectrometer. Besides, it is very difficult to determine precisely elements and its compositions if there is only a small amount of one of the overlapped elements. In practice, EDS is most often used for qualitative elemental analysis, simply to determine which elements are present and their relative abundance. Depending on the specific needs of the investigations, quantitative results may be advised to use the electron microprobe. In some instances, however, the area of interest is simply too small and must be analyzed by TEM (where EDS is the only option) or high resolution SEM (where the low beam currents used preclude WDS, making EDS the only option). In this thesis, the sample composition was analyzed by a Hitachi FESEM S4800 microscope system integrated EDS at the Institute of Materials Science, Vietnamese Academy of Science and Technology. 2.3.5 Wavelength dispersive X-ray spectrometer (WDS) WDS was the original technique developed to precisely and accurately determine chemical compositions of micro-volumes (a few cubic microns) of "thick" specimens, and the instrument used is the electron microprobe. The - 21 - key feature of the electron microprobe is a crystal-focusing spectrometer, of which there are usually 3-5 different diffracting crystals. The WDS spectrometer can acquire the high count rate of X-rays produced at high beam currents, because it measures a single wavelength at a time. This is important for trace element analysis. In practice, it is advantageous to use the speed of EDS for an initial survey of an unknown sample because major elements will be rapidly identified. However, if trace elements are present they will not be identified, and it may be difficult to interpret complex overlaps which are common in EDS analysis. Following the initial energy dispersion survey, wavelength dispersion can be used to check for overlaps and to increase sensitivity for trace elements. A Jeol JXA 8800R electron probe microanalyzer at the Institute of Geology and Minerals was used in the present study. 2.3.6 Magnetization hysteresis loops Magnetization curve provides basic magnetic properties of a magnetic material. From the curve, one can estimate the saturation magnetization MS, the coercitivity HC, the magnetic anisotropy K, the magnetization remanence MR and the exchange bias field HE. Magnetic behavior can also be understood of microscopic structural properties. In the present study, measurements of magnetization curves were performed using a DMS 880 VSM system at the ITIMS. The magnetic field used in the present study was up to 13.5 kOe along both the parallel and perpendicular directions. For these measurements, the background resulted from any source such as the sample holder and the substrate was subtracted. Before each measurement, a standard Ni sample (with total magnetic moment of 3.799 emu) was always used to calibrate the system. For the measurement of the hysteresis loops at low temperatures, a tube attached in the VSM and a thermocouple is placed inside the tube - 22 - together with the heating coil. By evaporating liquid nitrogen and simultaneously adjusting the current for the heating coil, one can control the system with the temperature accuracy of about 5 degrees. 2.3.7 Magnetization-temperature curve Magnetization-temperature curve were carried out by a VSM system (described in the previous subsection). Temperature was controlled by evaporating liquid nitrogen (in the range of low temperature) or blowing pure nitrogen gas (in the range of high temperature), and simultaneously adjusting the current for the heating coil. In the present study, the measurement was performed in the temperature range from 120 to 320 K and the step of 5 K. 2.3.8 Magnetic force microscope & atomic force microscope Observations of magnetic domains were carried out using a NT-MDT Solver magnetic force microscope (MFM) at the College of Technology, Vietnam National University, Hanoi. The tip used in the present study was coated by CoCr alloy with the coating thickness of 40 nm, the curvature radius of 30-40 nm and the cone angle less than 30 degrees. Before each measurement, it was magnetized along the direction perpendicular to the sample surface. The same tip was used to observe the MFM and AFM images. The surface roughness was determined to be less than 2 nm. - 23 - Chapter 3 3. EXPERIMENTAL RESULTS 3.1 Introduction In this chapter, the results of crystallographic and magnetic properties of [MnPd/Co]10 multilayer thin films are presented. The crystallographic properties characterized by XRD, FESEM and AFM. Meanwhile, the magnetic properties, in particular parallel and perpendicular exchange biases and anisotropy, are characterized by VSM and MFM. The aim of the present thesis is to study the perpendicular exchange bias effect. Therefore, the investigation and comparison between parallel and perpendicular exchange biases is necessary. The perpendicular magnetic anisotropy is also important due to its contribution to the effect. Some measurements were carried out at room temperature for a better understanding of physical origin of the perpendicular anisotropy and also perpendicular exchange bias. The magnetic properties of the multilayers are discussed in the next chapter in conjunction with the structure. 3.2 Crystallographic structure 3.2.1 Glancing incident X-ray diffraction Fig. 3-1 shows the θ/2θ scan X-ray diffraction pattern of [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5 nm) as-deposited multilayers. -24- 25 30 35 40 45 50 55 60 65 70 0 200 400 600 800 1000 1200 1400 ( 2 2 0 ) ( 2 0 0 ) (b) (c) (a) ( 1 1 1 ) Fig. 3-1. X-ray diffraction spectra of [MnPd(10 nm)/Co(x nm)]10 multilayers, (a) x = 2.5 nm, (b) x = 3.5 nm, (c) x = 4.5 nm. 2θ (deg.) I n t e n s i t y ( a . u . ) - 25 - 3.2.2 Cross-section observation Shown in Fig. 3-2 is the cross-section image of [MnPd(10 nm)/Co(7.5nm)]10 as-deposited multilayer. Fig. 3-2. Cross-sectional view of [MnPd(10 nm)/Co(7.5 nm)]10 as- deposited multilayer. 