原子島在金屬/半導體介面的成長研究 (以鈷/銀/鍺(111)為例)

全文

(1)Department of Physics, National Taiwan Normal University Doctoral Thesis. 國立臺灣師範大學物理學系博士論文. Study of atomic island growth on metal/semiconductor interfaces (as example of Co/Ag/Ge(111)). 原子島在金屬/半導體介面的成長研究 (以鈷/銀/鍺(111)為例). 指導教授(Supervisor)：傅祖怡(Tsu-Yi Fu) 研究生(Ph. D. Student)：黃筱嵐(Xiao-Lan Huang) 中華民國一百零一年七月.

(2)

(3) Contents. Abstract. Chapter 1 Introduction............................................................................1 1-1 Surface reconstruction...................................................................................1 1-2 Ferromagnetic metals on the semiconductor surface......................6 1-3 Structure transformation..............................................................................7 1-4 Strain relaxation...............................................................................................9 1-5 Electronic structure......................................................................................10 1-6 Previous studies in our lab........................................................................11. Chapter 2 Experimental Details.........................................................13 2-1 Principle of scanning tunneling microscopy (STM).......................13 2-2 Principle of Auger electron microscopy (AES).................................16 2-3 Principle of low energy electron diffraction (LEED)......................18 2-4 The experimental setup and process....................................................20.

(4) Chapter 3 Results and Discussion......................................................25 3-1 Thermal evolution of Co on the coexisting Ag/Ge(111)(Ag-. ) and. 3-2 Structure of Co-. -. (Ag-. ) phases.........25. nanoislands on Ag/Ge(111)-. surface...............................................................................................................43 3-3 Strain–relaxation in the structure of CoAg/Ge(111)-. islands on. surface.................................................................59. 3 - 4 E l ect ro n i c st r u ct u re o f Co i sl an d s g ro w n o n t h e Ag/Ge(111)-. surface.................................................................75. Chapter 4 Conclusion.............................................................................85. Reference...................................................................................................89.

(5) Abstract. The thermal reaction of Co on Ag/Ge(111)-. /. phases was studied by scanning tunneling microscopy, low energy electron diffraction, and Auger electron spectroscopy. Firstly, we address on the controversies over the chemical composition of Co islands. by. examining. -. the. phase,. Ag/Ge(111)-. thermal as. well. reaction as. the. of. Co. on. coexisting. phase. From the study, one finds that Ag atoms. shift from. phase to. phase because of the. interaction between Co and the surface. The fact suggests that it is on the surface where Ag-less phase ( Ag-richer phase. ) transforms into. . Secondly, we proof that. periodicity is composed. of Co-Ge alloy, whereas. periodicity is composed of pure Co. Thirdly, we realize that it is -. preventing Co from diffusing into substrate. when annealing the surface at the temperature between 320 K and 730 K. It. is. Ag/Ge(111)-. known. that. -. islands. grown. on. surface are in hcp structure with a (11-20). orientation. The island evolution involves the shape transformation.

(6) Abstract. of a unit cell from parallelogram into rectangular. Meanwhile, the shape of the island shifts from hexagonal to stripe. In additions, it is identified that Co-. islands grow along two crystallographic. directions: pseudo-[0001] and pseudo-[1-100]. We observe a lateral shift between the topmost and the underlying bilayers for islands which grow along pseudo-[0001] direction. On the other hands, no lateral shift is perceived for those growing along pseudo-[1-100] direction. In terms of the strain–relaxation of CoAg/Ge(111)-. islands grown on. surface, we analyze the images taken by. scanning tunneling microscopy. From the studies, one realizes a common fact that. -. islands adopt a more compact. arrangement than Ge(111) substrate does, whereas each Coisland is different in the degree of atomic compactness. Yet, we do not observe any distinct relationship between strain–relaxation and the island height. In addition, we identify three different groups of islands. from. analyzing. the. correspondence. between. the. strain–relaxation and the island size: (i) small islands (less than 80 nm2) with fixed inter-row distances in high atomic compactness, (ii) small islands with unfixed inter-row distances, and (iii) big islands (bigger than 80 nm2) with fixed inter-row distances in less compact.

(7) Abstract. atomic arrangement, as compared to the first two groups. Based on the obtained information, we propose the model that explains the relationship between the strain–relaxation and the island size. Regarding electronic structure, we study phase,. -. phase,. -. -. island,. and. island by means of scanning tunneling. spectroscopy at room temperature. Similar to the one acquired from. -. Ag/Ge(111)-. the. spectrum. obtained. from. structure reveals a shoulder at 0.7 V, which. indicates that Ge adatoms were donated to the electronic states of the Ag-driven phase. However, the electronic spectrum taken from the. -. island shows a large number of peaks,. which indicates the complex bonding between. -. island and the substrate. In addition, the spectra obtained from the Co-. island grown on the step demonstrate a number of peaks. at negative sample bias, which is different comparing to those taken from the Coexplains. the. island located on the terrace. The phenomenon various. Co-substrate. interactions,. which. are. accompanied with the growth of Co islands at different areas of the stepped surface..

(8) Abstract. Keywords: Co; Ge(111); Ag; STM; epitaxy; phase transformation;.

(9) Chapter 1 Introduction. 1-1 Surface reconstruction Although people have been studying the reconstruction of epitaxy on semiconductor surfaces over the past decades, it is the subject that worth investigating continuously. The different interaction between the adsorbate atoms and the covalent semiconductor contributes to various reconstructed structures. Table 1-1-1 to Table 1-1-4 show the surface restructured by epitaxy, which is formed on Si(100), Si(111), Ge(100), and Ge(111) substrates respectively [1-174]. The different. experimental. conditions, (e.g. coverage of deposit, annealing temperature, etc.), lead to the various kind of reconstruction formed on the surface. For example, under the condition of 0.3 ML Ag coverage on Ge(111)-c. substrate,. reconstruction is formed at. annealing temperature 420 K, whereas. reconstruction is. established at annealing temperature 830 K [62, 146-161]. It is noticeable that the substrate dominated the reconstruction. For example,. reconstruction is usually formed for epitaxy. on Si(100) and Ge(100) substrates [1-13, 34, 36-46, 50-51]. On the other hands,. and. reconstruction is formed for.

(10) 1-1 Surface reconstruction. epitaxy on Si(111) and Ge(111) substrates [3, 35, 52-60, 62-85, 88-94, 97-141, 143-161, 164-168, 170, 173-174].. Table 1-1-1 Reconstruction of epitaxy on Si(100) substrate. Deposits clean. reconstruction ;. ;c. ;c. ;. ;. ;. ;. [1-3]. Ag. ;c. As. ;. ;. ;. ;. ; [4-10]. ; [11-13]. Au. c. ;. B. c. ; [28-29]. C. c. ; [30-33]. ;. Ge. ; [34]. In. ;. (or. ;. Pb. ;2. ;. Sb. ; [13, 39-42]. );. ; [35] ;c. ;c. 2. ; [36-38]. ; [14-27].

(11) 1-1 Surface reconstruction. Table 1-1-2 Reconstruction of epitaxy on Ge(100) substrate. Deposits. reconstruction. clean. c. ;. ;. Au. c. ;. ; [45-46]. In Pb. ;4. ; [3, 43-44]. ;. ;. ; [47-48] ;c. ;“. ;” (probably c. ;) [37,. 49]. Sb. ; [50]. Si. ; [51]. Table 1-1-3 Reconstruction of epitaxy on Si(111) substrate. Deposits clean. Ag. reconstruction ;. ;. c. ;. ;. ;. ;. ;. ;. ;. ; [56-60] ; [61]. Au. ;. B. Bi. ;. ;. [3, 52-55]. As. Ba. ;. ;. ;. ;. ; [80-85] ;. ; [86-87] ; [88-89] 3. ; [62-79]. ;.

