Department of Physics, National Taiwan Normal University Doctoral Thesis
The study of silicene and iron growth on semiconductor based substrate.
Supervisors : Tsu-Yi Fu, Wen-Chin Lin
致謝辭
從2009 年開始進入師大擔任 IYPT 助理到系上的實驗助教，隨後也跟著進
入博士班就讀，不知不覺就在這裡經歷了七年的光陰，感謝在擔任助理與實驗助 教時，陳俊明助教在工作上協助我許多。賈至達教授鼓勵我就讀博士班，還有同 儕呂杰翰博士的互相扶持。後來也進入傅祖怡教授的原子解析度顯微術實驗室 (ARML)繼續進行我的博士研究，在剛進入 ARML 時，林榮君與黃筱嵐博士也給 了我很多研究方向建議。同時，也感謝林文欽教授讓我在他的實驗室同時進行其
他方面的研究，並給予我經濟上的支助。還有吳啟彬教授帶著我使用STM 給我
許多寶貴的經驗。在博士研究上傅祖怡與林文欽教授也給了我許多指導，讓我有 許多機會前往外國參加學術研討會開闊視野。在這段期間，在日本東大的林俊良 教授也給了我許多的協助，對於我許多學術上的問題，都能夠詳細的回覆，尤其 是論文寫作方面，很細心的幫助我修改許多錯誤與提供許多寶貴的意見。在系上 也感謝駱芳鈺與邱顯智教授適時的關心。也感謝蘇維彬教授抽空前來擔任我的口 試委員。
在進入ARML 實驗室後，碩士生周明寬讓我開始能夠熟悉 STM 的使用，謝
伯宜開始進行矽烯的實驗，各方面的儀器參數校正多虧了有伯宜讓實驗能夠很快 的步上軌道，你對實驗的謹慎與細心讓我刮目相看，佳原與泰龍不眠不休的進行 實驗，使得儀器能夠取到非常好的影像數據。而新血盧奕宏、曾崇瑋與陳弘儒
使得實驗能夠繼續順利的進行。另外也感謝 FIM 的學弟妹，張琬喻、黃詮友、
石婕廷、蕭靜瑜與陳藝芳生活上的關心。也感謝在林文欽教授的實驗室的學弟妹 們，博士班同學蔡承叡經驗上的交流，許詮喆與張博鈞協助我儀器上的架設，碩 士班徐凱霖協助實驗的進行，也感謝實驗室其他成員黃瀚元、陳宜樺、陳昱全、
張昀穎各方面的協助。另外也感謝何依黛平常的學術討論。
最後也感謝大學時期的朋友們的鼓勵，楷雯、雅婷、穆仁、簡暐、孔鼎、郁 凱、幸宜、伯霖長期的鼓勵。五年的博士班生活因為有大家而精彩，共勉之。
Contents
Abstract ... 4
Chapter 1 Introduction ... 6
1.1. Si(111)-(7×7) ... 7
1.2. Ge(111)-c(2×8) ... 8
1.3. Si-Ag and Si-Ge reconstructed surface ... 10
1.4. Silicene ... 12
1.5. MoS2 ... 15
1.6. Iron growth ... 16
1.6.1. Iron silicide... 16
1.6.2. Iron germanide ... 16
1.6.3. Iron/MoS2 interface ... 17
1.6.4. Fe/Pd/C60 growth ... 18
Chapter 2 Experimental details ... 20
2.1. Scanning Tunneling Microscopy (STM) ... 20
2.2. X-ray Photoemission Spectrum (XPS) ... 22
2.3. Magneto-Optical Kerr effect (MOKE) ... 23
2.4. The experimental setup and process ... 25
2.4.1. Instruments ... 25
2.4.2. Substrate preparing ... 26
2.4.3. Experiment process ... 27
Chapter 3 Result and discuss ... 29
3.1. Silicene growth ... 29
3.1.1. Si grow on Ag-Si (√3×√3)𝑅30° reconstructed surface ... 29
3.1.2. Si grow on Ag-Ge (√3×√3)𝑅30° reconstructed surface ... 34
3.1.3. Si grow on Ag(111)/Si(111) substrate ... 38
3.2. Fe growth ... 51
3.2.1. Fe/Si(111) substrate ... 51
Chapter 4 Conclusion ... 97 Reference ... 100
Abstract
Part I Silicene growth
Using scanning tunneling microscopy (STM), we studied the formation of Si monolayer grown on (√3×√3)R30° Ag-Ge(111) and Ag-Si(111) reconstructed surface, respectively. Thereafter, we also increase Ag thickness, where is formed 6~12 ML Ag(111) layer, to grow silicene. On √3Ag-Si(111), deposited Si exchange with Ag atom to form the new √3Ag-Si(111) islands without forming new Si monolayer. On √3Ag-Ge(111), the Ag-Si exchange behavior is suppressed by stable bonding of Ag and below Ge(111) substrate. We measure the isolated Si monolayer with mixing √3×√3 and 2×2 superstructures on the top layer. From the demonstrated ball model, the Si monolayer found in this study is very possible to consist of honeycomb structure. On Ag(111)/Si(111) surface, we have measured classical silicene superstructures, such as 4×4, √13×√13-I, 2√3×2√3. Unlike growing on the single crystal Ag(111), where the discontinuous silicene sheet formed between various superstructures, the continuous silicene sheet formed between various superstructures on 6~10 ML Ag(111)/Si(111) by Ag domain rotation and shift because of low Ag(111) unstable on Si(111) substrate.
Part II Iron growth
In this part, we investigate Fe growth on various substrate, which are Fe/Si(111), Fe/√3Ag/Si(111), Fe/Ge(111), Fe/√3Ag/Ge(111), Fe/MoS2, and Pd/(Fe/Pd/C60)x/Au/Al2O3 system. In Fe/Si(111), we measured the thermal evolution of Fe-silicide from γ- FeSi2 to β- FeSi2, and 𝛼-FeSi2 at the highest temperature by STM. In particular, the growth of β-FeSi2(011)//Si(111) is different with previous Fe-silicide studies. Besides, the isolated √3Ag-Si buffer layer causes the formation of 𝜀-FeSi without appearing on Si(111) substrate at the same temperature.
In Fe/Ge(111), the deposited Fe formed Fe-Ge clusters at RT. When Fe ratio is increased by increasing temperature, Fe-Ge clusters evolve into (2×2)
reconstruction by increasing the temperature and disappeared at 640 K. With Ag buffer layer, only nanoparticle growth occurred and 3D islands were formed early at 570 K.
In Fe/MoS2, we measured the surface morphology, magnetism and chemical states of Fe/MoS2 by STM, MOKE, and XPS, respectively. Fe deposition on the MoS2 substrate resulted in a nanoparticle array with the particle size ranged a few nanometer (~3 ± 1 𝑛𝑚). For low-coverage Fe deposition < 6 ML, nanoparticles were well-separated and long-range magnetic anisotropy was absent at room temperature. When the Fe coverage increased, in-plane magnetic anisotropy was observed and the magnetic coercivity increased monotonically. The depth-profiling XPS measurement of Pd/2 ML Fe/MoS2 also confirmed the dominance of the pure Fe state at the interface. The increase in Fe coverage changed the morphology from a nanoparticle array to a continuous coverage, leading to the onset of the ferromagnetic ordering and the transition from a continuous surface oxidation to a bilayer structure.
