鈷薄膜成長於鎢(111)與皺化鎢{112}上的晶體結構與磁性研究

全文

(1)國立台灣師範大學物理學研究所碩士論文指導教授：林文欽教授 (Prof. Wen-Chin Lin) 宋克嘉博士 (Dr. Ker-Jar Song). 鈷薄膜成長於鎢(111)與與皺化鎢{112}上上的晶體結構與磁性研究 Growth, Structure and Magnetism in Co/W(111) and Co/Faceted Pd/W{112}. 研究生：蔡承叡 (Cheng-Jui Tsai) 撰. 中華民國 98 年 8 月.

(2) “Dedicated to everybody about helping me in my master life.”.

(3) Growth, Structure and Magnetism in Co/W(111) and Co/Faceted Pd/W{112} Cheng-Jui Tsai. Abstract The crystalline structure and magnetism of Co/W(111) and Co/faceted Pd/W{112} surface are studied with low energy electron diffraction (LEED), Auger electron spectroscopy (AES), temperature programmed thermal desorption (TPD) and magneto-optical Kerr effect (MOKE). We find that Co/W(111) has revealed in-plane ferromagnetic ordered for room temperature growth mode and both of in-plane and perpendicular magnetism for low temperature growth mode. Co undergo well epitaxial on W(111) because of the small lattice mismatch in a specific orientaion. Co/faceted Pd/W{112} reveals two-step hysteresis loops and we infer that alloys are formed. There is ∼40 Oe reduction in the magnetic coercivity of Co/faceted Pd/W{112} when 1.8 langmuir hydrogen is dosed with 1.8 Langmuir..

(4) Acknowledgements. ACKNOWLEDGEMENTS. 說來也覺得不可思議，碩士生涯居然走到了尾聲。身埋在這兩年的日子裡，時常覺得度日如年，更看不到畢業的盡頭，但每當回頭去細數時便又會發現時光流逝的是如此迅速，又得和時間來一場拉距戰。這兩年就在這既慢且快的步調中一步步抵達終點，現在的我似乎能夠理解“不要急，但是要快”的意涵了。碩士生涯能夠如此準時畢業，首先要感謝我的指導教授林文欽老師。老師在我無處可去的時候收留了我，成為老師的第一個學生，雖然一直以來對實驗室的生活步調不太能接受，但經過兩年的實驗磨練，從一個連電錶都不太會用的門外漢，到現在可以獨立的設計實驗的想法，這都得感謝林文欽老師的指導。從老師身上體會到最多的是做事的效率，只是這點我一直都沒能做得很好。另外老師也提供了一個機會讓我到原分所做實驗，一般的學生是很難有這樣子的經驗的。再來要感謝我的另一個老闆宋克嘉博士，在宋老師底下做實驗讓我學到其他同學都無法學到的實驗技術，儀器壞了可以讓我自行設法修理，別的老師大概不可能放手到這種地步吧。在實驗上宋老師提供了許多寶貴的經驗和想法，而且由於老師的學識淵博，每回和老師的交談過程裡都學到不少知識，更甚的是，宋老師也會提供一些養生保健的概念以及做人處事的態度，讓我在學識、品德上都得到提升。另外要感謝輔大物理系的劉建楠教授，每當在我遇到困難想找人討論時，劉老師總是不厭其煩的和我討論。謝完師長，再來就是同學了。首先要感謝的人是昌文宗同學，昌文宗帶領我快速的融入師大.

(5) 環境當中，那段一起把實驗室從廢墟變成能夠赤腳走路的過程，一起討論物理的漂亮和噁心的地方，一起煩惱著未來想要做什麼，這樣子的經歷我會一直記住的。莊孟勳同學讓我見識到，這世界上還真的是無奇不有啊！雖然你常把儀器搞爆，但如果不是這樣我也沒有機會學到這麼多儀器上的知識。另外要感謝林宜樺同學，林宜樺同學從我在輔大的時候就在課業上幫助我許多，還真是情義相挺，真應該頒發一個“惠我良多”的匾額給妳(笑)。邱傑振同學和林彥穎同學，雖然我們相處的時間不長，但是大家還是都會互相支援，不知道以後還有沒有機會再一起合作。實驗上要感謝林亦成學弟的大力相助，學弟相當的任勞任怨，讓我在最後實驗的階段能夠順利的完成。最後要感謝亞靚以及台大生統室的同學，亞靚總是在聽我抱怨實驗上的不順遂，讓我的壓力有個可以傾訴的對象，而歡樂生統室的同學讓我也感染了他們的快樂氣氛，苦悶的實驗之餘也能有快樂的心情。最後最後還是要感謝我的父母，因為他們的支持才能讓我無後顧之憂的唸我所喜愛的物理。兩年似乎真的是太短了，還有許多想學的東西還來不及去碰就已經畢業了。未來的路還很遙遠，在這裡要再一次感謝那些在我碩士生涯中曾經幫助過我的人，沒有你們就沒有現在的我，即使未來我們相遇的機會變少了，我依然會記得你們。.

(6) Contents. 1 Introduction. 2. 1.1. Basic properties about W(111) substrate . . . . . . . .. 2. 1.2. Motivation . . . . . . . . . . . . . . . . . . . . . . . . .. 9. 2 Introduction to Basic Concepts of the UHV System. 11. 2.1. Why Should We Need UHV . . . . . . . . . . . . . . .. 11. 2.2. Gas Release from Solids . . . . . . . . . . . . . . . . .. 13. 2.2.1. Diffusion . . . . . . . . . . . . . . . . . . . . . .. 14. 2.2.2. Vaporization . . . . . . . . . . . . . . . . . . . .. 15. 2.2.3. Leaks . . . . . . . . . . . . . . . . . . . . . . .. 15. 2.2.4. Desorption . . . . . . . . . . . . . . . . . . . . .. 16. 2.2.5. Permeation . . . . . . . . . . . . . . . . . . . .. 21. 2.2.6. Backstreaming . . . . . . . . . . . . . . . . . .. 22. How Could We Get UHV . . . . . . . . . . . . . . . . .. 22. 2.3. 3 Introduction to Experimental Instruments 3.1. 25. Pumping Systems . . . . . . . . . . . . . . . . . . . . .. 26. 3.1.1. Mechanical Pump . . . . . . . . . . . . . . . . .. 26. 3.1.2. Turbo Molecular Pump . . . . . . . . . . . . . .. 27. 3.1.3. Titanium Sublimation Pump . . . . . . . . . . .. 28. 2.

(7) Contents. 1. 3.1.4 3.2. Ion Pump . . . . . . . . . . . . . . . . . . . . .. 28. Measurement Instruments . . . . . . . . . . . . . . . .. 30. 3.2.1. Auger Electron Spectroscopy . . . . . . . . . . .. 30. 3.2.2. Low Energy Electron Diffraction . . . . . . . . .. 31. 3.2.3. Magneto-Optical Kerr Effect . . . . . . . . . . .. 37. 4 Experimental Results and Discussion. 42. 4.1. Experimental Method . . . . . . . . . . . . . . . . . . .. 42. 4.2. Thickness Calibration of Co Films . . . . . . . . . . . .. 44. 4.3. Structure I : Co/W(111) . . . . . . . . . . . . . . . . .. 48. 4.3.1. Low Temperature Growth of Co on W(111) . .. 48. 4.3.2. Room Temperature Growth of Co on W(111) .. 51. 4.3.3. Discussion of in-Plane Crystalline Structure of. 4.4. 4.5. Co/W(111) . . . . . . . . . . . . . . . . . . . .. 54. Magnetism I : Co/W(111) . . . . . . . . . . . . . . . .. 56. 4.4.1. RT Growth of Co/W(111) . . . . . . . . . . . .. 56. 4.4.2. LT Growth of Co/W(111) . . . . . . . . . . . .. 57. Structure and Magnetism II : Co/faced Pd/W{112} . .. 59. 4.5.1. LT Growth of Co/faced Pd/W{112} . . . . . .. 59. 4.5.2. Hydrogen Effect on Co/faceted Pd/W{112} . .. 65. 5 Conclusion. 67. Bibliography. 68.

(8) Chapter 1 Introduction 1.1. Basic properties about W(111) substrate. Figure 1.1: LEED patterns of 1.1 PML Pd/W(111) at an incident energy of 101 eV, which had been annealed at 1000 K for 3 min. [10]. In this section, we will review some properties about W(111). The thermal equilibrium crystal shapes are predicted polyhedral [5] [14], i.e. when a body is obtained enough heat energy, it will be changed its morphology to a convex body. Dr. Song et al. firstly found a faceted 2.