(12) 1-1 Surface reconstruction. Deposits. reconstruction. C Ca. ; [90] ;. Ga Ge In. ;. ;. ;c ;. ;c ;. Li. ;. Mg. ;. ; [95-96]. ;. ;. ; [35,. ; [125-126] ;c ;. ;. ;. ; [127-129]. ; [130-136]. ; [125, 137]. Pb. ;. Sb. ; [139-141]. Sr. ;. 97-123] ; [124]. Na. ; [91-92]. ; [93-94]. K. Mn. ;. ;. ; [138]. ;. ;. 4. ;. ; [142].

(13) 1-1 Surface reconstruction. Table 1-1-4 Reconstruction of epitaxy on Ge(111) substrate. Deposits clean. Ag As. Reconstruction c. ;. ;. ;. ;c. ;. ; [3, 55,. 143-145] ;. ;. ;. ;. ; [62,. ;. ; [164-165]. 146-161] ; [162-163]. Au. ;. Bi. ; [166]. C. ;. Ga. ; [168]. ;. ; [167]. ;. In. ; ;. K. ; [143]. Li. ;. Mn. ; ;. ; [143] ; [170]. Na. ; [143]. Sb. ;. ;. Sn. ; [173]. Pb. ;. ; [171-172]. ; [174]. 5. ; ;. ; [169].

(14) 1-2 Ferromagnetic metals on the semiconductor surface. 1-2 Ferromagnetic metals on the semiconductor surface Grown. on. semiconductor. surfaces, the. thin. films of. ferromagnetic d-transition metals are receiving growing attentions because of their potential in the application of spintronics devices [175-182]. Ge, as the semiconductor, has promising properties for microelectronic technology, as one is able to control the characters of Ge doped with Mn though applying a small voltage [183]. Since Ge(111)-c. is to believe the most thermally stable structure for. Ge, we choose the surface as the semiconductor substrate in our research [184]. In addition, we select Co as the ferromagnetic d-transition metal due to its high magnetization at room temperature [185]. The previous study have demonstrated that the thin film of Co grown on Ag/Ge(111) surface exhibits magnetic properties , as opposed to the non-ferromagnetic Co/Ge(111) surface [186-189]. For the phenomenon, J.S. Tsay et al. have suggested that the intermediate Ag layer acts as a buffering layer for the thin film of Co, which prevents a chemical interaction between the deposited Co atoms and Ge (111) surface. To avoid a direct metal-semiconductor reaction, and to improve the overall magnetic efficiency, one selects Ag as a buffering layer in this research.. 6.

(15) 1-3 Structure transformation. 1-3 Structure transformation hcp film fcc film fcc film. fcc substrate. Fig. 1-3-1 The model of the growth of Co films on the Cu(111) substrate. Much research about determining the structure of Co epitaxy has been conducted [190-195]. Co films grown on Cu(111) substrate are forming fcc structure with lower coverage, whereas developing hcp structure with higher coverage that consists of 10 monolayers or above [195], as the model shown in Fig. 1-3-1. Both structural types have been revealed in those Co films grown on Au(111) surface [192-194]. Since we knew the fact that the thickness of the film and the substrate are the two factors determining the formed structure of Co film, it is of interest to establish a structural model for Co epitaxy. Based on the knowledge, there is the possibility to manipulate the process of growing island at atomic level according to the 7.

(16) 1-3 Structure transformation. profound understanding of its mechanisms. Much previous research has been performed about CoAg/Ge(111)-. islands grown on. surface. Yet, for the topic of interest, there is. lacking of information that could contribute to the elaboration of a suitable model. From the earlier studies devoting to the subject of Co epitaxial growing on metal surfaces, it is to believe that the substrate structure and thickness of deposited film determine the appearance of a specific Co phase [190-195]. The mentioned phenomenon is resulted from the two possible crystallographic structures of. -. islands, whether in fcc or hcp. In the. research, we will compare both structural types, hcp and fcc, and to find the universal structure of Co-. 8. islands..

(17) 1-4 Strain relaxation. 1-4 Strain relaxation To avoid possible problems encountered during the fabricating processes, it is a must to understand the mechanisms that control the growth of the film at the beginning of the course. For the reason, much experimental effort has been focused on investigating nano-sized objects, which are the components involved in a film preparation at the early stages [196-207]. In addition, the strain and the stress influence the structure of the growing object [208-209]. The related subject involves the structural parameters changes (e.g. lattice constant) depending on the thickness and the horizontal size of the island. Opposing to the changes regarding lattice constant with overlaying thickness, those concerning island size cannot be monitored by using the techniques of low energy electron diffraction or medium energy electron diffraction [209-212]. Thus, for this research, we observe the alterations in structural parameters by studying the horizontal size of the island using scanning tunneling microscopy (STM).. 9.

(18) 1-5 Electronic structure. 1-5 Electronic structure It is known that STM image contains the information of topography, as well as of electronic structure. Although the feature of STM image is influenced by the electronic structure of semiconductors, in general, it is determined by the topography in terms of metallic materials due to their large number of free electrons. By using scanning tunneling spectroscopy (STS), and comparing the STM images obtained at different sample bias, one can distinguish the electric conductivity of the surface. We conduct STS investigation and constant current-imaging tunneling spectroscopy (CITS) investigation on the Co islands on Ag/Ge(111)-. system. They are the two experimental. techniques that have been successfully applied for the inspection of the. various. metal/semiconductor. and. semiconductor/. semiconductor systems [213-221]. Based on certain peculiarities found in both STS and CITS data, we argue that it is the electronic properties of the surface influencing the topography of the appeared nanostructures.. 10.

(19) 1-6 Previous studies in our lab. 1-6 Previous studies in our lab [222-226] ─ Structure 1 ( ─ Structure 2 (. ) ). Fig. 1-6-1 The illustration shows the percentage of two structures as a function of Co coverage. [223] Much previous studies have been devoted to the growth of Co islands on Ag/Ge(111)-. [222-226]. By STM technique, we. find that the size of island with either. or. periodicity is depending on the annealing temperature and the amount of Co coverage (see Fig. 1-6-1 and Fig. 1-6-2) [222-223]. Comparing the two types of islands, CoxGeyformed by 1–2 atomic layers, whereas Co-. is typically is composed of 3–5. atomic layers, which is higher than the former kind. From the observation, it is to believe that we are able to manage the growth. 11.

(20) 1-6 Previous studies in our lab. of Co-island and to achieve the structures of desired properties by controlling the two factors respectively [224].. Annealing Temperature (℃). Fig. 1-6-2 Size control by means of annealing temperature for variant coverage is illustrated in this figure. For the coverage of 0.35, 1.4, 2.1 and 2.8 ML, which the fitting lines are almost parallel indicating a similar trend, the average island size increases with the raising temperature. For the coverage is 3.5, 4.2 and 4.9 ML, which the fitting lines tend to be horizontal, the size is nearly fixed with the change in temperature. [224]. 12.

(21) Chapter 2 Experimental Details. 2-1 Principle of canning tunneling microscopy (STM) [227-231] In quantum mechanics, Schrödinger’s equation describes an electron with energy E moving in a potential. . (2-1-1).. is the wavefunction, representing the state of the electron at a location z.. U(z) E Sample. Tip. Fig. 2-1-1 A tunneling junction. The wave function of electron decays at the region where the potential is bigger than the energy of electron. The solution of the Schrödinger’s equation is represented as follows when the energy of electron E is smaller than the potential U (see Fig. 2-1-1). (2-1-2), where.

(22) 2-1 Principle of scanning tunneling microscopy (STM). (2-1-3). Considering the entire sample states regarding the possible energy intervals (eV), the tunneling current is representing as (2-1-4). The local density of states (LDOS) at a location z with the energy E for a sufficiently small є is defined as (2-1-5). The LDOS shows the number of electrons per unit volume per unit energy. From Eq. (2-1-2), Eq. (2-1-4), and Eq. (2-1-5), tunneling current can also be expressed as (2-1-6). The LDOS can be obtained by calculating the derivative of the current (2-1-7). For transmission probability, it has been realized that some distance and energy dependency can be removed from the derivative of the current through dividing. 14. by . The quantity.

