鐡錳鐡薄膜成長於銅三金之晶體結構與磁性

全文

(1)國立臺灣師範大學物理研究所碩士論文. 指導教授：林文欽博士. Crystalline Structure and Magnetism of Fe/Mn/Fe/Cu3Au(001) 鐡錳鐡薄膜成長於銅三金之晶體結構與磁性研究. 研究生：邱傑振撰中華民國 98 年 6 月.

(2) 誌. 謝. 首先要感謝的是林文欽老師，從入學的暑假一直到畢業，整整二年寒暑從未間斷的辛勤教導，不僅於課業上給予指導，更教曉我做人做事的態度，老師對我的照顧及付出，點滴在心，無限感激！另外，要感謝台大物理系林敏聰教授在研究上的啟發及教導，並提供良好的實驗設備及環境。感謝在同步輻射中心做實驗的過程中，魏德新博士及陳悅來博士在各方面的提示和幫忙，還有口試委員宋克嘉博士、郭建成教授所提供之寶貴意見，使論文得以更加完善。台大物理系奈米磁學研究團隊中，要特別感謝王柏堯學長，一路走來不管在研究方向、實驗態度以及儀器維護給予悉心嚴格的指導與幫忙。另外，感謝學長紀乃友、莊程豪、翁聖勳、林盈志以及其他成員於研究方向及報告表達方面給予寶貴意見。感謝師大物理系同窗莊孟動、林彥穎、蔡承叡以及低維度磁學研究團隊的其他成員於課業與生活上不吝給予協助，與你們相處的日子感到無限的歡樂。最後，感謝林楹璋、邱清源、許紘瑋以及同步輻射中心奈米科學小組的老師及同仁在實驗和生活上給予大力支持。特別感謝我的父親邱裕亮先生與母親賴美招女士，如果不是你們從小到大無怨無悔的支持與包容，我將無法有今日的一切，感謝我的弟弟傑明，一路走來對家庭的付出讓我沒有後顧之憂。曉貞姐，感謝.

(3) 妳於百忙之中能傾聽我的不快，祝妳事事順心。當然還要感謝我的女朋友伊庭不離不棄的照顧，總是在我無助時，適時給予我安定的力量。最後，要感謝的人實在太多，如有被遺忘的朋友在此亦一併感謝，感謝大家於這二年來對我的照顧。在此也祝福所有還在學的朋友們都能夠在求學的路上找到合適自己的方向！ 2009 年 6 月 26 日.

(4) Crystalline Structure and Magnetism of Fe/Mn/Fe/Cu3Au(001) Jie-Jhen Ciou. Abstract The growth, crystalline structure and magnetism of Fe/Mn/fcc-like Fe/Cu3 Au(001) are characterized by medium energy electron diffraction (MEED), low energy electron diffraction (LEED) and magneto-optical Kerr effect (MOKE). Compared with Fe/Mn/Cu3 Au(001), there are some interesting findings in magnetism of Fe/Mn/fcclike Fe/Cu3 Au(001). With the fcc-like Fe buffer layer, the crystalline structure of Mn/fcc-like Fe/Cu3 Au(001) is still face-centered cubic (fcc) structure at low coverage. The key factor responsible for the extraordinarily enhanced coercivity in perpendicular direction of the Fe overlayer and shifted critical thickness of spinreorientation transition (SRT) is attibuted to the perpendicular magnetization of fcc-like Fe buffer layer..

(5) TABLE OF CONTENTS Abstract 1. Introduction 1 2. Apparatus 2.1 UHV System in NSRRC………………………………………….…..3 2.2 Low Energy Electron Diffraction (LEED)…………………….…......4 2.2.1 Theory……………………………………...……………….…......4 2.2.2 Basic components of the standard LEED apparatus………....8 2.2.3 LEED-I/V………………………………………………………….9 2.3 Auger Electron Spectroscopy (AES)…………….…………….….....11 2.3.1 Theory……………………………………………………………11 2.3.2 Basic components of the standard AES apparatus……………….12 2.4 Magneto-Optical Kerr Effect (MOKE)……...………………………13 2.5 Medium Energy Electron Diffraction (MEED)……………………...15 3. Background 18 3.1 Two-dimensional lattices and superlattices....…………………….…18 3.2 Film deposition and growth modes.………………………………19 3.3 Artificial fcc Mn films…………………………...…………………..20 3.4 Hysteresis loop……………………………………………………....21 3.4.1 Hard magnetic material……………………………..………...….22 3.4.2 Soft magnetic material……………………………………………23 3.5 The exchange interaction between FM and AFM materials………...24 3.6 Magnetic anisotropy…………………………………………………25 4. Experiment and Result 29 4.1 Experiment procedure……………………………………………….29 4.2 Growth and Structure of Mn/Fe/Cu3 Au(001)……………………30 4.3 Growth and Structure of Fe/Mn/Fe/Cu3 Au(001)…………………....35 4.4 Magnetic properties of Fe/Mn/Fe/Cu3Au(001)……………………...37 5. Discussion 46 6. Conclusion 51 Bibliography 52.

(6) Chapter 1 Introduction The study of magnetic materials has been a major study in science due to its widerange applications. For example, with the decrease in track size in sensor heads of higher recording density hard disk drives, instability of exchange coupling between ferromagnetic (FM) / antiferromagnetic (AFM) bilayers becomes an important issue. Two common phenomenons of this coupling are exchange bias and an increased coercivity of the FM layer due to contact with the AFM layer. These two properties of the FM/AFM coupling are utilized in modern data storage technology for pinning FM layers. For example, a unidirectional coercivity enhancement has been discovered in a temperature range below the blocking temperature in Co/CoO bilayers. Below this temperature range, the usual shift of the center of the M-H loops to the negative or antiparallel to the cooling field direction is found. Another important issue is spin-reorientation transition (SRT). Owing to the reduced dimension, the competition between interface anisotropy and volume anisotropy would be a key factor in magnetic ultrathin film system. For example, Co/Pt(111) and Fe/Pt(111) with low coverage reveal perpendicular anisotropy due to perpendicular interface anisotropy. With increasing of the film thickness, the in-plane shape anisotropy contributed from dipole-dipole interaction dominates the anisotropy energy of magnetic thin film. Therefore, this kind of magnetic thin films reveals SRT from perpendicular to in-plane direction with the increasing of film thickness. This thesis is broadly divided into three parts, in first chapter several experimental instruments will be introduced, including the UHV system at beamline 05B2 in National Synchrotron 1.

(7) Chapter 1. Introduction. 2. Radiation Research Center (NSRRC) combined with low energy electron diffraction (LEED), Auger electron spectropy (AES), magneto-optical Kerr effect(MOKE), and medium energy electron diffraction (MEED). Some general background knowledge involves thin film growth, crystalline structure and magnetism will be introduced in chapter two. Detail experiments and results on Fe/Mn/fcc-like Fe/Cu3 Au(001) trilayer will be described in chapter three. Two most important results are coercivity enhancement and spin-reorientation transition due to the fcc-like Fe buffer layer effect. For the exchange coupling between ferromagnetic (FM)/ antiferromagnetic (AFM) bilayers, the fcc-like iron buffer layer with perpendicular magnetization may enhance the exchange coupling between Fe overlayer and Mn layer indirectly and in the meantime, the critical thickness of spin reorientation transition is shifted to higher coverage..

(8) Chapter 2 Apparatus 2.1. UHV System in NSRRC. Figure 2.1: Ultrahigh vaccum system combined with MEED, LEED, AES, and MOKE at 05B2 beamline in NSRRC Due to the large ratio of surface to bulk atoms in ultrathin films and nanoparti3.

(9) 2.2. Low Energy Electron Diffraction (LEED). 4. cles, the surface contamination is crucial to the crystalline structure and magnetic properties. Therefor all the experiments included in the thesis are performed in an ultrahigh vacuum (UHV, base pressure∼2×10−10 ). Basically the experiments are carried out in a multi-functional chambers, as shown in Fig. 2.1. All the experimental apparatuses are combined in it so that every process can be in situ performed . The pumping system consists of an ion pump and a turbo pump combined with two mechanical pump as its fore-pumping. After baking the chamber to about 110-130 ∘. 𝐶 for more than 24 hours, the ultrahigh vacuum (UHV) can be reached. A 4 or 5. dimensional manipulator with the capabilities of cooling and heating samples is also equipped in the UHV system. Medium energy electron diffraction (MEED) is used to monitor the growth condition and to calibrate the thickness of thin films. Auger electron microscopy (AES) is used to check the surface element composition and also helps to calibrate the film thickness. Low energy electron diffraction (LEED) and LEED-I/V are used to characterized the lateral and vertical crystalline structure. Besides, there are also MOKE, sputter gun, residual gas analyzer (RGA), evaporation guns equipped in the chamber for the measurements of magnetic properties, sample cleaning, leak test, and evaporation etc.. 2.2 2.2.1. Low Energy Electron Diffraction (LEED) Theory. Low-Energy Electron Diffraction (LEED) is a technique that is used to investigate the surface crystalline order. Since the electrons do not penetrate deeply into material, the diffraction pattern that only arises from the atoms at the surface. In order to understand LEED images we need to deal with two-dimensional Bravais Lattices. In general, a crystal is a periodic arrangement of atoms or molecules that can be represented by a unit cell. A real vector 𝑅 and a reciprocal lattice vector 𝐾 can be respectively written as.