(9) 1.1. Basic properties about W(111) substrate. 3. phenomenon in Pd/W(111) by experiment (figure 1.1) [10] The Pd-covered films are stable when the thickness is <1 PML. If the Pd thickness is above 1 PML, the system is situated in bilayer growth mode, these results are confirmed by local volume potentials calculation [6]. From the results of calculation, the bilayer oscillations are apparent in relaxed surface energies for the Pd thickness >1 PML. Upon annealing above 700 K, a 3 ML (corresponds to1 PML) Pd film on W(111) facets into a 2 ML Pd film on W{112} and it is concerned with the bilayer growth mode. If the Pd coverage is excessed 1 PML, it would be formed 3D clusters with the faceted process (figure 1.2). [11]. Figure 1.2: STM image of faceted Pd on W(111). We can see three {112} planes of facet and some 3D clusters of Pd on W(111) when Pd coverage excess 1 PML. [11]. From others studies indicate that clean W(111) surface is stable, i.e. there are no facets appearance after annealing. T. E. Madey et al. indcut by experiment that some elements like Pd, Rh, Ir, Pt and Au,.

(10) 1.1. Basic properties about W(111) substrate. 4. which can induce faceting on W(111) and the common point of them is Pauling electronegativity >2. If the annealing time is prolonged to 9 min, there are two different types of facets {112} and {110} existent simultaneously (figure 1.3). [4]. Figure 1.3: Top view of the STM image of Pd facets. {110} and {112} surfaces is existent simultaneously by prolonged annealing. [4]. From early theoretical study [5], the faceted condition can be attributed to surface energy difference ∆Esurf ace between [111] and x Theorem 1.1 (Faceted Condition) ∆Esurf ace. γ x = − γ[111] A < 0 cosθ. (1.1). where γx is the total surface energy density corresponding to the orientation x, A the total surface area and θ the angle between x and [111]..

(11) 1.1. Basic properties about W(111) substrate. 5. We can also define the surface energy density σx corresponding to the area of a unit cell Ax , which corresponds to the orientation x, and from the geometrical relationship about bcc lattice    1, h + k + l is even Ay δa (a · a) = , δ[hkl] =  Ax cosθ δb (a · b) 2, h + k + l is odd. (1.2). we can substitute (1.2) into (1.3) and get. 3 σ{112} − σ[111] Ncell 2 3σ[110] − σ[111] Ncell. ∆E(111)→[112] = ∆E(111)→{110} =. (1.3) (1.4). where Ncell is the number of surface sites on the topmost layer of the (111) surface. ∆Esurf ace is independent on the reference energy of the overlayer Pd because 1 PML Pd coverage has the same adatoms on (111) and faceted {112} surface. Next, we can express another representation — surface formation energy σA/B of the substrate B covered by deposition films A. σA/B = (σB + HA/B ). (1.5). where σB is the orientation dependent surface energy of the clean substrate B and HA/B is the heat of formation of adsorbing a certain number of adatoms A on B. The faceted phenomenons of above discussions are called the global faceting, which energy is efficient enough for the thermal equilibrium shape to be formed. If the surface energy density γ is highly.

(12) 1.1. Basic properties about W(111) substrate. 6. Figure 1.4: An example of hill-and-valley faceting : oxygen-covered tungsten surface. [3]. anisotropic but the surface lacks for efficient heat energy, then the crystal shape forms pyramid-like structures and they are called hilland-valley faceting. [3] Figure 1.4 is an example about hill-and-valley faceting, oxygencovered W(111) surface, the oxygen provides the high surface energy anisotropy to induce faceting. Hill-and-valley faceting is observed in the annealing temperature ranges 860–1300 K and 1640–1800, as shown in figure 1.5. [3] The number of steps of a facet decreases with increasing annealing temperature until the temperature reaches to 1400 K. In the ranges of 1400–1600 K, the hill-and-valley faceting is transformed to the global faceting and the vertex is truncated. In the higher temperature ranges (1640–1800 K), the global faceting is transformed to the hill-and-valley faceting again, and the number of steps with raise to increase of the annealing temperature. From the further study of the system of oxygen-covered W(111) [2], A. Szczepkowicz and R. Bryl control the faceted morphology by quench-.

(13) 1.1. Basic properties about W(111) substrate. 7. Figure 1.5: The FIM images of oxygen-induced hill-and-valley faceting on W(111). Number of steps decreases with the increase of temperature in the ranges of 800–1400 K. [3]. Figure 1.6: The FIM images of oxygen-induced global faceting on W(111). Truncated vertex area increases with the increase of temperature.) [2].

(14) 1.1. Basic properties about W(111) substrate. 8. ing the sample to study the thermal evolution of the vertex (figure 1.6). They also claim that this thermal evolution satisfies the prediction of the 3D Ising model of type-B (figure 1.6).. Figure 1.7: The 3D Ising model of two types of thermal evolution. T0 is the vertexrounding temperature and T1 the edge-rounding temperature. [2]. In the case of Co/W(111), pseudo-layer-by-layer growth of thick pseudomorphic Co films is observed at 380 K and transition to the (6 × 6) structure are revealed at 780 K [8]. Co/W(111) is found experimentally to be ferromagnetism in in-plane direction at 380 K when Co thickness exceeds 7.6 psedomorphic ML (∼2.3 PML) . About the pseudomorphic structure, we have a new point of view to explain it from our data, and we will reveal others properies about Co on W(111) and faceted Pd/W{112} in our study..

(15) 1.2. Motivation. 1.2. 9. Motivation. Special Structure and its Corresponding Magnetism From previous studies, different growth temperatures of Co on W(111) shall result in different crystalline structure. [8] Co/W(111) exhibits 1 × 1 structure for low temperature growth at 380 K. Under high temperature condition at 710 K the structure transformation to 6 × 6 are observed, see figure 1.8. So that we can control the condition of growth temperature to explore more new structure and corresponding magnetic properties.. Figure 1.8: LEED patterns of Co on W(111) at growth temperature (a) 380 K for 1 × 1 and (b) 710 K for 6 × 6 structure. [8]. Spin-Polarized Field Emission Tip Because of low surface energy and low work function of W(111) [9] [12], it is appropriate for field emission tip [13]. Faceting is a well known phenomenon, which Pd are deposited on W(111) surface and via ≥ 700 K annealing for 3 minutes. [10] Going through the faceted proceed, sample surface will formed lots of {112} pyramids in which.

(16) 1.2. Motivation. 10. the vertex is more sharpness and beam spot is more brightness than W(111). According to these points of view, if we can deposit Co on faceted system by layer-by-layer growth mode, Co/faceted Pd/W{112} might produce a faceted surface associate ferromagnetism to get spinpolarized field emission tips. Hydrogen Effect. Figure 1.9: The system of Fe/faceted Pd/W{112} (a) with hydrogen dose at 2×10−9 torr and corecivity increase with exposure time. (b) The pressure in Clean chamber is 7 × 10−11 torr, dirty chamber and CO chamber is 2 × 10−9 torr. [1]. Figure 1.9 shows that in our previous study of Fe/faceted Pd/W{112}, there is an interesting phenomenon that coercivity will be enhanced with H2 dosing [1]. Although the mechanism of this phenomenon is unknown, we can check that hydrogen effect are universal either in thin magnetic films on faceted Pd/W(111) or a special case of Fe for further progress via Co experiment..

(17) Chapter 2 Introduction to Basic Concepts of the UHV System 2.1. Why Should We Need UHV. From thermodynamics about ideal gas, the flux of the number of molecules per unit area is 1 Γ = nv 4. (2.1). where n is the density of molecular numbers, v the mean speed of molecules, and from the equation of state P = nkT r 8kT v= mπ. (2.2) (2.3). where P is the pressure, k the Boltzmann constant, T the temperature and m the molecular weight of ambient gas molecules. We can find that when the experimental temperature is fixed, then chamber pressures are proportional to the number of molecules in the chamber. 11.