At last, we report on the hybridization-induced large X-ray magnetic circular dichroism (XMCD) of carbon in Pd-Fe-C60 composite thin films. The samples were prepared by repeating sequential deposition of C60, Fe and Pd for five times on Au/Al2O3(0001) substrate in an ultrahigh vacuum (UHV) chamber. The Pd-Fe-C60 composite thin films were investigated by MOKE, XMCD, and Raman spectroscopy. The composite thin films revealed in-plane anisotropy. After annealing the sample at 527 K, the Kerr signal became weak and the magnetic coercivity was decreased. Then, considerable XMCD signal was observed at the carbon K-edge, but relatively small XMCD signal appeared at Fe L2,3-edge. In contrast to the XMCD spectrum of mixing transition metal and C60 system in previous study, we observed that the carbon will induce strong XMCD signal. These observations indicate the hybridization-induced magnetic moment in carbon and possible reduction of magnetization in Fe.
Keywords: silicene, iron, semiconductor, 2D material, ferromagetic
Chapter 1 Introduction
The spintronic material is revealing the application in nanotechnology. The combination of spintronic materials, such as magnetic transition metals, with semiconductor(SC) is crucial in technical advancement and warrants detailed study. In this thesis, we study a serious of Fe growth on semiconductor surface, such as Fe/Si(111), Fe/Ge(111) with(without) Ag buffer layer, and Fe/MoS2. In Si- and Ge-based semiconductor, the formation of Fe silicide and germanide with rich magnetic property can be the ferromagnetic (FM) electrode for spin injector1-3 and detector4 of the future spintronic device(Fig. 1.1). Besides, the interface between the FM and SC is also considered.5 In the latest decade, two-dimensional (2D) layered materials, such as graphene, MoS2 and silicene, have been shown to possess unique electronic and magnetic properties for future applicationstran6,7-11, which can be used as channel for spin current in spintronic device. Therefore, the combination of spintronic materials with 2D materials is crucial in technical advancement and warrants detailed study.
Thus, we investigate the growth mode and magnetic properties of FM electrode, and the behavior of FM/SC interface. Besides, we try to fabricate silicene on Si- and Ge-based semiconductor because of the mature Si based industry technology. The controllable FM electrode is also considered, where we combine Fe, Pd, and C60 as electrode to study the ferromagnetic contribution of each material. Next, we introduce the staffs we used in the experiments.
1.1. Si(111)-(7×7)
Silicon, the group Ⅳ semiconductor with 1.11 eV energy gap, is the diamond structure. Si atoms were combined together by covalent bonds with sp3 binding. When Si bulk was cleavage, the covalent bonds were partially broken along the cleavage surface. The phenomena of the surface atoms rearranging to decrease the surface energy are called the surface reconstruction. Si(111)-(7×7) is one of the surface reconstruction on Si(111) surface when annealing Si(111) at least at 650 K in the ultra-high vacuum environment. The surface forms 7×7 periodic structure can be measured by LEED shown in Fig. 1.2(a). After STM technology was invented, we can measure the real 7×7 atoms arrangement shown in Fig. 1.2(c). Fig. 1.2(d) and (e) are the top and side views of the Si(111)-(7×7) dimer ad-atom stacking fault (DAS) model by Takayanagi12 et al.. The DAS structure is composed sequentially from the top to bottom layer by ad-atom, reconstruction, and bulk layers. The bulk layer remains Si(111)-(1×1) structure. The reconstruction layer is symmetry along the dimer line. In the left site, the bottom Si atoms sticking on the top Si atoms of the bulk layer is called fault site. In the right, the configuration is the same as bulk structure is called un-fault side. By sp3 bonding breaking on the top of reconstruction layer and forming dangling bonds, the ad-atom layer rearranges into (2×2) periodic in a triangular regime with 6 ad-atoms in each site. In STM images shown in Fig. 1.2(b) and (c), the protrusions are ad-atoms in DAS model. Besides, the fault side is slightly lighter than un-fault site in filled-state STM image because of package stacking between reconstruction and bulk layer.
Fig. 1.2 Si(111)-(7×7) (a)LEED pattern, and (b)filled-state and (c)empty-state STM images.
(d)Top view and (e) side view of DAS12 model.
1.2. Ge(111)-c(2×8)
In integrated circuit technology13, the earliest transistors were fabricated using Ge14. Germanium is also group Ⅳ semiconductor with 0.66 eV energy gap smaller than Silicon. Ge is also the diamond structure and the lattice constant a0 is 5.64 Å, which is slightly larger than Si. The common surface reconstruction of Ge(111) is Ge(111)-c(2×8) reconstruction. Fig. 1.3(a) is the LEED result of Ge(111)-c(2×8) reconstruction. Fig. 1.3(b) and (c) are the filled- and empty-states STM images, respectively. The rest atoms can be measured
c(2×8) unit cell is formed by local 2×2 and 2× 3 unit cell. Each c(2×8) unit cell contains 4 ad-atoms and rest atoms. Ge(111)-c(2×8) reconstruction is related simple surface reconstruction without dimers and stacking fault as Si(111)-(7×7).
Fig. 1.3 Ge(111)-c(2×8) (a)LEED pattern, and (b)filled-state and (c)empty-state STM images.
(d)Top view and (e) side view of the ball model.
1.3. Si-Ag and Si-Ge reconstructed surface
Here, we introduce another surface reconstruction which formed by pre-depositing low coverage of Ag (≤ 1 ML) on Ge(111)-c(2×8) or Si(111)-(7×7) reconstructed surface and annealing to variant temperatures. Fig. 1.4(a) and (b) are the Ag-Si(111)15 and Ag-Ge(111)16 reconstruction phase diagrams, respectively. The diagrams show the common ( 3× 3)𝑅30° reconstruction phase for both Ag-Si and Ag-Ge. The ( 3× 3)𝑅30° reconstruction presents the atomic flatness and significant surface parabolic band structure17, which presents two dimensional free electron gas (2DEG) property. The early model of ( 3× 3)𝑅30° Ag-Si atomic arrangement is honeycomb chained triangle (HCT) model18,19, where the top and side view are shown in Fig. 1.5(a) and (b), respectively. The HCT structure is made up by three sub-layer: bottom bulk, Si trimer, and top Ag trimer. Si atoms formed the Si trimer on the bulk Si(111), and each Si atom of Si trimer combines with one Ag atom by ionic covalent bond to form Ag trimer, which is 0.73 Å higher than Si trimer. In the 3 unit cell, two Ag trimers mirror-symmetry arranged. The 1 ML ( 3× 3)𝑅30° Ag-Si contain 2/3 ML Ag. Fig. 1.5(c) is the STM image showing the 3 honeycomb structure.
In our study, 3 Ag-Ge20 is also considered as HCT model.
Fig. 1.4 (a)Ag-Si(111)15 and (b) Ag-Ge(111)16 surface reconstruction phase diagrams.
Fig. 1.5 ( 3× 3)𝑅30° Ag-Si reconstruction: (a)top and (b)side view of the HCT18 ball model. (c)Empty-state STM image.
In Ag-Ge reconstruction diagram, 4×421,22 reconstruction could be formed by lower Ag coverage (≤ 0.7 ML) without existing in Ag-Si reconstruction phase.
Fig. 1.6(a) is the filled-state STM image of the Ag-Ge 4×4 reconstruction surface, which presenting the obviously two triangular structures with different brightness, which one side is Ag atoms, and another side is Ge atom. Fig.
1.6(b) is the ball model to show that Ag and Ge atoms in each triangular site.