(10) 2.2. Low Energy Electron Diffraction (LEED). 5. ⃗ = 𝑟1 𝑎⃗1 + 𝑟2 𝑎⃗2 + 𝑟3 𝑎⃗3 𝑅 ⃗ = 𝑘1 𝑏⃗1 + 𝑘2 𝑏⃗2 + 𝑘3 𝑏⃗3 𝐾 The relation between real and reciprocal bases is. 𝑏⃗1 = 2𝜋. 𝑎⃗2 ⊗ 𝑎⃗3 𝑎⃗1 ⋅ (𝑎⃗2 ⊗ 𝑎⃗3 ). 𝑏⃗2 = 2𝜋. 𝑎⃗3 ⊗ 𝑎⃗1 𝑎⃗1 ⋅ (𝑎⃗2 ⊗ 𝑎⃗3 ). 𝑏⃗3 = 2𝜋. 𝑎⃗1 ⊗ 𝑎⃗2 𝑎⃗1 ⋅ (𝑎⃗2 ⊗ 𝑎⃗3 ). Next we need to know what happens when a wave is scattered from a crystal. If the incident waves are in phase then the scattered waves will also be in phase, and we will get constructive interference when the scattered waves have a mutual phase shift of a integral number of wavelengths. The path length that the lower wave has traveled longer than the upper wave is given by 2𝑑 sin 𝜙 and the infterference condition or Braggs law (Fig. 2.2) is. Figure 2.2: Illustration of Bragg’s Law. 2𝑑 sin 𝜙 = 𝑛𝜆,. 𝑛∈𝑍. This means that if we measure the scattering angles we can obtain information about perpendicular distance in the crystal. We take a look at two scattering centers with mutual distance 𝑑 that are hit by an incident plane wave in direction ⃗𝑛 with wave vector 𝑘 (Fig. 2.3). The wavelength of the wave vector is then given by 𝑘 = 2𝜋/𝜆..

(11) 2.2. Low Energy Electron Diffraction (LEED). 6. The plane wave is scattered into the direction 𝑛⃗′ with wave vector 𝑘 ′ . Scince the scattering is elastic, e.g. 𝑘 = 𝑘 = 2𝜋/𝜆, so the wavelength of the wave vector is invariable except its direction. Constructive interference occurs when the difference in path length must be an integer number of wavelength.. Figure 2.3: Illustration of two scattering waves. ′ 𝑑 cos 𝜙 + 𝑑 cos 𝜙 = 𝑑⃗ ⋅ (⃗𝑛 − 𝑛⃗′ ) = 𝑚𝜆 where m is and integer. mutiply with 2𝜋/𝜆 𝑑⃗ ⋅ (⃗𝑛 − 𝑛⃗′ ) = 2𝜋𝑚 for integer m We now look at a row of atoms with mutual spacing R in a Bravais lattice, ⃗ ⋅ (⃗𝑛 − 𝑛⃗′ ) = 2𝜋𝑚 for integer m and all Bravais lattice vector R 𝑅 ⃗ = (⃗𝑘 − 𝑘⃗′ ) must This is consistent with the condition that the scattering vector 𝐾 be a reciprocal lattice vector, e.g. ⃗ = 𝑘1 𝑏⃗1 + 𝑘2 𝑏⃗2 + 𝑘3 𝑏⃗3 𝐾 All above lead us the Laue condition: The scattering vector 𝑘⃗′ − ⃗𝑘 must be a ⃗ We will have constructive interference if this condition reciprocal lattice vector Δ𝐺. is fulfilled. ⃗ = 𝑘⃗′ − ⃗𝑘 𝐺 The LEED pattern is an image of the reciprocal lattice. Those points in reciprocal space that fulfill the two-dimensional Laue condition result in intensity maxima..

(12) 2.2. Low Energy Electron Diffraction (LEED). 7. These can be explained in a graphical way (Fig. 2.4 and Fig. 2.5) by means of the Ewald construction. The sphere radius represents the wave vector ⃗𝑘 of the. Figure 2.4: Illustration of Ewald sphere incident electron beam, and diffracted beams with wave vector 𝑘⃗′ appear wherever a reciprocal lattice rod intersects the Ewald sphere. The diffracted pattern thus reflects the symmetry of the surface unit cell, and the separation between the beams is inversely proportional to the interatomic distance. If the circle (sphere) intersects a reciprocal lattice point then fulfilling the Laue condition. We therefore get diffraction in all direction where 𝑘⃗′ ends on a reciprocal lattice vector. From the Ewald sphere construction it is easy to show geometrically that the Laue condition are equivalent ⃗ and to Braggs law. If the change of the wave vector by scattering on atoms at 𝑑𝑖 ⃗ is given by 𝐺 ⃗ = 𝑘⃗′ − ⃗𝑘, then the phase shift between the two scattered waves is 𝑑𝑗 ⃗ − 𝑑𝑗). ⃗ The total amplitude scattered from all atoms is then a sum of ⃗ ⋅ (𝑑𝑖 given by 𝐺 the individual scattering events with appropriate inclusions of the phase differences. If we have n atoms in the unit cell, each with a scattering strength then the total amplitude 𝑆𝑘 scattered from the unit cell is 𝑆𝑘 =. 𝑛 ∑. ⃗ ⃗. 𝑓𝑗 (𝑘) exp𝑖𝐺⋅𝑑𝑗. 𝑗=1. This is called the structure factor or structure amplitude. In this case the scattered intensity is equal to ∣𝑆𝑘 ∣2 ..

(13) 2.2. Low Energy Electron Diffraction (LEED). 8. Figure 2.5: Relationship between LEED and Ewald sphere. Figure 2.6: Diagram of LEED apparatus. 2.2.2. Basic components of the standard LEED apparatus. The common LEED apparatus is composed of a hemispherical phosphorescent screen with a fixed electron gun aligned along the central axis of the screen. The crystal sample is positioned at the centre of the hemispherical screen. With this arrangement, diffraction beams leave from the surface of the crystal travel towards the screen. The incident electron beam is emitted from the electron gun behind the hemispherical fluorescent glass screen and strikes the sample through a hole in the screen. Before the electrons impact the screen they must pass through a retarding field energy analyzer. Standard modern LEED optics are illustrated schematically in.

(14) 2.2. Low Energy Electron Diffraction (LEED). 9. Fig. 2.6. The screen is metallic, coated in a phosphorescent material and biased to 4 − 6 𝑘𝑉 . Therefore electrons are accelerated directly on to the surface coating and those which impinge on the screen lead a spot proportional to the beam intensity. The metallic nature of the screen allows the incident electrons to be conducted away and avoiding the undesirable charging. The electron gun is made of a LaB6 filament placed inside a metallic cylinder called wehnelt. Electrons emitted from the filament must be accelerated to an anode which is at a positive potential with respect to the filament. Electrons near the filament have low kinetic energies compared to the total potential drop and so will tend to follow field lines and also can be adjusted by the wehnelt voltage. However, as the electrons are continuously accelerated towards the anode, their kinetic energy may become so much greater through the anode aperture. Because the anode is virtually positive respect to the cathode, strong deviation of the field lines only occurs quite near the anode in a region where electrons are approaching their maximum kinetic energy. Following are a system of electrostatic lenses (A,B,C,D). If we choose the poper potential of the wehnelt, the incident electron beam is possible to be focused by the electrostatic lenses. The LEED Optics typically consist of four hemispherical grids concentric with the luminescent screen, each containing a central hole through which the electron gun is inserted. The first grid is on ground potential and therefore around the sample is a field free region. The next two grids are set to the retarding voltage which is slightly lower than the kinetic energy of the electrons emitted by the electron gun. It repels almost all the inelastically scattered electrons and the elastically scattered electrons pass the next grid which is set to ground voltage. Then they accelerated towards the luminescent screen (set to a high positive voltage). Behind the screen there is a view port in the UHV system so that the LEED pattern can be observed directly or recorded with a CCD camera.. 2.2.3. LEED-I/V. In real experiments, since the incident direction of electron beam is nearly normal to the substrate, by consideration of electron mean free path, the low- energy electrons will penetrate several atomic layers and then be scattered. In this situation,.

