鈷在矽(111)和鍺(111)表面與雙羧基紫精在銅(100)表面上的表面結構變化研究

全文

(1)國立臺灣師範大學物理研究所博士論文指導教授 (Advisor)：傅袓怡 (Tsu-Yi Fu)、Klaus Wandelt Doctoral dissertation. 鈷在矽(111)和鍺(111)表面與雙羧基紫精在銅(100)表面上的表面結構變化研究 Investigations structural transitions of the cobalt on the Si(111) and Ge(111) surfaces and dicarboxylated viologens on the Cu(100) surface. 研究生：蔡松麟 (Sung-Lin Tsay) 中華民國九十八年 (2009).

(2) Acknowledgement I am very sincerely grateful to all people who help me to complete this thesis. First, I would like to express my greatest appreciation and thanks to my supervisor Prof. Tsu-Yi Fu for supporting me the opportunity to contact surface science and doing the researches of the UHV-STM in the Institute of Physics, National Taiwan Normal University. I am deeply indebted to Prof. Fu for her helpful advice, experiment technology, and fund support during my PhD period. In first part of this thesis, these good results are also contributed by PhD student of Jiun-Liang Lin, and master student of Jang You Guo, Jr Guei Gau, Wen Chen Chen, Min Hua Lin, Jiun Rung Chen, and Yue Shian Jang. In the investigation meanwhile, I also would like to thank Xiao-Lan Huang, Hui-Ya Tsai, Jia-Lun Jiang, Qi-Jun Wang, Jing-Shu Hu, Peng-Zhe Zai, Xuan-Zhen Hong, Yi-Feng Chen, Yu-Xian Li, and Zhi-Hao Zhao students to encourage me and help me to complete this thesis. Second, thank Prof. Jyh-Shen Tsay very much for conducting the EC-STM field. I acquired the opportunity of two times to join the group of Pro. Dr. Klaus Wandelt to study the investigation of dibenzyl viologens on the Cu(100) surface from the help of Dr. Peter Broekmann during two months in the Institute of Physical and Theoretical Chemistry, Bonn Unversity. I am very grateful to Dr. Peter Broekmann and PhD student Duc-Thanh Pham for their valuable suggestions, excellent experiment, and fruitful discussions. Last year, I got the scholarship from the German Academic Exchange Service (DAAD) and National Science Council of ROC to go to the group of Pro. Dr. Klaus Wandelt again to study the topic of phase transition of carboxylated viologens on the Cu(100) and HOPG surfaces during eight months. I am also greatly indebted to Prof. K. Wandelt, my co-supervisor, who gives me his help and time in listening to me, discussing to me, helping me work out my problems, and carefully correcting my thesis a lot. Finally, I greatly appreciate my family give me endless love and support me achieve the PhD degree. I also have the deeply gratitude to my girlfriend for accompanying my side whether happy or sad.. ii.

(3) 摘要此論文主要分為兩個部分，第一個部分的實驗，為鐵磁性物質鈷原子被蒸鍍在矽(111)-7×7 與鍺(111)-c(2×8)表面上，隨不同加熱退火的溫度處理後，利用超高真空掃描穿隧顯微鏡(UHV-STM)與低能量電子繞射儀(LEED)來研究鈷島成長行為。第二個部分的實驗藉由電化學掃描穿隧顯微鏡(EC-STM)與循環伏安法 (Cyclic Voltammetry, CV)，研究有機分子雙羧基紫精 (Dicarboxylatic Viologens) 在銅(100)與 HOPG 表面上隨著溶液偏壓而發生相變的行為。相變會因為外在環境的影響而發生，如同利用加熱退火使得鈷島産生結構的改變，與溶液在外加偏壓下驅使 Violognes 分子重新排列形成新的排列。當鈷被蒸鍍在矽(111)-7×7 與鍺(111)-c(2×8)表面上，在鈷鍍量低時，表面上出現因為鈷矽與鈷鍺形成合金時所出現的缺陷，為了確認這些缺陷的確是因為鈷與矽形成反應，把鈷鍍在低溫的矽表面，在回溫過程時發現當鈷與矽形成反應時造成缺陷的出現，此反應出現在 126K 到 130K 之間。600K 的加熱退火之後，會出現 Co5Ge7 的合金，另外鈷在鍺上有機會形成具有√13 × √13R14°週期性的表面結構，但是出現機率小，所以此結構的週期在 LEED 上沒有呈現出來，之前也沒有任何的研究發現。當鈷與矽或與鍺形成反應後所呈現的磁性強度遠比單純鈷還來得小，所以銀被選為緩衝層避免鈷與底下的基底反應。在鈷鍍量低時，鈷無法形成 2 維並且具有週期性的結構在矽(111)-√3 × √3表面上，然而鈷能夠形成 √13 × √13R14°與 2×2 的表面結構在鍺(111)-√3 × √3表面上在經過加熱退火的處理後。只有當鈷鍍量大於 1.8 單層之後，才有辨法形成 1×1 的表面結構在銀/ 矽(111)-√3 × √3表面上。因為銀/矽(111)-√3 × √3面具有未滿足的電子態，所以使得鈷在低鍍量時，無法形成具有週期的表面結構，鈷傾向滿足那些未滿足的電子態來降低表面自由能而不是形成 2 維島，但是當鍍量增加時，鈷不受到那些未滿足的電子態的影響而形成 1×1 的表面結構，除此之外基底鍺也扮演著重要的角色使得鈷能夠形成具有表面結構的 2 維島，因為同樣是√13 × √13R14°的表面結構能夠在 Si(111)-7×7 與矽(111)-√3 × √3表面形成，進一步發現√13 × √13R14° 與 2×2 的表面結構跟加熱退火的溫度和鈷的鍍量有關係造成兩種結構的相變。為了避免因為銀/矽(111)-√3 × √3表面上的未滿足電子態而使得鈷無法形成具有週期性的 2 維鈷島，約 6 單層鍍量的銀鍍在 100K 的矽(111)-7×7 表面上，再經過加熱退火至 300oC 後，銀能夠形成平坦的表面在矽(111)-7×7 表面之上，然後再鍍上鈷，發現依然無法形成具有週期性的 2 維鈷島，所以是否能夠形成具有週期性的 2 維鈷島主要是受到鍺基底的影響。濃度 0.1mM 的雙羧基紫精在銅(100)表面上發現 6 種排列的模式，雙羧基紫精在氧化態(+2)時程現出平均分佈的點陣排列與傾斜的排列，在此氧化態時雙羧基紫精的含氮雙芽基平面平行表面，因為+2 的氧化態與基底陰離子層氯 /銅 (100)-c(2×2)之間的作用比分子之間還來得強，在還原態(+1)時分子之間傾向以面對面的方式排列( π-π )形成條紋的圖案，主要發現了亞穩態與鬆散的條紋排列和 iii.

(4) 緊密的條紋排列與雙分子排列，這麼多的排列方式主要是因為此雙羧基紫精具有 7 個長度的烷基鍵，因此在空間與分子之間作用力的影響下造成了多樣的排列。更進一步發現雙層的堆積在亞穩態時就已經形成了，這是因為氮與氧之間的氫鍵作用力所造成的，所以當溶液濃度升高至 1.0 mM 時，分子多層堆積的成長被發現，此外也發現了不同陰離子層對分子的排列造成影響，當陰離子層為氯離子時，可以發現然亞穩態，然而當陰離子層為溴時，並沒有發現此態。. iv.

(5) Contents List of abbreviations Abstract 1. Theoretical background ...................................................................................... 1 1.1 Scanning Tunneling Microscopy (STM).................................................... 1 1.2 Low Energy Electron Diffraction (LEED) ................................................ 5 1.3 Introduction to electrochemistry ............................................................... 6. Part I: The magnetic material cobalt on the Si(111) and the Ge(111) surfaces dependent on annealing temperatures ........................................11 Chapter 1 Introduction .................................................................................. 12 Chapter 2 Experiments, introduction of the Si(111), the Ge(111), and the Ag-√ × √ surfaces..................................................................................... 15 2.1 Experiments...................................................................................... 15 2.2 Introduction to the Si(111)-7×7 and the Ge(111)-c(2×8) surfaces .. 17 2.3 The formation of the Ag-√ × √ layer on the Si(111)-7×7 and the Ge(111)-c(2×8) surfaces .............................................................. 20 Chapter 3 Co on the pure Si(111)-7×7 and Ge(111)-c(2  8) surfaces dependent on annealing temperatures .......................................................... 21 3.1 Submonolayer Co on the Si(111)-7×7 surface ................................. 21 3.2 Submonolayer Co on the Ge(111)-c(2×8) surface ........................... 22 3.3 Submonolayer Co on the Si(111)-7×7 surface at low temperature . 24 Chapter 4 Co on the Ag/Si(111)-√ × √ and the Ag/Ge(111)-√ × √ surfaces dependent on annealing temperatures............................................ 29 4.1 Co on the Ag/Si(111)-√ × √ surface ............................................ 29 4.2 Co on the Ag/Ge(111)-√ × √ surface .......................................... 37 4.3 Comparison of the Co growth behavior on the Ag/Si(111)-√ × √ surface versus the Ag/Ge(111)-√ × √ surface ............................. 46 4.4 Co on the flat Ag/Si(111)-1×1 surface .............................................. 52 Chapter 5 Conclusion .................................................................................... 56. Part II: The study of the surface structures and phase transitions of the viologens on the Cu(100) and the HOPG surfaces ...................... 57 Chapter 1 Introduction .................................................................................. 58 Chapter 2 Viologens and experiment ............................................................ 62 2.1 Introduction to viologens ................................................................. 62 i.