(18) 2.1. Why Should We Need UHV. 12. Conventionally, the degree of vacuum has been divided into several ranges as follow : (i) Low vacuum : atmosphere–10 torr (ii) M edium vacuum : 10–10−3 torr (iii) Highvacuum : 10−3 –10−6 torr (iv) V eryhigh vacuum : 10−6 –10−9 torr (v) U ltrahigh vacuum : 10−9 –10−12 torr (vi) Extreme ultrahigh vacuum : < 10−12 torr. If we assume the probability of the adsorption of gases is almost surely, i.e. Pr [adsorp.] = 1, then the adsorption rate of gases Γadsorp. is. Γadsorp.. 1 P = nv = 4 4kT. r. 8kT P ∼ 3.52 × 1022 √ mπ mT. (2.4). The typical size of solid is about 1015 atoms/cm2 , so that if we want to get 1 hour for the clean sample at room temperature, the base pressure of the chamber must be less than 1.93 × 10−10 torr (m ∼2). We can get this result from a more simple idea, let us consider the definition of Langmuir. Definition 2.1 (Langmuir) 1 Langmuir is defined by chamber pressure multiply by the exposure times (sec), which is equal to 1 × 10−6 torr · sec It mean that how many numbers of adatoms are adsorbed on the surface in ML during 1 sec. So that we can get a similar result from.

(19) 2.2. Gas Release from Solids. 13. Figure 2.1: Gas sources in the vacuum system.. P =. 1 × 10−6 = 2.78 × 10−10 torr 3600. (2.5). Note: For some surface (ex: Au or oxide), because those of low activity, the probability of absorption is not almost everywhere (Γadsorp. < 1 4 nv),. 2.2. the clean time will be prolonged.. Gas Release from Solids. Firstly, we consider an issue that when the system exists flowing out, it must be associated with intake. There are various gas sources in the chamber and they are illustrated with the figure 2.1..

(20) 2.2. Gas Release from Solids. 2.2.1. 14. Diffusion. Diffusion is a phenomenon about transport of one material through another. In the vacuum system, gases could diffuse from the interior wall of the chamber to vacuum. The diffusion rate of transport through the bulk to the surface usually dominant the whole diffusion process. Consider the gas dissolves in a solid wall, which has a uniform initial concentration C0 . The outgassing rate is. q = C0. D πt. | 4DC0 = d |. 12 ". ∞ X. 2 2 # nd 1+2 exp − 4Dt n=1 {z }. origin at the center of the solid ∞ 2 2 X. (2n + 1) π Dt exp − d2 n=0 {z }. (2.1). origin at one surf ace. and the diffusion constant D (m2 /sec) is. D = D0 exp(−. Eact. ) kT. (2.2). where Eact. is the thermal activation energy of the diffusion gas in the solid, d the thickness of the material (m). C0 is also represented the internal pressure of the gas dissolved in the solid (P ascal). In the limit as we approach t = 0, the rate of gas release from the surface is q = C0. D πt. 21. 1. ∝ t− 2. (2.3).

(21) 2.2. Gas Release from Solids. 15. When t → ∞, then 2 4DC0 π Dt q= exp − 2 ∝ exp(−aDt) d d. (2.4). From (2.3) and (2.4), we can find that the diffusion rate decreases slowly at first, and becomes to decay rapidly when t ≥ d2 /24D. 2.2.2. Vaporization. A vapor is a gas above its condensation temperature. Vaporization is a thermal stimulation process that molecules occur phase transition from solid (or liquid) phase into vapor phase. The pressure of the vapor over the surface in equilibrium is the vapor pressure of solid (liquid) and it is supported that temperatures of the solid (liquid) and the vapor are the same. In equilibrium, Γrelease , the rate which of molecules release from the surface is expressed with (2.1) and (2.4). 2.2.3. Leaks. Leaks are the holes on the chamber or the splits which are located on between the vacuum components and the chamber. In general, the chamber pressure is lower than atmosphere, so that gases in air could be leaked in the chamber through the opening. The leaking rate is usually a constant. We could realize that the vacuum system is leak or not from the mass spectrometer or the Residual Gas Analyzer (RGA) data..

(22) 2.2. Gas Release from Solids. 16. There are three sections about the ion sources, a mass analyzer, and a detector in RGA. Firstly, the gases are ionized by e-beam from a hot filament, subsequently, these ions are classified according to mass by the mass analyzer. Finally, the detector counts all kinds of the number of mass and reveals the spectrum. By the way, the pressure meter — ion gage has a similar principle of operation. The difference between RGA and ion gauge is that ion gauge has no mass analyzer. The unbaked chamber has a great deal of 18 signal (H2 O). If the chamber is baked, but the pressure is still bad (> 10−9 torr) and there are existing 18, 28(N2 or CO), and 48(CO2 ) peaks, then the chamber must be took leak test. We chose the helium gas as the test source, and previously, controlled its flow rate with 2 bubbles within 1 second by immersed the leak duct in water. For leak test, we wrapped two garbage bags around the chamber and sprayed the helium gas into the bags. If there was existing the helium signal in one bag, then we could reduce the investigation of range by this dichotomy until to found the leak position. Note: The maximum signal is H2 peak in our chamber at UHV. 2.2.4. Desorption. Desorption is an excited process, which when molecules previously adsorb on the interior surface of the chamber at first, and they are stimulated and released from the surface to vacuum by the external energy source. Adsorbate molecules might originate from the gas out of the chamber via permeation to inner surface. So that we have to realize.

(23) 2.2. Gas Release from Solids. 17. two kinds of adsorption process — “physisorption” and “chemisorption” before to discuss the desorption.. Figure 2.2: Potential energy diagram of physisorption and chemisorption. x is the distance between adsorbate molecules and surface, Ep the heat of physisorption, Ec the heat of chemisorption, Ea the activation energy of chemisorption and Ed = Ea + Ec the desorption energy.. Physisorption originates from the van der Waals interaction of energy < 40 M J/(kg-mole). The chemical properties of adsorbate molecules and the surface have no change. For instance, it is analogous to the gas condensates on the surface, i.e. physisorbed molecules are removed quickly with ambient temperature and don’t hinder pumping. Chemisorption has a strong interaction but a sorter interaction distance between the surface and adsorbate molecules than physisorption. In chemisorption, eletrons can be exchanged between chemisorbed molecules and surface atoms, and chemisorbed molecules can be dissociated to atoms and bonded on the surface. Figure 2.2 shows the potential energy diagram of the physisorption and chemisorption. Now consider a thermal desorption process without readsorption,.

(24) 2.2. Gas Release from Solids. 18. we can define the 0th order desorption rate as shown below : Definition 2.2 (0th Order Desorption Rate) 4. F (t) =. dσ = ν0 eEv /kT = ν0 e−Ed /NA kT dt. (2.5). where σ is the molecular surface density, ν0 the frequency factor (constant) of the zero order desorption, Ev the latent heat of vaporization and NA the Avogadro’s constant. It is described the desorption process of multilayers of molecules. These molecules depart from a surface saturated with large quantities of vapor, just like a glass of water come to boil. When the coverage is about one monolayer, it can be described by the 1st order desorption process : Definition 2.3 (1st Order Desorption Rate) dC(t) e−Ed /NA kT F (t) = = C(t) dt τ1 = ν1 e−Ed /RT C(t) = −K1 C(t) 4. (2.6) (2.7). where C(t) is the molecules surface concentration, τ1 the vibrational frequency of a molecule in an adsorption site, and ν1 the frequency factor of the 1st order desorption. K1 can be expressed with the average residence time τr τr =. 1 K1. (2.8). We can get a more realizable representation by integerating (2.3).

(25) 2.2. Gas Release from Solids. dC(t) = C0 K1 e−K1 T = C0 K1 e−t/τr dt. 19. (2.9). This equation implies that the desorption rate decreases rapidly with a few time τr . For instance, the Ed in the system of H2 O/metal is ∼96M J/(kg-mole), therefore, τr is ∼ 105 , it implies that water is the most difficult gas to remove without baking. If the gas dissociates on adsorption and recombines on the surface before desorption, this process can be defined the 2nd order desorption rate : Definition 2.4 (2nd Order Desorption Rate) dC(t) −K2 C02 2 F (t) = = −K2 C (t) = dt (1 + C0 K2 t)2 4. (2.10). where K2 is contains exp(−Ed /kT ). Equation (2.10) shows that the total desorption time is longer than 1st order case by 1/t2 . The thermal desorption process can be figured out that we input the heat energy to the system which has molecules-molecules and molecules-surface interface. Because the desorption energy Ed of these interfaces are not equal, we can expect that there are two kinds of desorption rates in the system and them are corresponded to the desorption temperatures. So that we can use this idea to determine our sample thickness, the method is called T emperature Programmed thermal Desorption (TPD)..