(15) 2.2. Low Energy Electron Diffraction (LEED). 10. the interference and scattering within the different layers will also contribute to the intensity of diffraction spots. The setup of LEED-I/V (Fig. 2.7) is used for measuring the interlayer distance by recording the (0, 0) beam intensity varying with the incident energy of the electron beam. From the Bragg condition and the de-Brogile relation, we have: ℎ. 2𝑑 cos 𝜃 = 𝑛𝜆 = 𝑛 √. 2𝑚(𝐸𝑘 − 𝑉 ). with the vertical interlayer distance 𝑑, the kinetic energy of the incident electron beam 𝐸𝑘 , and the potential cost for electrons to escape from the atoms. Therefore 𝐸𝑘 = 𝑛2. ℎ2 +𝑉 8𝑚𝑑2 cos2 𝜃. If 𝐸𝑘 and 𝑛2 are taken as the Y and X axis respectively, we can fit the peaks of LEED-IV curve, as shown in Fig. 2.7(c). What has to be reminded of is that not all the peaks come from the Bragger condition, some of them are due to multidiffraction. Thus we have to pick out the most appropriate peak that can fit the equation.. Figure 2.7: (a) Illustration of Bragg diffraction. (b) LEED-I/V curve of bulk Cu3 Au(100) measured at 100 K. (c) The LEED-I/V curve of bulk Cu3 Au(100) fitting.

(16) 2.3. Auger Electron Spectropy (AES). 2.3 2.3.1. 11. Auger Electron Spectropy (AES) Theory. The principle of the Auger process is illustrated in schematically in Fig. 2.8. It is an atomic non-radiative emission process, mediated by electrostatic interaction. When an atom is irradiated either by high energy photons or electrons, with subsequent core hole formation, it rearranges its electronic structure such that deep initial hole in the core level is filled by an electron from one of the outer shells. This transition may be accompanied by the emission of a characteristic X-ray photon or a Auger transition. If an electron beam of energy 𝐸𝑝 of the order of 𝑘𝑒𝑉 𝑠 incidents. Figure 2.8: Energy level diagram of AES process on the surface of a solid, a continuous spectrum of electron energies ranging from 0 𝑒𝑉 to 𝐸𝑝 can be detected. Peaks of variou mechanisms will be seen in the spectrum. A sharp peak at the energy 𝐸𝑝 and a broad one between 0 and 200 𝑒𝑉 correspond to respectively the elastic scattering and the emission of the secondary electrons as the result of a cascade inelastic scattering inside the solid. Weak peaks can be seen are associated with Auger processes between them described in the following. For an incident electron with the energy 𝐸𝑝 collides a single atom, the inner shells involved in the processes resulting in the emission of an Auger electron are denoted A, B, and C (Fig. 2.8). As the consequence of the collision, an electron from the inner shell A is expelled from the atom, leaving a single ionized atom whose electronic configuration is far from its ground state. The hole left behind is illustrated by an open circle. Levels, B and C, are all shifted downward equivalently. To minimize the energy, an electron from the level B, experiences a decay into the empty state in the.

(17) 2.3. Auger Electron Spectropy (AES). 12. shell A. Following this process, a photon with the corresponding energy is ejected and may be absorbed by an electron from the other shallow level. As the hole moves upward, the shell B is set at the initial energy, since it affected by the fully screened nucleus charge. Then if the energy emitted following the B to A transition will be absorbed by an electron in the C shell, it will leave the atom and the energy gain overcomes the vacuum level. The electron emitted into vacuum is called the Auger electron, whose energy depends on the binding energies of the levels involved in the transitions. It is obvious that the Auger electron collects the whole information of the inner electronic configuration which is determined by the atomic number. The Auger energy of the transition ABC of the element determined by the difference of the total energy before and after the transition.. 2.3.2. Basic components of the standard AES apparatus. Figure 2.9: Diagram of AES apparatus The standard equipments for AES (Fig. 2.9) consist of electron gun, energy analyzer and data processing electronics. The electron gun produces the primary electron beam with a typical energy of 1 to 5 𝑘𝑒𝑉 . In the Auger electron spectra, Auger peaks are superimposed on a rather lager continuous background. This background can be removed by differentiating the energy distribution function 𝑁 (𝐸). The current 𝐼(𝑉 ) collected by the semispherical collector (screen) with the retard-.

(18) 2.4. Magneto-Optical Kerr Effect (MOKE). 13. ing potential 𝑉𝑟 is related to the distribution of Auger number 𝑁 (𝐸) as ∫ ∞ 𝑁 (𝐸)𝑑𝐸 𝐼(𝑉𝑟 ) ∝ 𝑒𝑉𝑟. 𝑑𝐼(𝑉𝑟 ) ∝ (𝐸) 𝑑𝑉𝑟 If a modulating voltage 𝑉𝑚 sin 𝑤𝑡 is applied to the retarded voltage, then 𝐼(𝑉𝑟 ) becomes𝐼(𝑉𝑟 + 𝑉𝑚 sin 𝑤𝑡). In Taylor expansion: ′′. 𝐼 (𝑉𝑟 ) 2 2 (𝑉𝑚 sin 𝑤𝑡)2 + . . . 2! ′′ 𝐼 (𝑉𝑟 ) 2 ′ 𝑉𝑚 + . . .] cos 2𝑤𝑡 + . . . = 𝐼(𝑉𝑟 ) + [𝐼 (𝑉𝑟 )𝑉𝑚 + . . .] sin 𝑤𝑡 − [ 4 ′. 𝐼(𝑉𝑟 + 𝑉𝑚 sin 𝑤𝑡) = 𝐼(𝑉𝑟 ) + 𝐼 (𝑉𝑟 )𝑉𝑚 sin 𝑤𝑡 +. because 𝑉𝑚𝑟 , the higher order terms for 𝑉𝑚 can be neglected. From equation of (3.2), the amplitude for the collected current component with characteristic frequencies 𝑢 ′. and 2𝑢 are proportional to 𝑁 (𝐸) and 𝑁 (𝐸).. 2.4. Magneto-Optical Kerr Effect (MOKE). If a linear polarized light is incident into a ferromagnetic sample, since of the different reflection coefficients of right and left circular polarization components, the reflected beam will become elliptical polarized. This phenomenon is so called magneto-optical Kerr effect. The angle between the primary axis of the elliptical polarization and the linear polarization is called Kerr rotation, and the ellipticity of the elliptical polarization is called Kerr elliptical, as shown in Fig. 2.10. Let. Figure 2.10: Schematic display of DC MOKE.

(19) 2.4. Magneto-Optical Kerr Effect (MOKE). 14. 𝑟+ 𝑒𝑖𝜃+ and 𝑟− 𝑒𝑖𝜃− stand for the reflection coefficients of right and left circular polarization, respectively. The Kerr rotation and Kerr ellipticity can be illustrated as − 𝜑𝐾 = − 𝜃+ −𝜃 and 𝜀𝐾 = 2. 𝑏 𝑎. =. 𝑟+ −𝑟− , 𝑟+ +𝑟−. respectively. Both of them are proven to be. proportional to the magnetization of sample. thus by measuring 𝜑𝐾 and 𝜀𝐾 with cyclic applied magnetic field, we can get the hysteresis loop. In general, there are three types of MOKE measurement. Each of them has different geometry of the magnetization and the light path, as shown in Fig. 2.11. In the polar Kerr effect, the magnetization lies in the plane of incidence and is perpendicular to the surface. In the longitudinal Kerr effect, the magnetization lies in the plane of incidence and is parallel to the surface. In the transverse geometry, the magnetization is perpendicular to the plane of incidence and on the surface. In magnetic ultrathin film, the. Figure 2.11: Different geometry for MOKE measurement Kerr signal is so small that the noise may result in significant effect. Therefore, in our experiments, a modulator is added between the polarizer and the sample such that the modulated signal can be taken by lock-in technique with a larger ratio of signal to noise. The schematic illustration is shown in Fig. 2.12. This method is called AC MOKE. Fig. 2.13 shows the in situ experimental apparatus of polar and longitudinal MOKE. The intensity of the reflection beam is taken by a photodiode, and then amplified by the preamplifier. By the preamplifier, the signal of current from the photodiode is expanded and transformed to voltage signal. Next the voltage signal is sent to the lock-in amplifier. The oscilloscope is used to check the modulation of the laser beam..

(20) 2.5. Medium Energy Electron Diffraction (MEED). 15. Figure 2.12: Schematic display of AC MOKE. Figure 2.13: Diagram of MOKE apparatus. 2.5. Medium Energy Electron Diffraction (MEED). Medium Energy Electron Diffraction (MEED) is a surface sensitive technique, which allow us to measure properties of the sample surface during the growth process. As the names of MEED , suggest the most different characteristics between them are the incident energy and incident angle. In MEED, incident electron beam.

