利用電壓及氫化反應可逆地控制材料磁性

全文

(1)國立台灣師範大學物理研究所博士論文. 利用電壓及氫化反應可逆地控制材料磁性 Reversible control of magnetism by applying voltage and hydrogenation. 指導教授：林文欽研究生：張博鈞. 博士撰. 中華民國一百零七年四月.

(2) 致謝. 致謝我進入林文欽老師的實驗室已經是六年前的事情了。這六年來我能學到許多事情，最主要的原因就是因為老師的包容與不斷地給我嘗試的機會，讓我能從反覆的嘗試中汲取到許多經驗。而且在研究方向上，老師也提供許參考意見，讓我的研究一直以來都還算順利。我也要感謝承叡學長，他在我碩一碩二時教了我很多物理知識，也幫我解答了許多疑惑，還教了我許多關於咖啡的知識。英奇學長晚上時常陪我留在實驗室，也會一起吃學校餐廳，另外，學長當時做的投影片被當作經典，其中一些素材我到現在也會使用。我在這六年中，除了老師，最該感謝的就是銓喆了。除了時常切磋討論各種五花八門的議題之外，在實驗忙不過來時也是最好的幫手，而且也教會我許多新儀器，最重要的是他時常提供了我一些同一事物但不同角度的看法，令我反思自己。還有跟我差了兩屆的昱全、凱霖和宜樺。因為有你們的努力，我才能在你們建立的根基上發展，後面許多的實驗結果都和你們的努力脫離不了干係。再來就是澤銘了，明明只是大學專題生卻異常的認真努力。一開始跟著我度過了一段不短的實驗撞牆期，後來有能力獨自實驗就讓我們的實驗進度飛快，最後也有得到結果，並發在不錯的期刊上。這些都遠遠超過了一般大專生會有的表現。最後就是宗佑、昀穎、政良，還有怡欣，跟你們相處時的喜怒哀樂，都豐富了我的生活經驗。每當想起這些回憶時都能提醒我心智上的不成熟，感謝大家陪伴這個脾氣不穩定又時常做蠢事的我度過這些時光。. ii.

(3) 中文摘要. 中文摘要在我們的電壓控磁的研究中顯示了，在鐵/氧化鋅結構中，可以藉由適當的電壓促使鐵的矯頑場減小 15%–20%，而且只要一移除電壓，矯頑場就會回復至原本的大小。另外，如果施加過量的電壓，則可以導致鐵/氧化鋅介面上的鐵氧化成氧化鐵、三氧化二鐵或四氧化三鐵等不同的氧化態，並且使樣品的矯頑場上升。各種氧化態的比例會取決於鐵和氧化鋅的厚度比例，且一旦氧化後是不可回復的。但就算發生了介面氧化，還是可以藉由適當的電壓來調降矯頑場的大小。在氫化控磁的研究中，我們觀察了鈷-30%鈀-70%的合金在不同氫氣氣壓下的磁性變化。原本無法觀察到磁域的樣品，在足夠的氫氣壓力下(0.2 bar)，會在樣品磁性翻轉時觀察到磁域，同時磁致曲線也會變得方正。我們依據在不同的固定外加磁場和氫氣濃度下所畫出的磁化翻轉曲線，可以推論出最小的翻轉單位 Barkhausen volume 會隨著氫氣壓力上升而變小。同時，氫氣壓力上升也會使得樣品的脫釘磁場變大，導致磁致曲線的矯頑場在 0.2 bar 以上的氫氣壓力會逐漸增加。此外，我們也研究了氫氣對於多層膜磁性耦合的影響。在 Pd/Fe/Pd/Fe/MgO(001) 中，也許是基板表面傾斜的緣故，第一層的鐵只有單軸磁異相性，到了第二層的鐵才出現雙軸磁異相性。當外加磁場和第一層的磁易軸的夾角接近 90 度時，兩個鐵磁層之間的耦合力會和外加場競爭。當外加場不夠強時，上層的鐵磁會被下層的鐵磁吸引並翻轉致同磁化方向。而且當樣品吸附氫氣後，鐵磁耦合會增強，需要更大的外加場才能打破耦合。因此當外加場在 6 Oe 時，氫氣的脫吸附能使上層的鐵磁在外加場和底層鐵磁之間扭轉。關鍵字：氧化鋅、電壓、磁性、介面氧化、氫氣、磁域、鐵磁耦合. iii.

(4) Abstract. Abstract In our study, the magnetic coercivity (Hc) of Fe/ZnO heterostructure was significantly enhanced by 2–3 times after applying a suitable current. This Hc enhancement originates from the Fe-oxidation at the Fe/ZnO interface induced by direct current heating. Depth-profiling X-ray photoemission spectroscopy analysis confirmed the formation of FeO, Fe3O4, and Fe2O3 close to the interface region, depending on the Fe thickness. Furthermore, the magnetic coercivity of Fe/ZnO heterostructure monotonically decreased as a relatively small voltage was applied. The reversibility of this effect was demonstrated by cyclically changing the bias voltage from 0 to 6–9 V; the Hc decreased 15%–20%. As thick Fe-oxide gradually formed at the interface by using a larger direct current heating, the Hc increased and the Fe/ZnO heterostructure still demonstrated a similar voltage-induced the reduction of Hc. In part two, the hydrogenation effect on the magnetic domain formation and domain wall velocity of 25 nm Co30Pd70 alloy thin films grown on SiO2/Si(100) substrates was investigated using magneto-optical Kerr microscopy. There was no domain wall motion observed in vacuum, but the nucleation and domain wall motion was formed and intensified with hydrogen pressure increased. Not only the domain formation but also the reversal motion was changed. Finally, series of reversal time constant 𝜏 , Barkhausen volume V, and depinning field under various magnetic field and hydrogen pressure were deduced from the reversal curve fitting and domain wall velocity. In part three, we deposited Pd/Fe/Pd/Fe multilayer film on MgO(001) substrate. The double loop was due to the pinning of bottom Fe uniaxial easy axis which was 90 degrees to the magnetic field instant of the antiferromagnetic coupling of two Fe layer. The hydrogenation would enhance the magnetic coupling of two Fe layers. If we set the magnetic field at appropriate magnitude, the magnetism of top Fe would be switched between extra field and bottom during the absorption or desorption of hydrogen. Key word: Zinc Oxide, Voltage, Magnetism, Interface Oxidation, Hydrogen, Magnetic Domain, Ferromagnetic Coupling. iv.

(5) Catalogue. Catalogue 致謝................................................................................................................................ ii 中文摘要....................................................................................................................... iii Abstract ......................................................................................................................... iv Catalogue ....................................................................................................................... v 1.. Introduction ........................................................................................................- 1 -. 2.. Basic Concepts ...................................................................................................- 2 -. 3.. 4.. 2.1. Magnetic hysteresis loop........................................................................- 2 -. 2.2. Magneto-optical Kerr effect ...................................................................- 3 -. 2.3. Electrostriction effect .............................................................................- 4 -. 2.4. X-ray photoelectron spectroscopy(XPS) ...............................................- 6 -. 2.5. Hydrogen absorption ..............................................................................- 7 -. 2.6. Magnetic domain ................................................................................ - 10 -. 2.7. Ruderman-Kittel-Kasuya-Yosida (RKKY) theory.............................. - 11 -. Experimental Instruments ............................................................................... - 12 3.1. Ultra-high vacuum system and Electron Beam Evaporator ................ - 12 -. 3.2. Magneto-Optical Kerr Effect (MOKE) ............................................... - 15 -. 3.3. Kerr microscope .................................................................................. - 18 -. 3.4. Pulsed laser deposition (PLD) ............................................................ - 20 -. 3.5. Photoluminescence (PL) ..................................................................... - 21 -. Voltage effect on Fe/ZnO ................................................................................ - 22 4.1. Sample preparation ............................................................................. - 22 -. 4.2. Current induced Magnetic coercivity enhancement ........................... - 25 -. 4.3. XPS data.............................................................................................. - 28 -. 4.4. Voltage-induce reversible coercivity reduce ....................................... - 29 v.

(6) Catalogue 4.5 5.. 6.. Summarize two phenomenon .............................................................. - 32 -. H2-promoted Domain wall motion.................................................................. - 34 5.1. Hydrogen absorption induced hysteresis loops changed .................... - 34 -. 5.2. Magnetic domain reversal changed by hydrogenation ....................... - 36 -. 5.3. Time-dependent magnetic domain reversal ........................................ - 38 -. 5.4. Domain wall velocity .......................................................................... - 45 -. 90˚-Rotation of Fe Using Hydrogen ............................................................... - 49 6.1. Growth and crystalline structure of Pd/Fe/MgO(001) ........................ - 49 -. 6.2. Magnetic coupling in Fe/Pd/Fe ........................................................... - 53 -. 7.. Discussion and Questions ............................................................................... - 61 -. 8.. Summary ......................................................................................................... - 63 -. 9.. Bibliography ................................................................................................... - 64 -. 10. Curriculum Vitae ............................................................................................. - 69 -. vi.