(26) 2.2. Gas Release from Solids. 20. Figure 2.3: The e-beam bombardment process.. In our experiments, we could heat the sample directly by a constant heating rate and detected the signals of mass of the sample by the time-evolution RGA signals. If the Co thickness was about 1 PML on W(111), the thermal desorption process should be the zero order desorption and it was corresonded to one peak for the time-evolution RGA data. If the Co thickness was >1 PML, it included the zero and the first order desorption rates in the desorption process and could be existed two peak in RGA signals. More details are in the section 4.2. For others stimulated desorption processes, we have more interesting to the electron-stimulated desorption phenomenon. First, the accelerative electron beams bombard on the surface, and the electrons of the adsorbed molecules are excited. Second, the excited electrons produce a repulsive potential between the surface and the molecules and the molecules desorb as a neutral or an ion. This principle has some useful applications, we can use this idea to clear the chamber (figure 2.3) or.

(27) 2.2. Gas Release from Solids. 21. Figure 2.4: The e-beam evaporator.. the substrate. We pass through the current in the filament and float it to 1000 V comparing with the chamber, then the molecules on the interior wall of the chamber could be kicked out and pumped. A similar application is the evaporation gun (figure 2.4), we just substitute the deposition source for the chamber wall and the ground is exchanged to the shielding. 2.2.5. Permeation. Permeation is a three-step process about gases (1) adsorb on the exterior surface of the chamber (2) diffuse from the outer surface to the inner surface of the chamber (3) desorb from the inner surface of.

(28) 2.3. How Could We Get UHV. 22. the chamber to vacuum. 2.2.6. Backstreaming. Backstreaming is defined as the transport of fractions of the pumping fluid from the pump to the chamber. There are some contributions from the pump like the evaporation condensed on the walls of the pump, or the oil vapor of the mechanical pump back flow in the chamber. Nesseldreher indicates that backstreaming of heavy oil fragments through turbo molecular pumps when their rotational velocity decreased to 40% at rough vacuum pumping [15]. Fortunately, turbo has a precaution system when the angular frequency is less than its maximum by 80%. Although TSP or ion pumps do not backstream, there are some weak points that they can produce H2 , carbon oxides or others contamination.. 2.3. How Could We Get UHV. The gas in the chamber can be expelled by various chemical pumps and physical pumps. In our laboratory, we have two kinds of chemical pumps which are T itanium S ublimation Pump (TSP) as well as ion pump, and physical pumps are mechanical pump and turbo molecular pump. All of the pumps have those own operation pressure region, so that we have to turn on these pumps orderly. The following are the steps of returning from air to UHV :.

(29) 2.3. How Could We Get UHV. 23. Step 1 Checked all of the ports had been screwed and all of the filaments didn’t turn on. Step 2 Turned on the mechanical pumps, and waited about 20–30 minutes. Step 3 Turned on the turbo molecular pumps and checked that they had reached to full speed (Ex : 1500 Hz for Pfeiffer D35614 ASSLAR). Note: If turo molecular pumps can not arrive at full speed, it might imply that the chamber has leaks. Step 4 At this step, we could turn on the ion gauge and checked the pressure that was dropping off steadily. The pressure should be ∼ 10−6 torr by now. Step 5 Baked the chamber. Note: We had to avoid crossed touching the heating types and didn’t wind the heating types around the rotation stage. Aluminum foil should cover the chamber uniformly. When we turned on the autotransformer before, checked the power load of the laboratory. Step 6 Checked for every heating type and autotransformer below their maximum load. If the temperature of the chamber is stable, then waited about 1 day. Note: The baking temperatures of the chamber and the rotation stage should be lower than 200 K and 100 K, respectively..

(30) 2.3. How Could We Get UHV. 24. Step 7 Turned on the filament of e-beam evaporators without high voltage for outgass, and channeled the RGA and the ion gauge to degas mode. Then used the e-beam bombardment for several hours. Now the pressure should be < 10−7 torr, outgassed the TSP. Step 8 When the pressure is 10−8 –10−9 torr, turned on the filament of LEED and AES until the pressure was raised to maximum and back to the initial value. Note: We should control the filament current to avoid the pressure that over the usage of region of the instruments. Step 9 Turned off the e-beam bombardment and others filaments, then turned on the ion pump on and off for sputtering the residual molecules at the cell until the pressure didn’t raise, then turned on it continuously. Note: Don’t turn on the e-beam bombardment and ion pump simultaneously. Step 10 Stopped baking and waited the chamber temperature to back to the RT. Degassed the e-beam evaporator sources by turned on the filament current and HV. Step 11 Cooled down the substrate and the cold trap, then cleared the substrate and ran the TSP, the UHV system would be got..

(31) Chapter 3 Introduction to Experimental Instruments. Figure 3.1: Multi-functional MOKE chamber in Laboratory 327, Institute of Atomic and Molecular Sciences, Academia Snica. The best pressure is < 6 × 10−11 torr.. 25.

(32) 3.1. Pumping Systems. 3.1 3.1.1. 26. Pumping Systems Mechanical Pump. Figure 3.2: A schematic picture of sectional view of the mechanical pump.. Working Ranges : atmosphere–10−3 torr A typical mechanical pump is the rotary vane pump, figure 3.2 shows its working mechanism. There are a couple of vanes that be mounted to the center of rotor and be separated by a spring. The rotor is located in a circular cavity and their center are offset, so that it can rotate in the cavity and vanes are still contact with cavity wall by compressing the spring. When the mechanical pump starts to work, the gas from the chamber enters the inlet port and increases in volume by the inlet pressure. The gas in the exhaust port is compressed by.

(33) 3.1. Pumping Systems. 27. the rotor and vane, and subsequently expels it to the air through the discharge valve. 3.1.2. Turbo Molecular Pump. Working Ranges : 10−2 –10−10 torr. Figure 3.3: Sectional diagram of the turbo molecular pump.. Turbo molecular pump has a molecular turbine like air turbines of an airplane. This molecular turbine has a series of rotor-stator disks and can reach to the rotor speed at 1500 Hz. The pumping process is that molecules are compressed by momentum transfer from blades to molecules at high speed. Although one stage of rotor-stator disk has low compression ratio, turbo molecular pump has a high compression ratio by cascade several stages which products all ratios of total stages. Turbo often couples with mechanical pump because the exhaust pressure remain in the molecular or transition region, and.

(34) 3.1. Pumping Systems. 28. it can’t be repelled to atmosphere. Therefore, a backing pump is needed. 3.1.3. Titanium Sublimation Pump. Figure 3.4: A schematic picture of sectional view of TSP.. Working Ranges : 10−9 –10−10 torr Titanium is a surface getter for active when it deposits on the surface in some thin film layers. If we directly heat the Ti filament by 50 A, Ti will sublimate and deposit on the wall of the cold trap. Hydrogen has a highly sticking coefficients at 80 K, i.e. TSP is a nice pump for catching the hydrogen. 3.1.4. Ion Pump. Working Ranges : 10−4 –10−10 torr.

(35) 3.1. Pumping Systems. 29. Ion pump has two advantages better than TSP that ions are more reactive with surface than neutral molecules and it can be embedded in the pump walls by a sufficient energy. This is the reason that why do we couple TSP and ion pump together in our chamber.. Figure 3.5: The pumping mechanism of the ion pump.. The pumping mechanism of the ion pump has five steps. The first step, in order to guide electrons within anodes, it creates a ∼ 1200 Gauss magnetic field by permanent magnets. The second step, it applies a ∼ 5600 V high voltage to generate a cloud of electrons and are trapped in anodes by magnetic field. The third step, when the molecules drift into anodes, they are bombarded by the cloud of electrons and are ionized to positive ion. The fourth step, ionized positive ions are accelerated by high voltage to sputter the cathode and be neutralized by surface charge transfer. The final step, cathode materials are ejected to the anode, then they become a surface getter.

(36) 3.2. Measurement Instruments. 30. like TSP.. 3.2. Measurement Instruments. 3.2.1. Auger Electron Spectroscopy. AES is a tool that it is used to investigate the surface composition. In our study, we use it to check the surface contamination and the thickness calibration. The latter is discussed in section 4.2.. Figure 3.6: KL1 L2,3 Aguer transition. (a) Ionization. (b) L1 −→ K transition. (c) Auger electron emission.. Consider the sample surface is bombarded by an electron beam, if the beam energy is higher than binding energy of an electron, which is confined in a core level (K shell ) of the sample. This electron is kicked out of atom and the whole atom will situate in excited state. The ionized atom is unstable and next an electron which comes from a higher energy level (L1 shell) that it occupies the hole in the K shell..