3.3 Magnetic properties The magnetic properties of the present sample were characterized by using the VSM and MFM. Based on the VSM measurements with the field cooling process, one can estimate the exchange bias field as well as the anisotropy constants. The magnetic anisotropy was also investigated through the domain structure observed by MFM for some typical samples at room temperature. Some hysteresis loop measurements were carried out at room temperature in order to understand the origin of the perpendicular magnetic anisotropy. - 26 - 3.3.1 Domain observation Observation of the domains by MFM at the sample surface of [MnPd(10 nm)/Co(3.5 nm)]10 as-deposited multilayer is shown in Fig. 3-3. Fig. 3-3. MFM image of [MnPd(10 nm)/Co(3.5 nm)]10 as-deposited multilayer. 3.3.2 Magnetization hysteresis loops at low temperature Before the magnetization hysteresis loop measurements, the samples had to undergo the so-called field cooling (FC) process. First, MnPd/Co multilayer deposited onto Si(111) substrate was heated to T = 590 K and kept at that condition for 5 minutes. Then the sample was cooled down to room temperature in the presence of a magnetic field of 5 kOe - 27 - (called the cooling field HFC) applied either in the film plane (parallel direction) or normal to the plane (perpendicular direction). This process was realized in a vacuum chamber with the pressure better than 2 × 10-5 mbar. After that, the sample cooled in a field of 5 kOe between the two poles of the VSM from room temperature down to the measurement temperature. Finally, the hysteresis loops were measured with the applied field direction as the same as the cooling field at cryogenic temperature below the blocking temperature TB ~ 240 K, namely at T = 120 K (see Fig. 3-4). Samples with the different thicknesses of Co and MnPd layers were studied. Parallel and perpendicular hysteresis loops measured at 120 K were shown in Fig. 3-5 for the series of samples with tCo varied from 2.5 to 10 nm while tMnPd is fixed at 10 nm and in Fig. 3-6 for another series of samples with the variation of tMnPd from 3.5 to 30 nm while keeping tCo at 3.5 nm. HFC H HFC H Fig. 3-4. Schematic diagram of measurement configurations for samples at 120K. Here, the measurement field direction (H) is the same as the cooling field (HFC). - 28 - -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel tCo = 2.5 nm M (a .u .) H (kOe) -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo =3.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 4.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 5.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 7.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 10 nm Fig. 3-5. Parallel and perpendicular hysteresis loops measured at T = 120 K for [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) field cooled multilayers. - 29 - Fig. 3-6. Parallel and perpendicular hysteresis loops measured at T = 120 K for [MnPd(y nm)/Co(3.5 nm)]10 (y = 3.5, 5.5, 7.5, 10, 15.5, 30 nm) field cooled multilayers. -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tMnPd = 3.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tMnPd = 5.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tMnPd = 7.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) tMnPd = 10 nm H (kOe) -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tMnPd = 15.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tMnPd = 30 nm - 30 - 3.3.3 Magnetization hysteresis loops at room temperature Magnetization hysteresis loop measurements of the samples processed at different conditions were carried out at room temperature (see Fig. 3-7). MnPd/Co multilayer deposited onto Si(111) was heated to 590 K and kept at that condition for 5 minutes. Then, the sample was cooled down to room temperature in the presence of a magnetic field of 5 kOe applied either normal to the plane (perpendicular direction) (Fig.3-7-(a)) or in the film plane (parallel direction) (Fig. 3-7-(b)). These samples processed at the same conditions as that described in the previous subsection. Some others were annealed at 590 K for 5 minutes, and then cooled down to room temperature in zero field (so-called zero field cooling ZFC) (Fig. 3-7-(c)). Besides, as- deposited samples were also used for these measurements (Fig. 3-7-(d)) in order to investigate systematically the effect of the field cooling and also annealing process. It should be noted that hysteresis loops of each sample were carried out in both the parallel and perpendicular directions at room temperature. The hysteresis loops of different samples processed at the same conditions are depicted from Fig. 3-8 to Fig. 3-11. - 31 - d) Sample as- deposited H H c) Sample cooled in the zero field H H b) Sample cooled in the parallel field. H HFC H a) Sample cooled in the perpendicular field. HFC H H Fig. 3-7. Schematic diagram of measurement configurations at room temperature. Here, HFC denotes the cooling field direction and H denotes measurement field directions. Note that all samples were measured in two different directions. - 32 - -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 2.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 4.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 5.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 7.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 10 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 3.5 nm Fig. 3-8. Parallel and perpendicular hysteresis loops measured at room temperature for [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) multilayers cooled in the field perpendicular to the plane. - 33 - Fig. 3-9. Parallel and perpendicular hysteresis loops measured at room temperature for [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) multilayers cooled in the field parallel to the plane. -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 2.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 3.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 4.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 5.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 7.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 10 nm - 34 - Fig. 3-10. Parallel and perpendicular hysteresis loops measured at room temperature for [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) multilayers cooled in the zero field. -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 2.