(7) Introduction. 1. Introduction Magnetic materials were widely used in memory element and sensing element.[40-43] For example, the critical core of magnetoresistive random access memory (MRAM) was a magnetic tunnel junction (MTJ).[44-47] But all the magnetism of magnetic thin film was constant after deposited. It was a goal for researcher that a magnetism of thin film could be changed by any external cause like the resistance of semiconductor changing by bias voltage. Some people combined multiferroic materials like PZT or BTO and magnetic thin film to reach voltage-control of magnetism.[48-51] Some paper showed that spin current can reverse the magnetization in a magnetic tunnel junction like CoFeB/MgO/CoFeB structure.[52-55] But there was still no people using ZnO to control magnetic film. ZnO had been widely studied as a semiconducting and photoelectric material for using in technical applications. ZnO also exhibited ferroelectric responses, making it suitable for devices in microelectromechanical and communication.. [1, 56-57]. Some paper showed that ZnO will become ferromagnetic after doping some transitionmetal. [2, 58-59] In our research, the combination of ZnO and Fe nano film achieved the reversible voltage-controlled change in Hc. In addition, hydrogen absorption induced magnetism in Pd/Co system changing had been a subject of research in our lab[16, 28, 38,39]. No matter the multilayer system or alloy system showed the obvious coercivity change after absorbing hydrogen, and turn to the original state after hydrogen desorption. It had a potential in applications for hydrogen sensors. Other paper showed that the hydrogen absorption induced the charge transfer in Pd. [3] But a much clear evidence to show what happen in CoPd system after hydrogen absorption was needed. Here we used Kerr microscope to observe the domain motion and domain wall velocity change after hydrogen absorption in Co30Pd70 alloy, and deduced the change of unit reversal volume and depinning field. Furthermore, ferromagnetic multilayer was an important issue due to the application potential for electronic sensor or element, so our research also touched the hydrogenation in Fe/Pd/Fe multilayer system. We analyzed the coupling behavior of two Fe layers by Land T-MOKE with different angular of sample, and showed that the absorption and desorption of hydrogen could make the top Fe layer reverse like a switch.. -1-.

(8) Basic Concepts. 2. Basic Concepts 2.1. Magnetic hysteresis loop. Hysteresis loop will be observed at the sample which can be polarized under electric or magnetic field and hard to be reversed until the reversed field large enough. It shows some information Here we used MOKE to measure the hysteresis loop of magnetic thin film. The Fig. 1 is the general image of hysteresis loop measured by MOKE. The Xaxis shows how large the magnetic field was, and the Y-axis shows the light intensity (or Kerr signal) which is related with magnetization.. Fig. 1. Schematic graph of MOKE loop. The blue curve is the initial magnetization curve.. At the begin, the magnetization is polarized by field, so the light intensity is increased with large and large field. At the right side of the figure, the intensity is not risen with the magnetic field, which means the saturation state of the sample. Of course the least field of this horizon line is the saturation field. When the field decrease to zero, the magnetization is not going to zero but keeping an amount between zero and saturation state. This state is called remanence state. The remanence shows how large the polarized amount will keep after the field is removed. If the remanence is equal to saturation state, that means this polar anisotropy has the lowest energy at the field direction. On the other hand, if the remanence is less than saturation state, it shows the anisotropy is higher at this direction. When the field is going to be negative, the polarization will not -2-.

(9) Basic Concepts be reversed immediately until the magnetic field overcoming the potential obstacle in the sample. There should exist a moment that the sum polarization is zero in field direction. And the amount of field corresponded to the moment is named coercivity. This three quantities: saturation, remanence and coercivity are most important data to describe a hysteresis loop. Although there still have some detailed description like the shape of hysteresis loop is square or tilted. But that is not the point of this section. We will discuss that in the experiment and result.. 2.2. Magneto-optical Kerr effect. When light reflated from metal, it will enter metal for few to dozens of nanometers deep (in the case of red light) and then reflate out. If the metal is magnetized by magnetic field, that means the permittivity of the metal is different between the magnetic polarization direction and other direction, the different polarization of light will through unequal equivalent path and get phase difference. In condition of linear polarized light, we can separate the polarization in to two part of perpendicular and parallel to the magnetized direction. They will get phase difference and become elliptically polarized light after reflate from metal. The opposite magnetic field will make phase difference become reversed, so the polarization direction of ellipse will be twist. After pass the second polarizer of MOKE, the different elliptically polarized light will make different intensity in light detector.. Fig. 2. This graph shows how two linear polarized light compose variety polarization with different phase difference.. -3-.

(10) Basic Concepts. 2.3. Electrostriction effect. Electrostriction is observable in all dielectric materials. This phenomenon is happened when the electric dipole in crystal get displacement by exposed to an external electric field. The displacement will accumulate through the whole structure and cause a strain in the direction of the field. The electrostriction material also has hysteresis loop, which is same as magnetic hysteresis loop, but the x-axis is electric field and the y-axis is strain. Different with the hysteresis loop of piezoelectricity, the hysteresis loop of strain is proportional to the external electric field. The strain of electrostriction effect is approximate squared relationship of electric field when the external electric field is large enough. But if we only use a small electric field, the strain will proportional to the field.. Fig. 3. x-E and D-E hysteresis loops of VDF(65)/TrEF(35) copolymer.[27]. The strain is linear to electric field if the polarization is not switched by electric field. The direction of strain will be twisted when polarization is switched by field. If the shape of electrostriction hysteresis loop is not squareness, which means the twisting of strain is not obvious in the loop, the shape of strain-electric field loop will be like a butterfly.. -4-.

(11) Basic Concepts. Fig. 4. x-E and D-E hysteresis loops of uniaxially drawn PVDF.[27]. Electromechanical response is also observed on ZnO. The amount of strain in 300 nm ZnO is about 1 nm with 10 V. In other paper, the magnetism of magnetic thin film on PZT can be change by the strain of PZT, and the amount of strain is about 1 nm. So we will research the effect of voltage-induced magnetism change on Fe/ZnO structure.. Fig. 5. The electrostriction effect in ZnO doped 1.2% Fe. [4]. -5-.

(12) Basic Concepts. 2.4. X-ray photoelectron spectroscopy(XPS). X-ray photoelectron spectroscopy is a way to measure what is the element in the sample and know the bonding state of the element. XPS requires keeping in high vacuum when it is working. XPS spectra is obtained by irradiating a sample with a beam of X-ray which has enough energy to excite the inner electron. After conquering the binding energy of the atom, the excess energy will be the kinetic energy of this excited electron. Finally, we just need the velocity of the electron to know the binding energy. The formula is as follows:. 𝐸𝑏𝑖𝑛𝑑𝑖𝑛𝑔 = 𝐸𝑝ℎ𝑜𝑡𝑜𝑛 − 𝐸𝑘𝑖𝑛𝑒𝑡𝑖𝑐 Each atom has unique binding energy, and this binding energy will shift with different molecular binding. If an atom loss some electron to bind with other atom, for example, transition metal bind with oxygen, the binding energy will be larger due to attenuation of shielding effect. On the contrary, get more electron in molecular binding will let the energy spectrum shift to the side of less energy. So the XPS not only can detector what substance it is, but also can measure the charge number of the substance.. Fig. 6. The principle of XPS. The X-ray excite a photoelectron out of atom. The kinetic of the photoelectron will be observed to calculate the binding energy of the atom.. -6-.

(13) Basic Concepts. 2.5. Hydrogen absorption. Hydrogen absorption has been researched in resent year because it is a popular issue about green energy using in the future, there for how to store hydrogen and what will happen become an important topic. A lot of researches show that palladium is a preferred metal to store hydrogen, and some of them also indicate a variety of physical properties changing after hydrogen absorption. Because the electron affinity between hydrogen atom and palladium is large than hydrogen itself, hydrogen molecule will be separate into two atoms due to the lattice constant of palladium large than hydrogen. The hydrogen atoms are so small so that they can diffusion into palladium and stay between palladium atoms.. Fig. 7. Schematic representation of the principal kinetic steps involved in Pd hydriding.[5]. Reflectivity change is a phenomenon after hydrogen absorption. Some research show that Pd/Y thin film could became transparent due to hydrogen absorption.[6] Some paper record that palladium will lose some electron when hydrogen stay at the surface of palladium.[7] Even some paper show that the lattice of palladium will change if the pressure of hydrogen is large enough.[5] In our research, we combine palladium and cobalt and expect that the magnetism of cobalt will be changed by hydrogen absorption of palladium.. -7-.