(37) 3.2. Measurement Instruments. 31. The energy loss of this electron from level L1 to level K is described by the energy level EL1 − EK or the binding energy EBK − EBL1 , where Ei and EBi are energy level and binding energy corresponding to the ith state, respectively. This energy loss is transformed to another electron, which stays in a higher level L2 (or L3 , they are co-expressed with L2,3 ) than L2 and can be ejected to the vacuum with kinetic energy AES . Ekinetic. AES Ekinetic = EL1 − EK − EBL2,3 − Φ. = EBK − EBL1 − EBL2,3 − Φ. (3.1) (3.2). where Φ is the work function of surface. This is called the KL1 L2,3 Auger transition. Since all of elements have different binding energies in KL1 L2,3 process, we can determine the surface composition by AES. 3.2.2. Low Energy Electron Diffraction. Owing to atoms can rearrange on the surface of bulk due to the free boundary condition, the crystalline structure of the surface and it corresponding properties are different from it bulk phase. Dynamics of surface atoms is illustrated by figure 3.7. LEED is a useful instrument to explore the surface crystalline structure, it construction and operation are shown in figure 3.8. Electrons are accelerated and collimated to an elecron beam by the electronic part, the beam energy is in the region of 0–1000 eV and the sample is grounded to avoid the net charged effect. Next this electron beam.

(38) 3.2. Measurement Instruments. 32. Figure 3.7: Fundamental atomic processes. (a) Detachment from an island. (b) Attachment at an island. (c) Interdiffusion. (d) diffusion on a terrace. (e) Nucleation. (f) Attachment at a step. (g) Deposition or adsorption on a terrace. (h) Deposition or adsorption on an island. (i) Desorption.. is focused on the sample surface and be scattered backwardly from the sample to the screen through the optical part. Optical part is a typical retarding field analyzer (RFA), it is constructed by four hemispheric and concentric meshes. The first mesh is grounded to ensure that electrons are traveling in a field-free zone when they entry into the repeller mesh. The second and third mesh are repeller meshes that are applied a negative potential (-V ) to filter electrons with energy less than eV. The end mesh is grounded to avoid the efficient capacitance effect between the repeller and the screen. The screen is biased at a high positive potential about 4–5 kV for development. The diffraction process is suffered from a difficult interaction between electrons and lattices (many-body interaction for charged particles). We usually use the elastic scattering theory to model the whole diffraction process. Nevertheless, the elastic theory is “fault” from it hypothesis that it assumes the interaction between electrons and.

(39) 3.2. Measurement Instruments. 33. Figure 3.8: A schematic picture of LEED.. lattices is weakly ! Only (0,0) beam (normal incidence) is satisfied with kinematic theory (Bragg diffraction) and the spot position is not changed with the incident beam energy. This is the reason that we can’t exactly figure out the crystalline structure of the surface from LEED patterns. In experimental discussions, we only roughly assume that LEED patterns are corresponded with lattices. Next the discussion will use the primary solid state theory. From solid state theory, the lattice is described by 3D Bravais lattice and lattice points are deonted by Rn =. 3 X. ni ai. (3.3). i=1. where ai is the basis lattice vector in real space and ni ∈ Z. However, in order to describe the crystalline structure of surface, the three dimension lattice must be reduced to two dimension with respective to.

(40) 3.2. Measurement Instruments. 34. some certain planes like (111) (the direction of a1 + a2 + a3 ). All of 3D Bravais lattice points are projected to these planes and form a new 2D Bravais lattice. We can use a new basis lattice vector to describe the point position as R=. 2 X. mi ai. (3.4). i=1. We usually call this 2D Bravais lattice is in-plane lattice, the normal direction of in-plane is out of plane. Next, we can connect 2D basis vectors with 3D by a linear transformation. a1 = M11 a1 + M12 a2.  ,. . a2 = M21 a1 + M22 a2. a1 a2. . . ˆ =M. a1 a2.  . (3.5). where (3.5) is called the nomenclature of the Park-Madden matrix notation. The 2D reciprocal lattice vector is defined by. ghk = hb1 + kb2 ,. h, k ∈ Z. (3.6). where bj are reciprocal basis vectors, these reciprocal basis vectors have the orthogonality relation with lattice basis vectors ai as follow : b1 = 2π

(41)

(42). a2 × n n × a2

(43) , b2 = 2π

(44)

(45)

(46) a1 × a2

(47) , a1 × a2

(48). ai · bj = 2πδij ,. i, j = 1, 2. (3.7). (3.8). where n is a unit vector in the direction of out of plane. From (3.7),

(49)

(50) we can obtain the length of basis vector

(51) bi

(52).

(53) 3.2. Measurement Instruments.

(54)

(55)

(56) bi

(57) =

(58)

(59) 2π

(60) ai

(61) sin(a1 , a2 ). 35. (3.9). It implies that the length of the reciprocal vector is inversely proportional to the real space vector with respective to itself. Figure 3.9 indicates the crystalline structure of body-centred cubic (bcc) lattice.. Figure 3.9: The unit cells of bcc (111) surface in 2D and 3D lattice.. Kinematic Theory : From solid state theory, the scattering wave amplitude at a field point r is given by eik·r ψ =

(62)

(63)

(64)

(65) f (ko , k)ei(k−k0 )·rj r. (3.10). where ko and k are wave vectors of the incident wave and the scattered wave, respectively. f (ko , k) is the atomic scattering factor and rj the.

(66) 3.2. Measurement Instruments. 36. position of an atom. If there are more than one atom in a lattice point, then rj = rlattice + rc , the total amplitude is. Figure 3.10: Illustration of kinematic sacttering of electrons.. Ψtotal. eik·r X i(k−k0 )·rc X i(k−k0 )·rlattice fc e =

(67)

(68)

(69)

(70) e r c |{z} | {z } lattice {z } | A. SG. (3.11). G. where G is the lattice factor and SG the structure factor, which is a function of k and k0 . Finally, the total diffraction intensity is given by.

(71)

(72)

(73)

(74)

(75)

(76)

(77)

(78)

(79) Ψtotal

(80) 2 =

(81) A

(82) 2 ×

(83) SG

(84) 2 ×

(85) G

(86) 2. (3.12). Figure 3.10 illustrates the process of kinematic sacttering of elec

(87)

(88)

(89)

(90) trons. If we consider the elastic scattering case (

(91) k0

(92) =

(93) k

(94) ), SG is a furnction of ∆k = k − k0 only.

(95)

(96) 2 Note that J =

(97) G

(98) is called the interference function and it is implied the 2D Laue condition. Let rlattice = n1 a1 + n2 a2 with ni = 0–Ni , i = 1, 2, where Ni are the number of unit cells within the.

(99) 3.2. Measurement Instruments. 37. electron beam coherence width in directions of a1 and a2 , respectively. From (3.11), we can obtain. J=. 1) sin2 N1 (∆k·a 2 1) sin2 (∆k·a 2. ×. 2) sin2 N2 (∆k·a 2 2) sin2 (∆k·a 2. (3.13). J is a maximum when. ∆k · ai = 2αi π, 3.2.3. i = 1, 2 and αi ∈ Z. (3.14). Magneto-Optical Kerr Effect. Magnetic Anisotropy Ferromagnetic materials present spontaneous magnetization below the Curie temperature Tc . Their magnetization have some preferring directions in space, which are called the easy axes. The direction of easy axis is observed to be dependent on the intrinsic symmetry of crystalline structure and the sample morphology. The magnetic anisotropy energy is introduced as the energy difference between two different magnetization axes as in-plane and perpendicular and can be expressed as follows :. M ag Eanisotropy ≈ Kef f V sin2 θ. Kef f. Ksurf ace + Kinterf ace = Kvolume + − 2πMs2 d. (3.15). where Kvolume is a thickness independence volume term, Ksurf ace and Kinterf ace are surface and interface terms, respectively, −2πMs2 is the.