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 3.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 4.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 5.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 7.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 10 nm - 35 - Fig. 3-11. Parallel and perpendicular hysteresis loops measured at room temperature for [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) as-deposited multilayers. -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 2.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 4.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 5.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 3.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 7.5 nm -15 -10 -5 0 5 10 15 -1.0 -0.5 0.0 0.5 1.0 Perpendicular Parallel M (a .u .) H (kOe) tCo = 10 nm - 36 - 3.3.4 Temperature dependence of magnetization in MnPd/Co multilayers The multilayer of [MnPd(10 nm)/Co(3.5 nm)]10 was heated up to 590 K and kept at that condition for 5 minutes. After that, the sample was cooled down to room temperature in the zero field. This process was carried out in a vacuum chamber with the pressure better than 2×10-5 mbar. The magnetization – temperature curve of this multilayer was recorded by first cooling the sample from room temperature to 120 K in the zero magnetic field, then applying the magnetic field of 2500 Oe and warming the sample up to 320 K in the presence of the field and recording the moment in this warming cycle (Forward). Next, keeping the field while the sample is cooled from 320 K down to 120 K and recording the moment (Backward). It is noted that the applied field is perpendicular to the plane. The obtained curve is shown in Fig. 3-12. 120 150 180 210 240 270 300 330 0 50 100 150 200 250 300 Forward Backward M (e m u/ cm 3 ) T (K) ZFC H = 2500 Oe Fig. 3-12. Magnetization – temperature curve of [MnPd(10 nm)/Co(3.5 nm)]10 multilayer in the presence of a field of 2500 Oe. - 37 - Chapter 4 4. DISCUSSIONS 4.1 Introduction This chapter is to discuss the results of crystallographic and magnetic properties of [MnPd/Co]10 multilayers presented in the previous chapter. The behaviors of exchange bias in both the parallel and perpendicular directions and magnetic anisotropy will be studied. After that, based on the results of parallel and perpendicular exchange biases and the magnetic anisotropy, we propose a phenomenological picture to qualitatively explain the perpendicular exchange bias coupling mechanism. 4.2 Crystallographic structure Crystallographic properties are obtained from XRD patterns and cross- section observation using a FESEM. They are discussed here due to their relation to the magnetic properties of the multilayers. 4.2.1 Glancing incident X-ray diffraction Fig. 3-1 shows the θ/2θ scan X-ray diffraction patterns for as-deposited [MnPd(10 nm)/Co(x nm)]10 (x = 2.5, 3.5, 4.5) multilayers grown onto Si(111) substrates held at ambient temperature. It is observed that in the samples, MnPd is polycrystalline with fcc structure. Apart from the peak (111), other peaks of MnPd are also observed as (200) and (220) for the sample with the smallest tCo (tCo = 2.5 nm). However, they are found to be very weak and unobservable with increasing tCo. Meanwhile, almost no peak for Co can be observed showing that Co layer may be formed with low crystallinity. - 38 - 4.2.2 Cross-section observation Shown in Fig. 3-2 is cross-sectional view observed by FESEM microscope on the fresh broken pieces of the as-deposited [MnPd(10 nm)/Co(3.5nm)]10 multilayer thin film. One can see that there are 20 layers corresponding to alternate MnPd and Co through the dark and bright contrast (as shown in Fig. 2-2). From the bottom to top, cumulative waviness increases gradually. The effect is usually observed in multilayer thin films, especially in metallic ones [63]. 4.3 Magnetic properties The magnetic properties are discussed based on the magnetic measurements. From the hysteresis loops, one can estimate the exchange bias field as well as the anisotropy constants. Besides, the anisotropy property can be also qualitatively estimated through domain structure observed by MFM. As shown in Fig. 3-5 and Fig. 3-6, the negative shifts of the hysteresis loops show that the exchange bias effect is found in the [MnPd/Co]10 multilayers in both the parallel and perpendicular directions. From the hysteresis loops, the exchange bias field (HE) and the coercivity (HC) can be extracted by using definitions: HE = |HSU + HSD|/2 (Eq. 4-1) HC = (HSU - HSD)/2 (Eq. 4-2) Where, HSU and HSD are the switching fields in the upward and downward branches, respectively, of the hysteresis loop. Besides, the effective magnetic anisotropy (Keff) can also be readily deduced from the hysteresis loops. Hence, the uniaxial magnetic anisotropy (KU) is quantitatively calculated as shown later. The largest exchange bias fields are found to be extremely high, 1750 and 1650 Oe in the parallel and perpendicular cases, respectively. - 39 - It is interesting to note that the preferred orientation of the magnetization perpendicular to the plane is observed. The result also means that the uniaxial magnetic anisotropy is present in the samples. Its value is expected to be large due to taking into account the demagnetization field anisotropy for the case of perpendicular applied magnetic field. 4.3.1 Domain observation As shown in Fig. 3-3, domain structure observed by MFM at the samples surface is clearly seen. This domain structure again confirms the perpendicular magnetic anisotropy existing in the studied multilayers. We note that the domain size is in order of several hundreds of nanometers. 4.3.2 Thickness dependence of exchange bias Effects of the MnPd and Co thicknesses on perpendicular and parallel exchange biases will be more quantitatively analyzed in the next subsections in order to have an insight into the phenomena. 