(6) 2.2 Experiment ....................................................................................... 64 Chapter 3 Results and Discussions of the self-assemble layers of carboxylated viologens on the surfaces ......................................................... 69 3.1 Chlorine anion modified Cu(100) surface ....................................... 69 3.2 Cyclic Voltammetry measurements (CV) ........................................ 71 3.3 Overview of all phases ...................................................................... 75 3.4 Dot array .......................................................................................... 78 3.5 Oblique row phase............................................................................ 82 3.6 Metastable phases............................................................................. 86 3.7 Stripe Pattern ................................................................................... 91 3.8 Phase transition from the dot array to the stripe pattern without the metastable phases ....................................................................... 98 3.9 Closed Stacking Stripe Pattern ........................................................ 99 3.10 Chloride Desorption and Dimer phase .........................................104 3.11 Determining the orientations of the alkyl chains for all phases ...107 3.12 Summary .......................................................................................108 Chapter 4 The bilayer formation of carboxylic viologens and the influence of the chlorine and the bromine anion layers on the growth behavior of the thin film ......................................................................................................... 111 4.1 Dilute viologen electrolyte ............................................................... 111 4.2 Dicarboxylated viologens on HOPG ............................................... 114 4.3 The initial stacking of the bilayer formation .................................. 116 4.4 High concentration carboxylated viologens (1.0 mM) on the Cu(100) surface ............................................................................... 118 4.5 Carboxylated viologens mixed with the bromic acid (0.1 mM) on the Cu(100) surface ....................................................................122 4.6 Summary .........................................................................................125 Chapter 5. Conclusion ..................................................................................127 2. Conclusions of all works ...................................................................................129 References .............................................................................................................132 Publication list ......................................................................................................140. ii.

(7) List of abbreviations AC APBs CDR CITS Co-Ge Co-Si CV DAS DBV DC EC FSs FHUC HCT HER HOPG It LDOS LEED. Alternating Current AntiPhase Boundaries Copper Dissolution Reaction Current Image Tunneling Spectroscopy The compound of the Co and the Ge The compound of the Co and the Si Cyclic Voltammetry Dimer Adaom Stacking 1,1′-dibenzyl-4,4′-bipyridinium Direct Current ElectroChemical Ferromagnetic Semiconductors Faulted Half Unit Cell Honeycomb Chain Trimer Hydrogen Evolution Reaction Highly Ordered Pyrolytic Graphite The parameter of the tunneling current Local Density Of States Low Energy Electron Diffraction. l-√ MV N-H O-H Ubias. Lower − √ × √ layer 1,1′-dimethyl-4,4′-bipyridinium Hydrogen bonding of the Nitrogen and the Hydrogen Hydrogen bonding of the Oxygen and the Hydrogen The parameter of the sample bias. u-√ UHUC UHV RT RFA SAM STM −( − V. W. mode. Upper − √ × √ layer Unfaulted Half Unit Cell Ultra High Vacuum Room Temperature Retardation Field Analyzer Self-Assembled Monolayer Scanning Tunneling Microscopy 1,1'–Bis(7-carboxyheptyl)-4,4'-bipyridinium Volmer -Weber mode. ). iii.

(8) Abstract In this thesis, the investigations include two parts. The first part (Part I) is the research of the magnetic material cobalt on the Si(111) and the Ge(111) surfaces dependent on annealing temperatures studied by ultra high vacuum-scanning tunneling microscopy (UHV-STM) and low energy electron diffraction (LEED). The second part (Part II) is the research of the surface structures and phase transitions of viologens on the Cu(100) and HOPG surfaces dependent on the applied potentials studied by cyclic voltammetry (CV) and electrochemistry scanning tunneling microscopy (EC-STM). The aims of both subjects are focused on the observation of the surface structure and morphology at stable thermal/kinetics conditions. There are two major topics in part I. The first topic describes the Co absorbed on the pure Si(111)-7×7 and Ge(111)-c(2×8) surfaces. The initial reaction of the Co with the Si substrate happens at the temperature range from 126 to 130K. The defects of the Co-Si compounds are different from the intrinsic defects of the Si(111)-7×7 surface. The Co-Si compounds also decrease the brightness of the neighboring adatoms compared to the intrinsic defects of the Si(111)-7×7 surface. Therefore, the Co on the pure Si(111)-7×7 and Ge(111)-c(2×8) surfaces forms the Co-Si and the Co-Ge compounds at room temperature appeared as the dark region of the defect-like feature. The silicon atoms can separate on top of the Si(111)-7×7 surface after annealing to 400oC, and then the Si(111)-7×7 surface structure disappears. The Co5Ge7 alloy is observed on the Co/Ge(111) surface after annealing to 600K. Further, Co atoms can form the √13 × √13R14° periodic surface structure, but the structure is unfavorably formed compared to the Co5Ge7 alloy. The compound formations of the Co-Si and the Co-Ge result in a lower magnetic property than bulk Co. Therefore, the silver buffer layer is introduced on the intermediate layer between the Co and the Si(111) and Ge(111) surfaces as described at the second topic. At low Co coverage, the Co can form periodic surface structures of the √13 × √13R14° and the 2×2 on the Ag/Ge(111)-√3 × √3 surface. For the Co/Ag/Si(111) case, the Co forms a cluster shape both on the Ag/Si(111)-√3 × √3 and on the flat Ag/Si(111)-1×1 surfaces at low Co coverage. Further, the average size and height of Co clusters on the flat Ag/Si(111)-1×1 surface are almost independent on annealing temperatures from room temperature to 300oC. The reasons for the Co/Ag/Si(111) surface without periodic surface structure are due to the unsaturated states on the Ag/Si(111)-√3 × √3 surface and the weaker interaction of the Co with the Si(111) surface than the Ge(111) surface. Dicarboxylated viologens mixed with a 10 mM HCl on the Cu(100) and HOPG surfaces was studied in different redox states. At the beginning, a 0.1 mM violgens on the Cu(100) surface is investigated. The dicationic viologens show the dot array and iv.

(9) the oblique row phases. The radical viologens exhibit the metastable phases, a stripe pattern, the closed stacking stripe pattern, and a dimer phase. The stacking configuration of the dicationic viologen core plane is preferred to be face-on on the surface and that of the radical viologen is formed by π − π stacking with the neighboring viologens. Because dicarboxylated viologens bear long alkyl chains and carboxylic acid groups at the ends of the alkyl chains as illustrated by (HOOC-(CH2)7-V-(CH2)7-COOH), the complex interactions are considered to be the reason of forming various phases on the Cu(100) surface. The experiment of a 0.1 mM dicarboxylated viologens on the HOPG surface without the influence of an anion layer is to confirm the existence of a bilayer formation due to the intermolecular interaction of the hydrogen bonding. The high viologen concentration (1.0 mM) on the Cu(100) surface shows the effects of the chloride and bromide anion layers reflecting on phase transition and the multilayer growth behavior due to the hydrogen bonding interaction and the polarizability, respectively. The effect of the anion layers is consistent with by a 0.1 mM viologens mixed with a 10 mM KBr on the Cu(100) surface.. v.