(23) 2-1 Principle of scanning tunneling microscopy (STM). is thus being widely used to identify the density of states of each STM result. There are a large number of STM studies being conducted to investigate the semiconductor surface with a clean and adsorbate covering. From which, one is indicated that the topic is highly interested by the field. It is also to suggest that STM is the reasonable approach to such system. In addition, one knew the fact that the electronic structure of a reconstructed semiconductor surface may exhibit differently according to its local condition. Realizing the fact that the structure formed by any of the components ─ adatoms, dimers, or other complicates, is developed on the surface in order to minimize the number of dangling bonds in it, one confirms that STM is an outstanding technique used for studying the reconstructed semiconductor surface.. 15.

(24) 2-2 Principle of Auger electron microscopy (AES). 2-2 Principle of Auger electron microscopy (AES) [232] Auger electron primary electron B. C. K A Fig. 2-2-1 As electron A in K-shell being ejected by the electron beam, electron B in L1-shell transits to K-shell to fill the empty space. The energy released from this transition is desorbed by electron C in L3-shell. The ejected electron C is the Auger electron. In AES, the sample is exposed to a primary electron beam with the energy of 3 keV. The electron beam ejects the electron A located in K-shell. To fill up the empty space, the electron B in L1-shell moves to K-shell. The electron C in L3-shell adsorbs the energy released due to the transition of electron B (see Fig. 2-2-1). Then, electron C departs from the atom. The corresponded kinetic energy of electron C is given by (2-2-1).. 16.

(25) 2-2 Principle of Auger electron microscopy (AES). represents the binding energy of electron in K shell, whereas. and. stand for the binding energies in L1 and L3. shells respectively. The ejected electron C is the Auger electron. Since the amount of kinetic energy of an Auger electron only relied on the binding energies of electrons in the atom, it is valued as the characteristic of an atom.. 17.

(26) 2-3 Principle of low energy electron diffraction (LEED). 2-3 Principle of low energy electron diffraction (LEED) [233] The wavelength. of electron can be represented by de Broglie. equation (2-3-1). where m symbolizes the mass of the electron, and. indicates. the velocity of the electron. Constructive interference appears when the scattered waves contain the path differences of multiple wavelengths. from the. neighboring lattice. The represented equation is as follows, (2-3-2), or (2-3-3). stands for the incident angle of the primary wave, whereas is the directions of the backscattered waves. The distance a is measured between the periodically arranged scatters (see Fig. 2-3-1). Letter n symbolizes the integer denoting the order of the diffraction. In two-dimensional space, the diffractive pattern is considered as the extension of de Broglie equation. It also corresponds to the reciprocal space of surface. On the other hands, the periodicity in real space is obtainable. 18.

(27) 2-3 Principle of low energy electron diffraction (LEED). Fig. 2-3-1 The filled balls represent the lattice points in the real space. The matter waves of electron interfere forming the pattern that corresponds to the reciprocal space.. 19.

(28) 2-4 The experimental setup and process. 2-4 The experimental setup and process The experiment is performed in two separated UHV chambers. One is equipped with Omicron variable temperature scanning tunneling microscopy (VT-STM) (see Fig. 2-4-1). The other is set up with the four-grid Retard Field Analyzer LEED, and AES (see Fig. 2-4-2). A concentric hemispherical analyzer (CLAM 2 VG) is used to inspect the data obtained from AES. For the research, we use p-type Ge(111) wafer (1 to 10 Ω-cm resistivity with 500 μm thickness) as the. substrate.. A. K-cell. dispenser. and. a. well-collimated. e-bean-bombardment type evaporator were for the depositions of Ag and Co respectively. In addition, all STM images presented in this paper are acquired at the temperature of 300 K.. Fig. 2-4-1 The chamber with STM.. 20.

(29) 2-4 The experimental setup and process. Fig. 2-4-2 The chamber with AES and LEED. 2-4-1. Preparation -. for. clean. surface,. and. -. surface,. the. Ag/Ge(111)-. surface (i) Ge(111)-c. surface: for the experiment, Ge(111)-c. surface is cleaned in situ by repeating the procedure applying Ar + bombardment (1.0 keV), followed by annealing the sample at the temperature between 920 K and 1070 K. (ii) Ag/Ge(111)-. surface: the Ag/Ge(111)-. surface is prepared by exposing the Ge(111)-c. substrate to Ag. atoms of 1 ML coverage followed by annealing at temperature ranging from 720 K to 770 K. (iii) Ag/Ge(111)Ge(111)-c. /. surface: after annealing. surface with the Ag submonolayer coverage at the 21.

(30) 2-4 The experimental setup and process. temperature of 670 K or 770 K, one prepares the region where -. and Ag/Ge(111)-. coexisted.. 2-4-2 Processes for Co deposition and annealing For Co deposition, as well as the corresponded annealing condition, three types of experimental processes are applied to the experiment in chap. 3-1. Process Type 1: Co is deposited at the sample temperature of 300 K. The substrate is annealed after Co deposition in situ by direct annealing (Omicron VT-STM). The STM tip traces the same area during annealing the surface, as shown in Fig. 2-4-3. The method we used to determine the temperature is illustrated in chap. 2-4-3.. Before Annealing. During Annealing. The detected area. After Annealing STM tip. Fig. 2-4-3 The detected area was traced during annealing. Process Type 2: Co is deposited at the sample temperature of 670 K. The substrate is annealed by the heating plate. The temperature of the substrate is measured with a K-type thermocouple.. 22.

(31) 2-4 The experimental setup and process. Process Type 3: Co is deposited at the sample temperature of 300 K. The substrate is annealed by the heating plate after Co deposition. The temperature of the substrate is measured with a K-type thermocouple. Process Type 3 is applied for the experiment in chap. 3-2, chap. 3-3, and chap. 3-4 as well. Two processes for Co deposition on Ag/Ge(111)surface are developed. One is the substrate being annealed at 670 K during Co deposition (Process Type 2). The other is the substrate being annealed at the same temperature (670 K) after Co deposition (Process Type 3). The experimental results one obtained from the two processes are similar. That is, the structures of Co islands performed in the two processes are the same. In addition, the area covered by Co islands is limited although the amount of Co coverage applied is as much as 4.9 ML [224]. The percentage of Co covered area is saturated at 50% as annealing process happened during Co deposition (Process Type 2). In contrast, the percentage of Co covered area is saturated at 95% for annealing after Co deposition (Process Type 3). That is, the only difference is that the saturated area of Co islands is lager when annealing the sample during Co deposition (Process Type 2) than after Co deposition (Process Type 3).. 23.

(32) 2-4 The experimental setup and process. 2-4-3 Temperature determination for in situ direct annealing at VT-STM stage (Process Type 1) The temperature of substrate is measured by infrared thermometer for the temperature higher than 850 K. For the temperature lower than 850 K, it is estimated by the linear relationship between the Napierian logarithm of the inverse of resistance. and the inverse of temperature. for. semiconductor when the temperature is lower than 850 K [234-235], as shown in Fig. 2-4-4.. 0.0 -0.1 -0.2 -0.3 -0.4 1.00 1.04 1.08 1.12 1.16 1.20. Fig. 2-4-4 The linear relationship between. and. is. applied to extrapolate the temperature of the Ge substrate when it is lower than 850 K.. 24.