Fig. 1.6 Ag-Ge 4×4 reconstruction: (a) filled state STM image and (b) top view of the ball model.22
1.4. Silicene
In 2004, graphene23, the two dimensional carbon allotrope, was first discovered by exfoliate graphite. Graphene was started to be researched for its abundant properties24. Silicon, the same group Ⅳ element is also expected the two dimensional honeycomb structure as graphene, is called silicene. The theoretical25-27 results of the energy versus lattice constant shown in Fig. 1.7 indicated that freestanding silicene prefer to constitute as the buckled structure, which energy is lower than the planar structure. Unlike the planar graphene with sp2 bonding between C atoms, the mixing of sp2-sp3 bonding28 between Si atoms causes the buckled silicene. Besides, silicon bulk is not like graphite, which could be used to produce graphene by exfoliating. Thus, Lelmi et al.29 attempt to use molecular beam epitaxy (MBE) method depositing Si atoms on Ag(111) substrate, where substrate temperature was held at 520 K in 2010.
The 2 3×2 3 𝑅30° superstructure was measured by STM and considered as silicene structure, shown in Fig. 1.8. Despite the result is controversial, it open a new experimental direction to fabricate the silicene. Thereafter, variant silicene superstructures by its flexibility of the buckled property, were observed on Ag(111) with distinct substrate temperature and coverage in succession.
The superstructure phase diagram30 is shown in Fig. 1.9. The STM combining with density function theory (DFT) simulation also presents the well-fitting consequent for silicene superstructures on Ag(111)31, which shown in Fig. 1.10.
In the electronic respect, free-standing silicene is also considered to exhibit the Dirac fermion band structure27 as graphene. The stronger spin orbital interaction of silicene is due by its buckled characteristic32. The theoretical studies show several fascinating electronic properties, such as superconductivity33, tunable energy gap34, giant magneto-resisitance(GMR)35, and topological insulating states36. Fig. 1.11 presents the simulated phase diagram of silicene electronic properties under varying applied electric and magnetic field. Very recently, the first free-standing silicene FET37 was fabricated in agreement with predictions of Dirac-like ambipolar charge transport38.
Fig. 1.7 Energy versus hexagonal lattice constant of 2D Si and Ge are calculated for various honeycomb structures27.
Fig. 1.8 The first fabricated silicene29 on Ag(111) with 2 3×2 3 𝑅30° superstructure. The STM images of (a)Ag(111) and (b)silicene. (c) Proposed ball model of silicene in (b).
Fig. 1.9 LEED pattern of phase evolution of silicene on Ag(111)30.
Fig. 1.10 The STM images and DFT simulation of 4×4 and 13× 13R13.9°
silicene superstructure.31
1.5. MoS2
Unlike other 2D material, such as silicene or graphene with pure Si and C atoms, molybdenum disulfide (MoS2) is made up with Mo and S atoms. Fig.
1.12(a) and (b) show the top and side view of MoS2 structure. The single MoS2
layer contains the top and down S atom sub-layers and the middle Mo atom sub-layer.The single MoS2 sheet can be exfoliate from the MoS2 bulk such as graphene exfoliating. The direct band gap is measured in the monolayer comparing to the indirect band gap in bulk structure39. In recent studies, the MoS2 nano-sheet have demonstrated magnetic response40, piezoelectricity41, and the strain engineering of a band structure8, revealing in nano-devices. Fig.
1.12(c) shows the empty-state STM image of the cleaved MoS2 surface in our laboratory. The lighter and darker protrusions show the top S and middle Mo atoms, which are conformed to theoretical calculation42.
Fig. 1.12 Crystal structure of MoS2: (a) side (b) top view of the ball model. Red frame is the trigonal prismatic (2H) unit cell structure. (c) The empty-state STM image of MoS2.
1.6. Iron growth 1.6.1. Iron silicide
Fe-Si phase diagram43 shows the complicated Fe-Si alloy bulk structure, such as 𝛼-, β-FeSi2 and 𝜀-FeSi structure. The growth of Fe on Si-based substrate also formed γ-FeSi2, which is undistorted CaF2 structure without existing in Fe-Si phase diagram. 𝛼-FeSi2 is metallic distorted CaF2 structure.
β-FeSi2 is semiconductor (Eg=0.8~0.9 eV) distorted CaF2 structure, which can be used in the device application44,45. The heteroepitaxial growth of β-FeSi2 is β-FeSi2(110)/Si(111) and β-FeSi2(101)/Si(111).46 𝜀-FeSi is cubic B20 structure with Eg~50 meV. Hattori et al. summarized schematic structure-magnetic phase diagram for Fe thin film (0~100 nm) growing on Si(111) at variant annealing temperature (300~1200 K) 47. When the scale of Fe silicide is reduced from thin film to nano size (diameter <10 nm), the phase diagram is different with the thin film. The formation of the nano size Fe silicide can be used as electrode of spintronic device with its variant magnetic and electronic properties.
1.6.2. Iron germanide
In the last decade, semiconductor technology, with the high carrier mobility48,49 and low melting point of Ge, was expected to advance according to Moore’s law49,50 In addition, the development of FE and SC coupling was critical in advancing electronics. The Schottky barrier of FE/SC interfaces51,52 is crucial in spin injecting contacts53 without using insulting layers, which precludes the instability associated with Ge oxides50. Fe has a stronger magnetic moment than Co and Ni do54, a property used in designing high-density storage devices with specific magnetic properties, such as antiferromagnetic Fe/Cu2N55. Fe–Si-base alloys have been widely and tenaciously investigated56-58. Although Fe–Si and Fe–Ge alloy structures are similar, Fe/Ge alloys exhibit certain unique properties, such as the helimagnetic property of B20 FeGe films59,60 used in skyrmion devices without FeSi films; moreover, the magnetic coercivity of Fe/Ge(100) is larger than that of Fe/Si(100)61. Furthermore, Fe–Ge alloys with bulk and thin films differ in
diffusion63, changes the surface morphology to form FeGex surface alloy structures corresponding to the magnetic properties. Regarding the complex surface behavior of Fe/Ge(111), inserting(√3×√3)R30° Ag–Ge buffer layers is expected to reduce Fe–Ge reactivity and magnetic dead layers. Fu et al.65 reported that Ag buffer layer resisted the Fe induced defect on the c(2×8)–
Ge(111) substrate at low temperature. Tomaszewska et al.66 showed the interaction of deposited Ni and c(2×8)-Ge(111) were prevented by the Ag buffer layer. The reconstruction of 1 ML (√3 ×√3)R30° Ag-Ge buffer layer, which confirmed as in-equivalent triangular (IET)20 structures, is highly thermal stability below 850 K16,67. Different Fe growth structures are expected because of the low interaction in the Fe–Ag phase diagram68. The structure and magnetic properties of continuous Fe–Ge thin film under specific temperature had been investigated in recent years62,69-71. For the high density of electronic devices, the research of isolated FM clusters was beneficial for spintronic and memory storage application such as the artificial skyrmion Co nanodot on Co/Pd substrate72. In our experiment, to investigate the isolated Fe–Ge clusters, we measured the surface morphology of the low Fe coverage on Ge(111) and Ag buffer layer and explored the thermal evolution of FeGex
structure during annealing.
1.6.3. Iron/MoS2 interface
Most studies have focused on defect, strain, or transition metal doping-induced magnetism of MoS27,10,11. Only a few studies on MoS2-supported magnetic nanoparticles or thin films have been reported.