(10) 1. Theoretical background In this chapter, the basic concepts of scanning tunneling microscopy, low energy electron diffraction, and electrochemistry are introduced. 1.1 Scanning Tunneling Microscopy (STM)1,2,3 The observation at the recent research of the surface on the nano scale plays an important role to see the atoms because the piezoelectric material made this “observation” possible. The STM experiments can be carried out in air, liquids, and vacuum conditions. In gas media, the samples need to be very thin (< 2 nm) or highly conduction oxide films to support enough conductivity without charging effects. Because the oxide films are able to be removed by sputtering or annealing the surfaces in vacuum, the STM in vacuum serves almost kinds of the conductors and the semiconductors. For the STM experiments in liquids, acidic and alkaline liquids with high purity water (e.g. conductivity < 18 M Ω  cm ; TOC < 5ppb) are used and the sample is kept in liquids to avoid the oxide film formation. The STM technique is applied on the observation of the surface from several micrometers to the atomic level, the information of the surface electronic structure, and the manipulation of atoms on the nano-scale. The STM is the advanced product of the functional concept of the tunneling which arises from the quantum mechanism. The tunneling quantum mechanism is that the electron tunnel across a narrow vacuum gap from the tip to the surface or from the surface to the tip as the distance between the tip and the surface is approaching ~ 10 Å. The small bias potential which normally is in the range from 2 mV to 2V supports the necessary potential difference between the tip and the surface for the tunneling occuring. The basic physical concept is best understood by the sample case of one-dimensional tunneling. For a particle with the energy E = 1/2 V0 incident on the square barrier from the left this is shown in Figure 1-1. Figure 1-1. A particle with V0/2 energy is incident on the V0 potential barrier. The potential barrier = V0 at the range of 0  x  s . At the other ranges the potential = 0.. The Schrödinger equation is writing as (  2 / 2 m ) ''  [V ( x)  E ]  0 1.

(11) And the solutions are.  e  ikx  re ikx (x<0)   ( x)   Ae k ' x  Be k ' x (0<x<s)  te ikx (x>s)  Where  k  (2 mE )1/2 ;  k ' =(2m[V0  E ])1/2 By solution of the. ( ), the transmission coefficient can be obtained as. 16k 2k '2 2k 's T  t  2 '2 e (k  k ) 2. which shows the probability for the electron tunneling between the two conductors as an exponential function of the barrier width s. By decreasing the wide of the barrier, the tunneling current signal can be increased. The wide of the barrier s represents the distance between the two conductors along the z direction. Because of the exponential dependence relation, the lateral resolution can be enhanced. For example, as the tip leaves the atomic position, the signal immediately decrease to show the clear atomic position. The most general expression for the tunneling current density J at 0K from a conductor to another is given by. J  J 0[ e A where A . .  (  eV )e A.  eV. ]. 4s 2m e 1 s ; J0  ;   s 2  ( x)dx 2 s 1 h 2 (s). which corresponding symbols are shown in Figure 1-2.4. Figure 1-2. The energy diagram of an arbitrary barrier describing the parameters: ( ) means the barrier height, ∆ =. −. is the thickness of the part of the. barrier above the Fermi level, and ( ) indicates the potential energy of an electron between the two surfaces and .4 2.

(12) The tunneling current density J is the net current density. The flowing from the electrode ( +. 1 ). to. electrode. 2. is. and. the. opposite. flowing. is. . In the case of the small voltage, low temperature, and the most. ideal tip as a singular point probe, the tunneling current can be simplified to the proportional relation as ∝ ∑ | ( )| ( − ) = ( , ) the symbols of which are indicated by Figure 1-3.. Figure 1-3. Energy diagram describing the wave function overlap leading to tunneling occurrence between two electrodes. The and mean the work function and the wave function of the electrode in the left side. is the wave function of the electrode in the right side electrode. The extra applied potential is eV.4 The surface local density of states (LDOS) is described by the quantity of ( , ) at at the position of the point probe. Therefore, the STM image is the contour map of the surface LDOS. The effect of the bias at positive or negative potential for the current direction is shown in Figure 1-4. At positive bias (a), the tunneling current flowing is toward the unoccupied states of the surface. At opposite bias (b), the tunneling current flows out from the occupied states of the surface to the tip. By this treatment, the surface electronic structure can be identified. The dashed line represents the empty wave function above the Fermi level.. 3.

(13) ( a ) Metal. Vacuum. Semiconductor. (b ). Ec EF. eVT. EC. Tip. Tip Surface Surface Bulk DOS Surface DOS. Figure 1-4. (a) and (b) illustrate the tunneling current flowing direction at positive and negative biases.2 There are the three operation modes during the scanning. The three operation modes are the constant current mode (a), constant height mode (b), and current image tunneling spectroscopy (c). (a) The constant current mode In this mode, the tunneling current is fixed by the feedback signal to drive the z piezoelectric stick to adjust the distance between the tip and the surface resulting in the increasing or decreasing of the tunneling current to achieve the set tunneling current as the surface morphology changes. Because this mode relies on the feedback system, the speed of the scanning process is lower than the constant height mode. (b) The constant height mode The constant height mode means that the distance between the tip and the surface is kept constant and the signal of STM image represents the magnitude of the tunneling current. Therefore, the scanning speed is faster than the constant current mode. Further, the fast scanning speed has the larger opportunity to observe the motion traces of surface atoms, but it is also easier to crash the tip if the surface morphology is rough. (c) The current image tunneling spectroscopy (CITS) The current image tunneling spectroscopy is used to detect the LDOS on the surface. The method is to fix the height at a set tunneling current. Subsequently, the bias is changed within a set range (i.e. from -2.0 V to +2.0 V). And then the energy diagram of the tunneling current as a function of the bias is obtained. By drawing the dI⁄dV plot, the peak positions are corresponded to the particular electronic states.. 4.

(14) 1.2 Low Energy Electron Diffraction (LEED) The LEED technique is available for the surface structural analysis by the elastic scattering of the electrons. The diffraction spots are corresponded to the surface period. The maximum in the diffraction intensity at any angle satisfies the formula of nλ = d sinθ where λ ≈ 150⁄V Å, n is an integer, d is the lattice constant, and the angle θ = sinψ − sinϕ as indicated by Figure 1-5(a).. (a). (b ). Figure 1-5. The illustration shows the diffraction process in real space at the surface (a) with corresponding Ewald sphere construction in reciprocal space (b).2 The LEED pattern is the Fourier Transform space. The relation between the lattice spaces and the Fourier spaces are given by the formulas       2 a  a  2 a  a  2 a  a  a1*   2 3 ; a2*   3 1 ; a3*   1 2 a1  a2  a3 a1  a2  a3 a1  a2  a3  * where ai is in the reciprocal space and ai is in the real space. It is convenient to understand the reason of the LEED pattern formation from the Ewald sphere construction in the reciprocal space as shown in Figure 1-5(b). In reciprocal space, the conditions of the two dimensional scattering diffraction are ⃗ = ⃗ − ⃗ where ⃗ = ⃗ . The radius of the Ewald sphere is ⃗ . As the circle of the Ewald sphere intersects with a vertical line, the intersection point corresponds to the particular diffraction direction and satisfies the diffraction conditions resulting in the LEED spots.. 5.

(15) 1.3 Introduction to electrochemistry5,6 The reaction O+ne-  R where O is the oxidized state and R is the reduced C (0, t) state is describe by the Nernst equation E = E + 2.3RT nF log C (0, t) (E is related to the standard potential E of the couple O/R, and the surface concentrations are O (C (0, t)) and R (C (0, t)). The simple reaction processes are mass electron mass O bulk  transport    O electrode , O electrode  transport    R electrode , and then R electrode  transport    R bulk .. The simple reactions are not including a adsorption, the phase formation, a phase transition, and the coupled chemical reactions. For adsorption isotherm several isotherms are considered such as (a) The Langmuir isotherm (a fundamental isotherm disregarding lateral interactions), (b) the Frumkin isotherm (an interaction isotherm considering lateral interactions), (c) the Temkin isotherm (a heterogeneous surface isotherm), (d) the Flory-Huggins-type isotherm (a substitutional isotherm) and so on. The three modes of mass transport are namely (a) the diffusion, (b) the convection, and (c) the migration. The diffusion is the result of a concentration gradient. The convection is an intrinsic phenomenon of the solution such as stirring, flowing the solution, rotating, and vibration the electrode. The migration comes from the electrical field effect. Normally, (b) and (c) can be reduced by using a stationary solution and increasing the concentration of the electrolyte. The flux of mass transport is J(x,t)=-D. C ( x, t ) zFDC  ( x )   C ( x, t )V ( x, t ) where the first term is from the x RT x. diffusion, the second is from the convection, and the last term is from the migration (D is the diffusion coefficient cm2/s, V ( x, t ) is the hydrodynamic velocity, F is the Faradic constant, and z is the charge). The diffusion term is mainly attributed to mass transport due to the J(x,t)=-D Fick’s second. law. C ( x, t ) formula called the Fick’s first law. By the x. C ( x, t )  2 C ( x, t ) , D t x 2. i (t )  nFAD0 C 0 ( b ) / ( D0t )1/2 where (πD t). /. the current. is obtained. from. is the diffusion-layer thickness and A. is the surface area. The concentration of the solution as a function of the distance from electrode surface is illustrated in Figure 1-6. The illustration shows the concentration gradient decreasing as the waiting time increases which results in a small peak intensity in the peak CV.. 6.