(21) 2.5. Medium Energy Electron Diffraction (MEED). 16. strikes a single crystal surface at a glancing angle, forming a diffraction pattern on a screen as shown in Fig. 2.14. The electrons with 3-5 𝑘𝑒𝑉 order energy for MEED. Figure 2.14: Schematic display of the MEED measurement are focused with glancing angle 3∘ − 5∘ for MEED. Then, the electrons are scattered by the periodic potential of the crystal surface, which results in a characteristic diffraction pattern on the screen. The combination of glancing incidence and strong electron-substrate interactions reduces the penetration depth of incident electrons to a few monolayers. Compared to LEED, 3-100 𝑘𝑒𝑉 electrons are used in MEED. This results in a mean free path in the 1-10 𝑛𝑚 range, which is substantially greater than that used in LEED. However, surface sensitivity is maintained via the glancing incidence geometry which ensures that the normal component of the incident electron wave vector is small. Therefore, the penetration depth will be small. As shown in Fig 2.14. the diffracted intensity is displayed directly on a screen to be detected instantly. Furthermore, MEED arrangement in UHV chamber allows it to be used conveniently for in-situ and real-time observation of MBE thin film growth process. By monitoring the specular spot ((0,0) beam) intensity during deposition, growth information can be obtained in real time for feed back control. In a good layer-bylayer growth system, the MEED intensity reveals regular oscillation behavior at the condition of constant deposition rate. This effect qualitatively correlates the beam intensity and the surface roughness. In general, the decay of MEED intensity is.

(22) 2.5. Medium Energy Electron Diffraction (MEED). 17. attributed to the roughness of the sample surface. However, in other growth modes such as island growth and step flow-growth, MEED oscillation do not reveal the obvious oscillation..

(23) Chapter 3 Background 3.1. Two-dimensional lattices and superlattices. In principle, the surface region of a crystal is a three-dimensional entity; relaxations and reconstructions usually extend into the crystal by more than one atomic layer. Moreover, the experimental probes in surface experiments, even slow electrons, usually have a non-negligible penetration depth but compared to subsurface layers, the topmost atomic layer is always predominant in any technique based on the use of electrons or atoms. As a result, each layer of atoms in the surface is intrinsically inequivalent to other layers and the only symmetry properties which the surface posses are those which operate in a plane parallel to the surface. Although the surface region is three-dimensional, all symmetry properties are two-dimensional. The basic surface lattice can be described by a set of two-dimensional translational vectors: 𝑅𝑚𝑛 = 𝑚𝑎1 + 𝑛𝑎2 where (m,n) denotes a pair of integers numbers, and the 𝑎𝑖 are the two unit mesh vectors. As seen before, the topmost atomic layer could be rescostructed with different periodicity compared to the one of the bulk; in this case one speaks of a superlattice i.e. a lattice which is superimposed to the basic surface one. The surface net of this topmost atomic layer can be written in terms of the previous defined vectors as: 𝑏1 = 𝑚11 𝑎1 + 𝑚12 𝑎2 𝑏2 = 𝑚21 𝑎1 + 𝑚22 𝑎2 18.

(24) 3.2. Film deposition and growth modes. ⎡ ⎣. 𝑏1 𝑏2. ⎤. ⎡. ⎦=𝑀⎣. 𝑎1 𝑎2. 19. ⎤. ⎡. ⎦. and 𝑀 = ⎣. 𝑚11 𝑚12 𝑚21 𝑚22. ⎤ ⎦. the ratio between the areas of the two unit meshes is Δ𝑀 =. ∣𝑏1 ⊗ 𝑏2 ∣ ∣𝑎1 ⊗ 𝑎2 ∣. which can be used to characterize the relation between the superlattice and the basic surface lattice. If Δ𝑀 is an integer then one has a simple superlattice; if its a rational number the superstructure is referred to as a coincidence lattice and finally if detM is an irrational number the superstructure is called an incoherent lattice.. 3.2. Film deposition and growth modes. Figure 3.1: Basic modes of thin film growth (a) Islands growth mode or VollmerWeber (VW) mode (b) Layer by layer growth mode or Frank-Van de Merve mode (FM) (c) Layer and islands growth mode or Stransky-Krastanov (SK) mode The aim of this paragraph is to introduce the thin film growth on the surface of a substrate and various growth modes will be described from a phenomenological point of view. The films growth mode is to determine its structural and magnetic properties. If both the film and the substrate are of the same material one speaks of homoepitaxy while, on the other hand, of heteroepitaxy. Three different growth modes can be qualitatively classified (see also Fig. 3.1):.

(25) 3.3. Artificial fcc Mn films. 20. 1. islands growth mode or Vollmer-Weber (VW): small clusters are nucleated directly on the substrate surface and then grow into islands of the condensed phase when the adatoms (or molecules) are more strongly bound to each other than to the substrate. 2. layer by layer growth mode or Frank-Van de Merve (FM): the first adatoms form a complete monolayer on the surface and covered with a somewhat less tightly bound second layer because the atoms are more strongly bound to the substrate than to each other. 3. layer and islands growth mode or Stransky-Krastanov (SK): the last mode is an mixing process characterized by both 2D layer-by-layer and 3D island growth and islands start to form after one or more atomic layers are completed. One of the key factor to understand the different growth modes is the difference between lattice parameters of the film and the substrate. As a matter of fact once the atoms of the films start to deposit on the surface they aim to reproduce the two-dimensional pattern of the substrate; doing this they inevitably increase the elastic-deformation energy of the film due to the difference with the lattice parameter of the substrate. If the difference is small then the film can grow with the lattice parameter of the substrate even to higher thicknesses (pseudomorphic growth); on the contrary, if the difference is large, the film starts to grow with his own lattice parameter from the very first layers. In the intermediate case there could be a transition between these two growth modes at a critical thickness, when the elastic stress becomes greater of the adhesion force between the film and the substrate.. 3.3. Artificial fcc Mn films. Bulk Mn exhibits complicated structural transitions at various temperatures. 𝛼-phase (58 atoms per cubic cell) Mn is stable below 1000 K, and reveals antiferromagnetic (AFM) ordering with the phase transition temperature (T𝑁 ) < 100 K. 𝛽-phase (20 atoms per cubic cell), 𝛾-phase (face-centered cubic: fcc), and 𝛿-phase (body-centered cubic: bcc) Mn also sequentially appear at the temperatures higher than 1000 K. Due to the high temperature, the magnetic ground states of bulk Mn.

(26) 3.4. Hysteresis loop. 21. materials of the above structures are unable to be characterized. How to prepare Mn samples with stable fcc or bcc structure near room temperature thus becomes an important issue. About thirty years ago, the 𝛾-phase bulk Mn samples were prepared by rapid quenching of Mn with dilute concentration of Cu, Ni, Pd, or Fe with T𝑁 =540 K. However not until the last decade, 𝛾 and 𝛿-phase pure Mn films are prepared by choosing the proper single-crystalline substrates like Cu(100), Fe(100), and Ag(100) etc.. Due to the complexity and variety in the structure of Mn materials, the magnetic properties of Mn with different structures still attracts many theoretical and experimental efforts even up to now. In recent studies, Mn ultrathin films grown on Cu3 Au(100) substrate at room temperature reveal the well-defined 𝛾-phase crystalline structure. A transition from face-centered cubic (fcc) to facecentered tetragonal (fct) structure is also observed. By capping the Fe overlayer, the AFM properties of 𝛾-Mn films can be detected through the exchange bias effect in the Fe/Mn bilayers. But so far the studies on the interrelation between the crystalline structure, surface morphology and magnetism in such 𝛾-Mn films are still absent. Experimentally, the growth temperature possesses crucial effects on the crystalline structure, morphology, and even the magnetic properties of thin films. In general, growth at LT highly enhances the surface roughness and the morphological changes occur with increasing thickness. In surface magnetism, the magnetic properties including the surface anisotropy, crystalline anisotropy etc. are strongly dependent on the surface morphology and crystalline structure. Thus the magnetic behavior of ultrathin films can be affected by the growth temperature seriously. For example, Fe/Cu3 Au(100) films reveal a fcc to bcc structural transition and simultaneously the perpendicular to in-plane spin-reorientation transition (SRT) at a critical thickness of 3.5 and 5.5 monolayer (ML) for room-temperature (RT) and low-temperature growth (LT), respectively.. 3.4. Hysteresis loop. The macroscopic magnetic properties of a ferromagnet are characterized by the hysteresis loop and determine the suitability of magnetic materials for a given ap-.

(27) 3.4. Hysteresis loop. 22. plication. The hysteresis loop is obtained by applying a magnetic field 𝜇0 𝐻 to the specimen and measuring the ensuing change of the magnetic polarization M in field direction. Starting from the initial demagnetized (𝑀 = 𝜇0 𝐻) state the polarization increases with increasing field and finally reaches the saturation polarization (𝑀 = 𝑀 𝑠𝑎𝑡 ) When the magnetizing field is reduced to zero from the saturated state the sample remains magnetized. This polarization at zero field is called remanence 𝑀𝑟 . It can be returned to zero by applying a reverse magnetic field of strength 𝜇0 𝐻𝑐 known as the coercive field which is, therefore, the measure for the magnets resistance against demagnetizing fields. Further increase of the reversed applied field magnetizes the sample to saturation in the opposite direction (𝑀 = −𝑀 𝑠𝑎𝑡 ). A large spontaneous polarization 𝑀𝑠 is a prerequisite for high 𝑀𝑟 values, whereas a large magnetocrystalline anisotropy constant may result in large coercivities. In principle, ferromagnetic substances can be classified into soft and hard magnetic materials depending on how easily the material can be (de)magnetized as illustrated in Figure 3.2.. 3.4.1. Hard magnetic material. Hard magnets provide stable permanent magnetic fields (after exposure to a magnetic field) and create surface poles without continuous expenditure of electrical energy. They are characterized by high coercivity, high remanence, and a high maximum energy product. The maximum energy product (𝐵𝐻𝑚𝑎𝑥 ) represents the magnetic energy per unit volume which can be maximally stored by a hard magnet, thereby specifying the performance or strength of a permanent magnet. 𝐵𝐻𝑚𝑎𝑥 is defined as the maximum rectangular area within the 𝐵(𝐻) = 𝜇0 𝐻 + 𝑀 hysteresis loop in the second quadrant. Typical hard magnetic materials are based on RE-TM compounds, hard ferrites, AlNiCo, and CoPt/FePt..