Kamaratos et al. reported the distribution of intercalated Fe, Ni, and Pd atoms in the bulk of MoS2 crystals using Auger electron spectroscopy (AES) and Ar+ sputtering with room temperature growth and 1200K post-annealing73. Ni and Pd atoms tend to diffuse and distribute uniformly in the bulk of MoS2 crystals, whereas, Fe atoms tend to accumulate between the molecular layers and likely form a FeMo2S4 compound73. Durbin et al. reported the interaction of Cr with the MoS2(000l) surface by using core-level and valence-band soft X-ray photoemission spectroscopy (XPS)74. At RT, the vapor-deposited Cr formed relatively flat films on the MoS2(000l) surface. Cr underwent a limited chemical reaction with the MoS2 substrate to form a chromium sulfide and Mo metal74. Bulicz et al. reported that pulsed laser deposited Co was accompanied by the
formation of metallic Mo and a net depletion of S at the substrate surface75. Mascaraque et al. reported the electronic structure and the surface reactivity of Co deposited on MoS2(000l) by using angle-resolved photoemission spectroscopy. Their results indicates a partial reaction of Co with MoS2 to produce CoS and metallic Mo76. Frindt et al. used exfoliated monomolecular layers of MoS2 as substrates for the preparation of fine particles of Ni, Co, and Fe magnetic materials77. Some of these particles have magnetizations and coercivity values suitable for use in information storage technologies77. In our previous study of Co and Fe thin films deposited on 2D materials, the highly oriented pyrolytic graphite (HOPG), the Ar+ sputtering-induced surface defects led to a more uniform distribution and higher density of particle nucleation78,79. This considerable effect of surface defects on the growth behavior induced a stable canted magnetization. However, detailed investigations, such as that of surface morphology and magnetic properties, on the magnetic transition metal deposited on the MoS2 surface have not yet been reported. In this study, Fe nanoparticles and continuous films were deposited on cleaved MoS2
substrates. The growth and surface structure of Fe coverage were investigated using scanning tunneling microscopy (STM). In addition, the magnetic property and chemical state near the surface and interface region were characterized, providing valuable information for the application of hybrid devices by combining magnetic and 2D materials.
1.6.4. Fe/Pd/C60 growth
In previous, the study of Fe/C60 system presents unique ferromagnetic property, where the coercivity is increased by C60 concentration at RT.80 The strong temperature-dependent M-H squareness is measured in Fe–C60 films with high C60 concentration due to the nano-size grain effects.80 Besides, the result of C60 over-layer enhanced perpendicular anisotropy of Co thin film, which implied the controllable magnetic anisotropy orientation of organic-material–ferromagnetic systems.81 Low spin orbital interaction in organic spintronic systems reduces spin relaxation82 to transport the spin electron, such as C6083.
hysteresis loop, where the coecivity on the order of 50 Oe and Curier temperature is larger than RT.87 Besides, the hydrogen adsorption of C60 is increasing after Pd doping.88 Thus, we fabricate Fe/Pd/C60 multilayer with combining the coupling of Fe-Pd, Fe-C60, and Pd-C60 to measure the magnetic property under various situation, such as sample annealing or exposing in H2
gas. If C60 were destroyed into carbon nanotube by Pd or Fe doping or annealing, the magnetization of carbon will be induced by defect in C6089 or hydrogen absortion90,91.
Chapter 2 Experimental details
2.1. Scanning Tunneling Microscopy (STM)
STM is one of the scanning probe microscopy. Fig. 2.1(a) is the scheme of the STM. When the tip approaches the surface of the sample into quantum tunneling effect distance, the electron has probability tunneling through the vacuum into the tip (surface) from surface (tip) by applying bias voltage (Vb) forming the tunneling current (It). When It enter into the controller, the controller commands the tip by piezoelectric material, and the feedback will determine the distance between the tip and sample. After X-Y plane mapping, we can get the STM image referring to the surface morphology.
Fig. 2.1(b) is the scheme of the tunneling effect between tip and sample.
The electron wave function obeys the schrödinger equation with energy E in vacuum barrier potential U(z) is:
− ℏ 2𝑚
𝑑7
𝑑𝑧7𝜓 𝑧 + 𝑈 𝑧 𝜓 𝑧 = 𝐸𝜓 𝑧
When E is smaller than U(z), the tunneling current becomes:
𝐼?= 𝐼@𝑇 = 𝑒𝐴 𝜓@ 7(ℏ𝑘
𝑚)𝑇 = 𝐶𝑜𝑛𝑠𝑡.× 𝜓@ 7𝑒J7KL , 𝑤ℎ𝑒𝑟𝑒 𝐾 = 2𝑚(𝑈 − 𝐸) ℏ
Thus, tunneling current exponentially decays with the barrier width z.
Considering to the work function Φ of probe and sample, and the Vb shown in Fig. 2.1(c), K can be modified as
𝐾 = 2𝑚(Φ?+ ΦS − 𝑒𝑉U) ℏ
In general, Φ is about 4~5 eV and the value of K is about 20 nm-1. So the variation of z by 0.1 nm, the tunneling current will decay or increase 10 times.
Thus, STM exhibits an excellent vertical resolution to reach the atomic scale measurement.
In Fig. 2.1(c), when applying bias voltage Vb on the tip, electron at occupied state with energy (ε − e𝑉U), the density of state 𝜌?(ε − e𝑉U), and Fermi-Dirac probability 𝑓?(ε − e𝑉U) tunneled into the sample surface at unoccupied state at energy ε , the density state 𝜌S(ε) , and Fermi-Dirac
92
We can consider 𝜌?(ε − e𝑉U) as the constant and count for 𝐼SJ?. The tunneling current is simplified as:
𝐼?JS ∝ 𝐶×𝜌?× ^_`𝜌S(ε)
@
𝑑ε
𝜕𝐼?JS
𝜕ε ∝ 𝜌S(ε)
In the semiconductor, the cleaved surface usually forms a reconstructed surface, such as Si(111)-(7×7) and Ge(111)-c(2×8). The reconstructed surface presents the specific local density of state (LDOS). The STM images not only show the surface morphology, but also provide the electronic information of the surface. Therefore, we can research semiconductor related system by STM and STS technology. In the thesis, all STM images presented were acquired at RT in the constant-current mode and analyzed using WSxM software93. To present the clear STM images, the STM images were proceed by below function: FFT filter, lattice, and equalize to reduce the noise, correct the lattice constant, and adjust the image contrast to enhance the fine period structure.
Fig. 2.1 (a)The scheme of the STM. (b)Schematic picture of the tip-vacuum-sample one-dimensional tunneling effect. (c)Schematic of a tip-vacuum-sample tunneling junction.
2.2. X-ray Photoemission Spectrum (XPS)
X-ray photoemission spectrum is the surface composition and chemical state measurement technology. The X-ray transparent depth is 2~10 nm depending on X-ray energy. Fig. 2.2(a) shows the process of electron transition in the XPS measurement. When the X-ray is incident into the inner core of the surface atoms, the electron in 1s state was excited and escaped into the vacuum by photoemission effect. The escaped photoelectron will be detected by hemispherical energy analyzer to obtain the kinetic energy EK of the electron, where EK can be written as:
𝐸K = 𝒽𝓋 − 𝐸U− Φ
𝒽𝓋 is the X-ray energy and Φ is the working function of the surface and the electron detector. Thus, we can calculate the binding energy 𝐸U of L, K, M level electron by the known value of EK, Φ, and 𝒽𝓋. The binding energy is the characteristic of elements. Fig. 2.2(b) shows the Fe 2p XPS of Fe/MoS2 results in section 3.2.3. Each peak indicated the binding energy value of the element.