(100) 3.2. Measurement Instruments. 38. volume shape anisotropy and Ms is the saturated moment per unit volume. Magnetic Hysteresis loop Hysteresis loop is one of the most distinctive experimental facts of ferromagnetism. It is obtained by applying to the sample a cyclic magnetic field H and by recording the ensuing change of the magnetization M along the field, where M is defined as the average magnetic moment per unit volume. Hysteresis loops may be of many different shapes, thus it’s important to get some parameters to characterize the loop properties. Two quantities of particular importance are the remanent magnetization Mr and the coercivity Hc . Remanent magnetization (or remanence) Mr represents the magnetization obtained by applying a magnetic field and then removing it, and Hc is the field needed to bring the remanence to zero. The variety of the hysteresis loop shape is the direct consequence of the possible magnetic domain structure. The mechanisms of the hysteresis loop are magnetization rotation, which needs strong applied field to overcome the energy barrier. Magneto-Optical Kerr Effect 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 reflection beam will become elliptical polarized. This phenomenon is so called M agneto-Optical K err E ffect (MOKE). The angle between the primary axis of the elliptical.

(101) 3.2. Measurement Instruments. 39. polarization and the linear polarization is called Kerr rotation, and the ellipticity of the elliptical polarization is called Kerr ellipticity, as shown in figure 3.11.. Figure 3.11: Schematic illustration of magneto-optical Kerr effect. After reflected from the ferromagnetic sample, the linear polarized laser beam becomes elliptical polarized. This setup is called DC MOKE.. Let 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 follows :.

(102) 3.2. Measurement Instruments. 40. Theorem 3.1 (Kerr Rotation) φKerr = −. θ+ − θ− 2. (3.16). Theorem 3.2 (Kerr Ellipticity) 4. Kerr =. r+ − r− b = a r+ + r−. (3.17). where a and b are semi-major and semi-minor axes, respectively. Both of them are proven to be proportional to the magnetization of sample. Thus by measuring φKerr and Kerr 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 ( the direction of magnetic field) and the light path, as shown in figure 3.12.. Figure 3.12: Different geometry for MOKE measurement.. In the polar Kerr effect (or perpendicular), the magnetization lies in the plane of incidence and is perpendicular to the surface. In the longitudinal Kerr effect (or in-plane), the magnetization lies in the plane of incidence and is parallel to the surface. In the transverse.

(103) 3.2. Measurement Instruments. 41. geometry, the magnetization is perpendicular to the plane of incidence and on the surface. In magnetic ultrathin films, the Kerr signal (DC MOKE) is so small that the noise may result in significant effect. Therefore, in our experiment, 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 figure 3.11. Due to the difference with the DC MOKE shown in figure 3.13, this method is called AC MOKE.. Figure 3.13: Schematic illustration of AC MOKE..

(104) Chapter 4 Experimental Results and Discussion 4.1. Experimental Method. Figure 4.1: A experimental flow chart. Two path for three different sample.. Figure 3.1 shows the experimental flow about our study work. At first step, in order to get a clear suface of W(111) we used directly 42.

(105) 4.1. Experimental Method. 43. heating by passing 50A current though it for 8 second. After flashing was proceeded, the sample was double checked by Auger electron spectroscopy (AES) low energy election diffraction (LEED) until oxygen and carbon could not be observed. In fact, some paper points out that carbon still exists in deeper region of bulk by several annealing process in W(111) substrate. So that W(111) sample was prolonged heating at 1000 K in an oxygen pressure of ∼10−9 torr when the chamber was just recovered UHV from air. Second, we deposited Co on W(111) or faceted Pd/W{112}) as we need. In the case of Co/W(111), we select two different growth temperatures of 300 K (RT) without annealing and 80 K (LT) with subsequent 300 K annealing for 5 min. After sample was prepared, we characterized Co thickness and crystalline structure by AES and LEED, respectively. Finally, we took MOKE data for magnetism analysis. In the Co/faceted Pd/W{112} case the different things we did from LT case of Co/W(111) is Pd deposition before Co and via 1000 K anneal with 10 minutes to get {112} three-sided pyramids in faceted Pd surface, then we took LEED patterns to check if the three faceted diffraction spots was formed..

(106) 4.2. Thickness Calibration of Co Films. 4.2. 44. Thickness Calibration of Co Films. Figure 4.2: Co/W AES ratio v.s. deposition time. This plot is our first strategy about relationship between Co thickness and deposition time.. In our experiment, the thickness calibration of Co was done by temperature programmed thermal desorption (TPD). Co is deposited by M olecular Beam E pitaxy (MBE) method with EFM3 evaporator, which provide stable deposition rate so that we can infer Co thickness from desorption time after determined the TPD diagram (or TDS, thermal desorption spectrum). But we don’t know the precise Co thickness and deposition rate from TPD at first. We can control just only the deposition time and to find the relationship between Co/W AES ratio and corresponding deposition time. We can see from figure 4.3 that AES ratio of Co/W is.

(107) 4.2. Thickness Calibration of Co Films. 45. Figure 4.3: AES peak-to-peak signal. The signal is dealed with differentiation.. 4. AES RCo/W =. Co peak-to-peak value W peak-to-peak value. (4.1). When Co thickness is increasing with deposition time, the corresponding peak-to-peak value are enlarge but W signal are reduced. So AES if RCo/W is large, it is mean that Co thickness is thick. figure 4.2 shows. the AES ratio versus deposition time diagram. We take TPD data after getting AES ratio subsequently. Now we have the second plot about thickness calibration, the TPD of Co as shown in figure 4.4. Figure 4.4 shows that the black curve defines the 1 PML and the area under TPD curve is meant the quantity of Co element. So only one thing we did is just integrating this curve to determine others precise thickness by the area ratio with respect to 1 PML. Then we.

(108) 4.2. Thickness Calibration of Co Films. 46. Figure 4.4: TPD of Co/W(111). The black curve is corresponding to 1 PML, others are decided by the ratio of area under curve with respect to the 1 PML..

(109) 4.2. Thickness Calibration of Co Films. 47. can plot the figure 4.5 and get Co thickness from AES ratio soon after deposition in the future experiment without sample desorption. Further more, we can find deposition rates are stable in generally from figure 4.6. The deposition rate curve is satisfied with. θ = 0.4t. (4.2). approximately, where θ is Co thickness and t is deposition time. It is implied we can control Co thickness roughly by choosing deposition time to get any PML as we need and double check from AES ratio diagram.. Figure 4.5: AES ratio v.s. Co thickness. We can decide the thickness from this diagram after scanning AES..

(110) 4.3. Structure I : Co/W(111). 48. Figure 4.6: Co thickness v.s. deposition time. The red line is fitting curve. In general, we can find the deposition rates are stable with 0.4 PML/min.. 4.3 4.3.1. Structure I : Co/W(111) Low Temperature Growth of Co on W(111). In this section, we introduce the structure of low temperature and room temperature growth of Co on W(111). Figure 4.7 is LEED images of Co/W(111) of LT growth mode. We selected beam energies at 120 and 130 eV to observe the crystalline structure variation with Co coverage layers. The first row of figure 4.7 is clean W(111) substrate for contrast. We can check all LEED images in figure 4.7 to make sure that in-plane lattice constant of Co/W(111) remain invariant comparison with clean W(111) surface, it implies that Co thin films is pseudomorphic (ps) growth on W(111). This result is consistent with.

(111) 4.3. Structure I : Co/W(111). 49. previous study about growth temperature at 380 K. Moreover, even if Co thickness is 6.7 PML, the 1 × 1 crystalline structure has not be changed.. Figure 4.7: LEED patterns of LT growth of Co on W(111) for various Co thickness from 2.8 PML to 6.7 PML with beam energy at 120 and 130 eV. a and b spot intensity change with Co thickness.

(112) 4.3. Structure I : Co/W(111). 50. Consider the LEED spots for first 6-fold symmetry inner the picture, we define this cycle to be cycle A and the cycle B is outside the cycle A. At cycle A and B we choose 2 corresponding spots a and b to analysis the perpendicular lattice constant. Let us focus on Co coverages of 2.8 and 3.8 PML at 120 eV beam energy, a/b LEED spot intensity ratio are different from each other. At 2.8 PML, the LEED spot intensity of a is larger than b, but at 3.8 PML, it’s seems like perpendicular lattice constant various with Co thickness because of intensity of spot b is larger than a. Figure 4.8 show that at 2.8 PML Co structure is different from others thickness and the ratio decay with thickness increase.. Figure 4.8: a/b LEED intensity ratio with different beam energy. 2.8 PML has different structure comparison with others thickness. The curves tend to converge to 0..

(113) 4.3. Structure I : Co/W(111). 51. Note that figure 4.8 is “NOT” I-V LEED diagram ! Since the chosen beam energys are discrete and the spots are not (00) beam, so we can say the perpendicular crystalline structure has changed, but we can’t explain completely what the perpendicular crystalline structure has changed. For this reason, we have to do more experiments to figure out the details in the future. 4.3.2. Room Temperature Growth of Co on W(111). Figure 4.9: LEED images of RT growth of Co/W(111). In-plane crystalline structure has no change comparison with LT growth mode..