(16) Figure 1-6. The plot describes the concentration of the solution as a function of the distance from electrode surface.6 The C0 is the concentration at the surface and the value is 0, and Cb is the concentration of the bulk. For the potential-sweep case, Figure 1-7 shows the peak of CV formed following the potential evolution due to the surface concentration changing. The maximum current is given as i p . nFAD0C0 (b, t ) where C0(b,t) is the concentration of the bulk . solution and δ is the width of diffusion layer. Not only the charges transport but also the absorption/desorption process on the surface results in a peak appearing due to a surface concentration changing. Therefore, the CV measurement provides important information prior to the STM measurement.. Figure 1-7. The surface concentration variation by the sweeping potential results in the peak appearing of CV.6 7.

(17) The electrical double layer is the important concept to describe the charge distribution and electrical potentials at the solution/solid interface. The interface always is constructed by the segregation of positive and negative charge in the direction normal to the surface. The charge segregation can occur through the adsorption of positive or negative ions at the interface with negative or positive potential. Here, the five models (the Helmholtz, Gouy-Chapman, GouyChapman-Stern, Grahame, and special adsorption models) are considered to illustrate different of the charge segregation cases at positive potential electrode surface as shown in Figure 1-8 to 1-10. The , , , and ( , , , and ) are the potentials (charges) of the metal surface, the interphase-solution phase boundary, the inner Helmholtz plane, and the outer Helmholtz plane, respectively. The schematic diagrams of the Helmholtz model and the Gouy-Chapman model are shown in Figure 1-8(a) and (b).. (a). (b ). The potential distribution. The potential distribution. The charge distribution. The charge distribution. Figure 1-8. The Helmholtz model (a) and The Gouy-Chapman model (b). The Helmholtz model is the simplest model which represents a complete monolayer layer with the opposite charge to the metal surface thereby forming positive-negative charge pair like a perfect electron capacity with electrostatic force shown in Figure 1-8(a). Here, the potential dependence on the distance x from the surface along the normal direction is linearly decreasing according to the Poisson equation (. d 2  ( x) d     const ). 2 dx 0 dx. 8.

(18) The Gouy-Chapman model only includes the diffusion space without the actual requirement of the ion absorption as shown in Figure 1-8(b). In this model, all ions are considered to be the point charges and are free to move. Because the electrode is at positive potential, the concentration of the ions with negative charge is higher near the electrode surface and then decreases in the x distance based on the equation. d 2  2n0 z 2e2    2      0e x with an exponential decay. 2 dx  0kT The Gouy-Chapman-Stern model is a combination of the Helmholtz and the Gouy-Chapman approaches as shown in Figure 1-9(a). The relation between the potential and the distance x within the doubled layer is also linear while it exponentially decays in the diffusion region. Thought the ions are considered to be the point charges at the diffusion region, a diluted concentration solution nearly fits the condition. Further, the absorbed ions can form an good electrical capacity on the surface, and hence the Gouy-Chapman-Stern model is widely accepted. As the absorption occurs, the absorption space is considered in two forms as indicated by Figure 1-9(b). The first is that the ions are in direct contact with the surface called the specific absorption and the plane through the centers of ions the plane is called the inner Helmholtz plane. The plane of the centers of the non-specifically absorbed ions is called the outer Helmholtz plane. The non-specific absorption means that the ions are indirectly absorb on the surface, namely with water molecules or other salvations molecules around them.. (a). (b ). The potential distribution. The charge distribution. Figure 1-9. Gouy-Chapman-Stern model (a) and the Grahame model (b). Figure 1-10 shows the illustration and the potential diagram of a specific absorption 9.

(19) model. The anions are strongly absorbed on positive potential electrode surface resulting in an excess charge in the inner Helmholtz plane due to the strong interaction between the anion and the surface. This specific anions absorption is called the anion layer in the inner Helmholtz plane. And then the non-specific absorption of the cations can occur on top of the anion layer by electrostatic forces. Therefore, the anion layer plays an important role to reduce the mobility of the cations such as the organic molecule on the surface for the STM observation in this work.. Figure 1-10 The specific absorption model of strong anions.. 10.

(20) Part I : The magnetic material cobalt on the Si(111) and the Ge(111) surfaces dependent on annealing temperatures. 11.

(21) Chapter 1 Introduction The research of the magnetic material on a semiconductor is the field of the ferromagnetic semiconductors (FSs) which has been a great attraction due to the merging of the electron spin and the carriers in the semiconductors since 1990s.7-12 The character of the spin-polarized current on a semiconductor surface can be applied the process of the read in/read out in the magnetic storage.13,14 The investigation aim is to understand the growth behavior of the materials, the ferromagnetic property in these systems, and further to develop the spintronic devices.15,16 The various parameters such as annealing temperature and the material coverage determine the surface phenomena of the atomic arrangement of the surface structure and the starting point of the kinetics to drive the diffusion and the nucleation motions of the absorbed materials. The metal-semiconductor interface also plays an important role to influence the epitaxial growth and to lead to special islands shapes (i.e. the two or the three-dimension islands and favorable thickness).17-19 Especially in the field of the self-assembled monolayer as well as the self-organized growth, and periodic surface reconstructions, these subjects become valuable to offer a fundamental understanding of the interaction between the material and the substrate, and the interfacial phenomena.20 The ferromagnetic material cobalt is chosen as the magnetic source because Co material is one of the important magnetic materials used in recording devices, has the largest magnetic anisotropic energy of the ferromagnetic transition metals, and the Co nanocrystals show a size-dependent structural property on a 3-4-nm-thick film of the amorphous carbon versus a uniform three-dimension superlattices on the highly oriented pyrolytic graphite (HOPG) substrate.21-23 In order to control the degree of the electron spin, the semiconductor surfaces of the Si(111) and the Ge(111) are good templates for the spin-based devices.21 The band gap features of the semiconductors provide the selection of the tunneling electrons through the tunneling effect instead of a superconductor.10 Further, the carrier lengths of semiconductors exhibit only tens of nanometers which enabled achieve the magnetoelectronics revolution.24 At room temperature, the previous reports showed an Auger peak shift due to the alloy formation of the Co-Ge and the Co-Si occurring in the process of depositing Co.25-27 Moreover, Co atoms easily dissolve into the silicon bulk and the germanium bulk.25 Co atoms penetrate more facile in a Ge substrate (~150oC) than in a Si substrate (~350oC) due to the higher surface defect concentrations of the vacancies and the interstitials.26 The Auger signals of the Co disappear without spreading at all during annealing from 200oC to 920oC on the Si and Ge substrates due to the alloy.28 The stable solid phase of the Co-Ge is formed after high temperature annealing above 650oC.29 The compound formations of the Co-Ge and the Co-Si reflect the decreasing 12.

(22) Kerr signals.30-32 The hysteresis loops are appeared until the cobalt coverage up to a 5.6 ML for the Co/Si(111) system and a 10 ML for the Co/Ge(111) case.31,33 The higher cobalt coverage needed for detecting the magnetic feature is due to the alloy formation at the interface. Thereby the topmost cobalt layers which can absorb on top of the stable alloy result in the enhanced hysteresis loops and do not react with the Si or Ge atoms. Based on the concept of the alloy formation acting as barrier at the interface between the magnetic materials and the substrate, silver is chosen as a buffer layer since the Ag and Co atoms are immiscible in the bulk phase diagram up to 600oC and Ag atoms are able to form a stable and well-ordered √3 × √3R30° reconstruction on the Si(111) and Ge(111) surfaces before 550oC.34-36 Therefore, we introduce the Ag buffer layer in Co-Si and Co-Ge systems to avoid Co atoms dissolve into the silicon and germanium bulks. Such treatment is also employed in other researches. The cobalt phthalocyanine molecules can diffuse on a Ag/Si(111)-√3 × √3 surface to form a highly ordered close packed array due to a reduce reactivity of the molecules with the Si(111) surface.37 The growth direction of naphthalene tetracarboxylic diimide (NTCDI) rows corresponds to the principal axes of the Ag/Si(111)-√3 × √3 surface.38 Perylene-3, 4, 9, 10-tetracarboxylic-3, 4, 9, 10- dianhydride (PTCDA) and NTCDI show the square, hexagonal, and row phases on a Ag/Si(111)-√3 × √3 surface.39 The Ag/Si(111)-√3 × √3 surface severs as buffer layer to avoid that the absorbed materials strongly react with the substrate resulting in forming well-ordered adlayer structure. The buffer layer of the Ag-√3 × √3 is a suitable template to impede the Co-Si and the Co-Ge compound formations and makes the absorbed material formed the order surface structure. Further, the Ag adatoms can form a 6×6 and a √39 × √39 phase on the Ag/Ge(111)-√3 × √3 surface.40 Co atoms on the Si(111) surface construct a 2×2 structure in alloy formation, but only with the presence of small areas on the surface.41 The 2×2 structure also can be induced by the indium adsorption on the Si(111)-7×7 surface.42 A √21 × √21 structure of the silver atoms appears on the Ag/Si(111)-√3 × √3 surface only at low temperature 150K and that of gold atoms on the Ag/Si(111)-√3 × √3 surface at the room temperature.42,43 Therefore, it is expected that Co can form a well-ordered surface structure on the Ag/Ge(111)-√3 × √3 and on the Ag/Si(111)-√3 × √3 surfaces due to a reduced alloy formation. The magnetic properties of the Co/Ag/Si(111) and Co/Ag/Ge(111) compounds are less apparent then the Co/Si(111) and the Co/Ge(111) cases in highly cobalt concentration.44,45 The reason is the silver buffer layer without forming thermally stable √3 × √3 structure and then the silver atoms segregate up to the topmost layer to reduce the magnetic properties.. 13.