(33) Chapter 3 Results and Discussion. 3-1. Thermal. evolution. (Ag-. (. of. Co. -. on. the. ) and. coexisting -. ) phases [236]. One prepares the. surface by depositing 300 K of 1. monolayer (ML) of Ag onto the Ge(111)-c. reconstruction,. then annealing the object at the temperature between 470 K and 830 K. The resulting surface is thus complete covered by the domains of the. phase. For the submonolayer regime,. several other Ag-induced phases have been previously introduced [62, 146-161]. From the earlier studies, one knows that depositing more than 0.1 ML of Ag onto the c. surface (the native. Ge(111) surface reconstruction) results in. domain. When. saturating the phase with more than 0.3 ML of Ag, we observe that the. domains and the developing. coexisted. Yet, comparing to the Ag atom per unit cell, both the. phase. phase that contains 1 and the. phases. contain fractional amounts of Ag atoms in their unit cell. Fig. 3-1-1 is the surface phase diagram for the Ag-terminated Ge (111) surface reproduced by LEED..

(34) 3-1 Thermal evolution of Co on the coexisting …. Annealing Temperature (K) 900 900 800 800 700 700. 8 2 2. 3,5. 4,6. 600 600. 1. 5 11 3 12. 4,6. 13. 1. 14. 14. 8. 7. 500 500. 7. 10. 10. 9. 400 400 300 300 0.2 0.2. 0.4 0.4. 00.6 .6. 1.0 1.0. 0.8 0.8. Ag coverage (ML) Ge-c. Ag-. Ag-. Ag-. (inset). Ag-. Fig. 3-1-1 The phase diagram of submonolayer Ag on Ge(111) surface. Our experiment results are marked with black labels and the results from others are marked with pink labels. The corresponding reference of the numbers are as follows: (1) [146]; (2) [147]; (3) [148]; (4) [149]; (5) [150]; (6) [151]; (7) [152]; (8) [153]; (9) [154]; (10) [155]; (11) [156]; (12) [157]; (13) [158]; (14) [159]. [236]. 26.

(35) 3-1 Thermal evolution of Co on the coexisting …. Fig. 3-1-2 (a) The STM image of Agwith size. /. surface. nm2. The sample bias is 1.28 V. (b) The LEED. pattern of Ag-. /. surface. The primary energy is. 58 eV. (c) The model LEED patterns of. and. periodicities. [236] Using the preparation method described, we fabricate the Ag/Ge(111) surface with coexisting regions of the the. and. phases. Fig. 3-1-2 (a) is the corresponding STM image.. Fig. 3-1-2 (b) shows the LEED pattern, whereas Fig. 3-1-2 (c) illustrates the model LEED pattern of Ag-. /. phase. with 0.3 ML of Ag coverage. From the illustrations, one observes that the small insets of the between the neighboring. phase are clearly resolved domains. It is to believe that the. 27.

(36) 3-1 Thermal evolution of Co on the coexisting …. domains contribute to the total surface insignificantly. Therefore,. hereafter. Ag-. /. we. refer. the. surface. as. an. surface, neglecting the presence of the. domains.. Fig. 3-1-3 (a) The STM image of Ag-. phase with size. nm2. The sample bias is 0.537 V. (b) The model of Agsize Ag-. phase. (c) The STM image of Ag-. phase with. nm2. The sample bias is 0.921 V. (d) The model of phase. [236] The obtained STM images of Ag-. /. show good. agreement to those published in early reports (see Fig. 3-1-3 (a). 28.

(37) 3-1 Thermal evolution of Co on the coexisting …. and (c)) [62, 146-161]. In addition, the models of AgAg-. and. phases (see Fig. 3-1-3 (b) and Fig. 3-1-3 (d)) reveal the. fact that there are more Ag atoms perceived in Agthan in AgAg-. phase. phase. Thus, it is reasonable to ascribe. phase as the Ag-richer phase.. a. b. Ag-4×4. Ag-4×4. Ag-4×4 Ag-4×4 Ag-. Ag-4×4. Ag-. ig. 3-1-4 (a) The STM image of Co on Ag-. /. surface at the annealing temperature of 320K. The sample bias is -1.492 V. (b) The STM image of Co on Ag-. /. surface at 380 K annealing temperature. The sample bias is -0.994 V. The size of both the STM images is. nm2. [236]. For Process Type 1, we perform in-situ observation on the changes. of. Co-covered. -. surface. in. the. morphology that are caused by annealing. Fig. 3-1-4 (a) shows the surface is covered with Co of 0.1 ML at 300 K, and then be annealed at the temperature of 320 K. The surface consists of two upper terraces and one lower terrace, both of which contain 29.

(38) 3-1 Thermal evolution of Co on the coexisting …. phase. From the image, one observes randomly distributed Co clusters among the upper and lower terraces, as represented by bright spots. Annealing the surface at 380 K (see Fig. 3-1-4 (b)), one finds. reconstruction, instead of the previously observed reconstruction, exhibits at the lower terrace. One also. notice that. pattern replaced. pattern in the. sizable area of the left upper terrace. From the examination, it is to believe that a moderate increase in temperature could result a transformation in phase from the. to the. .. Oppositely, we do not observe the transformation from the phase to the the ratio of. phase. Thus, it is to consider that. phase to. phase become smaller as. the annealing temperature increased from 320 K to 380 K. The phase diagram (see Fig. 3-1-1) shows that there is no transition from. phase to. phase occurs after. increasing the temperature. Namely, the ratio of. phase to. phase with an increasing temperature is larger when there is no Co deposits coexisted. On the contrary, we observe smaller ratio when there is Co deposits occurred. That is, as the transformation from Ag-less phase to Ag-richer phase appears after. 30.

(39) 3-1 Thermal evolution of Co on the coexisting …. properly increasing the temperature of the sample with Co deposits. Fig. 3-1-5 shows the structural Co islands formed on the surface that is annealed at the temperature of 480 K or 520 K. One observes that the Co islands get bigger with the increasing annealing temperature (see circles indicated in Fig 3-1-5 (a) and (c)). Also, at the steps, one notices the presence of Ge adatoms that the amount of which is larger when annealing the sample at the temperature of 520 K than of 420 K (see arrows indicated in Fig. 3-1-5 (b) and (d)). However, in previous report, the adatoms were not found for Co formed on pure were. found. only. surface [222-225]. These Ge adatoms on. the. -. Accompanied with the transformation from. surface. phase to. phase, the Ge adatoms are developed upon the broken dangling bonds in Ag-. structure. The Ge atoms later migrate. to the steps, which is similar to the Ge adatoms found at the monoatomic steps on Ge(111)-. surface [237].. 31.

(40) 3-1 Thermal evolution of Co on the coexisting …. Ag-. Ag-4×4. AgAg-4×4. Fig. 3-1-5 (a) The STM image of Co on Ag-. /. surface at the annealing temperature of 480 K. The sample bias is 1.530 V, and the size of the image is. nm2. (b) The inset. of the STM image in (a). (c) The STM image of Co on Ag-. /. surface at the annealing temperature of. 520 K. The sample bias is 1.249 V, and the size of the image is nm2. (d) The inset of the STM image in (c). The adatoms at steps are arrowed and the islands are circled. [236]. 32.

(41) 3-1 Thermal evolution of Co on the coexisting …. For Process Type 2, Co deposition is performed at the sample temperature of 670 K. When choosing studied temperature, we are motivated by the fact that. -. surface with a. submonolayer Co coverage produces the structure of islands that contains either. or. patterns on the tops,. which are usually formed when the surface is annealed at the temperature of 670 K [222-225]. In our experiment, islands with patterns are periodically developed on both Ag-. surface and Ag-. surface, and we can observe. islands formed on the boundaries of Ag-. and Ag-. surfaces as well (see Fig. 3-1-6). Ge adatoms are found at steps (see the arrow in Fig. 3-1-6). Yet, comparing to the example shown in Fig. 3-1-5, the sample demonstrated in Fig. 3-1-6 contains less the amount of Ge adatoms. It is to believe the most of Ge adatoms is consumed to form CoxGey-. islands at steps. Co atoms are given enough. energy to migrate to the steps, forming CoxGey islands at the temperature of 670 K. The temperature is also the condition when Co deposition is performed.. 33.