(14) Basic Concepts. Fig. 8. Calculated local DOS for (a) bulk Pd and (b) Pd(111) surface with and without a monolayer of H atom.[7]. Fig. 9. The magnetic hysteresis loop of post-annealed 5 nm Pd/4nm Co/5 nm Pd trilayers. The coercivity is increased after hydrogen absorption, and recover after desorption.[28]. -8-.

(15) Basic Concepts. Fig. 10. Magnetic hysteresis loops of 14 nm Co14Pd86/Al2O3(0001) was measured by using the magnetooptical Kerr effect for the perpendicular direction. The coercivity was increased and the shape of hysteresis loop became squareness.[16]. In our previous works, we tried the bilayer system of Pd combined with Fe, Co, Ni, and only the Kerr intensity change was observed.[38] In Pd/Co/Pd trilayers system, the clearly coercivity increased with hydrogen absorption was observed after annealing.[28] In addition, the CoPd alloy system was founded that not only the coercivity but also the remanence were increased after hydrogen absorption.[16] However, a much clear evidence to show what happen in CoPd system after hydrogen absorption was needed even though we got a lot of data on MOKE. So here we used Kerr microscope to get more detail from sample, and hope to analyze what happen in CoPd system after hydrogen absorption.. -9-.

(16) Basic Concepts. 2.6. Magnetic domain. Magnetic domain is the magnetization of sample dividing into many part and the polarization direction of each part can be separated in order to reduce magnetostatics energy. A ferromagnetic material with an evenly magnetization everywhere will create a magnetic field in the space surrounding itself. This means there has higher magnetostatics energy. Accordingly, the magnetization will split into two anti-parallel part, or even split into more part to make a round. This make magnetic field lines pass through each domain and reduce the magnetic field outside the round, also reduce magnetostatics energy.. Fig. 11. The reversing magnetic domain in Au/Fe/STO(100).. On the other hand, sometimes magnetic domain is also used to describe the process of magnetized area reversal. When a thin film with uniform magnetization is exposed under an increasing external magnetic field in opposite direction, there should has some point be reversed first. Then the reversal area diffuses from those point until all the area is influenced. We can call this shattered area magnetic domain during the reversal process. The tattoos and the speed of the diffusion are depending on material and defects. We can see the tattoos on the sample with square magnetic hysteresis loop. If the loop is tilted, it will be hard to see any domain.. - 10 -.

(17) Basic Concepts. 2.7. Ruderman-Kittel-Kasuya-Yosida (RKKY) theory. When two magnetic thin films are separated by a non-magnetic thin film, the coupling between two magnetic thin films will oscillate between ferromagnetic and antiferromagnetic as a function of the distance between the layers. This oscillation between ferromagnetic and antiferromagnetic is the Ruderman-Kittel-Kasuya-Yosida (RKKY) theory.. Fig. 12. The exchange interaction of Co layers in fcc Co/Cu(100) systems in terms of an oscillatory interaction function J(n). The open circles show the antiferromagnetic coupling and the close circles show the ferromagnetic coupling with different number of Cu layers between two Co layers. [29]. Fig. 13. Dependence on Pd thickness of the exchange coupling in 5.7 ML Fe/ Pd/ 10 ML Fe trilayers.[30]. The strength of exchange coupling will become very as the thickness of spacer layer increased. Generally, the exchange coupling effect only work in few ML of spacer material. But also some paper show that the exchange coupling can maintain the influence even the thickness of spacer layer increased larger than 10 ML.. - 11 -.

(18) Experimental Instruments. 3. Experimental Instruments 3.1. Ultra-high vacuum system and Electron Beam Evaporator. Our ultra-high vacuum system can be roughly divided into three parts: mechanical pump, turbo pump and vacuum chamber. Before going to vacuum, we need check all ports should be tightly sealed. First, turning on mechanical pump let the pressure going to about 10−3 𝑚𝑏𝑎𝑟. Actually, because there has an aluminum hose about 1~2 meter between mechanical pump and chamber, the pressure will fill an order between them. When the pressure of mechanical pump side is going to 1 × 10−3 𝑚𝑏𝑎𝑟, the pressure of chamber only going to 1 × 10−2 𝑚𝑏𝑎𝑟. But the turbo made by Pfeiffer is designed to use in 1 × 10−1 𝑚𝑏𝑎𝑟 of front pressure, so turning no turbo is fine for that pressure difference. The revolution of HiPace 80 turbo just need 2 minutes to attain 1500 Hz, and the working current will decrease from 3.7 ampere to 0.1~0.3 ampere. If turbo need more time or current to attain or keep the revolution, we need to check each port should be tight or send the turbo to be maintain. After turbo is full speed, we need to bake chamber to about 150°C at least a day so that the pressure will go to 1 × 10−8 𝑚𝑏𝑎𝑟. If not, the pressure will stay at 1 × 10−6 𝑚𝑏𝑎𝑟.. - 12 -.

(19) Experimental Instruments. Fig. 14. Ultra-high vacuum system with Electron Beam Evaporator.. In order to get better ultra-high vacuum, the chamber require cleaning the filament and source of electron beam evaporator. We have two kinds of electron beam evaporator in our lab. One can do target linear translation. The electron can hit the tip of source directly. In order to avoid the vapor of metal spray everywhere, there has a big copper shell around the evaporator to shield the vapor. Because the copper shell is hanging inside the chamber, so the contact area is small and it is hard to baking the shell. Therefore, we need light the filament few hours to baking the shell. The another one electron beam evaporator is using crucible. The target was melted in the crucible by electron beam at first used. The crucible is insulating from chamber by ceramics and hart to bake. So we need use electron beam to heat crucible for cleaning. After baking, we can remove all the heat source, and the pressure will go to 1 × 10−9 𝑚𝑏𝑎𝑟.. - 13 -.

(20) Experimental Instruments. Fig. 15. The Electron beam evaporator made by me. The main structure is a crucible and a filament.. Electron beam evaporator can be divided into election source part and target part. The election source part is a filament. It will have free electron when current pass and light the filament. The target has 1000V to attract those free electrons. The posited volt side connect to the target, and the negative side and ground connect to the filament. In other words, there has 1000V potential energy between filament and target, and that energy will change to thermal energy to heat target when free electron hit on it. It should be attended that we need to check that target is hot enough for deposit.. - 14 -.

(21) Experimental Instruments. 3.2. Magneto-Optical Kerr Effect (MOKE). A simple magneto - optical Kerr effect should have light source, two polarizers, electromagnetic coil and photodiode detector. The light source is solid-state laser for 4.5mW and the wavelength is 670 nm. The light of laser is polarized by first linear polarizer, and become elliptically polarized light after reflected from sample magnetized. Afterword, the second linear polarizer can be used for checking the change of elliptically polarization along with the switched magnetic field.. Fig.16. The magneto - optical Kerr effect system. At the top is no core MOKE for YIG measuring.. The polarizing axis of polarizer is close to the semi-minor axis of elliptical polarization. When the elliptical polarization is twisted due to the magnetization of sample changed by magnetic field, the projection of elliptical polarization on polarizer will be changed.. - 15 -.

(22) Experimental Instruments. Fig. 17 The elliptical polarization switched by magnetic field. The rad line is the angular of second polarizer.. In Fig. 17, there has two parabolas representing the light intensity about semi-minor axis of elliptical polarization. They will exhibit in turn with the switched magnetic field. The X-axis is the angle between the semi-minor axis of elliptical polarization and polarizing axis of polarizer. If the polarizer is fixed at positive angle side, the light intensity in positive magnetic field is stronger than the light intensity in negative magnetic field. The light intensity changing represent the high and low side of magnetic hysteresis loop. When the polarizer is turned to the negative angle side, the hysteresis loop will become upside-down. In special cases, there will be nothing when the polarizer angle is near the cross of two parabolas. Two elliptical polarization have same projection on polarizer and the light intensity will not be changed by magnetic field. So we should avoid this when we doing MOKE measurement.. Fig. 18. The schematic diagram of polar, longitudinal and transversal MOKE. - 16 -.

(23) Experimental Instruments MOKE can be categorize by the magnetic field direction with respect to the plane of incidence and the reflecting surface. The polar MOKE is the magnetic field perpendicular to the reflecting surface of sample, so someone will call it as perpendicular MOKE. When doing polar MOKE, we usually make the incident light much close to the normal incidence for getting better polar Kerr signal. The longitudinal MOKE is the magnetic field parallel to the incident plane and sample surface. Same as the polar MOKE, we also make the incident light much close to the sample surface for getting better longitudinal Kerr signal. If someone mention in-plane MOKE, that means the longitudinal MOKE. The transverse MOKE is the magnetic field parallel to the sample surface but perpendicular to the incident plane. In transverse MOKE, we may get a strange magnetic hysteresis loop if the easy axis of sample is parallel to the magnetic field but perpendicular to the incident plane. Only when the magnetic moment is switching by magnetic field, the magnetic moment will be parallel to the incident plane. So we will get to peak in transverse MOKE hysteresis loop at where the coercivity is in longitudinal MOKE hysteresis loop.. - 17 -.