(28) 3.4. Hysteresis loop. 23. Figure 3.2: (a) Typical hysteresis loop of hard and soft magnets. (b) 𝐵𝐻𝑚𝑎𝑥 is defined as the maximum rectangular area within the 𝐵(𝐻) = 𝜇0 𝐻 + 𝑀 hysteresis loop in the second quadrant.. 3.4.2. Soft magnetic material. Soft magnetic materials enable amplification of the flux produced by an electrical current considerably, therefore being important in any application involving a change in magnetic induction. They are characterized by low coercivity 1), high (initial) permeability which describe the response ofmagnetic materials to a smallmagnetic field, therefore indicating how much magnetic induction B is generated by the material in a given magnetic field strength H and low high-frequency losses. The characteristics usually go along with low conductivity and small magnetostriction which describes the change of the shape of a ferromagnetic specimen during the magnetization process. Soft magnets are based on Fe, FeVSi, FeVNi (permalloy), FeVCo ((su)permendur), soft ferrites, and metallic glasses..

(29) 3.5. The exchange interaction between FM and AFM materials. 3.5. 24. The exchange interaction between FM and AFM materials. The exchange anisotropy of FM/AFM systems was first discovered by Meiklejohn and Bean in 1956. They used ferromagnetic Co nanoparticles which were embedded in their native antiferromagnetic CoO layers. The Co/CoO systems were treated by field cooling procedure which the sample was heated and subsequently cooled down to below the Neel temperature T𝑁 with a sufficiently strong magnetic field being presented. The origin of the hysteresis loop of Co in Co/CoO systems was no longer centered at zero field (H = 0) which has a shift along the field axis. Independent of the direction of the horizontal shift of the hysteresis loop, it is always accompanied by an increase of the coercivity H𝐶 below the blocking temperature T𝐵 which is linked to the anisotropy of the AFM layer. However, the coercivity is also affected by the thicknesses and microstructures of the FM and AFM layers. This increase of H𝐶 is intuitively simple to understand. For an AFM with small anisotropy, when the FM rotates, it drags the AFM spins irreversibly, hence increasing the FM coercivity. For a large AFM anisotropy, the FM decouples because it cannot drag AFM spins, consequently the coercivity is reduced. Fig. 3.3(a) shows an FM/AFM system at a temperature T that is below the Curie temperature (T𝐶 ) of the ferromagnet and above the blocking temperature T𝐵 of the antiferromagnet (𝑇𝐵 < 𝑇 < 𝑇𝐶 ). The hysteresis loop of the FM/AFM system at this temperature is equal to the hysteresis curve of the pure ferromagnet plus a small paramagnetic part from the AFM and is shown in Fig. 3.3(b). The hysteresis loop is perfectly centred around the origin. The FM/AFM system is then cooled well below T𝐵 in presence of a magnetic field H (or in a state where the FM has a remanent magnetisation), which is usually aligned parallel to an easy axis of the FM in Fig. 3.3(d). The hysteresis loop of the FM/AFM system in this state is shifted along the field axis, generally in the direction opposite (negative) to the applied cooling field (Fig. 3.3(e)). Additionally, the hysteresis loop shows a strong increase of the coercivity H𝐶 ..

(30) 3.6. Magnetic anisotropy. 25. Figure 3.3: Comparison of the magnetic properties of (a) an FM/AFM system at a temperature T with 𝑇𝐵 < 𝑇 < 𝑇𝐶 and (d) the same system after field-cooling below the AFMs Neel temperature (with 𝑇 < 𝑇𝐵 < 𝑇𝐶 ). (b) The system shows a normal ferromagnetic hysteresis loop (c) an uniaxial anisotropy, indicated by a sin2𝜃behavior of the torque measurements (e) after field cooling below T𝐵 , shows a horizontally shifted loop with increased coercivity (f) and an unidirectional anisotropy with a sin 𝜃-behavior of the torque measurements.. 3.6. Magnetic anisotropy. Figure 3.4: Display of the angle between the magnetization M and the film surface normal. Ferromagnetic materials, such as Fe, Co, or Ni present spontaneous magnetization below the Curie temperature. In reality, these materials show their magnetization preferring some directions in space, which are called the easy axes. The direction of easy axis is observed to be dependent on the intrinsic symmetry of crys-.

(31) 3.6. Magnetic anisotropy. 26. talline structure and the shape of the sample. For example, in the latter, it might be parallel to the plane in thin film system. To discuss the mechanism of easy axis in detail, the magnetic anisotropy energy (MAE) is introduced as the energy difference between two different magnetization axes. Since the MAE (𝐸𝑡𝑜𝑡𝑎𝑙 ) depends on the orientation of the magnetization with respect to crystalline axes or its external shape, in a phenomenological model it can be decomposed into the crystalline anisotropy (𝐸𝑐𝑟𝑦𝑠𝑡𝑎𝑙 ) and shape anisotropy (𝐸𝑠ℎ𝑎𝑝𝑒 ). However, in the above discussion, the strain has been taken to be zero. If the strain does exist, it can induced magneto-elastic anisotropy (𝐸𝑚𝑎𝑔𝑛𝑒𝑡𝑜𝑒𝑙𝑎𝑠𝑡𝑖𝑐 ). Therefore, from these different effects on the magnetic behavior, the total MAE can be expressed as : 𝐸𝑡𝑜𝑡𝑎𝑙 = 𝐸𝑐𝑟𝑦𝑠𝑡𝑎𝑙 + 𝐸𝑠ℎ𝑎𝑝𝑒 + 𝐸𝑚𝑎𝑔𝑛𝑒𝑡𝑜𝑒𝑙𝑎𝑠𝑡𝑖𝑐 In the analysis of ultrathin film anisotropy, the MAE is defined as the energy difference between the perpendicular and parallel to the film plane. It appears that, neglecting high order terms, a uniaxial description is often sufficient : 𝐸 = 𝐾𝑒𝑓 𝑓 sin2 𝜃 = (𝐾𝑉 +. 2𝐾𝑆 ) sin2 𝜃 𝑡. E is the total energy per unit volume, 𝐾𝑒𝑓 𝑓 stands for the effective anisotropy energy including the contribution from various source and denotes the angle between the magnetization and the film surface, as shown in Fig. 3.4. The second order term 𝐾2 sin4 𝜃 is usually very small. But in some cases, for example when the first order term vanishes, the second order term will become much more important than in usual. 𝐾𝑆 indicates the contribution from the surface or interface per unit area. Meanwhile, 𝐾𝑉 denotes the contribution from the volume or bulk per unit volume. The factor 2 of the surface term comes from the assumption that the contribution from the interface and surface are nearly identical, and this should be modified in some cases. Since the system prefers low energy, positive 𝐾𝑒𝑓 𝑓 describes the case of easy axis perpendicular to the film surface (𝜃 = 0) and negative 𝐾𝑒𝑓 𝑓 describes the case of easy axis parallel to the film surface ( 𝜃 = Π/2).. In Fig. 3.5 and Fig. 3.6, the plots of the product 𝐾𝑒𝑓 𝑓 𝑡 versus t are shown for the cases of Co/Cu(100) and Ni/Cu(100), where 2𝐾𝑆 and 𝐾𝑉 stand for the.

(32) 3.6. Magnetic anisotropy. 27. Figure 3.5: Effective magnetic anisotropy per unit area per Co layer versus the Co layer thickness.. Figure 3.6: Effective magnetic anisotropy per unit area per Ni layer versus the Ni layer thickness. intercept at 𝑡 = 0 and the slope respectively. If 𝐾𝑉 and 𝐾𝑆 are of different signs, because 𝐾𝑆 decays with thickness as 1𝑡 , 𝐾𝑒𝑓 𝑓 will change sign with the increasing of thickness, which results in the SRT. This spectacular behavior is found in many magnetic ultrathin film system, such as Co/Pt(111), Fe/Cu(100), Fe/Cu3Au(100), Ni/Cu(100) et al., and will be discussed later in this section. The magnetocrystalline anisotropy arises essentially from the spin-orbit coupling, but also, to lesser extent, from the dipolar interactions. The symmetry breaking on the surface and interface induces the magnetocrystalline surface anisotropy, which may be responsible for the perpendicular magnetization in several systems such as ultrathin films Fe/Cu(100), Co/Ag(100) etc. However, the sign of the magnetocrystalline surface anisotropy is strongly system dependent, thus resulting in different magnetic behavior. The shape anisotropy, which is also called the demagnetization energy, arises from the dipolar interaction. Because the dipolar interaction decrease slowly as a function of the distance 𝑟 (like. 1 ), 𝑟3. the dipolar field experienced by a given moment depends.