(23) The aim of Part I of this thesis is to investigate the surface reconstructions of the Co on the Ag/Si(111)- √3 × √3 and Ag/Ge(111)- √3 × √3 surfaces dependent on annealing temperatures of the substrate, the cobalt coverage, and to compare them to the cases of the Co on the Si(111)-7×7 and Ge(111)-c(2×8) surfaces. The Part I of the thesis is divided into five chapters. Chapter 1 “Introduction and outline” describes the experimental motivation and each chapter content. Chapter 2 shows the experimental methods including preparation of the clean Si(111)-7×7 and Ge(111)-c(2×8) surfaces, and how to construct the silver buffer layer on them. Chapter 3 presents the growth behavior of Co on the Si(111)-7×7 and Ge(111)-c(2×8) surfaces without the silver buffer layer. The thermal evolution of the cobalt islands on the Ag/Si(111)-√3 × √3 and Ag/Ge(111)-√3 × √3 surfaces is shown in Chapter 4. The last Chapter 5 is the conclusion of this part of the thesis.. 14.

(24) Chapter 2 Experiments, introduction of the Si(111), the Ge(111), and the Ag-√ × √ surfaces The experiments are described in section 2.1 including the preparation of the clean Si(111)-7 × 7 and Ge(111)-c(2× 8) surfaces and the Ag- √3 × √3 layers on the Si(111)-7×7 and Ge(111)-c(2×8) surfaces. Section 2.2 shows the brief statement about the Si(111)-7×7 and the Ge(111)-c(2×8) surfaces. Section 2.3 represents the formation of the Ag-√3 × √3 layer on the Si(111)-7×7 and Ge(111)-c(2×8) surfaces. 2.1 Experiments The experiments are performed in an ultrahigh-vacuum (UHV) chamber equipped with a variable temperature STM (at base pressure less than 5×10-11 torr), two well-collimated e-beam evaporators for depositing high purity Ag and Co atoms, and the LEED instrument. Figure 2-1 shows the positions of the instruments in the UHV system. The number of (1) represents the STM for the surface observation, (2) the Retardation Field Analyzer (RFA)-LEED for the determination of the surface periodicity, (3) the sputtering ion gun (Ar+) for cleaning the germanium surface, (4) the Kundsen-cell for the Ag evaporator by heating the crucible, (5) the Molecule Beam Evaporator (MBE) of the Co by heating the rod evaporant with the accelerated electron, and (6) is the He flow cryostat for cooling down the sample to 100K by injecting liquid nitrogen. The apparatus for the fabrication tip is shown in Figure 2-2.. Figure 2-1. Picture of the UHV-STM system. (1) is the STM chamber, (2) RFA-LEED, (3) the sputtering ion gun, (4) the Ag evaporator, (5) the Co evaporator, and (6) is the nitrogen storage cryostat. 15.

(25) Tip holder. 2M KOH Figure 2-2. Illustration of the apparatus for fabrication of the tip in the UHV-STM. A tungsten wire with a 0.35 mm diameter is mounted into the tip holder and then immersed in the 2.0 M KOH solution. The electrode of the tungsten wire is connected to positive potential, and that of a stainless steel wire serves as negative potential. The reaction is Positive potential: W(s) +8OH -(aq)  WO -24 (aq) +4H 2O (aq) +6e Negative potential: 6H 2O (aq) +6e -  3H 2(q) +6OH Net reaction : W(s) +2OH -(aq)  2H 2O (aq)  WO 4-2 (aq)  3H 2(q). The initial applied voltage is 4 V and then decreased to 2 V to reduce the reaction speed at the DC mode. At the moment of the lower part of the tungsten wire dropping down, the voltage is cut off immediately to avoid etching the upper part of the tungsten wire which is used to probe. The magnifying glass is helping to observe the cut-off time. The clean Si(111)-7×7 surface is prepared by flashing to about 1200oC for several times and then annealing to 700oC for seven hours. The clean Ge(111)-c(2×8) surface was prepared by repeated cycles of ion bombardment (0.7 keV, Ar+ ) and annealing at 800oC for ten hours. In chapter 3, the Co is directly absorbed on the clean Si(111)-7×7 and Ge(111)-c(2×8) surfaces and then annealed from RT to 500oC to study the growth behaviors of the absorbed Co at different annealing temperatures. In chapter 4, the Ag-√3 × √3 surface was prepared before evaporating Co atoms on the surfaces. Therefore, depositing Ag of about 1 ML leads to a √3 × √3 reconstruction on the Si(111) and the Ge(111) surfaces by annealing to 500oC for 10 min. All surface 16.

(26) reconstructions are determined by LEED and STM. After that, we deposited an amount of Co atoms on the Ag/Si(111)- √3 × √3 and the Ag/Ge(111)-√3 × √3 surfaces at room temperature and annealed both surfaces for 1 hour from room temperature to 500oC in steps of 100oC. Subsequently, the growth behavior of the Co on the Ag-√3 × √3 surfaces at different annealing temperatures was studied. The Co coverage is determined from the flux monitor. As annealing temperature is above 500oC, the temperature is recorded by an infrared thermometer and the sample is annealed by the direct current controller. Below 500oC, the sample is annealed by a heat plate constructed by a loop of the tungsten wire and the temperature is detected by the K-type thermocouple. All samples were cooled with a rate of 2oC/min to room temperature. All STM measurements were performed at room temperature (RT) as well as all LEED experiments. All STM experiments are in the constant current operating mode and the tunneling current was fixed at 0.12 nA. 2.2 Introduction to the Si(111)-7×7 and the Ge(111)-c(2×8) surfaces The STM image of the Si(111)-7×7 surface is shown in Figure 2-3(a) and (b) for positive and negative potentials, respectively.. (a). (c). (b ). Figure 2-3. (a) and (b) show the Si(111)-7×7 surface at positive (+1.7 V) and negative (-1.6 V) biases, respectively. The image sizes of (a) and (b) are 6×6 nm2. (c) is DAS model of the Si(111)-7×7 surface.57. 17.