(24) Experimental Instruments. 3.3. Kerr microscope. In a nutshell, Kerr microscope is a MOKE which is combined with microscope and charge-coupled device camera, so it can take an image or even a video to record the magnetization reversal of sample instead of just a total intensity change of Kerr signal. The resolution of our Kerr microscope can be 216 nm each pixel by using the oil objective lens in room temperature measuring. But if we need to use the low temperature system, we should zoom out until 1μm each pixel (the image is about 450μm × 450μm). The low temperature system can measure a sample 1cm × 1 cm large at 90K, and also can be used to expose sample to hydrogen. So we can observe the domain motion change after hydrogen absorption on our CoPd alloy sample.. Fig.19. The Kerr microscope system.. There has two direction of magnetic field, in plane and perpendicular, can be used. At in plane mode, we can rotate the angular of sample and magnetic field. There also have three kinds of light mode can be chosen, so we can measure polar, longitudinal, transverse MOKE without moving sample. That is convenient to us for measuring the easy axis of Fe/Pd/Fe multilayers on MgO(001). - 18 -.

(25) Experimental Instruments. Fig. 20. The different mode of MOKE in the Kerr microscope.. Fig. 20 shows the different mode of MOKE in the Kerr microscope. There have 8 LED lights in Kerr microscope. With the combination of LED lights, we can use different mode of MOKE in our experiment. The plus sign means two direction of LED light will open at same time. The horizontal vector of light will be offset and we can get the pure polar Kerr signal. The minus sign means two direction of LED light will alternately open. The vertical vector of light will be offset and we can get the pure longitudinal or transversal Kerr signal.. - 19 -.

(26) Experimental Instruments. 3.4. Pulsed laser deposition (PLD). Pulsed laser deposition (PLD) is a kind of physical vapor deposition (PVD) technique using a high-power pulsed laser beam to heat a target what material we need to deposit on sample. The process of heating and depositing are in a high vacuum chamber. Instant of heating all of target, the pulsed laser can only evaporate the surface because the very short heating time prevents that the bulk of the target absorb the energy. The power of pulsed laser is 350 mJ per second. If the target is metal oxide, it will become metal ion and oxygen ion when the material is evaporated by laser. In order to make sure that the bonding of molecular we deposit on sample is same as the target, the oxygen is filling the chamber as 8 × 10−2 to 1 × 10−1 𝑚𝑏𝑎𝑟.. Fig. 21. The pulsed laser deposition system in Prof. Lo’ s lab.. - 20 -.

(27) Experimental Instruments. 3.5. Photoluminescence (PL). Photoluminescence (PL) is that any matter absorbs photons and then emit light. The electron in matter will excite to higher energy level after absorption of photons, then fall to original level. If there has energy band gap in the matter, light emission will happen in the relaxation process. The relaxation processes are various and the emission light may excite electrons to be re-radiated. Observation of photoluminescence at a certain energy can used to resolve the energy band gap of matter. So it is an appropriate instrument to observe the quality of semiconductor thin film.. Fig. 22. The Photoluminescence system in Prof. Lo’ s lab.. - 21 -.

(28) Voltage effect on Fe/ZnO. 4. Voltage effect on Fe/ZnO 4.1. Sample preparation. In our experiment, the Au/Fe/ZnO/Au multiple layers was prepared on Al2O3(0001) substrates by electron beam evaporator and pulsed laser deposition (PLD). The Au and Fe layers were deposited in ultra-high vacuum chamber with basic pressure of 3 × 10−9 𝑚𝑏𝑎𝑟. The ZnO layer was deposited at room temperature or 600 K with oxygen pressure of 8 × 10−2 𝑚𝑏𝑎𝑟 by PLD system. The energy of laser was 350mJ per second. The top and bottom of Au were used as electrodes when applying voltage. The parameter of E-beam heat evaporator was 1000V, 40mA for Au and 85mA for Fe. Each layer was deposited with different mask shown in Fig. 23.. Fig. 23 There are the masks and holders for fixing the Al2O3 substrate when depositing each layer. Number 4 was the bottom of the holder. Number 3 was at middle to fix substrate. Number 1 and 2 were at top as mask for depositing the top and bottom of Au. The mask for depositing ZnO was a baffle which is not in the picture.. - 22 -.

(29) Voltage effect on Fe/ZnO. Fig. 24. The process of depositing Au/Fe/ZnO/Au multiple layers on Al2O3 substrate.. First was Au at bottom as an electrode. Depositing Two strips of Au was for one more chance of measure because too much current will damage the sample. The second step is depositing ZnO at PLD chamber. We were used to make the area of ZnO is two thirds on substrate for ensuring the top Fe isolated from bottom Au. The final was depositing Fe and Au layer. Au was used as the protective layer to Isolate the Fe layer and air. The area of top layer was less than one third of substrate to prevent the contact with bottom Au. The size of the junction area is 1mm×1.5mm (The area each layer was overlap).. Fig. 25. Side view of Au/Fe/ZnO/Au multiple layers on Al2O3 substrate.[33]. - 23 -.

(30) Voltage effect on Fe/ZnO Fig. 25. showed the side view of the sample. Several different thickness Fe and ZnO sample were prepared. The thickness of Fe was 2.5, 5 and 10 nm and ZnO was 1~3 nm. Top 2nm of Au was protective layer, and the bottom Au was electrode.. Fig. 26. PL spectra of Fe/ZnO heterostructures measured before and after applying a current to induce the coercivity enhancement. [33]. Fig. 26 showed the photoluminescence(PL) spectra of Fe/ZnO heterostructures. Four emission peaks identified from the spectra were 2.65, 3.00, 3.20, and 3.35 eV. They represented oxygen-defect related, donor-acceptor pair, bound exciton, and free exciton emission lines. [8-10] The PL intensity was reduced after current annealing. The emission peak of free exciton emission was relatively reduced and the free exciton emission peak was increased. The above observations could be represented that oxygen in ZnO was lost or captured by Fe at the interface during annealing.. - 24 -.

(31) Voltage effect on Fe/ZnO. 4.2. Current induced Magnetic coercivity enhancement. The following discussion would focus on current annealing induced magnetic coercivity enhancement, but the independent variable would be voltage, not current. The current through the sample would reduce due to resistance increased during interface oxidation and became hard to show in graph, so the independent variable was showed in voltage.. Fig. 27. Normalized magnetic hysteresis loops of 5 nm Fe/2 nm ZnO structure was measured by longitudinal MOKE after applying different voltages. The coercivity gradually increased from 48 Oe after 5.5V and saturated at 180 Oe after 10 V applying. [33]. Fig. 27 showed normalized magnetic hysteresis loops of 5 nm Fe/2 nm ZnO structure measured by longitudinal MOKE after applying 0~12 voltages. The bias voltage was increased from 0 to 12V by each step of 0.5V. Hysteresis loops was measured after applying bias voltage 30 second. There was no any variety of hysteresis loops during the bias voltage increased to 5.5V. After 6V, the coercivity was increased from 48 Oe and saturated at 180 Oe with 9V. The shape of hysteresis loop became tilted but the ratio of remanence to saturation always maintained 100%. Bias Voltage higher then 9V didn’t make any change of hysteresis loop.. - 25 -.

(32) Voltage effect on Fe/ZnO. Fig. 28. (a) Magnetic coercivity, (b) current, (c) resistance, and (d) sample temperature was graphed as a function of bias voltage. [33]. Fig. 28 showed more information about what happened when the magnetic coercivity (Hc) enhanced. The Hc, current, resistance, and sample temperature was graphed as a function of bias voltage. The sample temperature was measured by k-type thermocouple contact with the junction area when the voltage was applied. As voltage was bellowed 5V, the Hc kept like a constant and the current increased stably with raised voltage, so this indicated that the Fe/ZnO heterostructure was ohmic conducting behavior. After the voltage was increased over 5V, irreversible changes of magnetism and conducting property were observed. As the voltage increased, the sample current and Hc increased first and then got to maximum. The maximum temperature heated by current at Fe/ZnO junction was about 500K, but the lower limit of temperature to enhance coercivity was about 400K which could be observed by compare the temperature and coercivity function. When voltage exceeded 6V, the temperature was higher than 400K, and the coercivity began to increase. As voltage transcended 7V, the increasing of resistance due to the Fe/ZnO junction altered resulted in the decreasing of current and heating temperature, but the coercivity still increased because the - 26 -.