When surface has the chemical reaction or binding to other elements, the peak will be slightly shifted. Thus, we can use XPS to observe the surface composition and chemical state.
Fig. 2.2 (a)The basic conception of XPS. (b) Fe 2p XPS of Fe/MoS2 section 3.2.3.
2.3. Magneto-Optical Kerr effect (MOKE)
Magneto-optic Kerr effect is used to measure the surface magnetized condition. The concept of the MOKE is that the linear polarized light reflects from a magnetized sample changing into ellipse polarized light. In macroscopic, the linear polarized light can be taken apart into right circular polarized light (RCP) and left circular polarized light (LCP). When linear polarized is incident into the sample, RCP and LCP can be effected by inner dielectric constant, respectively. The amplitude and phase of RCP(LCP) will be changed. The complex reflection coefficient is written as:
𝑟± = 𝑟±𝑒de± = 𝑛±− 1 𝑛±+ 1
Where ± refers to RCP and LCP light. The 𝑟± is the amplitude of the electric field and Δ± is the phase different between incident and reflect electric field. The Jones matrix expression of the electrical field 𝐸g of reflected ellipse light is:
𝐸g =h7i 𝑟j 1
𝑖 + 𝑟J 1
−𝑖 = h7i(gljg7 m) 1 𝑖gglJgm
ljgm
= h7i(gljg7 m) 1 ΦK
The rotation of 𝐸g to 𝐸d is called Kerr rotation, which can be expressed as the complex Kerr angle, where:
ΦK = 𝑖𝑟j− 𝑟J
𝑟j+ 𝑟J = 𝑖 𝑛j− 𝑛J
𝑛j𝑛J− 1= 𝜃K+ 𝑖𝜀K 𝜃K = 𝐼𝑚(oolJom
lomJp) = −p7(Δj− ΔJ) and 𝜀K = 𝑅𝑒(oolJom
lomJp) =gglJgm
ljgm
Here, 𝜀K is the ellipticity and 𝜃K is the Kerr rotation angle between long axis and origin linear polarized direction. Fig. 2.3 is the simple apparatus of the MOKE setup. The incident laser passes through the polarizer, which used to control the linear polarized direction of the incident laser, then the reflected laser passes through the second polarizer into the light intensity detector. By the theory calculation94:
𝐼g = 𝐼d[1 + 2𝜃K𝐽@(𝛿@) + 4𝜀K𝐽p(𝛿@) sin(𝜔𝑡) + 4𝜃K𝐽7(𝛿@) cos( 2𝜔𝑡)]
Where 𝛿@ is the small angle between second polarizer and long axis of the ellipse polarized light. Thus, we can get :
𝜃K = 𝐼g− 𝐼d 𝐽@(𝛿@)𝐼d Or by lock-in amplifier operate, we can get :
𝜀K = 𝐼{
4𝐽p(𝛿@)𝐼d, 𝑎𝑛𝑑 𝜀K = 𝐼7{
4𝐽7(𝛿@)𝐼d… … … (1)
In microscopic view, the displacement 𝐷 = 𝜀𝐸, where 𝜀 is the dielectric tensor. In non-magnetized sample, 𝜀 can be considered as a scalar constant, which non-diagonal value is zero in the tenor matrix. When the sample is under magnetized condition, 𝜀 will be wrote as :
𝜀 = 𝜀@
1 −𝑖𝑄€𝑚• 𝑖𝑄€𝑚‚ 𝑖𝑄€𝑚• 1 −𝑖𝑄€𝑚ƒ
−𝑖𝑄€𝑚‚ 𝑖𝑄€𝑚ƒ 1
𝑄€ is the Voigt parameter, which is appeared in the off-diagonal term of the dielectric tensor, and is linearly related to reduce magnetization (mi=Mi/Mr) of sample. Thus, 𝐷 = 𝜀𝐸 can be wrote as 𝐷 = 𝜀@(𝐸 + 𝑖𝑄€(𝑚×𝐸)), and the second term in the brackets is the Lorentz force form. The MOKE can be classified as longitudinal and polar types by the direction of (𝑚×𝐸), which is refer to in-plane and perpendicular magnetization of the sample, respectively.
In general, dialectic tensor is the form:
𝜀 = 𝛿d„ − 𝑖4𝜋
𝜔 𝜎d„, 𝑎𝑛𝑑 𝜎d„ = 𝜎pd„+ 𝑖𝜎7d„
The 𝜎d„ is the conductivity tensor, which can be induced by magnetization of the sample. The 𝜎d„ have the linear relationship to the magnetization. By theoretical calculation94, we can obtain:
𝜃K =4𝜋
𝜔 [𝐵𝜎pƒ‚+ 𝐴𝜎7ƒ‚
𝐴7+𝐵7 ] 𝑎𝑛𝑑 𝜀K = 4𝜋
𝜔 [𝐵𝜎pƒ‚+ 𝐴𝜎7ƒ‚
𝐴7+𝐵7 ] … … … (2) Where 𝐴 = 𝑛‡− 3𝑛𝑘7− 𝑛 𝑎𝑛𝑑 𝐵 = −𝑘‡ + 3𝑛7𝑘 − 𝑘
Thus, we can know 𝜃K and 𝜀K are related with magnetization of the sample. By the results of (1) and (2), we can measure ΔI(𝐼{ and 𝐼7{) variation without (with) lock-in amplifier versus the applied magnetic field to obtain the M-H hysteresis loop of the sample.
Fig. 2.3 Simple scheme of MOKE setup.
2.4. The experimental setup and process
2.4.1. Instruments
A variable-temperature scanning tunneling microscope (VT-STM;
Omicron), low-energy electron diffraction (LEED), Knudsen cell, and electron beam evaporator (EFM3; Omicron) equipped in an ultra-high vacuum (UHV) chamber with pressure below 2×10−11 Pa were used in the Fe/Si(111), Fe/Ge(111), and silicene experiments, which is shown in Fig. 2.4(a). The schematic picture of evaporate system for Fe/MoS2 and Fe/C60/Pd samples is shown in Fig. 2.4(b), where a STM (UNISOKU) and three depositing sources, Fe, Pd, and C60, are equipped in an UHV chamber with pressure below 2×10−8 Pa.
Fig. 2.4 Schematic picture of the instruments for (a)Fe/Si(111), Fe/Ge(111), and silicene systems and (b)Fe/MoS2 and Fe/C60/Pd systems.
2.4.2. Substrate preparing
(I)Si(111): Si(111) samples were cut from commercially available p-type wafers (1 - 10 Ω ∙ cm resistivity, 500 𝜇𝑚 thickness). The samples were ultrasonically rinsed in acetone and methanol three times, placed in the chamber and degassed through indirect heating up to 600 K for several hours until the UHV pressure decreased to a normal value. The samples were heated to 1500 K for 45s and cooled to 1170 K for 2 minutes in a cycle for 10 to 15 times. After finishing the cycles, the sample temperature was held at 970 K to 1070 K for 1h. Well Si(111)-(7×7) reconstructions formed after slow cooling, which were confirmed by the sharp (7×7) LEED patterns and STM images (Fig.
1.2).