(114) 4.3. Structure I : Co/W(111). 52. From identical analysis for RT growth of Co/W(111), in-plane lattice constant has no change because the positions of LEED spots almost the same as clean W(111), it imply that Co on W(111) is also pseudomorphic growth in RT growth mode. This result would not surprise us because it’s the same as 80 K and 380 K. Comparison with LT and RT case, we choose two nearest Co thickness about 3.7 and 3.8 PML for RT and LT growth mode, respectively, see figure 4.10.. Figure 4.10: LEED images of RT and LT growth of Co/W(111) with Co thickness 3.7 and 3.8 PML, respectively. For beam energy of 120 eV, perpendicular crystalline structure has different from growth mode by a/b intensity ratio analysis.. Figure 4.10 shows that in-plane lattice constant still does not change.

(115) 4.3. Structure I : Co/W(111). 53. but a/b ratio varies with different growth temperatures. For beam energy of 120 eV, the LT case indicates that spot a is brighter than b, but in the RT case, spot a is darker than b. It seems that even in the same thickness, RT and LT growth mode have a little different from perpendicular crystalline structure. From a/b LEED intensity ratio of RT growth of Co/W(111), it has the same trend that values of each curves are approach to zero as Co thickness increased. However, we can find some different from low Co coverages that RT growth mode has smaller intensity ratio than LT case, and RT growth mode is larger than LT case by two times at 3.7 PML.. Figure 4.11: a/b LEED intensity ratio of RT growth of Co/W(111). At low Co thickness, it has small ratio than RT case by two times at 3.7 PML. At high Co thickness, it has the same trend witch are approach to zero..

(116) 4.3. Structure I : Co/W(111). 4.3.3. 54. Discussion of in-Plane Crystalline Structure of Co/W(111). Figure 4.12: Top view of crystalline structure of bcc W(111).. Figure 4.13: Top view of crystalline structure of hcp Co(0001).. Let us discuss why Co/W(111) is pseudomorphic growth even in high coverages and why a/b ratios have identical behavior both in RT and LT case when Co coverage increased. Figure 4.12 and 4.13 show.

(117) 4.3. Structure I : Co/W(111). 55. the bulk crystalline structure of W and Co, respectively. From the formula of lattice mismatch where af is the lattice constant of deposition films, and as is the lattice constant of substrate, substituting parameters of W and Co into equation ??. ∆Co/W (111) = 2 ×. 2.51 − 4.47 = −56.16% 2.51 + 4.47. (4.2). and we can obtain the lattice mismatch ∆Co/W (111) , which is too large to convince us that Co could epitaxially grow on W(111) perfectly ! However, if we consider the situation on the left hand side of the figure 4.14, we can find. ∆0Co/W (111) = 2 ×. 3aCo − apW (111) 7.53 − 7.74 =2× = −2.75% 3aCo + apW (111) 7.53 + 7.74 (4.3). So far, we would understand why Co/W(111) is pseudomorphic growth, or more precisely, Co on W(111) is not pseudomorphic growth but epitaxial growth perfectly by three times lattice constant match. For similarly reason, we would understand why a/b ratios would decay to zero at large thickness, because LEED patterns are in reciprocal space and lattice constant is inversely proportional to itself in real space. On the other hand, the inner cycle A is corresponding to W(111) structure and outer cycle B is corresponding to Co structure, so when Co coverage is incresased, the LEED spots of Co(0001) would become sharply and of W(111) become cloudy, see figure 4.14..

(118) 4.4. Magnetism I : Co/W(111). 56. Figure 4.14: Left hand side show that Co structure is match with W(111) by rotated 300 and right hand side is corresponding LEED image.. 4.4 4.4.1. Magnetism I : Co/W(111) RT Growth of Co/W(111). For RT growth mode, the hysteresis loop in figure 4.15 illustrates that only in-plane Kerr signal is detected, i.e. the orientation of magnetization is lying on the in-plane direction. We can obtain the first hysteresis loop in figure 4.15 at 2.3 PML. From the reference [8], it indicates that Co/W(111) is ferromagnetically ordered at 380 K when Co thickness exceeds 7.6 ps monolayers (∼ 2.39 PML). It also point out that substrate atomic steps have a strong influence on the direction of easy axis and the magnetic domain structure, even though the weak in-plane magnetocrystalline anisotropy as shown in figure 4.16. Our experimental result at room temperature is the same as 380 K, which concludes that the in-plane magnetization of Co/W(111) only.

(119) 4.4. Magnetism I : Co/W(111). 57. Figure 4.15: MOKE of Co/W(111) growth at room temperature. Only the hysteresis loop in in-plane are observed when Co thickness exceeds 2.3 PML. Coercivity is enhanced with Co thickness increased. above 300 K growth temperature base on substrate crystalline structure. Note that coercivity in in-plane direction is enhanced with Co thickness increased. 4.4.2. LT Growth of Co/W(111). In the figure 4.17, perpendicular Kerr signal appears at LT growth mode, and coercivity is decayed with Co thickness increased. The hysteresis loop in perpendicular at 3 PML indicates that coercivity is larger than maximum external magnetic field. These results seems implied something new, but from LEED analysis in figure 4.8 and 4.11, as discussion in the section 3.3.1. We can’t deal with what’s going on.

(120) 4.4. Magnetism I : Co/W(111). 58. Figure 4.16: SPLEEM images of magnetic domain in Co on vicinal W(111). There are in-plane magnetization for Co/W(111) with growth temperature at 380 K only. Arrows in plot a—d denote the incident beam polarization direction, and the LEED pattern of vicinal W(111) is shown with magnetization direction and in-plane directionperpendicular to steps specified by the solid and dashed lines, respectively. [8]. in the crystalline structure between RT and LT growth mode from our data. We just could say that crystalline structure is changed by different growth temperature and maybe this difference results in the difference of magnetic behaviour. We could go through I-V LEED analysis to figure out this question..

(121) 4.5. Structure and Magnetism II : Co/faced Pd/W{112}. 59. Figure 4.17: MOKE of LT growth of Co/W(111). The difference from RT growth mode is that there are existence of perpendicular signal, and coercivity are decayed with Co thickness increased.. 4.5 4.5.1. Structure and Magnetism II : Co/faced Pd/W{112} LT Growth of Co/faced Pd/W{112}. In the Co/faceted Pd/W{112} case, LEED spots of faceted still clear when Co coverages below 2.5 PML, but at 3.8 PML as well as 4.8 PML we can’t obtain any periodic structure in these systems as shown in figure 4.18. So there are two straightforward inference that (i) Co films are stochastic growth on faceted Pd/W{112} surface and (ii) Co films are layer-by-layer growth on faceted surface at first, but come up unknown growth mode at higher thickness. So we can still see {112} diffraction spots at low Co coverages but LEED patterns become indistinct at high coverages..

(122) 4.5. Structure and Magnetism II : Co/faced Pd/W{112}. 60. Figure 4.18: LEED patterns of LT growth of Co/faceted Pd/W(111). We can see that faceted spots still clear when Co coverages below 2.5 PML but become cloudy above 3.8 PML. However, there are two key points in the Co/faceted Pd/W(111) system illustrated in figure 4.19. For one thing, the magnetic behavior is different from Co/W(111) system because of no perpendicular.

(123) 4.5. Structure and Magnetism II : Co/faced Pd/W{112}. 61. Figure 4.19: MOKE of Co/faceted Pd/W(111) with low growth temperature. This system has no perpendicular MOKE signal, and coercivity are enhanced with Co thickness increased. A more interesting phenomenon is “two-step” hysteresis loop. This phenomenon is attribute to surface and CoPd-W interface magnetic anisotropy.. signal, and coercivity are enhanced with Co thickness increased. Comparing with the coercivity of RT growth of Co/W(111) at 3.8 PML, faceted system is less than 27 Oe For another, the hysteresis loop in in-plane shows two-step loop shape. We are not amazed at the first point on account of diverse substrate often associate with various magnetic properties. The second point is really interesting because it implies that magnetic easy axes exist in two direction simultaneously when two-step loop occur. We can get some possible conclusions to accomplish this condition as follows: Case (i) : Morphology induced magnetic anisotropy.