(33) Voltage effect on Fe/ZnO temperature was higher than 400K. After the voltage was over 9V, the temperature reduced below 400K and stopped the increasing of coercivity.. Fig. 29. The magnetic coercivity exhibited as a function of (a) bias voltage and (b) current density in variety thickness of Fe/ZnO structure. In (b), the arrow showed the sequence of data about Hc with the variation of current density. [33]. Fig. 29 summed up the enhancement of coercivity in various Fe/ZnO structures. For each ZnO thickness, similar enhancement of coercivity always happened when the applied current density was large enough. In Fig. 29(a), the coercivity enhancement needed the voltage larger then 6-9 V in 2 nm ZnO system, but in 3 nm ZnO system, there needed the voltage larger then 12-14 V. it appeared that thicker the ZnO was, - 27 -.

(34) Voltage effect on Fe/ZnO larger voltage was needed to reach the same heating power. In Fig. 29(b), the coercivity evolution for each sample was plotted with the x-axis as the applied current density. Every sample always required its unique minimum current density for coercivity enhancement. The range were about 10–13 A/cm2, and the coercivity enhancement would stop when the current density reduced less than the minimum.. 4.3. XPS data. Fig. 30. The depth analyzing XPS of Fe measured at the indicated depth near the interface of Fe/ZnO. Two samples were deposited with same parameters but different thickness. The right part of graphs were before annealing and lift part were after annealing. The peaks of Fe, Fe +2, and Fe+3 were shown by the arrows. [33]. In order to find out the evidence to prove the contention of current annealing inducing interface oxidation, the depth analyzing XPS of Fe was measured at the indicated depth near the interface of Fe/ZnO. Fig. 30 showed the XPS spectra of 2.5 and 10 nm Fe/ZnO structure with the depths analysis. There had two junctions deposited on one sample simultaneously. One was annealed by current, and another one was just checked by MOKE. So we could compare the XPS spectra from each sample. Many previous - 28 -.

(35) Voltage effect on Fe/ZnO papers studied the characteristic XPS peak positions of Fe (707 eV), Fe+2 (709.9 eV), and Fe+3 (711.4 eV). [11-12] The peaks of Fe, Fe+2, and Fe+3 were shown by the arrows in Fig. 30. In Fig. 30(b), for 2.5 nm Fe system as deposited, the top of Fe layer was mostly pure metallic Fe, but the Fe near the interface of Fe/ZnO had been oxidized. The peak of Fe and Fe+2 simultaneously exist at depth. In Fig. 30(a), after the current annealing sufficient to induce an obvious Hc enhancement, a peak at 710.5 eV between Fe+2 and Fe+3 gradually became the main peak in XPS spectra when the depth is close to the Fe/ZnO interface. That meant there was Fe2O3 forming at the interface[13-15]. In the case of 10 nm Fe, the pure Fe peak was the main peak with a shoulder of Fe+2 in asdeposited sample shown in Fig. 30(d). After annealing, the peak of Fe+2 gradually became the main peak the XPS spectrum with increasing depth shown in Fig. 30(a). Comparing two sample after annealing, the thicker Fe film tended to form Fe+2 due to the shortage of oxygen because of the limited ZnO thickness, so only FeO was formed in 10 nm Fe/ZnO, rather than the Fe3O4. Fe2O3 and Fe3O4 could be ferromagnetic, and FeO could be antiferromagnetic. That was the reason the coercivity would be larger after junction annealed.. 4.4. Voltage-induce reversible coercivity reduce. In order to achieved the purpose that using voltage to influence magnetic coercivity reversibly, the thickness of ZnO was raised to 320 nm. In the previous studies, ZnO had obvious electrostriction effect[4], so we imitated all the detail in that paper to deposit ZnO. The sample Au (2 nm) / Fe (3 nm) / ZnO (320 nm) / Au (40 nm) / Al2O3 (0001) was prepared. The increased thickness of bottom Au was for reduced the resistance.. - 29 -.

(36) Voltage effect on Fe/ZnO. Fig. 31. (a) MOKE hysteresis loops with various bias voltages. (b) Summarized Hc values drawn as a function of bias voltage. The solid lines are guides for the eye. [34]. Fig. 31 showed the MOKE hysteresis loops measured at room temperature using a magnetic field along the in-plane direction with the applied bias voltag increased from 0 to 6 V. The magnetic coercivity (Hc) of the MOKE hysteresis loops decreased as the bias voltage increased, while the shape of hysteresis loop and the ratio of remanence were still invariant in Fig. 31(a). The Hc values drawn as a function of bias voltage from +6 V to -6 V in Fig. 31(b). The Hc was 112 ± 2 Oe without bias voltage. When the bias voltage increased from 0 to +3 V, the Hc decreased a little to 109 ± 2 Oe. While the bias voltage exceeded +3 V, the Hc intensely reduced and attaining a minimal Hc value of 93 ± 2 Oe at +6 V. There was the same variation about Hc reduced when the applied voltage was negative. So the Hc reduction was not related to the electric field direction. - 30 -.

(37) Voltage effect on Fe/ZnO. FIG. 32. (a) MOKE hysteresis loops were measured as the bias voltage was switched between 0 V and 6 V. (b) The value of Hc and switched bias voltage were shown as a function of time. [34]. Fig. 32 showed that the voltage-induced Hc reduction was reversible. The bias voltage was switched between 0 and 6V repeatedly and the MOKE hysteresis loops were recorded, as shown in Fig. 32(a). The hysteresis loop always deformed to same shape with 6V bias voltage applied. As shown in Fig. 32(b), the Hc switched between 112 ± 2Oe and 93 ± 2 Oe reversibly, which corresponded with the applied voltages between 0 and 6 V. Here we only used 6 V for the maximal voltage because that a large current generated by high voltage could heat the Fe/Zn junction and yielding further Feoxidation at the Fe/ZnO interface. This phenomenon would hinder the observation of voltage-induced Hc reduction, so the maximal voltage was set at 6 V. - 31 -.

(38) Voltage effect on Fe/ZnO. 4.5. Summarize two phenomenon. Fig. 33. MOKE hysteresis loops after different voltage annealing and under appropriate voltage to reduced coercivity. The hollow points represented that the loop was measured after annealing and without voltage. The solid points represented the loops was measured with appropriate voltage.. The previous data showed that the coercivity would be enhance due to the Fe/ZnO interface oxidation caused by enough current annealing. To investigate how voltageinduce reversible coercivity reducing would be after current induce interface oxidation, we annealed the sample by high voltage and measured hysteresis loops with suitable voltage. The Fe/ZnO junction was annealed under different voltage for 10min, and then measured the loop with appropriate voltage. The hollow points represented the loop measured after different voltage annealing, and the solid points represented the loops measured under appropriate voltage. Fig. 33 showed that even after annealing and interface oxidation, the voltage-induced coercivity enhancement was always effective.. - 32 -.

(39) Voltage effect on Fe/ZnO. Fig. 34. The summarized Hc values were shown as a function of bias voltage for as-deposited sample and the samples after 8V, 10V, and 12V annealed to the Fe/ZnO junction for 10 min. The solid lines are guides for the eye. [34]. Fig. 34 showed the summarized Hc values as a function of bias voltage. After applying 10 V for 10 min, the Hc value irreversibly increased from 112 ± 2 eV to 147 ± 4Oe, which was measured at V = 0. Because the resistance of Fe/ZnO junction was increased after the 10 V annealing, the sample could bear a higher voltage of at least 8 V without further Fe oxidation. The Hc reduced by voltage was shown as a function of bias voltage. After the 10 V annealing for 10 min, the Hc decreased from 147 ± 4 Oe to 112 ± 4 Oe as the bias voltage was increased from 0 V to 8 V. Certainly the Hc recovered to 147 ± 4 Oe after bias voltage was removed. After the 12 V annealing for 10 min, the Hc increased again to 189 ± 4 Oe measured at V = 0, and the junction could bear at least 9 V. The Hc reversibly decreased from 189 ± 4 Oe to 147 ± 4 Oe as the bias voltage was increased from 0 V to 9 V. The above experimental observations could be summarized as following. (1) When the applying voltage was larger enough to anneal the junction, the Fe/ZnO interface oxidation would begin and lead to the irreversible Hc enhancement. (2) If the applying voltage was not enough to heat up the sample, interface oxidation would not happen and the reversible Hc reducing was observed within the small voltage region. - 33 -.