4.3.2.1 Co thickness dependence of exchange bias Shown in Fig. 4-1 are the variations of the parallel and perpendicular exchange bias fields HE as a function of the Co thickness derived from the experimental curves in Fig. 3-5. We also present unidirectional anisotropy constant (JK) versus the Co thickness, which is defined as: 2 SCoE K MtHJ ××= (Eq. 4-3) - 40 - 2 3 4 5 6 7 8 9 1 0 1 1 0 1 0 0 0 2 0 0 0 3 0 0 0 4 0 0 0 P e rp e n d ic u la r P a ra lle l T = 1 2 0 K H C (O e) tC o (n m ) 2 3 4 5 6 7 8 9 1 0 1 1 0 5 0 0 1 0 0 0 1 5 0 0 2 0 0 0 P e rp e n d ic u la r P a ra lle l T = 1 2 0 K H E ( O e) tC o (n m ) 2 3 4 5 6 7 8 9 1 0 1 1 0 .0 0 0 .0 5 0 .1 0 0 .1 5 0 .2 0 P e rp e n d ic u la r P a ra lle l T = 1 2 0 K J K (e rg /c m 2 ) tC o (n m ) Fig. 4-1. The Co thickness dependence of perpendicular and parallel exchange bias fields (HE) , unidirectional anisotropy constant (JK) and coercitivity (HC). - 41 - Where HE is the exchange bias field, tCo is the Co thickness and MS is the saturation magnetization of Co layer which is in order of 320 (emu/cm3) in these measurements. A factor ½ in Eq. 4-3 is due to the fact that one layer has two interfaces. In this series of samples, tMnPd is fixed at 10 nm while tCo is varied from 2.5 to 10 nm. The maximum values of the parallel and perpendicular exchange bias fields in this series are very large (950 and 1650 Oe, respectively). As for the interfacial exchange energy, JK in the perpendicular cases is higher than that in the parallel ones. The maximum values are respectively 0.15 and 0.17 erg/cm2 for parallel and perpendicular exchange biases obtained at the highest tCo in the present study and there is no sign of leveling off. It seems, however, that the interfacial exchange energy in the parallel direction will cross over that in the perpendicular one as tCo passes a certain value larger than 10 nm. 4.3.2.2 MnPd thickness dependence of exchange bias Fig. 4-2 illustrates the variations of the parallel and perpendicular exchange fields with varying tMnPd as derived from the experimental curves in Fig. 3-6. In this series of samples, tCo is fixed at 3.5 nm while tMnPd is varied from 3.5 to 30 nm. The perpendicular exchange bias field increases as increasing tMnPd from 3.5 to 7.5 nm, and it decreases gradually with tMnPd value larger than 7.5 nm. Meanwhile, in the parallel cases, the situation is in the opposite, i.e., HE decreases as varying tMnPd from 3.5 to 7.5 nm and increases with tMnPd larger than 7.5 nm. At the same tMnPd value of 7.5 nm, the perpendicular exchange bias field reaches to a maximum value up to 1650 Oe while the parallel exchange bias field has a minimum value down to 250 Oe. The maximum HE in the parallel direction is up to 1750 Oe at the smallest tMnPd in the present study. - 42 - 0 5 10 15 20 25 30 0 500 1000 1500 2000 P e rp e n d ic u la r P a ra lle l H E ( O e) tM nP d (n m ) 0 5 10 15 20 25 30 0 1000 2000 3000 4000 5000 6000 P e rp e n d ic u la r P a ra lle l H C (O e) tM nP d (n m ) Fig. 4-2. The MnPd thickness dependence of perpendicular and parallel exchange bias fields (HE), coercitivity (HC). It is very interesting to note that by the optimum choice of the thicknesses of Co and MnPd layers in these two systems, we have obtained samples with “pure” perpendicular exchange bias effect, meaning that the parallel exchange bias field is much less than the perpendicular one. This property is rather - 43 - unique comparing with another work in [FeMn/FePt]10 multilayers in which both perpendicular and parallel exchange biases are always coexistent (see Phuoc et al. [59]). 4.3.3 Perpendicular magnetic anisotropy in MnPd/Co multilayers The present section focuses on the perpendicular magnetic anisotropy found in MnPd/Co multilayers. From the hysteresis loops, the effective magnetic anisotropy can be readily obtained from the area enclosed between the parallel and perpendicular magnetization curves [64]. It is well established that the effective magnetic anisotropy could be phenomenologically separated into a volume contribution KV and a contribution from the interfaces KS, and approximately described by: Co S Veff t KKK 2+= (Eq. 4-4) This relation just represents a weighted average of the magnetic anisotropy energy of the interface atoms and the inner atoms of a ferromagnetic layer of thickness tCo. The factor of 2 implies that one ferromagnetic layer is assuming to be bounded by the two identical interfaces. Eq. 4-4 is commonly used in experiment studies to determine the magnitudes of KV and KS by plotting the product of KefftCo versus tCo as in Fig. 4-3-(a). It should be noted that Eq. 4-4 could be rewritten as: SCoVCoeff KtKtK 2+= (Eq. 4-5) Combining Eq. 4-5 with the linear fit of the plot of KefftCo versus tCo, one can readily deduce that the slope of the fit line gives KV and the vertical axis intercept equal 2KS. Below a certain thickness (-2KV/KS), the interface anisotropy contribution outweighs the volume contribution, resulting in a perpendicularly magnetized system. - 44 - A negative Keff describes the case of a preferred orientation of the magnetization parallel to the plane. Inversely, a preferred orientation of the magnetization is perpendicular to the plane if Keff is positive. The negative slope indicates that a negative volume anisotropy KV, favoring parallel magnetization, while the intercept at zero Co thickness indicates a positive interface anisotropy, KS, favoring perpendicular magnetization. 4.3.3.1 Perpendicular anisotropy at low temperature As mentioned before, in Fig. 3-5 and Fig. 3-6, the perpendicular anisotropy was clearly evidenced. In the series of samples with varying tCo, the perpendicular easy axis transforms into the parallel one as tCo passed a critical value of 9 nm (see Fig. 4-3-(a)). Inversely, the easy axis switches from the parallel to the perpendicular direction for the series of samples with increasing tMnPd from 3.5 to 30 nm (see Fig. 3-6). It is remarkable that the spin reorientation in these multilayers is observed for the first time in our experiments. From the fit line shown in Fig. 4-3-(a), the value of |KV| is in the order of 106 (erg/cm3) and KS is about 0.6 (erg/cm2). Quantitative analysis on the perpendicular anisotropy of the thin films or multilayers can be based on the so-called uniaxial magnetic anisotropy (KU). This quantity can be calculated from Keff and MS by using definition: [64] KU = Keff + 2πMs2 (Eq. 4-6) The obtained result is shown in Fig. 4-3-(b). We note that the uniaxial magnetic anisotropy energy in the present samples is found to be large (in order of 106 (erg/cm3)) and slowly reduces with increasing tCo. It should be noted that the saturation magnetization of Co layers in the present study is of about 320 (emu/cm3) which is much smaller than the bulk Co value of about 1400 (emu/cm3) leading to the reduced |KV| for which the - 45 - major contribution is from the demagnetizing field energy. However, it is interesting that the low saturation magnetization of Co layers results in a reduced shape anisotropy, which is usually the main opponent of perpendicular magnetic anisotropy. It is mentioned that we have ignored the effect of the cooling field in calculating the anisotropy energies. However, to some extent, they are still invaluable and may be used to estimate the perpendicular anisotropy energy. 