(124) 4.5. Structure and Magnetism II : Co/faced Pd/W{112}. 62. Owing to faceted Pd/W{112} with many periodic three side of pyramids, if Co on faceted system is layer-by-layer growth, then we can argue that this sample has three magnetic easy axes by morphology induced anisotropy, and from among of two directions with strong anisotropy, so we can only find two-step loop without three-step. However, this point of view is not valid because two-step loop still exists even if LEED patterns is cloudy for high Co coverages at 3.8 PML and 4.8 PML.. Figure 4.20: CoPd alloys phase diagram. Co and Pd will form alloys for any Co-Pd ratio at our experimental temperatures [16].. Case (ii) : Two kinds of magnetic films coupling induced anisotropy It seems impossible that there are magnetic double-layers in our system since we have only one magnetic Co film. Even so, the figure 4.20 shows that Co and Pd form alloys for any Co-Pd ratio in.

(125) 4.5. Structure and Magnetism II : Co/faced Pd/W{112}. 63. bulk phase. So the problem now becomes to confirm that Co-Pd alloys are existence. At first, we can observe the figure 4.21(a) that Pd signal of the first monolayer with Co deposition is smaller than without Co deposition. It implies that Pd atoms of Pd-W interface are diluted with Co deposition. Second, we can obtain from figure 4.21(b) that Co signal of the first monolayer vanishes.. Figure 4.21:. TPD of (a) faceted Co/Pd/W{112} with Pd part and faceted. Pd/W(111). This diagram indicate that first monolayer peak is diluted with Co deposition;(b) Co/faceted Pd/W{112} with Co part and Co/W(111). It reveal that Co signal of the first monolayer disappear. Maybe we can interpret it as that Co covers on faceted Pd surface. But figure 4.22 indicates that desorption of Co and Pd happen simultaneously and note that the Co signal of the second monolayer is not.

(126) 4.5. Structure and Magnetism II : Co/faced Pd/W{112}. 64. symmetry and implies the existence of Co-Pd interface. In general, if the Pd signal of the second monolayer exists, it means that the binding of Pd-Pd interface is broken. Nevertheless, figure 4.22 shows the desorption of Co and Pd happen at the same time, if we figure it out to Co-Co and Pd-Pd desorption, then Co-Pd interface is not existent due to Pd remain just on the Pd-W interface.. Figure 4.22: TPD of Co/faceted Pd/W{112}. It show that Co and Pd desorption at the same time and the Co signal of the second monolayer is not symmetry.. However, we cannot exclude the case (ii) from above reasons because the interdiffusion between Co atoms and Pd atoms of Pd-W interface are possible. But this argument is relatively impossible because CoPd alloys should be existent everywhere from figure 4.20. So.

(127) 4.5. Structure and Magnetism II : Co/faced Pd/W{112}. 65. the better inference is that CoPd alloys form and the peak of Co and Pd of first and second monolayer is corresponding to CoPd-W and CoPd-CoPd interface, respectively. We can make a conclusion as the case (iii) : Case (iii) : Two-step in the hysteresis loop is caused by different anisotropy at surface and interface CoPd alloys. There is a similar result in Fe lms on ZnSe ??, the ZnSe side is closer to perfect Se-terminated at surface, and the other side is a nonprotected oxidized Fe-surface with different pinning conditions. In our case, the W-side is a non-flat faceted surface, and the other side might be a flatter surface at high coverage than W-side, so that we can find a similar two-step loop in our system. 4.5.2. Hydrogen Effect on Co/faceted Pd/W{112}. Figure 4.23 indicates that hydrogen can not affect coercivity efficiently in Co case comparing with Fe, see figure 4.16. In 2 PML Fe case, coercivity saturates from 100 Oe to 320 Oe when the sample with hydrogen dose above 10 minutes at 2 × 10−9 torr (∼1.8 L). In 3.8 PML Co case, coercivity decreases from 240 Oe to 200 Oe after hydrogen dose at 2 × 10−10 for 140 minutes (∼1.8 L). We could not distinguish hydrogen effect between Co and Fe case because our exposure pressure is too low to identify this effect as from hydrogen or cobalt oxide for long exposure time. Furthermore, for the lack of the data on 3.8 PML Fe sample with in-plane anisotropy, so that hydrogen effect in ferromagnetic films on Pd faceted surface is not confirmed..

(128) 4.5. Structure and Magnetism II : Co/faced Pd/W{112}. 66. Figure 4.23: MOKE of 3.8 PML Co/faceted Pd/W{112} with hydrogen dose at 2 × 10−10 . Coercivity is decreased from 240 Oe to 200 Oe after hydrogen..

(129) Chapter 5 Conclusion Owing to three times lattice constant match, perfect epitaxial growth of Co on W(111) is observed even if Co thickness is 6.7 PML for 80 K growth and ≥ 3.7 PML for 300 K growth. For identical reason, LEED patterns of inner and outer cycle is corresponding to W and Co, respectively, and inner W spots would become cloudy and disappear at high Co coverage. Co/W(111) is found to reveal in-plane magnetic anisotropy when grown at room temperature and both in-plane and perpendicular magnetic anisotropy for LT growth. The coercivity is enhanced with Co thickness for RT growth and decays with Co thickness for LT growth. These different magnetic behavior is attributed to different perpendicular crystalline structure. Co on faceted surface has two-step hysteresis loop and we infer that CoPd alloys are formed and two-step results from magnetic anisotropy of surface CoPd alloys and CoPd-W interface. Hydrogen effect is not confirmed because we can not tell if the coercivity is decreased by hydrogen or cobalt oxide.. 67.

(130) Bibliography [1] Yeh, Y.-C, Master thesis “Morphology magnetic properties of iron on faceted Tungsten(111) surface induced by palladium” (Nation Taiwan University, 2008) [2] Andrzej Szczepkowicz, Robert Bryl, Phys. Rev. B 71, 113416 (2005) [3] Andrzej Szczepkowicz, Robert Bryl, Surf. Sci. Lett. 559, L169 (2004) [4] C. H. Nien, T. E. Madey, Y. W. Tai, T. C. Leung, J. G. Che, C. T. Chan, Phys. Rev. B 59, 10335-10340 (1999) [5] C. Herring, Phys. Rev. 82, 87 (1951) [6] S. P. Chen, Surf. Sci. Lett. 274, L619-L626 (1992) [7] J. J. Kolodziej, T. E. Madey, J. W. Keister, J. E. Rowe, Phys. Rev. B 65, 075413 (2002) [8] K. L. Man, R. Zdyb, S.F. Hung, T. C. Leung, C. T. Chan, E. Bauer, M. S. Altman, Phys. Rev. B 67, 184402 (2003) [9] G. L. Kellogg, “Physics of Solid Surfaces” edited by G. Chiarotti, Landolt-B¨omstein (Springer-Verlag, Berlin, 1996) 68.

(131) Bibliography. 69. [10] Ker-Jar Song, Cheng-Zhi Dong, Theodore E. Madey, Langmuir 7, 3019-3026 (1991) [11] T. E. Madey, C. H. Nien, K. Pelhos, J. J. Kolodziej, I. M. Abdelrehim, H. S. Tao, Surf. Sci 438, 191-206 (1999) [12] T. M. Gardiner, H. M. Kramer, E. Bauer,Surf. Sci. 112, 181 (1981) [13] Tsu-Yi Fu, Lung-Chieh Cheng, C.-H. Nien, Tien T. Tsong, Phys. Rev. B 4, 113401 (2001) [14] M. Wortis, “Chemistry and Physics of Solid Surfaces VII ”, edited by R. Vanselow and R. Howe (Springer-Verlag Berlin, 1988) [15] W. Nesseldreher, Vacuum 26, 281 (1976) [16] “Introduction to Alloy Phase Diagrams”, edited by Hugh Baker (ASM International, 1992) [17] John F. O’Hanlon, “A user’s guide to vacuum technology”, (Joho Wiley Sons, Inc., 2003) [18] Yip-Wah Chung, “Practical guide to surface science and spectroscopy”, (Academic Press, 2001) [19] Friedhelm. Bechstedt,. “Principles. of. surface. physics”,. (Springer-Verlag Berlin, 2003) [20] “Surface and thin film analysis”, edited by H. Bubert and H. Jenett, (WILEY-VCH, 2002).

(132) Bibliography. 70. [21] R. Meckenstocka, D. Spoddiga, K. Himmelbauer, H. Krennb, M. Doi, W. Keune, Z. Frait, J. Pelzl, J. Magn. Magn. Mater. 240, 410–413 (2002).

(133)