(40) H2-promoted Domain wall motion. 5. H2-promoted Domain wall motion 5.1. Hydrogen absorption induced hysteresis loops changed. In our previous research, CoPd alloy system would change coercivity even the magnetic anisotropy after hydrogen absorption [16, 32]. But only the hysteresis loops could use to observe this phenomenon, no more measure could be used to analysis that. Now we have Kerr microscopy recently, so we could observe any change of magnetic domain after sample absorbed hydrogen. In this research, we prepared Co30Pd70 25 nm on SiO2/Si(001). The SiO2/Si(001) substrate was chosen not the Al2O3 because that the light passed through Al2O3 and reflected from the bottom would interfere the measurement of Kerr microscopy.. Fig. 35. (a) Longitudinal and (b) polar MOKE hysteresis loops was measured under vacuum and 0.2-0.8 bar H2 pressures. The dashed red curve in Fig.35(a) was measured after the recovery to vacuum and was almost same as the hysteresis loop of the initial vacuum condition (solid black curve). [35]. - 34 -.

(41) H2-promoted Domain wall motion The hysteresis loops of longitudinal direction and polar direction are shown in Fig. 35. The polar MOKE hysteresis loop was not saturated because the limit of our polar vacuum coil. In vacuum, the longitudinal hysteresis loop was tilted, and the coercivity was 700 Oe. In polar direction, the coercivity was 450 Oe and the hysteresis loop was tilted also, but the real coercivity should be larger than that. Under 0.2 bar H2, the coercivity reduced to 140 Oe in longitudinal direction and to 200 Oe in polar direction. The shape of hysteresis loops was become square and the remanence was increased to 100%. When the pressure of H2 increased from 0.2 bar to 0.8 bar, the remanence was still 100%, but the coercivity increased slightly. The coercivity increased from 140 Oe to 194 Oe in longitudinal and from 200 Oe to 435 Oe in polar direction. The dashed red curve in Fig. 35(a) was measured after the recovery to vacuum and was almost same as the hysteresis loop of the initial vacuum condition.. Fig. 36. The magnetic coercivity and remanence changed in longitudinal and polar direction under vacuum and different H2 pressure. [35]. - 35 -.

(42) H2-promoted Domain wall motion. 5.2. Magnetic domain reversal changed by hydrogenation. Fig. 37. Longitudinal Kerr images were taken with magnetic field near the coercivity under (a) a vacuum, (b) 0.2 bar H2, and (c) 0.8 bar H2. All Kerr images were taken at same area. The image size was 450 × 450 μm2 . [35]. - 36 -.

(43) H2-promoted Domain wall motion Those domain images were taken as magnetic hysteresis loops showed in Fig. 37 measuring, and the defect with ‘K’ shape at lower right corner Implied the measure was at same area. The image of Co30Pd70 in vacuum during reversal in (a) were shown. There was no observable domain contrast in the Kerr image due to the very small nucleation motion dominated the magnetization reversal. The nucleation was too small and reversed randomly so only the gray scale became darker and darker with the increasing of magnetic field. When the pressure of H2 was 0.2 bar, there was a lot of needle-like domain structure were seen at 0.91 Hc of magnetic field. The number and size of domains were increased when the magnetic field increased to 0.98 Hc. As the field increased to 1.01 Hc, the divided domain expanded until contact with neighbor domain, and the domain motion leaded the reversal. For the sample exposed to 0.8 bar H2, as showed in Fig. 37(c), the appeared domain was much larger and like diamond shaped with a magnetic field of 0.91 Hc, instant of needlelike that under 0.2 bar H2. As the magnetic field was increased to 0.97 and 1.0 Hc, there was less new nucleation domain appeared, and the reversal motion was replaced by expanding and merging of domain compared with the case that under 0.2 bar H2. From Fig. 37(a)-(c), we could make a conclusion that the hydrogen absorbed into Co30Pd70 made the number of domain decreased, the size of domain increased and the domain wall motion leaded the reversal gradually with the increased of H2 pressure. From vacuum to 0.2 bar H2, the observable domain formed and replaced the small domain, but the reversal was still dominated by nucleation motion until the magnetic field larger than 1.01 Hc. From 0.2 bar H2 to 0.8 bar H2, the size of domain became larger and the number of domain almost none increased with larger magnetic field, and the reversal leaded by the domain wall motion once the domain appeared.. - 37 -.

(44) H2-promoted Domain wall motion. 5.3. Time-dependent magnetic domain reversal. Fig. 38. Time-dependent longitudinal Kerr images taken under (a) vacuum, (b) 0.2 bar H2, and (c) 0.8 bar H2. The three series of Kerr images were taken over the same area at 1, 2, 3, and 6 s after the application of the magnetic field which was 0.9 Hc. The image size was 450×450 mm2. [35]. To analyze the hydrogen-mediated magnetic domain wall motion, the time-dependent magnetic domain reversal processes were monitored with the different magnetic fields and under different H2 gas pressures. Fig. 38(a) showed the series of Kerr images were taken under vacuum with an applied magnetic field of 0.9 Hc. No observable magnetic domain contrast came out at 1-6 s after the applied magnetic field, no matter how large the magnetic field was applied or the length of time we waited after the field was applied. - 38 -.

(45) H2-promoted Domain wall motion In Fig. 38(b), after exposed to 0.2 bar H2, a lot of needle-shape reversal nucleation appeared at 1 second after the applied magnetic field of 0.9 Hc,. Over the following 26 s, the nucleation number was nearly unchanged, which meant the density of nucleation was no increased, and the domain extended by domain wall motion. In Fig. 38(c), under exposed to 0.8 bar H2, reversed nucleation started to expand by domain wall motion at 1 s after applied magnetic field of 0.9 Hc. Compared with the condition of 0.2 bar H2 in Fig. 38(b), the sample exposed to 0.8 bar H2 had less nucleation density and large domain size, but domain wall motion dominated the magnetization reversal at the waiting under 0.2 and 0.8 bar H2.. Fig. 39. Time-dependent magnetization reversal curves were recorded under (a) vacuum, (b) 0.2 bar H2, and (c) 0.8 bar H2 after applied various reverse magnetic fields. The Kerr intensity was obtained from the sum of pixel from the Kerr images showed in Fig. 38. [35]. In order to get more information about the reversal, we made the Time-dependent magnetization reversal curves were recorded under different H2 pressure after applied various reverse magnetic fields. Before the magnetization reversal, the sample was saturated by a large positive field, and retained at the remanence state, then various negative magnetic field was applied to start the domain evolution. Fig. 39. showed the time-dependent magnetization reversal curves were recorded under (a) vacuum, (b) 0.2 bar H2, and (c) 0.8 bar H2 after applied various reverse magnetic fields. As previously proposed by Bruno et al., time-dependent magnetization reversal can be described by a kinetic model as follows: [17, 18] (1). 𝑀(𝑡) = 2𝑀 ⋅ 𝑒𝑥𝑝(− 𝑡⁄𝜏) − 𝑀 - 39 -.

(46) H2-promoted Domain wall motion Where M denoted the average saturation magnetic moment and τ was the time constant. In Fig. 39(b) and (c), the solid curves showed the fitting results of the time-dependent magnetization reversal data by using Eq. (1). When the larger reversal magnetic field was applied, the magnetization reversal completed faster and the smaller τ was obtained in the fitting. Only the series of data in Fig. 39(b) and (c) could be successfully fitted by used Eq. (1). That indicated the model could accurately describe the magnetization reversal behavior under 0.2-0.8 bar H2, which was discussed later in this section. However, it was difficult to fit the magnetization reversal curve measured in a vacuum, shown in Fig. 39(a). The magnetization reduced drastically immediately after applied the magnetic field, then slowly decreased with a nearly constant slope, and that was almost same under variant magnetic field. The magnetization reversal curves in Fig. 39(a) was totally different behavior from what described by Eq. (1). The reason might be shown in Fig. 39(a) and 39(a), where the magnetization reversal under vacuum was dominated by submicrometer scale nucleation without observable domain wall motion. The magnetization reversed immediately by nucleation motion after (< 1 s) the magnetic field was applied, and that was accounted for the intense jump in the first second of data shown in Fig. 39(a). When the Co30Pd70 alloy film was exposed to 0.20.8 bar H2, comparing to the nucleation motion under vacuum, domain wall motion was observed and dominated the magnetization reversal, so Eq. (1) can successfully fit the data curves in Fig. 39(b) and (c).. - 40 -.

(47) H2-promoted Domain wall motion. Fig. 40. Series of time constant τ under various magnetic fields and different H2 pressures were calculate from the curve which was fitted in Fig. 39 by Eq. (1). ln(τ) was linear with applied magnetic field H according to Eq. (4). The dots represented the experimental data and the solid lines were the results of linear fitting. [35]. The time constant τ depended on the activation energy EA for the magnetic flipping of each microscopic element compared with the thermal energy 𝑘𝐵 𝑇.[17, 18]  = 𝜏0 ⋅ exp(− 𝐸𝐴 ⁄𝑘𝐵 𝑇). (2). The activation energy EA was linear with the magnetic field H which was applied in the magnetization reversal process, as shown in Eq. (3). [17, 18] (3). 𝐸𝐴 = 𝑉 ⋅ 𝑀(𝐻 + 𝐻𝑝 ). The V was the Barkhausen volume which was reversed within a single activation motion in the magnetization. Hp was the field required for magnetization reversal without any other activation processes. We could observe that the time constant τ had the relation with the applied magnetic field H from the combination of equations (2) and (3). ln(τ) is linear with H. [17,18]] 𝑉⋅𝑀. 𝑙𝑛(𝜏) = − (𝑘. 𝐵 ⋅𝑇. ) ⋅ 𝐻 + 𝑙𝑛(𝜏0 ) −. 𝑉⋅𝑀⋅𝐻𝑝. (4). 𝑘𝐵 ⋅𝑇. - 41 -.