(27) The clean Si(111)-7×7 surface is prepared by high temperature annealing from ℃. ℃. 2 × 1 ⎯⎯ 5 × 5 ⎯⎯ 7 × 7.1 The Si(111)-7  7 is the thermal stable equilibrium structure at the end of the reaction pathways. The surface structure is observed with 12 adatoms and large holes at the corners of the Si(111)-7×7 unit cells from STM image. Further, the unit cell is divided into two equilateral triangles called the faulted half unit cells (FHUC) and unfaulted half unit cells (UHUC) resulting in unbalanced density of electron states at the special tunneling conditions. The Si(111)-7×7 reconstruction is described by the model of the Dimer-Adaom-Stacking-Fault (DAS) family of reconstructions referred to as the 7×7 DAS model as shown in the model of Figure 2-3.46,47 The DAS model of the 7×7 surface includes 19 broken bonds (12 on adatoms, 6 on rest-atoms, and 1 in the corner-hole) in a 7×7 unit cell. Before forming the 7×7 reconstruction, there are 49 broken bonds exiting in the same area for the Si(111)-1  1 surface and hence the Si(111)-7×7 surface efficiently decreases the surface dangling bonds to form the stable structure. The unfaulted domain means that the atomic stacking is following the sequence of A-B-C-A-B-C and so on in the direction normal to the surface as indicated by the half cell on the right side of the model, but the faulted domain is not allowing the stacking arrangement in the left half cell of the model. The unbalanced contrast between the faulted and unfaulted domains in the same unit cell is due to the DOS feature that the faulted stacking domain is associated with the layer resonances between the vacuum-solid interface and the last bulk-like layer of the crystal.48,49 Only as the bias is near the resonant bias, the contrast of the two halves of the unit cell become uniform. Therefore, the 2 nm height asymmetry between the two halves of the unit cell is confirmed. Further, a charge transfer from adatom dangling-bonds to rest-atom dangling-bonds is observed resulting in an only chlorine-terminated rest-adatom layer without adatoms.50-52 Because the melting point of the germanium is lower than that of the silicon, the clean Ge(111) surface has to be bombarded by argon ions to remove the contaminants and then annealed to 800oC to form the thermally stable c(2×8) reconstruction. The STM image of the Ge(111)-c(2×8) surface is shown in Figure 2-4(a). At negative bias potential, the rest atoms are observed in Figure 2-4(b). For a unit cell, there are four adatoms and rest atoms. The Ge(111)-c(2×8) surface has a larger charge transfer resulting in the larger band gap 1.2 eV compared to 0.45 eV of the Si(111)-7×7 surface.1 The reason is that the unit cell of the Si(111)-7×7 surface has 12 adatoms, but only 6 rest-atoms and thus incomplete charge transfers.1,53-55 Further, the bulk structures of the germanium and silicon are the same, but result in different ground-state reconstructions c(2×8) and 7×7 of the annealed Ge(111) and Si(111) surfaces, respectively. The decisive factor is the covalent radius.1 A lower atomic 18.

(28) number leads to a compression of the surface layer.1 The germanium has a covalent radius 4% larger than the silicon resulting in the greater compressive stress in a substitutional Ge/Si(111)-1×1 surface compared to Si(111)-1×1 surface.56. (a). (c). (b ). Figure 2-4. (a) and (b) show the Ge(111)-c(2×8) surface at positive (+1 V) and negative (-0.7 V) biases, respectively. The image sizes of (a) and (b) are 10×10 nm2. (c) is the model of the Ge(111)-c(2×8) surface.58 Figures 2-5(a) and (b) show the LEED patterns of the Si(111)-7 × 7 and Ge(111)-c(2×8) surfaces, respectively.. (a). (b ). Figure 2-5. The LEED patterns of (a) and (b) represent periodic surface structures of the Si(111)-7×7 and Ge(111)-c(2×8) surfaces at 67 eV and 58 eV, respectively.58,59. 19.

(29) 2.3 The formation of the Ag- √ × √ Ge(111)-c(2×8) surfaces. layer on the Si(111)-7 × 7 and the. The Ag-√3 × √3 layer on the Si(111) and the Ge(111) surfaces can be classified as the honeycomb chain trimer (HCT) model60,61, though the substrate structures of the Si(111)-7×7 and the Ge(111)-c(2×8) reconstructions are different. The HCT model is constructed by the Ag and Ge or Si trimers as shown in Figure 2-6(a). Every Si or Ge trimer (thin line) is surrounded by the six Ag atoms which form a hexagon called the Ag-hexagon. In the center of the neighboring three Ag-hexagons, there is one Ag trimer (thick line).62 A honeycomb √3 × √3 unit cell is outlined in the dashed line. The topographies of the Ag/Ge(111)-√3 × √3 and Ag/Si(111)-√3 × √3 surfaces are very similar in the range of several unit cells. However, the STM images of the larger regions are rather different as shown in Figure 2-6(b) and (d). We can find larger smooth √3 × √3 zones on the Ag/Ge(111) surface than on the Ag/Si(111) surface. The Ag/Si(111)-√3 × √3 surface is divided into the upper and the lower layers. When the bias is switched to negative voltage, the bright regions in Figure 2-6(c) disappear as compared to Figure 2-6(b). Therefore, we can make sure that the bright regions don’t come from contaminants. If we want to confirm Co clusters, STM images are measured at negative bias to avoid confusing Co atoms with the bright regions because the bright regions appear at positive bias. Further, Ag-√3 × √3 structure on the Si(111)-7×7 and Ge(111)-c(2×2) surfaces make all dangling bonds saturated by the silver termination.39 Therefore, the Ag buffer layer is a good template to avoid the Co-Si and the Co-Ge alloy.. (a ). (b). (c). Figure 2-6. (a) HCT model including the √3 × √3 unit cell (the dashed line) and the trimers of the Ag (thick line) and the Si (Ge) with the thin line. (b): the Ag/Si(111)-√3 × √3 surface (Ubias = -1.5 V). (c): the Ag/Si(111)-√3 × √3 surface (Ubias = +1.5 V). (d): the Ag/Ge(111)- √3 × √3 surface (Ubias = +1.2 V). Image sizes are 20.0×20.0 nm2.63. (d) 20.

(30) Chapter 3 Co on the pure Si(111)-7×7 and Ge(111)-c(2  8) surfaces dependent on annealing temperatures 3.1 Submonolayer Co on the Si(111)-7×7 surface Figure 3-1 shows the growth behavior of submonlayer Co dependent on annealing temperatures on the Si(111)-7×7 surface. As the Co is evaporated on the surface, the Co-Ge compound is immediately formed resulting in the liked defect regions at RT.64 With increasing annealing temperature the occupied surface area of the Si(111)-7×7 structure decreasing. Up to an annealing thermal valve of 400oC, the surface exhibits the disordered structure. At low annealing temperatures, not all Co clusters are diffused into the Si substrate and maintained on the surface since the unbroken Si(111)-7×7 structure is shown. No more the Si(111)-7×7 structure on the surface reflects that all Co atoms diffuse into the Si substrate and then Si atoms are segregated on the topmost layer after the high temperature annealing 400oC.65 Because the 2×2 and √7 × √7 surface structure of the CoSi2 are formed at last after 550oC annealing, both structures are not observed in Figure 3-1.64,66. Figure 3-1. Serial images of submonolayer Co absorbed on the Si(111)-7×7 surface from RT to 400oC. The bias and size of all images are +0.4 V and 50×50 nm2, respectively.67,75 21.

(31) 3.2 Submonolayer Co on the Ge(111)-c(2×8) surface A submonolayer Co absorbed on the Ge(111)-c(2×8) surface is shown in Figure 3-2. From (a) to (f), some higher islands are seen in large scale. The occupied surface areas of the higher islands do not vary with annealing temperatures and hence these islands are considered to be the contaminant such as the carbon islands which were not removed by the ion bombardment. There is no apparent changing in the large scale regions even after annealing to 900K. Next, the image size is zoomed in and shows the new surface phenomena in the STM images as shown in Figures 3-3 and 3-4. Figure 3-3 shows the Co-Ge compounds resulting in the darker regions after annealing to 600K. Figure 3-2. Serial images of submonolayer Co absorbed on the Ge(111)-c(2 × 8) surface from 400K to 900K. (a)-(f): Ubias = +1.5 V and 300×300 nm2.68. 22.

(32) Figure 3-3. The image (Ubias = +1.5 V and 40×40 nm2) shows that the defects are produced from the Co-Ge alloy after a 600K annealing. The insert image is magnified from the black square region.68 The darker regions also are observed by annealing to 500K. According to the magnetic property of the Co/Ge(111) surface disappears after annealing above 400K, the darker region structure is consider to be the stable phase of the Co-Ge alloy.69,70 The distance a of the neighboring atoms is calculated to be about 0.745 nm which is near to 0.764 nm i.e. the lattice constant of the Co5Ge7 alloy as shown in the insert image of Figure 3-3.71 The darker region with the structure of the Co-Ge compound is clearly seen, but the defect structure of the Co-Si compounds is unclear due to the defects of the Co-Si formed locally without condensation. After annealing to 700K, Co atoms form the special surface structure as indicated by the black circle in Figures 3-4(a) and (b). Here, the darker regions of the Co5Ge7 alloy is still found and the probability to observe the special surface structure is very lower compared to the darker region and other phases like the cloud shaped islands. The detailed structure of the special structure is shown in Figure 3-4(b). This structure plays an important role to reflect the growth behavior of the Co on the Ag/Ge(111)-√3 × √3 surface. More discussion about the structure is in sections 4.2 and 4.3. 23.