0 1 2 3 4 5 6 7 8 9 10 11 -0.5 0.0 0.5 1.0 1.5 Linear fit K ef f× t C o ( er g/ cm 2 ) tCo (nm) 2KS KV (a) T=120 K 2 3 4 5 6 7 8 9 10 11 0 1 2 3 4 K U (1 06 e rg /c m 3 ) tCo(nm) (b) T=120 K Fig. 4-3. (a) The plot of the product of Keff and tCo versus tCo and (b) the plot of KU versus tCo of [MnPd(10 nm)/Co(x nm)]10 (x =2.5, 3.5, 4.5, 5.5, 7.5, 10 nm) multilayers at 120K. - 46 - 4.3.3.2 Perpendicular anisotropy at room temperature The preferred orientation of the magnetization perpendicular to the plane is not observed in the as-deposited samples (see Fig. 3-8). For the samples with small tCo, there is not the preferred orientation of the magnetization. Meanwhile, the orientation of the magnetization parallel to the plane is preferred at large tCo. However, this property is improved considerably by annealing and the field cooling which will be discussed in details in the next subsections. Fig. 4-4-(a) shows a deviation from the linear behavior at small tCo. This effect is often present in the anisotropy studies of transition metal multilayers [64]. Many explanations have been given in this case. However, it is noted that this behavior was absent at low temperature. Therefore, a lowering of Curie temperature with the magnetic layer thickness, which is a well-known finite-size effect, can play an important role in the case of room temperature measurements [64]. As shown in Fig. 4-4-(b), KU is enhanced considerably by annealing and the field cooling. In some cases, the enhancement of KU makes its value overcome the demagnetization energy, therefore the preferred orientation of the magnetization switches from the in-plane to the perpendicular direction. The highest difference in KU (post and pre – processing) is observed up to 106 (erg/cm3) for the samples with tCo = 7.5 nm. 4.3.3.3 Effect of annealing on perpendicular anisotropy As shown in Fig. 4-4, the annealing process enhances the perpendicular orientation of the magnetization in comparison with the as-deposited samples - 47 - indicating that this process has influenced the crystal structure of the multilayers which is directly related to magnetic properties. 2 3 4 5 6 7 8 9 10 11 -1.5 -1.0 -0.5 0.0 0.5 1.0 Perpendicular FC Parallel FC ZFC As-deposited K ef ft C o ( er g/ cm 2 ) tCo (nm) (a) 2 3 4 5 6 7 8 9 10 11 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 Perpendicular FC Parallel FC ZFC As-deposited K U (1 06 e rg /c m 3 ) tCo (nm) (b) Fig. 4-4. Anisotropy energies of [MnPd/Co]10 multilayers which were treated at different conditions. (a) Plot of the product of Keff and tCo versus tCo and (b) plot of KU versus tCo at room temperature. - 48 - Many previously reported investigations have shown the origins of the perpendicular anisotropy in [Pd/Co] multilayers. The main contribution to the anisotropy is ascribed to the lattice mismatch between the layer Co and Pd [65, 66] and/or lowered symmetric property at the interface (Néel’s model) [67]. However, these phenomena often happen in the systems with the Co thickness of about few angstroms which is not in the case of our experiments where the Co thickness is order of nanometers. Another source of perpendicular anisotropy might come from c-axis preferred orientation of hcp Co. This assumption is also ruled out in our multilayers since XRD patterns show almost no peaks for Co. Since the composition analysis shows that the Pd composition in the AF layers up to 89 at. %, it is likely that interdiffusion between the MnPd and Co layers can give rise to the formation of interfacial alloys of Co-Pd. Therefore, we propose a schematic diagram of our multilayer structure as shown in Fig. 4-5. The CoPd alloy is known to have a large magnetostriction and large magnetic anisotropy [68]. Since the perpendicular anisotropy exists only as Co layer kept between two mono-layers, these interfacial alloys can experience significant tress due to lattice mismatch with MnPd layer. Thus, a strained interfacial alloy can be the origin of the perpendicular anisotropy in the multilayers. This suggestion is also consistent with the previous studies in Co/Pd [69, 70] and CoPd/Pd [71] multilayers. The interdiffusion is present even at room temperature [69], thus KU in the as-deposited sample may partly come from the strained interfacial alloy. The annealing process is understood as the enhancement of the interdiffusion and crystalline quality. It is worth noting that the preferred orientation of the magnetization of the as-deposited multilayer of [MnPd(10 nm)/Co(7.5 nm)]10 switched from - 49 - the in-plane to the perpendicular direction as the sample had been annealed at 590 K for 5 minutes in vacuum and then cooled down to room temperature whether or not magnetic field was applied (see Fig. 4-4-(a)). From the assumption above, the Co and MnPd thicknesses (tCo and tMnPd, respectively) are understood as nominal values. Simultaneously, the deviation from the linear behavior shown in Fig. 4-4-(a) also originates from the local composition variations of CoPd interfacial alloys which have TC depending on their composition. Unfortunately, it is very difficult to determine these compositions. Si substrate CoPd CoPd MnPd Co CoPd MnPd Co Fig. 4-5. Schematic diagram of multilayer structure after annealing. - 50 - 4.3.3.4 