(48) H2-promoted Domain wall motion So, if we could fit the curve by equation (1), we would get a series of τ values with the various magnetic fields under different H2 pressures. Fig. 40 showed that the 𝑙𝑛(𝜏) was linear with applied magnetic field H and could be fitted by Eq. (4). From the linear slope parameter −(V ⋅ M/k 𝐵 ⋅ T), we could deduce the Barkhausen volume V, because it was known that T = 300 K and M ≈ 0.7 ± 0.1 μ𝐵 /atom for Co30Pd70 alloy.[19, 20] The meaning of V values were discussed later in this paper.. Fig. 41. Time-dependent polar Kerr images were taken under (a) vacuum, (b) 0.2 bar H2, (c) 0.6 bar H2, and (d) 0.8 bar H2. The four series of Kerr images were taken over the same area at 2, 3, 4, and 6 s after the application of the magnetic field which was 0.92 Hc. The image size was 450×450 mm2. [35]. Because the 25-nm Co30Pd70 alloy film revealed polar magnetization reversal behavior, which was shown in Fig. 41 that the sample was coexistence of longitudinal and polar, so we also needed to investigate the domain wall reversal motion in polar moment. Fig. 41 showed the polar MOKE images recorded from 2 s to 6 s after applied reverse magnetic field with 0.92 Hc under different H2 pressure. No observable magnetic domain structure for the Co30Pd70 alloy film existed under vacuum, which was same as - 42 -.

(49) H2-promoted Domain wall motion the results of longitudinal measurement shown in Figs. 37 and 38. The gray scale of the polar Kerr images dark as the magnetic field applied without any observable contrasting structures, and slowly darker and darker with time. The reversal curve with different magnetic field also could not be fit by Eq. (3) because the reverse magnetic field only triggered magnetization switching and couldn’t spread nucleation sites or domain. Under 0.2 bar H2, as showed in Fig. 41(a), many small needle-like domains appeared after 2 s at left-hand side and then spread toward right-hand side. In Fig. 41(b), the nucleation under 0.4 bar H2 was considerably less than those under 0.2 bar H2 but much larger. Domain wall motion and domain merging also became faster and more clear. As the H2 gas pressure increased from 0.4 to 0.8 bar, the number of nucleation decreased and the domain wall motion was slightly faster than that at 0.4 bar. All the situations in polar motion were almost same as longitudinal.. Fig. 42. Series of time constant τ under various polar magnetic fields and different H 2 pressures were calculate from the curve which was fitted in Fig. 41 by Eq. (1). ln(τ) was linear with applied magnetic field H according to Eq. (4). The dots represented the experimental data and the solid lines were the results of linear fitting. [35]. By fitting the time-dependent magnetization reversal data with Eq. (1), The values of time constant τ could be deduced. Fig. 42 showed the ln(τ) was linear with applied magnetic field H according to Eq. (4), and from the slope of the fitting lines, the Barkhausen volume V can be calculated with the parameters of T = 300 K and M ≈ 0.7±0.1μ𝐵 /atom. - 43 -.

(50) H2-promoted Domain wall motion. Fig. 43. Barkhausen volume was respectively calculated from the fitting in Figs. 40 and 42 for the polar (open circle) and longitudinal (open square) magnetization reversal. [35]. Fig. 43 showed the Barkhausen volume V for the longitudinal and polar magnetization reversal, which were calculated from the fitting curves in Figs. 40 and 42. Because the thickness of Co30Pd70 alloy film was 25 nm, we could calculate the activation area in each magnetization reversal event by dividing V with 25 nm. In the polar Kerr measurement, when H2 pressure was increased from 0.2 to 0.8 bar, V decreased from 6800 ± 500 to 3000 ± 500 nm3. Identically, the activation area A decreased from 272 ± 20 to 120 ± 20 nm2. In the longitudinal Kerr measurement, V decreased from 4600 ± 500 to 3000 ± 500 nm3 when the H2 pressure increased from 0.2 to 0.8 bar. Identically, the activation area A decreased from 184 ± 20 to 120 ± 20 nm2. The deviation between polar and longitudinal measurements might be due to the reorientation of moment tilted angle. Nevertheless, the polar and longitudinal measurements showed same tendency that the V was decreased with H2 pressure. From Eq. (3), we could know that the activation energy EA in a single magnetization reversal event was proportional to V. Therefore, the hydrogen absorption in Co30Pd70 thin film induced the decrease of V suggested that the hydrogen content reduced the activation energy of single magnetization reversal event and promoted domain wall motion. On the other side, without enough H2 gas in the environment, large activation energy was needed to reverse a single magnetization. This phenomenon indicated that - 44 -.

(51) H2-promoted Domain wall motion reversing a single domain became harder in air or vacuum, so the domain wall motion was almost unobservable and the nucleation motion would dominate magnetization reversal.. 5.4. Domain wall velocity. Fig. 44. Time-dependent polar Kerr images under 0.8 bar H2 with magnetic field of 135 Oe were captured for observing domain wall motion. The inverted triangles showed the initial and final position of domain wall. (b) Summarized domain wall position X taken under 0.8 bar H2 with various magnetic fields plotted as a function of time. [35] - 45 -.

(52) H2-promoted Domain wall motion Fig. 44(a) showed the time-dependent polar Kerr images which were taken under 0.8 bar H2 with magnetic field of 135 Oe in the polar direction. The inverted triangles showed the initial and final position of domain wall. The defect positions on the left side in Fig. 44(a) showed that the position was fixed and the measurement was always in the same area. In Fig. 44(b), the DW position (X) was plotted as a function of time. A series of measurements were taken under 0.8 bar H2 with various reversed magnetic fields. The solid markings were the experiment data, and the lines were the linear fitted from the experiment data. As the magnetic field was larger, the slope of the fitting line was larger, which indicated that the expanding velocity of domain wall became faster. Because the magnetization reversal was dominated by nucleation motion under 0.2 bar H2, which denoted it was hard to find out an area only existing domain wall motion without nucleation motion, so here we didn’t discuss the domain wall expanding velocity under 0.2 bar H2.. - 46 -.

(53) H2-promoted Domain wall motion. Fig. 45. (a) Domain wall velocity values under different H2 gas pressure were shown as a function of the reciprocal of the reversal magnetic field. (b) ln(v) was shown as a function of 𝐻 −1⁄4 . The lines were fitted with the experiment data (solid notations) deduced by using Eq. (5), and the slope of the lines was H𝑒𝑓𝑓 = (𝑈𝑐 ⁄𝑘𝐵 𝑇)4 𝐻𝑐𝑟𝑖𝑡 . [35]. - 47 -.

(54) H2-promoted Domain wall motion Under 0.4-0.8 bar H2, domain wall motion dominated the magnetization reversal, so we could measure the velocity of domain wall in same area. Fig. 45 showed the domain wall velocity values under 0.4 ~ 0.8 bar H2 gas pressures as a function of the reciprocal of the magnetic field. The domain wall velocity was about 10-6 ~ 10-5 m/s, indicating that the DW motion was in a thermally activated creep regime. [19-21] At a temperature of 0 K, the depinning of magnetization only occurred with a very large magnetic field that should be larger than critical depinning field 𝐻𝑐𝑟𝑖𝑡 . At a finite temperature, even the magnetic field smaller than 𝐻𝑐𝑟𝑖𝑡 , the depinning of magnetization still occurred because of, and then the domain wall motion was observed. If the external force was very small but still enough to drive the domain wall with thermal energy, the domain wall velocity would be low and referred to as “creep.” [19-21] The domain wall velocity 𝑣 was shown as a function of the applied magnetic field H and could be described as the following equation. [23, 24] 𝑈. 𝐻𝑐𝑟𝑖𝑡 1/4. 𝑣 = 𝑣0 ⋅ 𝑒𝑥𝑝 [− (𝑘 𝑐𝑇) ( 𝐵. 𝐻. ). (5). ]. Where 𝑣0 was numerical prefactor, and 𝑈𝑐 was the scaling energy constant as the height of pinning energy barrier about reversal domain wall motion; and 𝐻𝑐𝑟𝑖𝑡 was the critical field, dependent on the depinning field. Magnetic domain wall velocity measurements show that the average energy barrier scales as (1⁄𝐻 )𝜇 with μ = 0.24 ± 0.04, which was agreement with theories giving μ = 0.25. [31] In Fig. 45(b), ln(v) was shown as a function of 𝐻−1⁄4 for 0.4, 0.6, and 0.8 bar H2. The lines were fitted with the experiment data (solid notations) deduced by using Eq. (5) , and the slope of the lines was H𝑒𝑓𝑓 = (𝑈𝑐 ⁄𝑘𝐵 𝑇)4 𝐻𝑐𝑟𝑖𝑡 . The increasing in H𝑒𝑓𝑓 indicated that the energy barrier about reversal domain wall motion was reduced and depinning field was increased by the H2 gas pressure. So the coercivity was increased with hydrogen pressure during 0.4bar to 0.8bar depending on the depinning field, and the domain wall velocity was increased depending on the decreased of energy barrier about reversal domain wall motion.. - 48 -.