(33) (a). (b). Figure 3-4 (a) shows special surface structure indicated by the black circle after annealing to 700K and (b) represents this surface structure in detail. (a): Ubias = +1.5 V and 80×80 nm2. (b): Ubias = +1.0 V and 15×15 nm2.68 3.3 Submonolayer Co on the Si(111)-7×7 surface at low temperature In order to confirm the dark region whether the alloy forms, the behavior of Co atoms on the Si(111)-7×7 surface at low temperatures is studied.72 After RT deposition of the Co, the STM images of Si (111)-7×7 surfaces show many dark spots as shown in Figure 3-5(b) compared to Figure 3-5(a).. (a). (b). Figure 3-5. STM images showing the clean Si(111)-7×7 surface (a) and the impurity Si(111)-7×7 surface (b) after Co deposited at room temperature. Size of images: 40×40nm2, and sample bias: -1.5 V.57 The defects should be a result from the reaction of the metallic Co atoms with the Si substrate. If a perfect Si(111)-7×7 surface can be prepared, we may proceed with the observations after the Co deposition. However a perfect Si(111)-7×7 surface is 24.

(34) hard to prepare, in fact, there are still 8 defects per 100 nm2 appearing on the clean Si(111)-7×7 surface, even though we do our best to treat the Si sample. To avoid confusing Co atoms causing the dark regions with the intrinsic defects of the Si substrate, the observations and statistics for the clean surface as shown in Case 1 of Table 3-1 where each site symbol is corresponds to the illustration of Figure 3-7. There are slight differences between the Si defects and the Co induced the dark regions as shown in Figure 3-6. The neighboring adatoms of the Co induced the dark spots are darker than those of the Si defects.. Figure 3-6. STM images comparing of the dark spots resulting from intrinsic Si defects and the Co-Si compounds. (a) Si corner defect, (b) Co induced corner dark spot, (c) Si center defect, (d) Co induced center dark spot. Size of images: 6×6 nm2, and sample bias: -2 V.73 Table 3-1. The table states the number of the defects or defect-like dots appearing at different sites for different cases. The site symbols are corresponded to Figure 3-7.73. 25.

(35) The unit cell of the Si(111)-7×7 reconstruction is divided into two different halves. One half of the cell contains a stacking fault (FHUC), the other one is unfaulted (UHUC).46 It has been shown that the FHUC is brighter than the UHUC in the STM images at negative bias due to different electronic states. Therefore, we divide the possible sites for the Co-Si reaction as shown in Figure 3-7. The differences of the defect appearances can be compared conveniently and summarized in Table 3-1. The intrinsic defects of the Si(111) surface appear at the UHUC more often. About 70% defects are found at the UHUC independent of whether the substrate is at RT (case 1) or low temperature 100K (case 4). After the RT deposition of the Co with a 0.15 ML (case 2), the defect-like spots increase to over 15 per 100 nm2 significantly. The addition of these defect-like spots should result from the reaction between Co atoms and the Si substrate. The estimated numbers of the Co induced spots are recorded in the case 3. Interestingly, this kind of the spots appears at the center sites of the UHUC with a high probability relative to other sites.. Figure 3-7. DAS model for Si(111)-7×7 in which we indicate different sites that we count corresponding to Table 3-1.73 Why do the Co induced the dark spots appear at the center sites of the UHUC more often? There are three possible explanations. The first, Co atoms prefer to stick at the site of impact, i.e., the sticking coefficient at the center sites of the UHUC is the highest. The second, Co atoms prefer to react with the Si at the sites, i.e., the activation energy of the Co-Si compound at the center sites of the UHUC is the lowest. The last, Co atoms prefer to diffuse to the sites and are trapped, i.e. the binding energy of the center sites of the UHUC is the greatest. To check these postulates, we performed the experiments at low temperatures at which the Co do not react with the Si. After deposition of a 0.02 ML Co at 100K (case 5), the bright dots which result from the metallic Co atoms instead of dark spots are observed. The distribution of the bright dots is rather similar for the FHUC and the UHUC, i.e. first postulate can be removed. 26.

(36) As the temperature increases slowly by not refilling the liquid nitrogen, the phenomenon from the bright dots to the dark dots in the STM images is observed. Of course, not all bright dots transfer to dark at the same temperature range. Especially for some bigger bright dots which may be formed by Co islands, they will react with the Si at higher temperatures. Figure 3-8 is an example of the STM images from 126 K to 130 K where we can find that some bright dots change to the defect-like dark spots as indicated by the white circles. As the temperature is increasing to the higher temperature, such as 135 K or 140 K, more and brighter dots change to dark spots. Unfortunately, to trace the changes in the images is difficult. This change from the bright to the dark dots cannot be found at the lower temperatures below 126K. Therefore, we can conclude that the reaction of Co and Si atoms starts to occur at the temperature range from 126 to 130K.. Figure 3-8. STM images showing Co adatoms change from the bright dots to the defect-like dark dots at the temperature range between 126 and 130 K.73. 27.

(37) At the initial stages of the Co-Si formation, a site preference phenomenon is not obvious. In Figure 3-8, the three events occur at an unfaulted corner site, a faulted corner site, and an unfaulted center site. The summarized statistic of table 3-1 also indicates that the reaction sites are rather random. The activation energy of reaction to Co silicide at different sites should be similar and hence the second postulate can be ruled out. The last possibility is that the center sites of the UHUC have the greatest binding energy. A previous study suggested that Co mobility on the surface is significantly less than in the Si bulk.74 We also find in Figure 3-8 that many bright dots keep their locations as the temperature increases from 126 to 130K. This means that Co atoms and islands cannot move in this temperature range. The surface diffusion energy is higher than the activation energy of the Co-Si compounds. Co atoms and islands diffuse around on the Si(111)-7× 7 surface if the temperature is high enough to overcome the diffusion barriers. Co atoms stay at the center sites of the UHUC for the longer time than at other sites because Co atoms are bound by the center sites of the UHUC stronger. Therefore, the reaction of the Co-Si compounds at the center sites of the UHUC is more probable as shown in the case 1 of Table 3-1 that we observe after the Co deposited at room temperature. In summary, the onset temperature of the Co-Si compound formation is found between 126 and 130K. The effect of the Si atoms segregation on top of Co layer is certified after the higher annealing temperature by the Co/Si(111) case. The surface structure of the Co5Ge7 alloy is confirmed in the darker regions from the STM images. Despite the lower observed probability of the special structure of the Co on the Ge(111)-c(2×8) surface, the effect of the structure formation will reflect the growth behavior of the Co on the Ag/Ge(111)-√3 × √3 surface to form a periodic surface structure rather than on the Ag/Si(111)-√3 × √3 surface at low Co coverage.. 28.

(38) Chapter 4 Co on the Ag/Si(111)-√ × √ and the Ag/Ge(111)-√ × √ surfaces dependent on annealing temperatures The sections 4.1 and 4.2 describe the investigation of the Co absorbed on the Ag/Si(111)-√3 × √3 and the Ag/Ge(111)-√3 × √3 surfaces with different cobalt coverages and annealing temperatures from RT to 500oC. The comparison of the Co on the Ag/Si(111)-√3 × √3 and the Ag/Ge(111)-√3 × √3 surfaces is represented in section 4.3. The last section 4.4 represents the Co on the flat Ag/Si(111)-1×1 surface without the Ag-√3 × √3 reconstruction to reduce the influence of the unsaturated states. In this chapter some STM images are repeated shown in different sections because it is easy to read without going back to see the figures. 4.1 Co on the Ag/Si(111)-√ × √ surface The LEED pattern and STM images of the Ag/Si(111)-√3 × √3 surface are shown in Figure 4-1(a) and (b) respectively. Figure 4-1(b) shows that the whole Ag-√3 × √3 surface covered with different height layers. One level can be seen as the numerous large holes possessing a bilayer depth and a flat bottom. The other level is irregular contour terraces with a shapeless distribution. This phenomenon is the same as the previous research, which pointed out that Ag atoms replace the Si atoms during annealing process.76 This phenomenon reflects that Ag atoms replace the original Si adatoms and make the substituted Si atoms rise up to product the new higher Si-layers. At the same time the substituted Si-layers become the vacancy islands. Finally, Ag atoms construct the √3 × √3 structure on the higher Si-layers and lower Si-layers. Thus, the Ag/Si(111)-√3 × √3 surface is divided into the upper and the lower √3 × √3 layers relative to the original Si-adatom layer. Furthermore, we can observe that the mechanism results in a two-level Ag-√3 × √3 layer with winding and long boundaries. We label the upper level layer of the Ag-√3 × √3 reconstruction as the u-√3 islands, and the lower level layer as the l-√3 surface. The edges between the u- √3 × √3 islands are considered to be the boundaries. The necks mean the connection between neighboring domains at the same level. The image contrast of the boundaries and necks depends on the sample biases as shown in Figure 4-1(c) to (f). In a comparison of Figure 4-1(c) and (d), the boundaries show up especially bright at positive bias. The necks are also brighter at positive bias as shown in Figure 4-1(e). Further, Figure 4-1(d) shows that the necks are darker at negative bias. Positive bias indicates the filled states, while negative bias represents the empty states. The changes in the STM image contrast means that there are the unsaturated states or inhomogeneous electronic density in the boundaries and the necks.77,78 Further, the boundaries are major along the 112 and [110] directions.77 The detail surface structures of the boundaries and the necks are discussing later. In order to avoid confusing the unsaturated states with Co atoms, the STM images for the Co 29.