(42) 3-1 Thermal evolution of Co on the coexisting …. a. AgCo-. Ag-4×4. b. Fig. 3-1-6 (a) STM image of Co islands on Ag-. /. surface. The sample bias is 0.791 V and the size of STM image is nm2. Co coverage is 0.2 ML. (b) The line profile along the blue line in (a). [236] In addition, the question whether an increase in Co coverage will translate into the noticeable differences in morphology of Co on Ag-. /. surfaces is worth studying. We then perform. the experiment following the similar process described (Type 2). The three images illustrate such relation, showing the different amount of Co coverage: 0.2 ML (Fig. 3-1-7 (a)), 0.6 ML (Fig. 3-1-7 (b)), 1.1 ML (Fig. 3-1-7 (c)). Observing the illustrations, one notices 34.

(43) 3-1 Thermal evolution of Co on the coexisting …. that the islands enlarge with an increased Co coverage. Meanwhile, the region of vanishes. domains gets smaller, and that eventually. with. the. -. Co. coverage. islands. along. of. 1.1. ML.. Yet,. some. with. the. islands. of. periodicity are formed at the same time.. Fig. 3-1-7 The STM images of Co islands on Agsurface with size. /. nm2. (a) The Co coverage is 0.2 ML and. the sample bias is 0.791 V. (b) The Co coverage is 0.6 ML and the sample bias is 0.833 V. (c) The Co coverage is 1.1 ML and the sample bias is 0.833 V. [236] To extract more quantitative information, we analyze a number of STM images in large scale. We are able to estimate the contribution of the phase area (. phase and. phase), as well as that of island area to the whole surface in different Co coverage (see Table 3-1-1). It is to believe that an enlarged Co island area leads to a dramatic decrease in the area of the. phase. 35.

(44) 3-1 Thermal evolution of Co on the coexisting …. Table 3-1-1 The percentage of the displayed area of Agphase, Ag-. phase, and Co islands with Co coverage.. [236] Percentage of Displayed Area AgAgCo islands. Co 0.2 ML. Co 0.6 ML. Co 1.1 ML. 25%. 5%. 0%. 65%. 75%. 70%. 10%. 20%. 30%. We argue that the Ag atoms beneath Co islands were removed, migrating to other Ag-. phase to form Ag-. phase. In. other words, there will be no Ag atoms preserved beneath Co islands. The previously removed Ag atoms will move to other area, causing phase transformation from Ag-. to Ag-. .. For Process Type 3, we measure the intensities of Auger signals of Ag (351 eV) and Co (775 eV) when annealing Ag-. /. surface with 1.6 ML Co deposit at the. temperature ranging between 300 K and 930 K. Fig. 3-1-8 shows the typical spectrum. Initially, the Ag signal drops as the temperature increases to a minimum at 570 K. Afterwards, the signal precipitously increases to a maximum at the temperature of 650 K, whereas its intensity corresponds to the value at the 36.

(45) 3-1 Thermal evolution of Co on the coexisting …. temperature of 300 K. It is to observe that the signal intensity decreases again at the temperature of 830 K because Ag desorbs from the surface. At the temperature of 930 K, the intensity eventually reaches zero, which indicates that all Ag atoms have desorbed from the surface. Referring to registered information, the LEED pattern at this temperature reveals c. periodicity,. which is the original for the Ge(111) surface’s reconstruction. Ag is valued as the buffer layer that prevents the diffusion of Co into substrate at annealing temperature ranging between 320 K and 730 K. The intensity of Co signal slowly decreases as the annealing temperature increases, and finally reaches zero at the temperature of 930 K. During the transition from. phase to. phase,. Ag atoms merely change their positions at the topmost layer. We thus argue that the amount of Ag atom on the topmost layer is conserved. It is to believe that the registered intensity of Ag signal is not sensitive to the process. Base on the study, one supposes that the observed alterations in the signal intensities reflect to other interaction between Co and the Aghave found that. -. /. surface. We. structure is developed. corresponding to the decreased Ag auger intensity at the. 37.

(46) 3-1 Thermal evolution of Co on the coexisting …. temperature of 570 K. The formation of Co-Ge alloy indicates that some Ge atoms have exchanged with Ag atoms. Thus, the depth of Ag atoms is deeper than the one before Co-Ge alloy formation, which suggests a decreasing in Ag Auger intensity.. AES intensity (a.u.). 6000. Ag 351eV Co 775eV. 5000 4000 3000 2000 1000 0. 300 400 500 600 700 800 900 1000 Annealing Temperature (K). Fig. 3-1-8 The Auger intensity of Ag and Co in related to the annealing temperature. [236] As the temperature increases above 650 K, the islands with periodicity. transform. to. those. with. periodicity. From the previous studies, one believes that Co islands with. periodicity are higher than those with. periodicity that reduces the area of Co islands [222-225]. Thus, there are more Ag atoms revealed on the topmost layer than at the. 38.

(47) 3-1 Thermal evolution of Co on the coexisting …. annealing temperature of 570 K. It results that the Ag Anger intensity reaches the amount similar to the one pre-annealing. Ag/Ge(111)-. and Ag/Ge(111)-. domains were. observed by LEED after Co diffusion into bulk at the temperature of 830 K, however, their boundaries located differently comparing to those before Co deposition. Fig. 3-1-9 and Fig. 3-1-10 respectively show the process of Co interaction with Ag/Ge(111)phase and Ag/Ge(111)-. phase. Although the formation of. islands commonly occurs in both phases, it involves different underlying. processes. according. to. each. condition.. The. interpretation presenting above demonstrates the stages of Co interacting with Ag-. or Ag-. surface. Below, we briefly. encapsulate the proposed scenario, which is specific to the phase of Ag/Ge(111) surface. (i) Co on Ag/Ge(111)-. surface (Fig. 3-1-9): Firstly, the. deposited atoms form clusters on the surface (a). Co atoms are prevented from diffusing into bulk by the intermediate Ag layer. When the temperature increases, Co clusters commence to nucleate (b). Secondly, annealing the surface, Ge atoms locally exchange with Ag atoms to form the alloy with Co atoms. This reaction leads to the creation of CoxGey alloy islands, which is proved by the observed. 39.

(48) 3-1 Thermal evolution of Co on the coexisting …. periodicity on the top and the drop of Ag AES intensity (c). Thirdly, Ag and Ge atoms exchange again so that the CoxGey-. islands are transformed into Co-. islands.. The new-formed islands are higher, whereas their lateral dimensions are smaller comparing to the counterparts (d). As the islands experience thermal degradation, the Co atoms diffuse into the Ge bulk (e). Further temperature increase leads to the gradual surface deterioration due to the fact that Ag atoms desorbed from the surface. Finally, the whole surface will be converted into the Ge(111)-c. surface, from which Co atoms are not detectable. (f).. a. b. T:300 K. T:420 K. Ge. Ge. c. d. e. f. T:570 K. T:650 K. T:830 K. T:930 K. Ge. Ge. Ge. Ge. Ag-. Ag atoms. CoxGey-. Co-. Fig. 3-1-9 The model of 1.0 ML Co coverage on Ag-. Co atoms surface.. The annealing temperature rising from 300 K to 930 K. [236]. 40.