(II)Ge(111): Ge(111) samples were cut from commercially available p-type wafers (1–10 cm resistivity, 500 𝜇 m thickness). The samples were ultrasonically rinsed in acetone and methanol three times, placed into the chamber and degassed through indirect heating up to 970 K for several hours until the UHV pressure decreased to a normal value. The samples were sputtered at 1.2–1.5 keV energy and 4×10−5 Pa for 1 h and directly heated to 970 K for 1 h. Well Ge(111)–c(2×8) reconstructions with terraces tens of nanometers wide formed after slow cooling and confirmed by STM and LEED (Fig. 1.3).
(III)(√𝟑×√𝟑)𝐑𝟑𝟎° Ag-Si(Ge) reconstructed surface: To form the √3Ag buffer layer, Ag atoms were deposited on the Si(111)-(7×7) (Ge(111)-c(2×8)) substrate using the Knudsen cell evaporator at RT. To guarantee the 1 ML (√3×√3)R30° Ag-Si(Ge) reconstruction formed on the Si(111)(Ge(111)) substrate, the sample was heated and maintained at 750 K for 5 min, and slowly cooled to RT. The over-much Ag atoms would desorb at 750 K16. The STM image and LEED pattern confirmed the well (√3 ×√3)R30° Ag-Si(Ge) reconstructions (Fig).
(IV)Ag(111) thin film surface: we deposit 6~12 ML Ag on the Si(111)- (7×7) substrate at 100 K. After slowly backing to room temperature, we can obtain the Ag(111) layer with atomic flatness.
2.4.3. Experiment process
Fig. 2.5(a) is the experiment process of Fe growth. We prepare clean Ge(111)-c(2×8) and Si(111)-(7×7) or (√3×√3)R30° Ag-Si(Ge) reconstructed surface in UHV chamber. Fe atoms were deposited using the electron beam evaporator at RT and annealed gradually at temperatures ranging from RT to 640 K. The temperature was measured using a K-type thermocouple. Thermal equilibrium between Fe atoms and the prepared substrates was attained after 30 min of annealing. The coverage ratio of Fe on the surface is decided by STM images. Fig. 2.5(b) is the process of silicene growth. We pre-deposit Ag on Si(111) substrate to form Ag(111) thin film or ( √ 3 × √ 3)R30 ° Ag-Si reconstructed surface in individual experiment. Si atoms were deposited using the electron beam evaporator. During depositing, the substrate kept at constant temperature. The temperature was measured using a K-type thermocouple. The depositing rate was 0.01 ML/min, which is calibrated by STM images.
Fig. 2.5 The experiment process of (a)Fe/Si, Fe/Ge growth and (b)silicene growth experiment
Fig. 2.6(a) is the experiment process of Fe/MoS2 system. MoS2 substrate is bought by commercial company. The samples were prepared in an UHV chamber with a base pressure of better than 3 × 10−9 Torr. The MoS2
substrates were cleaved in air, which is checked by STM, immediately transferred into the vacuum chamber, and then degassed at 400 K for at least 12 h. The Fe atoms were deposited using an e-beam heated evaporator. The surface morphologies of the bare MoS2 substrates and Fe/MoS2 films were in situ characterized using STM. In addition, the magnetic behavior in an ambient environment at RT was investigated by MOKE with the magnetic field in either perpendicular or in-plane directions. To study the surface Fe-oxide and the chemical interaction at the Fe/MoS2 interface, depth-profiling XPS investigation was used to confirm the chemical state of Fe, Mo, and S at various depths by combining an in-house XPS with Ar+ ion sputtering. Fig.
2.6(b) is the experiment process of Pd/(Pd/Fe/C60)/Au/Al2O3 system. We have prepared Au(~50 nm)/Al2O3 substrate and transferred into UHV chamber. C60, Fe, and Pd are sequentially deposited several round on the substrate then covered by Pd. The sample were ex situ measured by MOKE in air before and after annealing. The film composition is also checked by TEM. We also measured the sample by XMCD to understands the magnetic moment contribution for C60 and Fe.
Fig. 2.6 The experiment process of (a)Fe/MoS2 and (b)Pd/Fe/C60 system.
Chapter 3 Result and discuss
3.1. Silicene growth
In our experiment of silicene growth, we try to fabricate silicene, which can be the substrate of FM nano structure growth in the future work, on semiconductor based substrate, such as Si(111) and Ge(111). The fabrication of silicene is always depositing Si on single crystal Ag(111) in previous research, but it is hard to use single crystal Ag bulk as the substrate in industry technology because of its expansive price. Thus, we try to fabricate silicene on Si(111) and Ge (111) substrate with Ag wetting layer because of the mature technology in Si and Ge based semiconductor industry. We will show the results of Si growing on Ag-Si, Ag-Ge (√𝟑×√𝟑)𝑹𝟑𝟎° reconstructed surface, and Ag(111)/Si(111) in the following sections.
3.1.1. Si grow on Ag-Si (√𝟑×√𝟑)𝑹𝟑𝟎° reconstructed surface
The growth of silicene needs metallic substrate supporting because of its sp2-sp3 hybridized binding in silicene. The DFT calculation presents the strong interaction by p-d orbital hybridization between silicene and Ag substrate. Thus, the silicene growth on Ag reconstructed surface is expected to reduce the interaction between silicene and substrate by lower Ag ratio on the surface.
Here, we first choice Ag-Si (√𝟑×√𝟑)𝑹𝟑𝟎° reconstructed surface for silicene growth because of the wild applications of Si based semiconductor industry. After depositing Ag atoms on Si(111)- (7×7) reconstructions surface (S1), the top (7×7) Si ad-atoms are removed from Si(111) top layer (S1) to form a new layer (S2) on S1, and Ag atoms reacted with Si(111)- (7×7) reconstruction layer to form 3Ag-Si reconstructed surface. During forming 3Ag-Si on S1, the overmuch Ag atoms moved to S2 to form 3Ag-Si. Finally, the ratio of S2/S1, which is independent of the substrate temperature and Ag coverage, is approaching to 115. In our study, the initial ratio of S1 area is about 41%. Fig. 3.1(a) shows the clean 3Ag-Si reconstruction with two different level surfaces S2(high)and S1(low). White dash lines in Fig. 3.1(a) and (c) indicate the boundary of 3Ag-Si reconstruction. Fig. 3.1(b)-(d) are the results of 0.2 ML Si depositing on 3Ag-Si surface with different Ts. Comparing to the clean 3Ag-Si, the additional clusters appeared on three places at RT without periodic structures shown in Fig. 3.1(b), the circles A, B, and C are
represented Si clusters on S2 surface edge, 3Ag-Si reconstruction boundary and 3Ag-Si surface, respectively. When Si is deposited at 470 K shown in Fig. 3.1(c), the number of Si clusters reduced on the 3Ag-Si surface and the S2 edge then preferred to assemble at 3Ag-Si reconstruction boundary. The results implied that Si clusters drop from the edge top to side then Ag atoms hip-hop to the top forming a new 3Ag-Si S2 area. The notches (yellow circles) appeared at the S2 edge cause the S2 edge rougher than Fig. 3.1(b). When Ts
is increased to 570 K, the hole defects (blue circles) appear as shown in Fig.