Anomalous field induced anisotropy Studies on field-induced anisotropy in magnetic materials are not of particular interest in recent years. Naturally, one may expect that the cooling- field forces the easy axis direction parallel to the applied field. However, it is not in agreement with the present study. Anomalous field induced anisotropy at room temperature was observed for the first time in [MnPd/Co]10 multilayers. The uniaxial magnetic anisotropy was enhanced as the sample field cooled in the parallel magnetic field instead of that in the perpendicular one as usual. Meanwhile, the uniaxial anisotropy in the samples cooled in the perpendicular field is not improved comparing with that after annealing and zero-field cooling (see Fig. 4-4-(b)). The result shows that the field cooling direction has strongly influenced on the anisotropy property of the multilayers. We try to explain this anomalous phenomenon by estimating the effect of the field cooling on each layer of the sample. The MnPd and Co layers are eliminable because MnPd is paramagnetic at room temperature and Co is less crystalline. Since CoPd alloys have extremely large negative magnetostriction constants (λS = -1.5 × 10-4) [70], the parallel field can give rise to in-plane additional tensile stress compared with the zero field due to the enhancement of the misfit between CoPd interfacial alloy and MnPd layer. Meanwhile, the perpendicular field may not lead to the strong enhancement of this misfit because the strain of CoPd caused by the out-of-plane field is mainly in the normal direction. For the sample with largest tCo, which the parallel orientation of the magnetization after annealing is preferred, the anomalous effect is absent. In this case, the CoPd alloy layer might become stable. It is likely due to the fact - 51 - that the demagnetization field energy in the FM layer is higher than the uniaxial anisotropy. And therefore, it could give rise to the decline of the uniaxial anisotropy energy. From the analysis above, it can be seen that the anisotropy constants at low temperature are not completely exact due to the strong effect of the field cooling (see 4.3.3.1). The nature of the anomalous field induced anisotropy is stress induced anisotropy. 4.3.4 Temperature dependence of magnetization in MnPd/Co multilayers As shown in Fig. 3-12, three features are noticeable: (i) the ZFC curve exhibits a peak at 180 K; (ii) the forward and downward parts considerably depart from each other below this peak temperature and (iii) the reduction of the magnetization with increasing temperature. It is should be noted that the reduction of the magnetization is absent for the samples with large tCo. Besides, the magnetization contribution of CoPd interfacial alloy is considerable due to the polarization of Pd [69]. Appearance of a peak in the ZFC curve owes to blocking mechanism arising from a competition between the thermal energy and the magnetic anisotropy energy. Departure of the backward from the forward parts is suggestive of temporal relaxation, i.e., evolution of magnetization with time. The magnetization of the sample measured at small field perpendicular to the film plane decreases with increasing temperature. In the samples with small tCo, the reduction of the magnetization is found clearly, however it is not considerable in the samples with large tCo. Therefore, first of all, it is due to the reduction of the Curie temperature in Co thin films. Besides, the reduction of the magnetization partly comes from the CoPd interfacial alloys which - 52 - have local composition variations and the Curie temperature ranging from 1404 K for pure Co down to 130 K for a 3 at.% Co alloy [72]. 4.4 Explanation of exchange bias coupling mechanism Based on the studies on exchange bias and perpendicular anisotropy, we try to give a model to describe qualitatively the observed behaviors. The analysis of the origin of the perpendicular anisotropy in the previous section shows that it is due to the stressed alloying at the interface MnPd/Co. As experimentally observed in Fig. 3-5 and Fig. 3-6, the resultant easy axis is perpendicular to the plane in most of the samples, except some samples which parallel easy axis is favorable. Thus, there are two configurations of the FM spins corresponding to two preferred orientations of the magnetization. ¾ For the perpendicular-to-the-plane easy axis, the spins inside the FM layers and also the spins of FM layers at the interfaces are normal to the film plane. ¾ For the parallel-to-the-plane easy axis, the spins inside the FM layers lie in the plane of the film meanwhile the spins of the FM layers at the interfaces are canted with respect to the film plane. Because in this case, the demagnetization field energy is higher than the uniaxial magnetic anisotropy energy. As for exchange bias coupling, in the present study, the coupling between the FM layers and the AF layers is bilinear then the exchange bias fields will be proportional to the projection of the MnPd spins to the field cooling direction. Thus out-of-plane MnPd spin components are necessary to obtain perpendicular exchange bias, while in-plane MnPd spin components are necessary to obtain parallel exchange bias. Accordingly an out-of-plane loop shift would vanish, if the MnPd spins would completely lie in the plane. - 53 - Since the multilayers have not sharp interfaces due to the interdiffusion between the Co and MnPd layers, a phenomenological picture shown in Fig. 4-6 is reasonable. The fluctuations of the MnPd spins at the interfaces were attributed to the origin of the exchange bias effect. These fluctuations may originate from structural defects, roughness or exchange interaction to the FM layer at the interface. Regarding the MnPd thickness dependence of exchange bias, this behavior may be related to the thickness dependence of magnetic structure of MnPd layers at the interface. Obviously, there is the “conversion” of the exchange bias coupling energy between the perpendicular and parallel cases. It gives rise to the fact that the out-of-plane and in-plane AF spin components at the interface change as varying the MnPd thickness. At small