(49) 3-1 Thermal evolution of Co on the coexisting …. (ii) Co on Ag/Ge(111)-. surface (Fig. 3-1-10): Firstly, the. deposited Co atoms form clusters on the surface (a), as the case previously discussed. However, in contrast to the example on Ag/Ge(111)-. surface, the intermediate Ag-. layer. does not prevent the deposit from diffusion inward because half of the. phase is composed of Ge atoms only. Thus, a certain. number of Co atoms dissolve into bulk as the temperature rises. Meanwhile, Ag atoms were removed from the beneath of Co atoms, which leads to a partial conversion of the (. phase to the. phase (b). Secondly, further annealing the surface, Ge. atoms exceed the layer of Ag atoms to form CoxGey-. . It is. to suggest that the Ag atoms were forced to move to other locations at the topmost layer, which could cause the phase transformation from Ag-. to Ag-. . In other words, at this stage, it is the. formation of CoxGeyAg-. phase to Ag-. leads to transformation from phase (c). Thirdly, Ag and Ge atoms. exchange again so that CoxGeyinto Co-. islands are transformed. islands (d). As the islands encounter thermal. degradation, the increasing number of Co atoms diffuses inward through the Ge bulk. Finally, the surface is transformed back to. 41.

(50) 3-1 Thermal evolution of Co on the coexisting …. Ag-. phase (e). The Ag atoms desorbed from the surface,. forming Ge(111)-c. surface again.. a. b. c. d. e. f. T:300 K. T:420 K. T:570 K. T:650 K. T:830 K. T:930 K. Ge. Ge. Ge. Ge. Ge. Ge. Ag-. Ag-4. Ag atoms. CoxGey-. Co-. Co atoms. Fig. 3-1-10 The model of 1.0 ML Co coverage on Ag-. surface.. The annealing temperature is rising from 300 K to 930 K. [236]. 42.

(51) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. 3-2 Structure of Co-. nanoislands on Ag/Ge(111)-. surface [238]. b. a Co-. c. CoxGey-. Fig. 3-2-1 (a) -. nm2 STM image showing the coexistence of and. Ag/Ge(111)-. -. islands. surface. (b). hexagonal-shaped Co(11-20)-. grown. on. nm2 STM image of island of 0.4 nm height. (c). nm2 STM image of stripe-like Co(11-20)-. island of. 3.7 nm height. In (b) and (c), the angles between the atomic rows are given. [238] Observing the image in Fig. 3-2-1 (a), one notices the desirable reconstructions showing at the top of most Co islands grown on Ag/Ge(111)-. surface. Although the unwanted. reconstruction is also perceived, the amount of which. 43.

(52) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. is insignificant. According to the previous observation, the structure of the Ge(111) surface significantly influenced the structural properties of the Co islands grown on Ag/Ge(111)-. surface. [222]. It is the strong coupling between the Co atoms and the Ge(111) surface that contribute to the phenomenon. Thus, it is reasonable to neglect the influence of the Ag layer in the following consideration. Based on the condition, we propose a model for the Co-. phase growth on Ge(111) surface. For Fig. 3-2-1 (a), the Co-. islands adopt either hexagonal. or stripe-like shapes. Fig. 3-2-1 (b) and (c) demonstrate an insight into the inner structure of the individual island. Roughly inspecting Fig. 3-2-1 (b), the hexagonal island seems to contain unit cell similar to that of Ge(111)-. surface in parallelogram shape.. The phenomenon indicates that the substrate surface strongly influences the structure of the growing island. However, studying the image carefully, one notices that the angle of the intersection between the island-rows running in the directions shown in Fig. 3-2-1 (b) is appeared to be larger than the typical of Ge(111)surface (i.e., 70° vs. 60°). The example represented in Fig. 3-2-1 (c) even demonstrates the intersection angle at 77°.. 44.

(53) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. Further analyzing a number of images, we are confirmed that the above-mentioned discrepancy is distinct for the stripe-like islands, which are generally higher than their hexagonal counterparts. These observations have important implications on modeling the island growth because we can then speculate the island growth that involves a distortion of the island rows through referring the substrate surface rows. In addition, the degree of the distortion enhance with the increasing island height. Consequently, as the shape of the island unit cells changes noticeably (e.g. transforming from parallelogram to rectangular), island itself is undergoing a process of inner structure alterations. In other words, as their height grown, the islands are undergoing an evolution, transforming the form hexagonal-shape into stripe-shape. Based on the ideas mentioned above, we focus on the stripe-shaped islands as more evolved phase in an attempt to construct the model demonstrating the growth of the Co-. islands.. To ascribe a specific crystallographic structure to the Coislands, it is important to compare Co-. islands with typical. bulk Co, which is in hcp structure. From the previous researches, one knows that Co layers are also in hcp form when which is grown on metallic surface [190-191]. However, it was demonstrated that. 45.

(54) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. the Co layer can grow epitaxially on the Au(111) surface in the fcc structure when it is prepared under special conditions [192-194]. Moreover, Tonner et al. have shown that it is the different thickness of the deposited films influencing the established structure of Co layers [195]. From the observations and the previous studies, it is to believe that both fcc and hcp are possible structure of Co layer. Although the initial Co layer growing on clean Cu(111) surface proceeds in fcc phase, it will transform from the original structure to hcp structure as the film thickness increases beyond two layers [195]. Table 3-2-1 demonstrates the lattice parameters and the value of interlayer spacing from low-index planes with Co in bcc, fcc, and hcp structure respectively, as well as those from Ge(111)surface. Comparing the lattice parameters of the substrate and the various Co surface phases, we identify the Co-hcp(11-20) and the Co-bcc(111) planes by noticing their smallest lattice mismatch. To recognize the exact phase, it is a must to take account of the unit cell shapes displayed in Table 3-2-1 with those shown in the STM images. (see. Fig.. 3-2-1).. Realizing. the. fact. that. the. rectangular-shaped unit cells of hcp(11-20) plane and the typical shape of the evolved Co-. islands are similar, we conclude that. 46.

(55) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. the islands are oriented by hcp(11-20) plane rather than bcc(111) plane.. Table 3-2-1 Geometrical characteristics of low-index Co crystallographic planes in bcc, fcc, and hcp structures. [238] Lattice. Interlayer. parameters (pm). distance (pm). Co-fcc(100). 251×251. 177. Co-fcc(110). 251×355. 125. Co-fcc(111). 251×251. 205. Co-hcp(0001). 251×251. 204. Co-hcp(11-20). 434×407. 125. Co-hcp(10-10). 251×407. Co-bcc(100). 284×284. 142. Co-bcc(110). 284×402. 201. Co-bcc(111). 402×402. 82 164. Ge (111). 400×400. Crystallograph ic orientation. Lattice pattern. 47. 72 145. 82 245.

(56) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. a 11. [[11 11 -2 - 20]0 ] 2 2. 33. 44. 55 [ 00 00 0 [0 1]1 ]. [ 1[1 - 1-10000] ]. b. [0 0 0 1]. [1 -1 0 0]. [[1 1 11 --220] 0]. Co atoms in 1st, 3rd, 5th, ... layers Co atoms in 2nd, 4th, 6th, ...layers. Fig. 3-2-2 Schematic diagram of hcp structure. (a) Numbered balls represent atoms which belong to successive atomic layers, perpendicular to [11-20] crystallographic direction. The open circles refer to the layers with odd numbers, while the filled circles represent those with even numbers. (b) The top view of hcp(11-20) structure, the stacking sequence is ABABAB…. [238] 48.

(57) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. Fig 3-2-2 (a) illustrates a schematic diagram for the hcp structure. The lattice parameters of the Co(11-20)-. unit are. measured 407 pm along [0001] direction, and 434 pm along [1-100] direction. Observing the fact that the value for the interlayer spacing (125 pm) is by far lower than the lattice parameters (407 pm and 434 pm), it is to suggest that there is a bigger chance that the. islands. form. bilayered. structure. than. single-layered. construction. Considering the (11-20) surface, one notices from the schematic diagram that the stacking sequence is ABABAB…: one is represented in the first, the third, and the fifth layer as viewed from the top (layer A), whereas the other is demonstrated in the second, the fourth, and the sixth layer (layer B) (see Fig. 3-2-2 (b)). To verify the assumption, it is a must to study the line profiles. Firstly, we measure which along the Co-. islands randomly. chosen from a number of large-scaled STM images that were acquired under similar tunneling conditions. Secondly, based on their height, we number the islands in increasing order, starting the lowest. In case which two or more islands are measured with the same height, we ascribe them according to their successive numbers. From the obtained data, one notices that the majority of. 49.