3.1(d). The notches transfer into hole defects by S2 area expanding because of inhomogeneous expanding of S2.. After 670 K annealing, only some Si clusters assemble to the reconstruction boundary. Fig. 3.1(e) shows the result of surface area percentage versus Ts. The S1 percentage is kept near 29% for Ts
< 620 K after 0.2 ML Si depositing, but the area (S3) of Si clusters on S2
surface decrease linearly by increasing Ts. Although more Si clusters on S2
join to expand S2 area, the mechanism15 of forming 3Ag-Si reconstruction limits the S2 expanding. By increasing Ts, the Si clusters start to disappear from S2 edge then on the surface. The Si clusters on reconstruction boundary are hard to join the S2 area expanding. The S1 area percentage increases to 41
% at 670 K, which indicates the S2/S1 area ratio is back to the origin value, because the 3Ag-Si reconstruction temperature is about 600 K. The S2 edge remains rougher than origin of 3 Ag-Si, because Si atoms keep joining the edge to form 3 Ag-Si area.
Fig. 3.1 STM images of (a) clean 3 Ag-Si reconstruction surface (40×40 nm2, It=0.20 nA, Vb=0.7 V), STM images of 0.2 ML Si depositing on 3 Ag-Si reconstruction surface with substrate temperature at (b) RT (It=0.10 nA, Vb= -2.1 V), (c) 470 K (It=0.10 nA, Vb= -1.7 V), (d) 470 K (It=0.10 nA, Vb= -1.8 V), (e) 570 K (It=0.10 nA, Vb= -1.7 V), (f) 670 K (It=0.10 nA, Vb= -1.7 V), (g) the ratio of surface area with different level versus substrate temperature.
Next we compare the different Si coverage at 570 K. Fig. 3.2(a) and (b) show the results of 0.4 and 0.8 ML Si depositing on 3Ag-Si surface. The Si clusters on S2 edge almost assimilate into S2 layer at 570 K. The Si clusters only remain on the reconstruction boundary. When Si amount increases to 0.4 ML as shown in Fig. 3.2(a), the Si atoms assemble around reconstruction boundary and form new 3 Ag-Si islands. When the coverage is increasing to 0.8 ML, the area of 3 islands is larger than 0.4 ML and the bilayer 3 islands appeared as shown in Fig. 3.2(b). The monolayer and bilayer 3 islands were labeled as A and B. Fig. 3.2(c) shows the precise STM images of B island in Fig. 3.2(b). The STM images has been corrected by Si(111)-(7×7) lattice constants (insert frame of Fig. 3.2(c)) after contrast process and shows the clear structure period on B island surface. Fig. 3.2(d) shows the line profile of Fig. 3.2(c), which shows the 3 period (0.662 nm). Fig. 3.2(e) shows the line profile of 3 islands in Fig. 3.2(b). The results indicate the thickness of A
and B islands are 0.241 nm and 0.479 nm, respectively. Fig. 3.2(f) is the S1
and S3 area percentage versus Si coverage. The S1 area percentage decrease from 41% to 13% while the Si coverage is increasing from 0 to 0.8 ML.
Comparing to S1 percentage versus Ts of 0.2 ML Si coverage, the result implies that S2 ratio only depends on Si coverage. The S3 area percentage increases from 2% to 13% while Si coverage is increasing from 0.2 ML to 0.8 ML at 570 K. The increasing of Si on S2 area causes the forming of √3 islands.
The results imply that the Si atoms prefer to fill S2 layer then form √3 islands at the boundary line. When S2 is approaching to be fully filled, the remained Si clusters keep at √3Ag-Si reconstruction boundary then form √3 islands on the
√3 Ag-Si surface.
In summary, depositing Si on √3 Ag-Si only caused S2 area expanding without forming Si monolayer structure. Deposited Si atoms assembling as Si clusters dropped to S1 from S2 edge and Ag atoms hip-hop to the top forming
√3 Ag-Si surface. The S2/S1 ratio changed by Si coverage but was independent with the substrate temperature. The Si atoms prefer to locate on the reconstruction boundary. By increasing Si coverage, the Si atoms on the boundary formed monolayer and bilayer √3 islands at 570 K. In next section, we changed the substrate from Si(111) to Ge(111) and expect that Ag hip hop effect will be suppressed by the binding of Ag-Ge.
Fig. 3.2 The STM images of (a) 0.4 ML, (40×40 nm2, It=0.10 nA, Vb=-1.6 V) (b) 0.8 ML (40×40 nm2, It=0.10 nA, Vb=-1.8 V) Si depositing on 3 Ag-Si reconstruction at 570 K, the label A and B in (b) presented monolayer and bilayer 3 islands, (c) the zoom in B islands in (b) and correct by 𝑆𝑖(111) − (7×7) unit cell(insert figure), (d) the line profile of island B in (c), (e) the line profile of 3 islands in (c), (f) surface area percentage versus Si coverage.
3.1.2. Si grow on Ag-Ge (√𝟑×√𝟑)𝑹𝟑𝟎° reconstructed surface
In this section, we attempt to fabricate silicene on Ag-Ge (√3×√3)R30°
reconstructed surface, which supports an atomic flatness surface and unique surface electronic properties for silicene growth. In previous section of Si grow on Ag-Si (√𝟑×√𝟑)𝑹𝟑𝟎° reconstructed surface, Ag atoms prefer to hip-hop to the top layer forming a new Ag-Si (√𝟑×√𝟑)𝑹𝟑𝟎° reconstructed islands. We expect that Si grows on Ag-Ge (√𝟑×√𝟑)𝑹𝟑𝟎° reconstructed surface without Ag atoms hip-hop behavior.
Fig. 3.3 The empty-STM images (20×20 𝑛𝑚7) of (a)0.2 ML Si at RT (It=0.1 nA, Vb=1.80 V) (b) 0.2 ML Si at 370 K (It=0.2 nA, Vb=1.80 V) (c) at 420 K (It=0.1 nA, Vb=1.80 V) (d) 0.4 ML Si at 370 K (It=0.2 nA, Vb=-1.80 V) (e) 0.2 ML Si deposit on Ag-Si buffer layer at RT. (It=0.2 nA, Vb=-1.80 V) (f) Depth profile in (d) and (e).
Fig. 3.3(a)-(c) are the STM images of 0.2 ML Si atoms depositing at RT, 370 K, and 420 K, respectively. In Fig. 3.3(a), several pieces of assembled protrusions appear on the surface. The nearby area maintains √3 reconstruction, which indicates no site-exchange effect between Si and Ag or Ag hip-hop to the top layer95. Comparing to Fig. 3.3(e), where 0.2 ML Si grown
induce Ag hip-hop to the top layer, forming additional √3 Ag-Si islands. Our result is in a good agreement with the previous study. In Ag-Ge and Ag-Si reconstruction phase diagram (Fig. 1.4), higher thermal result shows the high thermal resistivity of a Ag-Ge compare to a Ag-Si surface to avoid Ag atoms segregating to the top forming new √3Ag-Si islands. On the other hand, the different Si structure observed on √3Ag-Ge(111) surface strongly suggests the site-exchange effect between Ag and Si is suppressed by the relatively strong Ag-Ge bonding and the deposited Si atoms remain on the top surface. Besides, if Si deposits on various temperature, which is larger than 400 K, the surface should form the √3Ag-Ge or Ag-Si reconstruction, but we measured the 2-D film structure in Fig. 3.3(a)-(c). Based on the above viewpoints, the protrusions on the √3Ag-Ge(111) surface were only considered as the structure of Si atoms. Fig. 3.3(a) showed that Si protrusions assembled as Y-shaped backbone with 3-fold symmetry. At 370 K as shown in Fig. 3.3(b), the number of piecemeal protrusions decreased and the Y-shaped backbone was filled by additional protrusions and reformed into pieces of triangular shape. To form larger Si monolayer, we increased Ts to 420 K, which shown in Fig. 3.3(c). The surface presents not only the larger Si monolayer but also additional protrusions and √3 Ag-Si islands (yellow color area in Fig. 3.3) on the Si monolayer. The formation of √3 Ag-Si islands indicates that the probability of Ag hip-hop effect as that on Si/√3Ag/Si(111)95 is also enhanced by increasing Ts. Besides, the √3Ag-Si islands were also formed by increasing the amount of Si atoms to 0.4 ML at 370 K as shown in Fig. 3.3(d), where the Si monolayer (blue color) and √3 Ag-Si islands (yellow color) coexisted. The insert frame in Fig. 3.3(d) showed the clear √3 structure on √3 islands as √3Si-Ag reconstruction. On the other hand, by comparing Fig. 3.3(d) with (e), the direct Ag hip-hop effect occurs on the √3Ag/Si(111) surface without forming Si monolayer, which is also differed from that on the √3Ag/Ge(111) surface. Fig.