tMnPd, the preferred orientation of the magnetization is in the film plane due to the fact that the strain between CoPd interfacial alloy and MnPd layer is negligible. It partly gives rise to the minority of the out-of-plane spin components. Therefore, the exchange bias field in the parallel case is stronger than the perpendicular one. Whereas, at larger tMnPd, the anisotropy property changes from the in-plane to the perpendicular direction, the majority of the out-of-plane components makes the perpendicular exchange bias predominate. However, after that, the exchange bias field in the parallel direction increases slowly and overcomes that in the perpendicular one which decreases gradually with increasing tMnPd. This behavior may be attributed to the AF spin reorientation at the interface which originates from a decrease of structural defects or rearrangement of the MnPd structure. Unfortunately, the experimental confirmation of this situation has not been carried out. Using this phenomenological picture, we try to explain the Co thickness dependence of exchange bias as keeping the MnPd thickness constant. At - 54 - a) FM AF FM AF b) Fig. 4-6. Schematic view of spin configurations of MnPd/Co multilayer: (a) perpendicular-to-the-plane easy axis and (b) parallel- to-the-plane easy axis. small tCo, the parallel and perpendicular exchange bias coupling energies (JK) increase with increasing tCo. It is likely related to the magnetic structure of the FM layer at the interface with the formation of the CoPd alloy. Unfortunately, - 55 - the composition and magnetic structure of the alloy can not be controlled because of the random intermixing. On the other hand, since the spins in the FM layers perpendicular to the film plane (see Fig. 4-6-(a)), they induce the spins arrangement in MnPd layers. The out-of-plane spin components in MnPd layers are larger than the in-plane one. Therefore, the exchange bias energy in the perpendicular direction is higher than that in the parallel one. However, with increasing tCo, the easy axis of the magnetization rotated from the perpendicular to the in-plane. The spin configuration in the FM layers changes as shown in Fig. 4-6-(b). Thus, the spin arrangement in MnPd layers also changes. The out-of-plane spin components reduce and the in-plane spin components increase gradually. Hence, the interfacial exchange bias coupling energy in the parallel direction will cross over that in the perpendicular one as tCo passes a certain value larger than 10 nm. From this phenomenological picture and the analysis above, we can explain why in the previously reported studies [60, 61, 62] and unpublished works by the Spintronics group (ITIMS), parallel exchange bias in MnPd/Co bilayers is large meanwhile perpendicular exchange bias is smaller. It is simple due to the restricted formation of CoPd interfacial alloy at the interface between the MnPd and Co layers. This could give rise to the spin canting at small angle to the film plane in Co layer at the interface. Thus, the out-of- plane spin components of MnPd layer at the interface are much less than the in-plane ones. - 56 - CONCLUSIONS AND FURTHER DIRECTION In the present work, the results on the exchange bias effect in Si/[MnPd/Co]10 multilayers are reported for the first time. ¾ Multilayers of Si/[MnPd/Co]10 were prepared successfully using the RF sputtering system. Structural characterization by XRD shows that MnPd layers are polycrystalline with fcc phase and Co layers may be less crystalline. ¾ The exchange bias effect was found in both the parallel and perpendicular directions. At T = 120 K, the largest exchange bias fields were extremely high, up to 1750 and 1650 in the parallel and perpendicular cases, respectively. The maximum value of JK is observed to be of 0.15 (erg/cm3) for parallel exchange bias and 0.17 (erg/cm3) for perpendicular exchange bias. Especially, we have obtained samples with nearly “pure” perpendicular exchange bias effect, meaning that the parallel exchange bias field is much less than the perpendicular one. ¾ Perpendicular magnetic anisotropy was also found in these samples. The easy axis direction strongly depends on both the MnPd and Co thicknesses. ¾ The origin of the perpendicular magnetic anisotropy was attributed to the formation CoPd interfacial alloy which causes a significant magneto-elastic effect. ¾ A phenomenological picture was proposed to explain the exchange bias effect. Fluctuations of the MnPd spins at the interface are presumed to be the key point in the perpendicular exchange bias mechanism. ¾ The studied results also show the anomalous effect related to field-induced anisotropy. The field cooling in the parallel direction enhanced the perpendicular anisotropy property instead of that in the perpendicular one. - 57 - ¾ In return, the strong perpendicular anisotropy plays a vital role in the behavior of parallel and perpendicular exchange biases, and vice versa, suggesting that one must take caution of the interplay between perpendicular magnetic anisotropy and exchange bias when studying perpendicular exchange biased systems. The study also gives a certain technological advance for applications of multilayer thin films since we can simultaneously control both strong perpendicular exchange bias and strong perpendicular magnetic anisotropy. More investigations are needed for a better understanding of the perpendicular exchange bias effect such as: determining the blocking temperature in both the parallel and perpendicular cases, investigating influence of the anomalous field cooling effect on exchange bias and also the origin of the perpendicular exchange bias effect and anisotropy in the interplay between them... - 58 - REFERENCES [1] W.P. Meikleijohn, C.P. Bean, Phys. Rev. 102 (1956) 1413. [2] D. Mauri et al., J. Appl. Phys. 62 (1987) 3047. [3] A.P. Malozemoff, Phys. Rev. B 35 (1987) 3679. [4] A.P. Malozemoff, Phys. Rev. B 37 (1988) 7673. [5] A.P. Malozemoff, J. Appl. Phys. 63 (1988) 3874. [6] K. Takano, R.H. Kodama, A.E. Berkowitz, W. Cao, and G. Thomas, Phys. Rev. Lett. 79 (1997) 1130. [7] W.H. Meiklejohn, C.P. Bean, Phys. Rev. 105 (1957) 904. [8] S. Gangopadhyay, G.C. Hadjipanayis, C.M. Sorensen, K.J. Kalbunde, IEEE