(58) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. islands have the equal height of either 320 or 535 pm (see Fig. 3-2-3). To proof the bilayered structure of Co-. islands, we. propose the strategy applying the discrepancy between the measured and the theoretical height. When the imaging objects are different in conducting properties, inconsistency occurs between the measured height (i.e., the discrepancy of 320 and 535 pm, 215 pm) and the theoretical height of the bilayered Co-(11-20) (i.e., 250 pm) due to the tip-caused effects. Moreover, one finds the above-mentioned. issue. even. more critical. in. case of. a. metal/semiconductor system.. 2000. Height (pm). 1500. 1000. 500. 0. 0. 100. 200. 300. 400. Island number Fig. 3-2-3 Distribution of height within a population of 500 Co-. islands. [238] 50.

(59) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. a pseudo[0001]. pseudo[1-100]. b. [0001. [1-100]. ]. 90 ∘. 60∘. c 1 2×2 1×1. 2 Fig. 3-2-4 Schematic diagrams of Co-(11-20) planes. The squares represent the positions of Co atoms in bilayers. The open circle corresponds to electron clouds of groups of four neighboring atoms. The pink filled circles refer to a group of six atoms in lower bilayer, and the yellow filled circles are the group in upper bilayer. (a) Array of parallelogram-shaped Co-. unit cells. Direction of. row shift is marked with arrows. (b) Array of rectangular Counit cells. (c) Array of parallelogram Cobackground of Co-. unit cells against a. lattice. Two different directions for the. island growth are marked with numbered arrows. [238] 51.

(60) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. As proposed above, Ge(111)-. surface is acting as a. template for Co island. Fig. 3-2-4 (a) represents the imagined configuration. of. the. surface,. which. is. an. array. of. parallelogram-shaped unit cells. The unit cells of Co island adopt the shape of the substrate surface in their early growing stage. Yet, referring to the substrate surface rows, the growing island rows will undergo a distortion as they evolved (see Fig. 3-2-4 (a), the direction of the distortion is shown by arrows). Eventually, the unit cells of Co island will adapt to the rectangular shape. Fig. 3-2-4 (b) shows the schematic structure of the resultant Coproposed model (i.e., Co-. . The. ) shows no correspondence in the. periodicities of Co island (i.e., Co-. ). However, referring to the. resent findings, it is to believe that under metal/semiconductor epitaxial system, the protrusions appearing on the STM image corresponds to the sites where the electron clouds of atoms group concentrated, rather than representing the individual atoms [239]. To identify the sites, we consider two alternative positions. The yellow filled circles represent the sites at which the electron clouds of six neighboring atoms group are likely to accumulate, whereas the open circles refers to the alternative site at which the electron clouds of four neighboring atoms group accumulated (see Fig. 3-2-4. 52.

(61) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. (c)). From the image, one observe that the open circles have a lower probability to be detected by the tip, as compared to the yellow filled circles. Therefore, in case when the open circles do not show on the STM image that the protrusions form a distinct pattern, it is because the represented sites are enclosed by the other position, which is indicated by the yellow filled circles. The arrangements of atoms are the same in the first bilayer (AB) and in the second bilayer (AB) because the stacking sequence of hcp(11-20) surface is ABABAB…. One may suppose that, in STM image, the protrusions in the first bilayer overlap the one in the second bilayer. However, we shall emphasize on the fact which the protrusions in the first bilayer do not overlap the one in the second bilayer as seen in Fig. 3-2-5. Namely, the atomic positions of the sites that the electron clouds are accumulated are not identical in the successive bilayers (11-20). Observing Fig. 3-2-4 (c), the pink filled circles represent the accumulated sites for the electron clouds of six neighboring Co atoms group positioned in the lower bilayer. Due to charge repulsion, the positions of the electron clouds of similar atoms group shift from the topmost bilayer (represented as yellow filled circles) to the sites, enclosing by the pink filled circles. In addition,. 53.

(62) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. one notice that there are two different directions for the island growth (see Fig. 3-2-4 (c), denoted as 1 and 2). Growing along direction 1, the rows in the topmost bilayer are consistent with respect to those in the underlying bilayer. For direction 2, however, the topmost and the underlying bilayers do shift with respect to each other. Fig. 3-2-5 contains illustrations of the resolved STM images of the Co(11-20)-. islands grown along direction 1 (see Fig. 3-2-5. (a)) and direction 2 (see Fig. 3-2-5 (b)), as well as their corresponding schematic diagrams (Fig. 3-2-5 (c) and (d)). For Fig. 3-2-5 (a) and (b), the rows in the topmost bilayers are indicated by black lines, whereas those in the underlying bilayers are suggested by white lines. Regarding Fig. 3-2-5(c) and (d), filled and open circles represent the protrusions in the topmost and underlying bilayers respectively. Observing the images, one notice a lateral shift between the atomic rows in the topmost and the underlying bilayers as the islands grow along direction 2, whereas no shift is perceived as the islands grow along direction 1.. 54.

(63) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. a. b. c. Fig. 3-2-5 (a). d. nm2 STM images showing Co-(11-20)-. islands growing along the pseudo-[1-100] direction, and (b) the pseudo-[1-100] direction. The black lines represent the rows in the topmost bilayers while the white lines refer to the rows in the underlying layers. Corresponding schematic diagrams are shown in (c) and (d), respectively. The filled circles refer to the protrusions in the topmost bilayer, whereas the open circles represent the atomic protrusions in the underlying bilayer. [238] From statistical analysis, it is to understand that 90% of the Co(11-20)-. islands grow along direction 2. The phenomenon. indicates that there is least lattice inconsistency between the newly grown island and its substrate growing along direction 2. Based on 55.

(64) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. our observation, the amount of lattice misfit between the unit cell of the rectangular-shaped Co(11-20)-. and the Ge(111)-1. is. 2% when the islands grow along the direction [0001], and 9% when which grow along the direction [1-100]. From the obtained results, we assign direction 2 as [0001] direction, whereas direction 1 as [1-100] direction. In addition, since ideal rectangular shape is not observed in the unit cells of the growing island, we ascribe the two directions. to. pseudo-[0001]. direction. direction.. 56. and. pseudo-[1-100].

(65) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. a. b 1st bilayer. [0001]. 2nd bilayer 3rd bilayer. Fig 3-2-6 (a) STM image showing the Co-(11-20)-. island. which is composed of three bilayers. (b) The differentiated image of (a). Lateral shifts between the successive bilayers indicate that the island grew along the pseudo-[0001] direction. [238]. 57.

(66) 3-2 Structure of Co-2×2 nanoislands on Ag/Ge(111)-√3×√3 surface. So far, we have focused on the Co-. islands composed of. two bilayers. For further understanding, one must also consider those Co-. islands constituting multi-bilayered systems. observed in the STM images. Fig. 3-2-6 shows the island composed of three bilayers. To obtain more detailed information about the islands growth, we study their structure appearing at the boundary between the successive bilayers with the focus on the grown direction of each bilayer. One observe a lateral shift at the boundary between the first and the second bilayers, which indicates that the bilayers have grown at the pseudo-[0001] direction. Moreover, the similar lateral shift has been found also at the boundary between the second and the third bilayers. Based on the information mentioned above, it is to believe that there exists only one preferential direction for a particular island to grow. Having analyzed a number of similar multi-bilayered islands, we conclude that it is correct for Co(11-20)Ge(111) surface.. 58. islands that are grown on.

(67) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. 3-3 Strain–relaxation in the structure of CoAg/Ge(111)-. islands on. surface [240]. a. b. c. Fig. 3-3-1 (a) An empty-state STM image (Vs=0.5 V, size= nm2). Areas enclosed with square boxes are shown in details in (b) and (c). (b) A typical image of Coof the. -. island. (c) STM image. surface shows the honeycomb. structure. In (b) and (c), the marks indicate the crystallographic directions along which line profiles were measured. [240]. 59.