(33) 3.6. Magnetic anisotropy. 28. significantly on the moments located at the boundary of the sample, and this results in the shape anisotropy. In the case of ultrathin films, the volume shape anisotropy 𝑉 𝐸𝑠ℎ𝑎𝑝𝑒 = −2𝜋𝑀𝑉2 sin2 𝜃. where 𝑀𝑉 is the saturated moment per unit volume. On the other hand the shape surface anisotropy contributes only weakly to the total surface anisotropy. For Fe, Co, and Ni, the volume shape anisotropy are larger than the volume magnetocrystalline anisotropy. Therefore the shape anisotropy usually dominates the volume term (without strain), and the negative sign of it results in the magnetization lying in the film plane. Magnetostriction is the phenomenon that the shape of a ferromagnetic specimen changes during the process of magnetization. Inversely, if a stress is applied to a ferromagnetic specimen, the direction of the magnetization will also be affected through the magnetostriction. This strain induced MAE is called the magnetoelastic anisotropy. Because the magnetoelastic anisotropy comes from the strain induced change in the magnetocrystalline anisotropy, its physical origin is also the spin-orbit interaction. Particularly in ultrathin films, where considerable strain may result from the epitaxial growth of the film onto the substrate having a different lattice constance, the strain induced anisotropy plays a very important role..

(34) Chapter 4 Experiment and Result 4.1. Experiment procedure. Figure 4.1: The flow chart of experiment procedure. The thin film preparation and measurements were carried out in a ultrahigh vacuum (UHV) chamber with the base pressure better than 2 × 10−10 torr. After cycles of 2 𝑘𝑒𝑉 Ar+ sputtering and annealing, the clean substrate of Cu3 Au(001) with well-ordered c(2 × 2) superstructure and flat surface was obtained. 2.5 ML Fe were sequentially evaporated onto the substrate with the substrate temperature of 300 K (RT). During the evaporation, the pressure was better than 9 × 10−10 torr. The growth was monitored by medium energy electron diffraction (MEED) with a beam energy of 2 keV and the grazing angle of 1∘ . From the periodicity of MEED oscil29.

(35) 4.2. Growth and Structure of Mn/Fe/Cu3 Au(001). 30. lations, the deposition rate was calibrated precisely. Then Mn and Fe layers were deposited in turn over the Fe/Cu3 Au(001) films at RT. After the film growth was finished, we checked the composition by Auger electron spectroscopy which different signals come from different elements. The lateral crystalline structure was characterized at 300 K by low energy electron diffraction (LEED). In this experiment, a 4 grid rear view LEED was used to take the LEED images. Besides, from the LEEDI/V curves, the average vertical interlayer distance (d⊥ ) of the film was determined using the kinetic approximation. Magnetic properties of the films were detected by magneto-optical Kerr effect (MOKE). The MOKE measurement was performed in both longitudinal and polar geometries with the modulation and lock-in technique. All the MOKE measurements of Fe/Mn/Fe trilayers were performed after a cooling process from 300 K to 195 K. A flow chart of this series of experiments is shown in Fig. 4.1.. 4.2. Growth and Structure of Mn/Fe/Cu3Au(001). Figure 4.2: MEED (0,0) spot intensity of Mn films grown on fcc-like Fe/Cu3 Au(001). Fig. 4.2 shows the (0,0) spot MEED intensity for Mn grown on fcc-like Fe/Cu3 Au(001). Mn grown on fcc-like Fe/Cu3 Au(001) at RT reveals apparent layer-by-layer growth even the thickness above 15 ML and the oscillation amplitude decreases gradually.

(36) 4.2. Growth and Structure of Mn/Fe/Cu3 Au(001). 31. Figure 4.3: Recorded intensity of MEED (0,0) and ( 21 , 12 ) spots for Mn films grown on Cu3 Au(001) at RT and LT (adapted from [8]) above 10 ML. Compared with Mn grown on Cu3 Au(001) without fcc-like Fe buffer layer, in Fig. 4.3 the MEED intensity of (0,0) spot for RT-grown Mn films also indicates layer-by-layer growth for 0 − 2 ML. After 2 ML, the oscillation amplitude is reduced and then gradually disappears after 6 − 7 ML. Obviously the number of oscillation of Mn grown on fcc-like Fe/Cu3 Au(001) is larger than that of Mn grown on Cu3 Au(001). This points out that the growth of Mn films with fcc-like Fe buffer layer may be better than directly grown on Cu3 Au(001). Fig. 4.4 shows the LEED patterns of Cu3 Au(001), 2.5 ML (fcc-like) Fe grown on Cu3 Au(001) at RT and various Mn films grown on 2.5 ML (fcc-like) Fe/Cu3 Au(001)at RT. Fig. 4.4(a) exhibits the c(2 × 2) structure of Cu3 Au(001). Fig. 4.4(b) the RT-grown 2.5 ML (fcc-like) Fe film reveals the p(1 × 1) LEED pattern. As shown in Fig. 4.4(c)-(g), the RT-grown Mn films appear the p(1 × 1) LEED pattern and the sharp p(1 × 1) LEED patterns exhibits the well-ordered crystalline structure. Figure 4.5 shows the LEED patterns of Cu3 Au(001) and various Mn films grown at RT. Figure 4.5(a) exhibits the c(2×2) structure of Cu3 Au(001). 0.5 and 1 ML RT-Mn films reveal p(2 × 2) structure and for 5 ML, a transition to c(2 × 2) structure is observed. With the larger coverage of more than about 9 ML, the RT-Mn films always reveal the p(1 × 1) LEED pattern.

(37) 4.2. Growth and Structure of Mn/Fe/Cu3 Au(001). 32. as shown in Figs. 4.5(e)-(h).. Figure 4.4: LEED pattern of (a) Cu3 Au(001) (b) 2.5 ML (fcc-like) Fe grown on Cu3 Au(001) at RT (c) various thickness of Mn grown on 2.5 ML (fcc-like) Fe/Cu3 Au(001) at RT. All the images are taken at RT with the beam energy equivalent to 150 eV. The LEED-I/V curves of (0,0) spot are also recorded for the analysis of interlayer distance (d⊥ ). Fig. 4.6 shows the LEED-I/V curves of RT-grown Mn films. The integers denote the order of the maximum conditions in Bragg interference. The indexed peak positions shift toward high energy both in RT-Mn films with increasing thickness. The shift toward high energy indicates a structural transition toward the smaller d⊥ . Fig. 4.7 shows the d⊥ deduced from the LEED-I/V curves. Mn films with low coverage reveal the d⊥ almost the same as the substrate (Cu3 Au(001, d⊥ =1.89 ˚ 𝐴). At 21.5 ML thickness, the d⊥ reduces to about 1.83 ˚ 𝐴. Since the LEED studies inform the coherence growth of Mn on Cu3 Au(001), the RT-grown Mn films are concluded to perform a structural transition from face-centered cubic (fcc) to a face-centered tetragonal (fct) and the critical thicknesses are ∼14 ML..

(38) 4.2. Growth and Structure of Mn/Fe/Cu3 Au(001). 33. Figure 4.5: LEED patterns of RT-grown Mn/Cu3 Au(001) with various coverages. All the images are taken at 100 K with the beam energy equivalent to 150 eV. (adapted from [8]). Figure 4.6: LEED-IV of various thickness of Mn grown on 2.5 ML (fcc- like) Fe/Cu3 Au(001) at RT.. In comparison with RT-grown Mn/Cu3 Au(001), Figure. 4.8(a) shows Mn films with low coverage reveal a d⊥ almost the same as that of the substrate Cu3 Au(001). At a higher thickness, the d⊥ is reduced to about 1.77 ˚ 𝐴. Since the coherence growth of Mn on Cu3 Au(001), the Mn films are concluded to perform a structural transition.

(39) 4.2. Growth and Structure of Mn/Fe/Cu3 Au(001). 34. Figure 4.7: Vertical interlayer distance of various thickness of Mn grown on 2.5 ML (fcc-like) Fe/Cu3 Au(001) at RT.. Figure 4.8: Vertical interlayer distance of various thickness of Mn grown on 2.5 ML Fe/Cu3 Au(001) at RT. (adapted from [8]) from a face-centered cubic (fcc) to a face-centered tetragonal (fct) structure and the critical thicknesses are ∼12 ML..