(55) 90˚-Rotation of Fe Using Hydrogen. 6. 90˚-Rotation of Fe Using Hydrogen 6.1. Growth and crystalline structure of Pd/Fe/MgO(001). In the above study, we explored the reaction about the magnetization reversal motion in CoPd alloy thin film after hydrogen absorption. But only one layer was hard to do something more useful like sensor or memory. So we want to change our field of research to the combination of magnetic interlayer exchange coupling and hydrogen absorption. Here we prepared Pd/Fe/Pd/Fe multilayers on MgO(001). Hydrogen molecules would be separated into atoms by the top Pd layer and diffused into the under layers when the sample was exposed to hydrogen pressure. The electronic structure of the mediate Pd layer would be change by exchanging electron with hydrogen atom, and the magnetic interlayer exchange coupling between the two Fe layers was expected to be modulated. But my job was researching the magnetic behavior of Pd/Fe/Pd/Fe multilayers and the switching process of two Fe layer. So here only discuss the magnetization switching behavior of Pd/Fe/Pd/Fe multilayers under different angle. The part of hydrogenation enhanced the magnetic coupling of two Fe layer wouldn’t be discussed too much.. Fig. 46. The schematic illustration of Pd/Fe/Pd/Fe multilayer film on Mg(001) substrate and the hydrogenation effect.[36]. - 49 -.

(56) 90˚-Rotation of Fe Using Hydrogen. Fig. 47. AFM topography images and line profiles of a pristine MgO(001) after 700℃ thermal annealing.. Fig. 47. showed the AFM topography images and line profiles of a pristine MgO(001) after 700℃ thermal annealing. The original MgO surface was covered by a lot of large structures that the size was 10 to 100 nm and the surface roughness was within 15 nm. After thermal annealing, the MgO(001) surface was became flat and the roughness was less than 1 nm. We prepared Pd/Fe/Pd/Fe multilayers deposited on flat MgO(001) structure by MBE system.. - 50 -.

(57) 90˚-Rotation of Fe Using Hydrogen. Fig. 48. (a) High-resolution TEM image near the Fe/MgO(001) interface region. (b) A magnified TEM image revealing the aligned crystalline structure of Fe(001) on MgO(001), as indicated by the red dashed line. [36]. As reported in previous study, the lattice mismatch in Pd/Fe/MgO(001) was sufficiently small for coherent epitaxial growth.[25] Following rotation through 45˚, the Fe(001) (lattice constant a𝐹𝑒 = 2.87 A° ) almost matched the MgO(001) (a𝑀𝑔𝑂 = 4.21 A° ), with evident mismatch of only approximately 4%.[25] Under this lattice matching condition, the bcc Fe <110> direction was aligned along the fcc MgO < 100 >. For fcc Pd(001) ( a𝑃𝑑 = 3.89 A° ) grown on bcc Fe(001), a small lattice mismatch of approximately 4% also ensured epitaxial growth. [25] Fig. 48 depicted a high-resolution TEM image near the Fe/MgO(001) interface. The magnified TEM image revealed the aligned crystalline structure of Fe(001) on MgO(001), as showed by the red dashed line. Even though the nanometer scale roughness of MgO, the atomic crystalline arrangement of Fe(001) was aligned along MgO(001).[36]. - 51 -.

(58) 90˚-Rotation of Fe Using Hydrogen. Fig. 49. (a) MOKE hysteresis loops of 2-nm Pd/3-nm Fe was measured at different azimuthal angle φ. (b) Polar illustration of the magnetic remanence (Mr)/magnetic saturation (Ms) ratio deduced from hysteresis loop. The magnetic remanence dependent on azimuthal angular showed the dominance of a uniaxial magnetic anisotropy. [36]. Before preparing Fe/Pd/Fe trilayer system, we explored the magnetic behavior of single-layer Fe on MgO(001). Fig. 49(a) showed MOKE hysteresis loops of 2-nm Pd/3nm Fe was measured at different azimuthal angle φ. At φ = 45˚ and 225˚, the shape of hysteresis loops was rectangular, and the remanence (Mr)/saturation (Ms) ratio was nearly 100%. At φ = 135˚, the hysteresis loops became tilted, and the Mr/Ms ratio decreased to nearly 10%, besides the magnetization reversal became slowly and gradually. In Fig. 49(b), the Mr/Ms ratio, which was deduced from hysteresis loop, was plotted as a function of the φ. The Fe(001) should be quadropole symmetry in the crystalline structure same as MgO(001), but the crystalline of Fe(001) on MgO(001) would have 45˚ rotation[25], and the bipolar symmetry was dominated by a uniaxial magnetic anisotropy energy, which might be due to the slight miscut of the substrate crystal orientation, so the easy axis was at 45˚ and 225˚.. - 52 -.

(59) 90˚-Rotation of Fe Using Hydrogen. 6.2. Magnetic coupling in Fe/Pd/Fe. Fig. 50. (a) Longitudinal and (b) transverse MOKE hysteresis loops of a 3-nm Pd/2-nm Fe/2-nm Pd/3nm Fe thin film on MgO(001) were measured with the different azimuthal angle φ from 0˚ to 180˚ by a step of 5˚. Split minor loops were observed in the region of φ = 135˚ ± 30˚. [36]. After we understood the uniaxial magnetic anisotropy energy in the Fe/MgO(001), we explored magnetic interlayer coupling in the Fe/Pd/Fe structure. Fig. 50. (a) Longitudinal and (b) transverse MOKE hysteresis loops of a 3-nm Pd/2-nm Fe/2-nm Pd/3-nm Fe thin film on MgO(001) were measured with the different azimuthal angle φ from 0˚ to 180˚ by a step of 5˚.. - 53 -.

(60) 90˚-Rotation of Fe Using Hydrogen In the longitudinal MOKE, hysteresis loops were always square from φ = 0˚ to φ = 90˚. When φ was about 105˚, a square loop appeared in the center and two minor loops appeared at two sides of the hysteresis loop. When φ approached 135˚, the center loop became flat until it was unobservable, and the loops at two sides seemed to move toward the center. At φ = 135˚, the longitudinal hysteresis loop evolved into a split double loop. In the transverse MOKE showed in Fig. 50(b), hysteresis loop was square at 0˚. The height of hysteresis loops was decreased and became almost unobservable when φ was increased from 0˚ to 45˚. When φ was increased from 45˚ to 90˚, the height of hysteresis loops increased once more, but it was turn left and right compared with φ at 0˚. When φ was going to 100˚, we observed strange hysteresis loop, which was included a square loop at center with two rotational symmetric loops. When φ was increased from 100˚ to 135˚, the two minor loops seemed to move toward the center loop. At φ = 135˚, the Kerr intensity of the hysteresis loop decreased and the loop was almost unobservable compared with others. Form the MOKE measurement, we could imagine that there was multi-symmetric magnetic behavior on Pd/Fe/Pd/Fe/MgO system. For example, in transverse MOKE, the hysteresis loop at 45˚ and 135˚ indicated the presence of a four-fold symmetry. However, in longitudinal MOKE, the magnetic behavior at 0˚ ~ 90˚ and 90˚ ~180˚ indicated the presence of a two-fold symmetry of magnetic anisotropy energy. Thus, a simple model about the magnetic easy axes was proposed to explain this complex magnetic behavior. In this double Fe layer system, the bottom-Fe layer was affected by the miscut of the substrate, so an uniaxial magnetic behavior was shown which was similar in Fig. 49. The top-Fe layer was far away from the substrate, so the feature of Fe(001) crystalline dominated and the easy axes became four-fold symmetry. Considering the two layer with different symmetry of easy axes, we could simulate the complex MOKE hysteresis loops shown in Fig. 51. and Fig. 52.. - 54 -.