3.3(f) shows the line-profiles along the lines in Fig. 3.3(c), (d) and (e), showing the heights of Si monolayer (level 1), protrusion on Si monolayer (level 2),
√3Ag-Si islands (level 3), and protrusion on √3Ag-Si islands (level 4). The results show the height of √3Ag-Si islands is near 0.25 nm which is conformed to √3Ag-Si height. The height of Si protrusion and Si monolayer is near 0.1 nm.
The single Si protrusion can locate on Si monolayer and √3Ag-Si islands. Here, we can know that Si depositing on √3Ag/Ge(111) surface prefers to form Si
monolayer at low Si coverage. When the temperature of substrate is increasing, the area of Si monolayer is increasing and √3Ag-Si islands are also appeared inside the Si monolayer. Additional Si protrusion can be located on new-formed √3Ag-Si islands. The deposited Si atoms were desorbed out surface after 620 K indicating low interaction between Si monolayer and the substrate. Nevertheless, the Si monolayer is at least stable below 520 K.
Fig. 3.4 The STM images of 0.4 ML Si deposit on Ag buffer layer at 370 k layer with (a)7.4×7.4 nm2, Is=0.2 nA, Vb=+1.8 V (b)Vb= -1.8 V, and depth profile of (c)line A and (d)line B in (a), (e) illuminated model of (a) and (b), (f) √3 unit cell, (g) 2×2 unit cell.
are the line-profiles showing two different periods of structures in Si monolayer.
Line A and B showed the periodicity of 0.692 and 0.772 nm which match with 2×2 and √3×√3 superstructures, respectively. By using √3Ag-Ge auxiliary grid, the location of protrusion can be determined. The protrusions in Fig. 3.4(a) and (b) reveal buckled Si atoms located on Ag atoms or in-between Ag atoms, respectively. The ball models of Si monolayer appearing in Fig. 3.4(a) and (b) is illustrated in (e). In the ball model, the Si monolayer grows with Ge(111) honeycomb lattice constant and stacks on √3Ag-Ge with different phase to bottom Ge(111) grid. The symbol Siup(down) presented the Si buckling up(down) atoms. Siup(+) atoms indicated that Siup atoms located on Ag atoms causing the bright protrusions in filled-states STM images; Siup(-) atoms indicate that Siup
atoms located in the interval of Ag trimer cause the additional bright protrusions in empty-states STM images. Fig. 3.4(f) and (g) showed the ball model of the 2×2 and √3×√3 unit cell causing by buckling property in Fig. 3.4 (e). The red area were the Ag atoms located area for IET(±) model and yellow (blue) frame presented the unit cell(unit honeycomb). In √3×√3 unit cell, each Si honeycomb owned one buckled Si, i. e. with vertical displacement from the surface. In 2×2 unit cell, each Si honeycomb ring owned two buckled Si atoms located on Ag atom (white) and between Ag trimers (gray), respectively. Two buckled Si atoms are opposite side in honeycomb ring. The 2×2 superstructure along Ge(111) orientation was the prioritized structure for deposited Si nucleating on √3Ag-Ge. After forming 2×2 core, the 2×√3 unit cell arranged along √3Ag-Ge crystalline orientation and formed Y-shape backbone shown in fig. Fig. 3.4(a). The √3×√3 superstructures grew around the Y-shaped structure with √3Ag-Ge orientation forming the triangular islands.
Using scanning tunneling microscopy, we studied the formation of Si monolayer grown on (√3×√3)R30° Ag-Ge(111) reconstructed surface. The results show that the Ag hip-hop effect, which may cause segregation of Ag on the top of Si monolayer, is suppressed because Ag atoms stably bound with Ge substrate. The isolated Si monolayer with mixing √ 3 and 2 × 2 superstructure increases its area by increasing substrate temperature from RT to 420 K. From the demonstrated ball model, the Si monolayer found in this study is very possible to consist of honeycomb structure. The (√3×√3)R30°
Ag-Ge(111) reconstructed surface can be a new substrate to grow silicene.
3.1.3. Si grow on Ag(111)/Si(111) substrate
For the rich research of silicene growth on single crystal Ag(111) bulk, the first silicene field effect transistor device37 was success fabricated and measured the carrier mobility is near 100 cm2/V1s1, which is far less than intrinsic carrier mobility by theory calculation96 ( 2.57x105 cm2/V1s1), in 2015.
What is happened between experiment and theory results? We know that silicene always forms various superstructure by varying unit cell orientation on Ag(111), such as 4x4 and 13× 13. Here, the boundary between these superstructures usually presents the defect condition.97 This is because the discontinuous silicene sheet between these superstructures on single crystal Ag(111) surface. Thus, we think the discontinuous silicene between superstructure cause the reduction of carrier mobility of silicene on single crystal Ag(111). In contrast, if we depositing Si on flexible Ag(111) surface, will the continuous silicene form between these superstructures? So, we deposit Ag on Si(111) forming Ag(111) thin film with 6 ~12 ML thickness to grow silicene.
In this section, we research the silicene growth on Ag(111) surface.
Before fabricating the silicene, we deposit 6~12 ML Ag on the Si(111)- (7×7) substrate at 100 K. After slowly backing to room temperature, we can obtain the Ag(111) layer with atomic flatness. Fig. 3.5(a) is the Ag/Si(111) LEED pattern, which shows the Ag(111) 6-fold symmetry pattern overlapping with Si(111) pattern. The outer oval spot presents the Ag(111) LEED pattern, which the angle between the center to edge of oval spot is about 7.4°. The ratio of Ag(111) and Si(111) LEED pattern is 1.33, which fits the lattice ratio(1.326) of Ag (0.288 nm) and Si (0.382 nm). Besides, the appearance of dim arc pattern indicated the distortion of Ag(111) growth, which is differ with single crystal Ag(111) surface LEED pattern98 shown in the inset of Fig. 3.5(a). The angle of the dim arc is about 19.7°, which is the same as Ag(111)/Sb-Si(111) consequent99 due to Ag(111) domain rotation. Here, the Ag(111) surface grows with Ag[110]//Si[110] and Ag[110]//Si[231] orientation for main spot and arc spot, respectively. Fig. 3.5(b) is the STM image of Ag(111)/Si(111) surface which is slowly annealing to 500 K. Fig. 3.5(c) is the line profile showing the