(40) 4.3. Growth and Structure of Fe/Mn/Fe/Cu3 Au(001). 4.3. 35. Growth and Structure of Fe/Mn/Fe/Cu3Au(001). Figure 4.9: (0,0) spot intensity of MEED of Fe grown on Mn/Fe/Cu3 Au(001).. Figure 4.10: LEED pattern of (a) 8 ML Fe (b) 10 ML Fe (c) 12 ML Fe (d) 14 ML Fe grown on 6 ML Mn/2.5 ML Fe/Cu3 Au(001) at RT. Fig. 4.9 shows the MEED oscillation of Fe films deposited on various RT-Mn films and apparently Fe films grown on Mn/fcc-like Fe/Cu3 Au(001) is layer-by-layer growth. Fig. 4.10 shows the LEED patterns of various thickness of Fe films grown on 6 ML Mn/2.5 ML (fcc-like) Fe/Cu3 Au(001) at RT. The c(2 × 2) structure of the Fe films are supposed to be the superstructure on surface. The dash lines indicate the spot positions are almost the same which means epitaxial growth. Fig. 4.11(a)-(c) shows the LEED-I/V curves of RT-grown Fe films. Fig. 4.11(d) shows the fitting results of LEED I/V curves, the interlayer distance of Fe films grown on Mn/2.5 ML (fcc-like) Fe/Cu3 Au(001) is about 1.51 ˚ 𝐴 so that the crystalline.

(41) 4.3. Growth and Structure of Fe/Mn/Fe/Cu3 Au(001). 36. Figure 4.11: LEED-IV of various thickness of Fe grown on 6 ML Mn/2.5 ML Fe/Cu3 Au(001) at RT. structure is body-centred tetragonal (bct) and apparently the vertical interlayer distance remains almost invariant in those coverages. Compared with Fe grown on Cu3 Au(001) (Fig. 4.12), the vertical interlayer distance (d⊥ ) of Fe films is almost the same as the substrate (Cu3 Au(001), d⊥ =1.89 ˚ 𝐴) at thickness lower than 3 ML and then performs a fcc-like to bcc-like structure transition. When the thickness of Fe films is larger than 7 ML, d⊥ is about 1.55 ˚ 𝐴. In our case, the d⊥ is almost the same as the substrate (Cu3 Au(001)) for the fcc-like Fe buffer layer and the d⊥ is 1.51 ˚ 𝐴 for the top Fe overlayer..

(42) 4.4. Magnetic Properties of Fe/Mn/Fe/Cu3 Au(001). 37. Figure 4.12: Two different average vertical interlayer distances related to fcc-like (full circles) and bcc-like Fe films (open circles), respectively. adapted from [2]. 4.4. Magnetic Properties of Fe/Mn/Fe/Cu3Au(001). Coercivity enhancement phenomenon Fe films grown on Mn/Fe/Cu3 Au(001) reveal different magnetic properties from those grown on Mn/Cu3 Au(001) at room temperature. In Fig. 4.13 the coercivity of magnetic hysteresis loop of 6.3 ML Fe/4.5 ML Mn/2.5 ML Fe/Cu3 Au(001) in perpendicular direction is greatly enhanced by the 2.5 ML fcc-like Fe buffer layer as compared with 6.3 ML Fe/4.5 ML Mn/Cu3 Au(001). The increase of H𝐶 is intuitively simple to understand. For an AFM (Mn) layer with small anisotropy, when the FM (Fe) layer rotates, it drags the AFM (Mn) layer spins irreversibly, hence increasing the FM (Fe) layer coercivity. In Fig. 4.14 the coercivity versus temperature indicates two important results. One is the coercivity in perpendicular direction enhanced by 2.5 ML fcc-Fe buffer layer with perpendicular magnetization. The possible explanation is that the perpendicular component of Mn layer spin configuration somehow ehanced by 2.5 ML fcc Fe buffer layer with perpendicular magnetization during growth. Thus the exchange coupling in perpendicular direction between Fe overlayer and Mn layer is enhanced, and the coercivity of the Fe overlayer is enhanced. The other is that the exchange coupling of 6.3 ML Fe/4.5 ML Mn/2.5 ML Fe in perpendicular direction is larger than 6.3 ML/6.5 ML Mn. Generally speak-.

(43) 4.4. Magnetic Properties of Fe/Mn/Fe/Cu3 Au(001). 38. Figure 4.13: Hysteresis loops of 6.3 ML Fe/4.5 ML Mn/2.5 ML Fe/Cu3 Au(001) and 6.3 ML Fe/4.5 ML Mn/Cu3 Au(001). At 196 K coercivity is extraordinary enhanced caused by wetting layer effect. ing, Mn layer with larger thickness results in larger exchange coupling between Fe overlyer and Mn layer. However the influence on excange coupling in perpendicular direction by fcc-like Fe buffer layer with perpendicular magnetization apparently remarkable..

(44) 4.4. Magnetic Properties of Fe/Mn/Fe/Cu3 Au(001). 39. Figure 4.14: Coercivity with respect to temperature. All the films were grown at 300 K. Observation of thickness dependent SRT The spin reorientation transition (SRT) phenomenon is an interesting subject in magnetism which describes the switch of magnetic easy axis. There are seveal different mechanism to induce SRT phenomenon. The thickness and temperature dependent spin reorientation transition of Fe/Mn/Fe/Cu3 Au(001) will be described. In Fig. 4.15 hysteresis loops of n ML Fe/4 ML Mn/2.5 ML Fe/Cu3 Au(001) was measured at 195 K. When the thickness of Fe overlayer less than 7 ML, the coercivity gradually increased with thickness due to more ferromagnetic moments. Then the magnetic easy axis was switched from perpendicular direction to in-plane direction. M𝑟 versus thickness curves (Fig. 4.16) also supports the same result.. The reason. why the magnetic easy axis aligns in different directions at different thickness is discussed in the following. When the thickness of top Fe layer is less than about 7 ML , the symmetry breaking on the surface induces perpendicular magnetization. However, when the thickness of Fe overlayer is larger than about 7 ML, the volume shape anisotropy which arises from the dipolar interaction forces the magnetic.

(45) 4.4. Magnetic Properties of Fe/Mn/Fe/Cu3 Au(001). 40. Figure 4.15: Hysteresis loops of various coverage Fe grown on 4 ML Mn/2.5 ML Fe/Cu3 Au(001). The onset thickness for SRT is about 7 ML.. Figure 4.16:. M𝑟 versus thickness curves of n ML Fe/4 ML Mn/2.5 ML. Fe/Cu3 Au(001). The transition region is 7 ML to 10 ML..

(46) 4.4. Magnetic Properties of Fe/Mn/Fe/Cu3 Au(001). 41. easy axis aligned in in-plane direction. In Fig. 4.17 hysteresis loop of 8 ML Fe/5 ML Mn/Cu3 Au(001) was measured only in in-plane direction at 195 K. When 2, 2.5, and 3 ML fcc-like Fe buffer layer was added individually between Mn film and Cu3 Au(001), the magnetic easy axis was switched from in-plane direction to perpendicular direction. The coercivity of those system with fcc-like Fe buffer layer seems almost the same.. Figure 4.17: Hysteresis loops of 8 ML Fe/5 ML Mn/n ML Fe/Cu3 Au(001). The magnetic easy axes switch from perpendicular direction to in-plane direction by fcc-like Fe buffer layers.. Observation of temperature dependent SRT In Fig. 4.18 hysteresis loops of 10 ML Fe/6 ML Mn/2.5 ML Fe/Cu3 Au(001) were measured at different temperature. At lowest temperature (195 K), the magnetic easy axis lay in perpendicular direction. With rising temperature, the magnetic easy axis started to switching from perpendicular to in-plane direction which meant the exchange coupling in perpendicular direction decreses. The key factor of temperature dependent SRT is the lattice vibration caused by rising temperature. At low coverage region, symmetry breaking which dominates the perpendicular magnetization is easily disturbed by lattice vibration..

(47) 4.4. Magnetic Properties of Fe/Mn/Fe/Cu3 Au(001). 42. Figure 4.18: Hysteresis loops of 10 ML Fe/6 ML Mn/2.5 ML Fe/Cu3 Au(001) measured at different temperature..

(48) 4.4. Magnetic Properties of Fe/Mn/Fe/Cu3 Au(001). 43. Overview of SRT. Figure 4.19: Hysteresis loops of m ML Fe/6 ML Mn/2.5 ML Fe/Cu3 Au(001) measured at different temperature..

(49) 4.4. Magnetic Properties of Fe/Mn/Fe/Cu3 Au(001). 44. Figure 4.20: Hysteresis loops of m ML Fe/4 ML Mn/2.5 ML Fe/Cu3 Au(001) measured at different temperature and 8 ML Fe/5 ML Mn/n ML Fe/Cu3 Au(001) measured at 195 K..

(50) 4.4. Magnetic Properties of Fe/Mn/Fe/Cu3 Au(001). 45. Figure 4.21: Hysteresis loops of m ML Fe/8 ML Mn/2.5 ML Fe/Cu3 Au(001) measured at different temperature and overview of SRT with a magnetic phase diagram..