(39) observation are measured at negative biases.79,80. Figure 4-1. (a) LEED pattern of the Ag/Si(111)-√3 × √3 structure (E = 58e V). (b) A large scale image of the Ag/Si(111)-√3 × √3 surface (100×100 nm2, Ubias = -2 V). (c) and (d) show the boundaries measured by +1.5 V and -1.5 V biases (20×20 nm2). (e) and (d) show the necks measured by +1.0 V and -1.0 V biases s (20×20 nm2).59,81 The boundary models for the Ag/Si(111)-√3 × √3 surface are built in Figure 4-2. Based on the HCT model, the width side of the rectangle regions of (a) and (b) are along the 112 and [110] directions indicated by corresponding black arrows, respectively, and the short lines between the silicon and sliver atoms represent the bonds. As the boundaries are formed due to the replacement effect, the rectangle regions are separated to be the higher and lower Ag-√3 × √3 layers, and hence the several bonds are broken resulting in the unsaturated bonds from the silicon atoms. 30.

(40) Figure 4-2. (a)-(d) are the HCT models of the Ag/Si(111)-√3 × √3 surfaces. The width side of the rectangle regions of (a) and (b) are along 112 and [110] directions indicated by corresponding black arrows and the lines represent the bonds. (c) and (d) show the possible broken bonds as the boundary formations on the regions. The detail structure of the necks is shown in Figure 4-3(a). The positions of the necks appear brighter than others at positive bias due to the dangling bonds. The line leaped over the both sides of the Ag-√3 × √3 domains is to identify the orientations and the position of the brighter spot is indicated by the circle in Figure 4-3(a). According to the relative positions between the line and the circle, the structure model of the necks is drawn in Figure 4-3(b). It is noticed that the brighter spot is fixed at the silicon atoms without bonding to the sliver atoms. Therefore, the brighter spots are the result of the dangling bonds of the silicon atoms. 31.

(41) Further, the intensity of the density of electron states is larger at the unoccupied states than at the occupied states and hence the boundaries and necks are brighter at positive bias than at negative bias as shown in Figure 4-1(c)-(f).1 Compared to the Ag/Ge(111)-√3 × √3 surface, there are no unsaturated states exiting. For instance Figure 2-6(d) measured at positive bias, the STM image of the Ag-√3 × √3 layer does not have the voltage dependence effect. Therefore, the Ag-√3 × √3 layer is formed directly on the Ge(111) surface without replacing effect from the silver atoms, and hence the complex boundaries and necks cannot be observed on the Ag/Ge(111)-√3 × √3 surface.. (a). (b). Figure 4-3. (a) shows the detail surface structure of the necks (Ubias = +1.0 V, and 10×10 nm2). (b) is the neck structure model. What do the unsaturated states at the boundaries and necks influence for the cobalt growth behavior on the Ag/Si(111)-√3 × √3 surface? Figures 4-4(a) and (b) show the STM images of a 0.2 ML and a 0.5 ML Co on the Ag/Si(111)-√3 × √3 surface at RT. When Co coverage is lower, Co clusters preferentially to absorb on the necks, and then rearrange along the necks to form the chain like formation. On the contrary, almost all of the u-√3 island boundaries distribute Co clusters (0.5 ML) in a chain like stacking. In fact, the numbers of the free neck positions are fewer than that of the free boundaries positions on the surface. As all neck positions are occupied by Co clusters, the boundary positions support the external Co clusters good residence for the condition of the high Co coverage. The next question is whether the electronic effect impacts the nucleation behavior of Co atoms or not and which positions are more stable for Co clusters (islands)? The answer is confirmed from the thermal evolution of a 0.1 ML Co on the Ag/Si(111)-√3 × √3 surface at the next paragraph.. 32.

(42) Figure 4-4. (a) Co/Ag/Si(111) surface with a 0.1 ML Co at Ubias = -1.0 V. (b) Co/Ag/Si(111) surface with a 0.5 ML Co at Ubias = -2.0 V at room temperature. Image sizes are 50×50 nm2. Thermal evolution of a 0.02 ML Co on the Ag/Si(111)-√3 × √3 surface from RT to 500oC is shown in Figure 4-5. As annealing temperature is increasing, the average size of Co clusters becomes bigger and higher. However, the growth of Co clusters mainly reflects itself in height. It is apparent that the occupied areas of Co clusters do not change much form 400oC to 500oC. Because the surfaces free energy that of the Ag γ = 1.30 J⁄m only half that of the Co (γ = 2.71 J⁄m ) 80,82, the occupied areas of Co clusters are rather maintained to avoid a surface free energy increasing. Oppositely, the average cluster height increases form 0.06  0.01 nm of Figure 4-5(d) to 0.21  0.01 nm of Figure 4-5(f). The main critical temperature is 300oC for the height of Co clusters increasing. Co clusters start to coalesce into the 3D islands after annealing to 400oC. After annealing to 500oC, almost all Co islands absorb on the boundaries instead of the necks or the terraces of the Ag-√3 × √3 surface. The stable Co shapes are 3D islands on the boundaries. Therefore, the density of states near the Fermi level of the boundaries and necks are the reason for Co clusters (islands) favoring growth around the boundaries and necks. In addition, the nearest neighboring distances between atoms of bulk Co and bulk Ag are 0.250, and 0.289 nm, respectively, resulting in the large mismatch of 13.1% between the Co and Ag lattices.83,84 The great surface strain may be the key factor for the surface nucleation and the surface diffusion in this heteroepitaxial system, and results in the 3D clusters growth mode for the Co deposited on the Ag/Si(111)-√3 × √3 surface.. 33.

(43) Figure 4-5 (a)--(f) Temperature dependent evolution of a Co deposit on the Ag/Si(111)- √3 × √3 surface from room temperature to 500oC (Ubias = +2.0 V, 0.02 ML Co). The inset profile of (f) shows 3D islands along the arrow direction. Image sizes are 100.0×100.0 nm2.81. 100oC. RT. 200oC. 300o 0.65. 0.50. 0.35 0. 400oC. 2. 4. 6. 8. 500o. Figure 4-6(a) shows the statistics of the average height of Co clusters (islands) on the surface and 4-6(b) shows the ratio of the areas of Co clusters (islands) occupying the boundaries and terraces from RT to 500oC. For annealing temperatures from RT to 300oC, the average height merely increases from 0.04 nm to 0.06 nm. Subsequently, the average height strongly rises to 0.11 nm and 0.22 nm after annealing to 400oC and 500oC, respectively. Co clusters start to nucleate to form the 3D islands. It is understood that the diffusion should be prior to the nucleation. According to Figure 4-6(b), the diffusion of Co clusters occurs from the terraces to the boundaries after annealing to 200oC. However, the ratio area hardly increases, apparently only from a 6.2 to a 6.5 after annealing to 300oC. The reason is that the boundaries are already filled with Co clusters. Therefore, the ratio almost doesn’t change. After annealing up to 500oC, a large amount of Co clusters prefers to be located on the boundaries rather than the necks. At the same time, the ratio increases to 14.0. This means that Co clusters readily diffuse and coalesce to form the 3D islands on the boundaries which also can be confirmed by the average height. Coalescence provides more free boundaries positions to accommodate more Co islands on the boundaries from the necks resulting in the ratio increasing.. 34.

(44) Figure 4-6. (a) The average height of Co clusters (islands) on the whole surface and (b) the ratio of the areas occupied by Co clusters (islands) on the boundaries and the terraces as a function of a temperature from RT to 500oC. All statistics are according to an analysis of the STM images of a 0.02 ML Co on the Ag/Si(111)-√3 × √3 surface. The illustration of Figure 4-6 is shown in Figure 4-7. At the beginning, Co clusters are distributed on the boundaries and necks. After annealing to 200oC, Co clusters at the necks and terraces diffuse to the boundary positions. And then Co clusters coalesce and become bigger and larger after 400oC annealing.. Figure 4-7. The illustration describes the mechanism of Co diffusion and coalescence dependent on annealing temperatures. When Co coverage increase to a 1.8 ML on the Ag/Si(111)-√3 × √3 surface and sample is annealed to 400oC, small local regions of the period structure appear as shown in Figure 4-8(a). In order to determinate this structure, the Co/Ag/Si(111) surface is investigated by the LEED and the STM. The LEED pattern of Figure 4-8(b) includes √3 × √3 and ring-like diffraction features.. 35.