Trans. Mag. 28 (1992) 3174. [9] Y. Wang, Y. Zhang, Y. Cao, M. Lu, J. Yang, J. Magn. Magn. Mater. (2007) in press. [10] W.H. Meiklejohn, J. Appl. Phys. 29 (1958) 454. [11] L.D. Bianco, D. Fiorani, A.M. Tesla, E. Bonetti, L. Signorini, Phys. Rev. B 70 (2004) 052401. [12] V. Papaefthymiou, A. Kostikas, A. Simopulos, D. Niarchos, S. Gangopadhyay, G.C. Hadjipanayis, C.M. Sorensen, K.J. Klabunde, J. Appl. Phys. 67 (1990) 4487. [13] J. Sort, J. Nogués, S. Suriđach, J.S. Muđoz, M.D. Barĩ, E. Chappel, F. Dupont, G. Chouteau, Appl. Phys. Lett. 79 (2001) 1142. [14] J. Sort, S. Suriđach, J.S. Muđoz, M.D.Barĩ, J. Nogués, G. Chouteau, V. Skumryev, G.C. Hadjipanayis, Phys. Rev. B 65 (2002) 174420. [15] J. Nogués, J. Sort, V. Langlais, V. Skumryev, S. Suriđach, J.S. Muđoz, M.D.Barĩ, Phys. Rep. 422 (2005) 65117. - 59 - [16] V. Skumryev, S. Stoyanov, Y. Zhang, G. Hadjipanayis, D. Givord, J. Nogués, Nature (London) 423 (2003) 850. [17] J. Eisenmenger, I.K. Schuller, Nat. Mater. 2 (2003) 437. [18] Y. Yamamoto, H. Nakagawa, U. Hori, J. Magn. Magn. Mater. 310 (2007) 2384. [19] O. Irglesias, X. Batlle, A. Labarta, J. Magn. Magn. Mater. 316 (2007) 140. [20] A. Mumtaz, K. Maaz, B. Janjua, S.K. Hasanain, M.F. Bertino, J. Magn. Magn. Mater. 313 (2007) 266. [21] K.N. Trohidou, M. Vasilakaki, L.D. Bianco, D. Fiorani, A.M. Testa, J. Magn. Magn. Mater. 316 (2007) e82. [22] N. Domingo, A.M. Testa, D. Fiorani, C. Binns, S. Baker, J. Tejada, J. Magn. Magn. Mater. 316 (2007) 155. [23] P.Z. Si, D. Li, C.J. Choi, Y.B. Li, D.Y. Geng, Z.D. Zhang, J. Magn. Magn. Mater. 142 (2007) 723. [24] H. Moradi, J. Magn. Magn. Mater. 310 (2007) e539. [25] Z.Y. Liu, J. Magn. Magn. Mater. 281 (2004) 287. [26] S. Bruck, E. Goering, Y.J. Tang, G. Schutz, A.E. Berkowitz, J. Magn. Magn. Mater. 310 (2007) 2316. [27] A.E. Berkowitz, K. Takano, J. Magn. Magn. Mater. 200 (1999) 552. [28] M. Kiwi, J. M. Lopez, R.D. Portugal, R. Ramirez, Solid State Com. 116 (2000) 315. [29] P.J. van der Zagg, A.R. Ball, L.F. Feiner, R.M. Wolf, P.A.A. van der Heijden, J. Appl. Phys. 78 (1996) 5103. [30] O. Petracic, Z.P. Li, I.V. Roshchin, M. Viret, R. Morales, X. Battle, I.K. Schuller, Appl. Phys. Lett. 87 (2005) 222509. - 60 - [31] S. Nicolodi, L.C.C.M. Nagamine, A.D.C. Viegas, J.E. Schmidt, L.G. Pereira, C. Deranlot, F. Petroff, J. Geshev, J. Magn. Magn. Mater. 316 (2007) e97. [32] M. Ali, C.H. Marrows, and B.J. Hickey, Phys. Rev. B 67 (2003) 172405. [33] V.P. Nascimento, A. Biondo, V.B. Nunes, E. BaggioSaitovitch, E.C. Passamani, Appl. Surf. Sci. 253 (2007) 6248. [34] I. Sasaki, R. Nakatani, K. Ishimoto, Y. Endo, Y. Shiratsuchi, Y. Kawamura, M. Yamamoto, J. Magn. Magn. Mater. 310 (2007) 2677. [35] C. Ning, K. Kannan M., Girt E., Farrow R.F.C., Marks R.F., Kellock A., Young A., Huan C.H.A., J. Appl. Phys. 87 (2000) 6647. [36] K. Imakita, M. Tsunoda, M. Takahashi, Appl. Phys. Lett. 85 (2004) 3812. [37] Z.W. Jiao, P.Z. Si, W.D. Jiang, Q.Wu, G.X. Ye, J. Alloys and Compounds (2007) in press. [38] N.C. Koon, Phys. Rev. Lett. 78 (1997) 4865. [39] T.C. Schulthess, W.H. Butler, Phys. Rev. Lett. 81 (1998) 4516. [40] P. Miltényi, M. Gierlings, J. Keller, B. Beschoten, G. Güntherodt, U. Nowak, K.D. Usadel, Phys. Rev. Lett. 84 (2000) 4224. [41] U. Nowak, A. Misra, K.D. Usadel, J. Appl. Phys. 89 (2001) 7269. [42] U. Nowak, A. Misra, K.D. Usadel, J. Magn. Magn. Mater.240 (2001) 243. [43] U. Nowak, K.D. Usadel, J. Keller, P. Miltényi, B. Beschoten, G. Güntherodt, Phys. Rev. B 66 (2002) 014430. [44] A. Misra, U. Nowak, K.D. Usadel, J. Appl. Phys. 93 (2003) 6593. [45] A. Misra, U. Nowak, K.D. Usadel, J. Appl. Phys. 95 (2004) 1357. [46] B. Beckmann, U. Nowak, K.D. Usadel, Phys. Rev. Lett. 91 (2003) 187201. - 61 - [47] D. Suess, T. Schrefl, W. Scholz, J.V. Kim, R.L. Stamps, J. Fidler, IEEE Trans. Magn. 38 (2002) 2397. [48] D.Suess, K. Krishner, T. Schrefl, J. Fidler, R.L.Stamps, J.V. Kim, Phys. Rev. B 67 (2003) 054419. [49] M. Kirschner, D. Suess, T. Schrefl, J. Fidler, J.N. Chapman, IEEE Trans. Magn. 39 (2003) 2735. [50] D. Lederman, R. Ramírez, M. Kiwi, Phys. Rev. B 70 (2004) 184422. [51] B. Kagerer, Ch. Binek, W. Kleemann, J. Magn. Magn. Mater. 217 (2000) 139. [52] Ch. Binek, B. Kagerer, S. Kainz, W. Kleemann, J. Magn. Magn. Mater. 226-230 (2001) 1814. [53] S. Matt, K. Takano, S.S.P. Parkin, E.E. Fullerton, Phys. Rev. Lett. 87 (2001) 087202. [54] C.H. Marrows, Phys. Rev. Lett. B 68 (2003) 012405. [55] F. Garcia, J. Sort, B. Rodmacq, S. Auffret, B. Dieny, Appl. Phys. Lett. 83 (2003) 3537. [56] J. Sort, F. Garcia, S. Auffret, B. Rodmacq, B. Dieny, V. Langlais, S. Suriđach, J. S. Muđoz, M. D. Barĩ, J. Nogués, Appl. Phys. Lett. 87 (2005) 242504. [57] S.S. Kim, J.Y. Hwang, J.R. Rhee, J. Magn. Magn. Mater. 310 (2007) 2310. [58] L. Sun, P.C. Searson, C.L. Chien, Phys. Rev. B 71 (2005) 012417. [59] N.N. Phuoc, T. Suzuki, IEEE Trans. Magn. 43 (2007) 897. [60] N.N. Phuoc, N. A. Tuan, N. P. Thuy, D. Babonneau and J. Rabier, Physica B: Condensed Matter 327 (2003) 385. [61] N.T. Nam, N.P. Thuy, N.A. Tuan, N.N. Phuoc and T. Suzuki, J. Magn. Magn. Mater. 315 (2007) 82. - 62 - [62] N.P. Thuy, N.A. Tuan, N.N. Phuoc, N.T. Nam, T.D. Hien and N.H. Hai, J. Magn. Magn. Mater. 304 (2006) 41. [63] M. Chládek, V. Valvoda, C. Dorner, C. Holý and J. Grim, Appl. Phys. Lett. 69 (1996) 1318. [64] M.T Johnson, P.J.H. Bloemen, F.J.A. den Broeder, J.J de Vries, Rep. Prog. Phys. 59 (1996) 1409. [65] K. Umeda, Y. Fujiwara, T. Matsumoto, K. Nakagawa, A. Itoh, J. Magn. Magn. Mater. 156 (1996) 75. [66] B.N. Engel, C.D. England, R.A. Van Leeuwen, M.H. Wiedmann, C.M. Falco, Phys. Rev. Lett. 67 (1991) 1910. [67] H. Takahashi, S. Tsunashima, S. Iwata, S. Uchiyama, Jpn. J. Appl. Phys. 32 (1993) L1328. [68] J. Carrey, A.E. Berkowitz, W.F. Egelhoff, D.J. Dsmith, Appl. Phys. Lett. 83 (2003) 5259. [69] J.I. Hong, S. Sankar, A.E. Berkowitz, W.F. Egelhoff Jr., J. Magn. Magn. Mater. 285 (2005) 359. [70] Sang-Koog Kim, Vladimir A. Chernov, Yang-Mo Koo, J. Magn. Magn. Mater. 170 (1997) L7. [71] Sang-Koog Kim, Jeong-Won Lee, Jong-Ryul Jeong, Sung-Chul Shin, J. Magn. Magn. Mater. 240 (2002) 543. [72] D.E. Nagle, P.P. Craig, P. Barrett, D.R.F. Cochran, C.E. Olsen, and R.D. Taylor, Phys. Rev. 125 (1962) 490.