(51) Chapter 5 Discussion Mn/fcc-like Fe/Cu3 Au(001) and Mn/Cu3 Au(001) From an intuitive aspect thin films grown on different substrate should appear different lattice orders but in our case Mn/fcc-like Fe/Cu3 Au(001) and Mn/Cu3 Au(001) present almost the same vertical interlayer distance. The thickness of the Fe buffer layer is 2.5 ML which means the vertical interlayer distance almost the same as the substrate Cu3 Au(001) (fcc, d⊥ = 1.89 ˚ 𝐴) (Fig. 5.1(a)). If the growth process is assumed to be epitaxial growth, although the interfaces are different, Mn films grown on fcc-like Fe/Cu3 Au(001) and Cu3 Au(001) should appear the similar lattice order. For Mn grown on Cu3 Au(001), Mn films with low coverage reveal a d⊥ almost the same as that of the substrate Cu3 Au(001) (d⊥ = 1.89 ˚ 𝐴) and the d⊥ is reduced to about 1.77 ˚ 𝐴 at a higher thickness. Since the LEED patterns indicates the coherence growth of Mn on Cu3 Au(001), the Mn films are concluded to perform a structural transition from a fcc to a fct structure and the critical thickness is about 12 ML (Fig. 5.1(b)). For Mn grown on fcc-like Fe/Cu3 Au(001), Mn films with low coverage reveal a d⊥ about 1.91 ˚ 𝐴 and the d⊥ is reduced to about 1.83 ˚ 𝐴 at a higher thickness. The Mn films perform a structural transition from a fcc to a fct structure and the critical thickness is about 14 ML (Fig. 5.1(c)).. 46.

(52) Chapter 5. Discussion. 47. Figure 5.1: Comparison of interlayer distance between Mn/fcc-like Fe/Cu3 Au(001) and Mn/Cu3 Au(001) (a) interlayer distance of Fe grown on Cu3 Au(001) adapted from [2] (b) interlayer distance of Mn grown on Cu3 Au(001) adapted from [8] (c) interlayer distance of Mn grown on Fe/Cu3 Au(001) Fe/Mn/fcc-like Fe/Cu3 Au(001) and Fe/Mn/Cu3 Au(001) The are two magnetic phase diagrams in Fig. 5.2. From the magnetic phase diagrams, which present the relationship between the direction of magnetic easy axis of Fe overlayer and thickness of Mn and Fe, for a fixed Fe overlayer thickness the magnetic easy axis of Fe overlayer tends to prefer in perpendicular direction with larger Mn thickness. On the other side, for a fixed Mn thickness the magnetic easy axis of Fe overlayer tends to prefer in in-plane direction with larger Fe overlayer thickness. In Fig. 5.2(b), the boundary of SRT (blue dash line) of Fe/Mn/fcc-like Fe/Cu3 Au(001) is shifted by 2-3 ML compared with that of Fe/Mn/Cu3 Au(001) (Fig. 5.2(a)). The shifted SRT boundary means that the mag-.

(53) Chapter 5. Discussion. 48. Figure 5.2: (a) Magnetic phase diagram of Fe/Mn/Cu3 Au(001). All data points were measured at 220 K. Provied by B. Y. Wang (b) Magnetic phase diagram of Fe/Mn/Fe/Cu3 Au(001). All data points were measured at 196 K. netic easy axis still aligns in perpendicular direction at larger coverage of Fe overlayer. Also points out the exchange coupling of Fe/Mn/Fe/Cu3 Au(001) is stronger than Fe/Mn/Cu3 Au(001). Although the different measured temperatures causes different degree of exchange coupling, we still believe that the buffer layer effect is the main factor. Magnetism of fcc-like Fe buffer layer/Cu3 Au(001) In Fig. 5.3 hysteresis loops of 2.5 ML Fe/Cu3 Au(001) in perpendicular direction were measured at both 300 K and 195 K. However if several coverages of Mn were deposited on 2.5 ML Fe/Cu3 Au(001), no hysteresis loop could be measured in both perpendicular and in-plane direction (Fig. 5.3). One possible reason is the alloy effect between Mn and Fe interface. From an intuitive aspect, the growth of Fe𝑥 Mn1−𝑥 alloy should reduce the thickness of fcc-like Fe buffer layer so that the T𝐶 of fcc-like Fe buffer layer is decreased..

(54) Chapter 5. Discussion. 49. Figure 5.3: Hysteresis loops of various coverage Mn grown on 2.5 ML Fe/Cu3 Au(001) measured at 300 K and 195 K. Conclusion of fcc-like Fe buffer layer effect In this experiemnt there are two obvious behaviors attributed to fcc-like Fe buffer layer. One is that the fcc-like Fe buffer layer with perpendicular magnetization induces spin-reorientation transition. In Fig. 5.4(a) the magnetic easy axis of 8 ML Fe/ 5 ML Mn/Cu3 Au(001) aligned in in-plane direction. Then if 2, 2.5 and 3 ML fcc-like Fe buffer layers were added, the magnetic easy axis switched to perpendicular direction (Fig. 5.4(a)). On the other hand, the coercivity of Fe/Mn/fcc-like Fe/Cu3 Au(001) is greatly enhanced compared with Fe/Mn/Cu3 Au(001). In Fig. 5.4(b) the coercivity of 6.3 ML Fe/4.5 ML Mn/2.5 ML Fe/Cu3 Au(001) is larger than those without 2.5 ML Fe buffer layer. All these phenomenons indicate that the fcc-like Fe buffer layer plays an important role in the magnetism of Fe/Mn/fcc-like Fe/Cu3 Au(001). The possible explanation is that the perpendicular magnetization of fcc-like Fe buffer layer may affect the spin configuration of Mn layer and somehow enhance the perpendicular component and also the exchange coupling between Fe overlayer and Mn layer is enhanced..

(55) Chapter 5. Discussion. 50. Figure 5.4: Fcc-like Fe buffer layer effect: (a) spin-reorientation transition (b) coercivity enhancement..

(56) Chapter 6 Conclusion ■ The crystalline structure for Mn on fcc-Fe/Cu3 Au(001) is concluded to fcc with d⊥ = 1.91 ˚ 𝐴 at low coverage and performs fcc to fct structure transition around 14 ML. The d⊥ for the fct structure is 1.83 ˚ 𝐴 at 21.5 ML. ■ There are three SRT phenomenons found in this experiment: thickness-dependent SRT, temperature-dependent SRT and buffer layer induced SRT. ■ With 2.5 ML fcc-Fe bottom layer, critical thickness of SRT is shifted 2∼3 ML and exchange coupling of Fe/Mn/Fe/Cu3 Au(001) in perpendicular direction is stronger than Fe/Mn/Cu3 Au(001).. 51.

(57) Bibliography [1] M.-T. Lin, J. Shen, W. Kuch, H. Jenniches, M. Klaua, C. M. Schneider, and J. Kirschner, Phys. Rev. B 55, 5886 (1997). [2] M.-T. Lin, J. Shen, W. Kuch, H. Jenniches, M. Klaua, C. M. Schneider, and J. Kirschner, Surf. Sci. 410, 290 (1998). [3] W. C. Lin, C. C. Kuo, C. L. Chiu, and M.-T. Lin, Surf. Sci. 478, 9 (2001). [4] W. C. Lin, B. Y. Wang, Y. W. Liao, Ker-Jar Song, and Minn-Tsong Lin, Phys. Rev. B 71, 184413 (2005) [5] W. C. Lin, P. C. Huang, K. J. Song, and Minn-Tsong Lin, Appl. Phys. Lett. 88, 153117 (2006). [6] W. C. Lin, S. S. Wong, P. C. Huang, C. B. Wu, B. R. Xu, C. T. Chiang, H. Y. Yen, and Minn-Tsong Lin, Appl. Phys. Lett. 89, 153111 (2006). [7] W. C. Lin, B.Y. Wang, T.Y. Chen, L.C. Lin, Y.W. Liao, W. Pan, N.Y. Jih, and K. J. Song, and Minn-Tsong Lin, Appl. Phys. Lett. 90, 052502 (2007). [8] W. C. Lin, T. Y. Chen, L. C. Lin, B. Y. Wang, Y. W. Liao, K. J. Song, and Minn-Tsong Lin, Phys. Rev. B 75, 054419 (2007). [9] B. Y. Wang, W. C. Lin, Y. W. Liao, K. J. Song, Minn-Tsong Lin , Surf. Sci. 600, 4517 (2006). [10] B. Feldmann, B. Schirmer, A. Sokoll, and M. Wuttig, Phys. Rev. B 57, 1014 (1998). [11] J. Nogu´ 𝑒s, I. K. Schuller, J. Magn. Magn. Mater. 192 203 (1999). 52.

(58) Bibliography. 53. [12] S. Matt, K. Takano, S. S. P. Parkin, and Eric E. Fullerton, Phys. Rev. Lett. 87, 087202 (2001). [13] W. Kuch, F. Offi, L. I. Chelaru, M. Kotsugi, K. Fukumoto and J. Kirschner, Appl. Phys. Lett. 86, 122504 (2005). [14] W. C. Lin, Master’s thesis, National Taiwain University (2000). [15] W. C. Lin, Doctor’s thesis, National Taiwain University (2006). [16] B. Y. Lin, Master’s thesis, National Taiwain University (2004)..

(59)