(68) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. Fig. 3-3-1 (a) is a typical empty-state STM image obtained after annealing Ag/Ge(111)-. surface with Co of submonolayer. coverage at 670 K. From the image, one observes the islands coexisting with those in. reconstruction. In. the inter-island area, a defect-free. surface structure is. fully conserved, which indicates that there is no chemical interaction occurs between the deposits and the surface. As suggested in our earlier paper, one must consider the influence from the substrate surface exhibited on the structure of the resultant islands, even though it is known that there is no chemical reaction between Co atoms and the substrate surface at the annealing temperature ranging from 320 K to 730 K [222-225]. One knows the fact that CoxGey-. is formed of Co-Ge alloy. and it is lower than its counterparts, Cothe idea, CoxGey-. in height. Based on. islands are expected to adapt the. atomic compactness of the substrate By analyzing a number of the Co xGey-. islands. regardless of size, we realize a lattice constant that is fixed to the amounts of 1.44 nm, the value similar to which of the substrate surface. However, for the CoCoxGey-. islands, which are higher than. islands, we expect less influence from the. 60.

(69) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. substrate on the inner atomic compactness. To proof the hypothesis, we define the following component so called strain–relaxation factor (hereafter relaxation):. (3-3-1). where and. and. -. -. are lattice constants of Co-. -. units,. respectively.. The. defined. relaxation is the measurement of the island atomic compactness with reference to the substrate surface. We have calculated 150 examples of Co islands with. periodicity randomly selected. from a large number of STM images. The method we applied to obtain the data needed determining the value is demonstrated as following. Firstly, we perform the line profiles of each investigated island at the. ,. , and. directions (see Fig. 3-3-1 (b)). Through the procedure, we are able to obtain three values from each island regarding to the corrugation periodicity. Secondly, we calculate the average corrugation period of each island. Thirdly, in order to calibrate the length scale, we find the corrugation period of the substrate surface by performing line profiles at the. ,. and. directions (see Fig. 3-3-1 (c)), and employing the 61.

(70) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. similar procedure as previously described. Fourthly, we evaluate the ratio of the island corrugation period to the substrate surface corrugation period. Finally, we multiply the known lattice constant of Ge(111)-. substrate surface by the calculated ratio to. find the island lattice constant. The process helps to minimize the tip-caused errors, which usually occur when obtain the distance using STM image.. Fig. 3-3-2 Relaxation factor vs. island height calculated according to Eq. (3-3-1) for a population of 150 Co-. islands. [240]. Fig. 3-3-2 shows a distribution of relaxation values according to the island height. Observing the negative value, it is to believe. 62.

(71) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. Co-. islands has a more compact atomic arrangement than. substrate does. There are only few islands that the relaxation amount reaches to or slightly above zero. Also, for the island with the height corresponding to 3 and 4 monolayers (1 - 2 bilayers as suggested in chap. 3-2), one observes a wide range of the relaxation change, reaching to. 12 % or. 16 % at the most. On the other. hands, it is to notice that the variations in relaxation for those islands with 5 ML of height (3 bilayers as suggested in chap. 3-2) are smaller, typically less than. %. However, having studied the. data, it is to believe that there is no distinct relation between the relaxation and the island height. From the observation, it is reasonable to neglect the effect of the island height in further considerations. Fig. 3-3-3 shows the relationship between relaxation value and island size. By analyzing the data, the following conclusions can be drawn. Firstly, although it is to believe that the size of the Coislands ranges from 5 nm2 to 925 nm2, most of the islands observed are less than 600 nm2. Also, roughly 30% of the islands are found smaller than 60 nm2. Secondly, a majority of the examples has a negative value in relaxation that is the same as the previous discussed relation. There are only a few islands that the relaxation. 63.

(72) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. amounts reach to or slightly above zero. It seems that the phenomenon occurs randomly without any preference to a specific island size. Thirdly, the amount of relaxation is less than. % for. the islands sizing between 80 nm 2 and 600 nm2. To simplify the discussion, we assume that there is a constant amount of relaxation for the sizing range. For most of the islands of sizes smaller than 40 nm2, the amount of relaxation ranges between. % and. There are only a few examples that the value exceeds – reaching to. %. %,. % at the extreme.. Fig. 3-3-3 Relaxation factor for a population of 150 Coislands. Inset shows the relation under interest for a population of islands with the size in the range from 5 nm2 to 130 nm2. [240] From our experiment, it is to believe that the atomic arrangement in Co-. islands is denser than which in the 64.

(73) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. substrate surface, with few exceptions. In addition, relatively small islands are more likely to adapt a more compact atomic arrangement comparing to which in those bigger islands. Similar conclusion is reached for the Co growth on Cu(100) and Cu(111) surfaces [241-242]. To find the answer for the question why smaller islands tend to display a larger relaxation than the bigger islands do, we refer to previous study, which indicates that Young's modulus decreases with grain size for the nanometer-size objects [243]. Consequently, it is easier to change the structural parameters in a smaller island than a bigger one. The fact may explain the relation between the relaxation and the island size mentioned above. To consider the phenomenon even further, we select a large number of Co-. islands that are less than 150 nm 2, as they are. the most representative examples in strain–driven effects. Two types of the Co(i) Co-. islands are classified: islands with the size smaller than 80 nm 2 show a. large variation in the relaxation (i.e. from – (ii) Co-. % to –. %).. islands with the size larger than 80 nm2 are likely. to show a constant relaxation, which is less than –. 65. %..

(74) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. To begin the discussion, one refers to the documents mentioning the theoretical framework of the island growth process [244-247]. From which, it is to consider the formation of islands as an efficient mechanism relaxing the strained heteroepitaxial layers. Approaching from the thermo-dynamical point of view, Tershoff and Tromp demonstrated that initially the strained islands have square shapes, which are governed by free energy of steps bounding the island [244]. However, it is found that after they reach beyond the critical size, the strained islands adapt a more elongated shape due to the dominant strain energy. On the other hands, Li et al. have demonstrated that it is the relative strength of step and the strain energy driving the evolution of island shape with increasing size [245]. More recently, it has been assumed that there exists a regime near the critical island size where the square islands and those elongated shapes coexisted [246-247]. Roughly inspecting Co-. islands, one is indicated that the. island shapes do not vary with island size. However, on close inspection, one observes a compact atomic configuration within the island, which is similar to the theory. Comparing the atomic arrangement of the two different island shapes with an identical area, one notices that the one of square island is more compact than. 66.

(75) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. the other of elongated island. Recalling the theory, the island shape does evolve from a square to a more elongated shape. In other words, the atomic arrangement of the island transforms from a more compact to a less compact structure, which is indeed shown in our experiment. In addition, it is to believe that step energy is influential to the compactness of atomic arrangement. For small islands, there is a minimum. step. free. energy. resulting. in. compact. atomic. arrangement, whereas for big islands, the predominated strain energy over the step energy, leading to a less compact atomic arrangement. Observing the coexistence of islands in different levels of atomic compactness, it is to suggest that the size of Co-. islands we manage to fabricate falls into the range close to. the critical size, which is the maximum value for the production of islands in compact atomic arrangement.. 67.

(76) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. a. b. Fig. 3-3-4. c. nm2 high resolution STM images of (a) ordered. and (b) disordered small. -. islands. The inter-row. distances in (a) are fixed, while in (b), the distance, between the rows labeled as 1 and 2, is larger than which between 2 and 3, and which between 3 and 4. (c). nm2 STM image of Co-. island with size 111 nm2. The inter-row distances are fixed. [240] For small islands, it is interesting to find out whether the differences in relaxation are translated to the differences in structure. We thus offer a detailed inspection on the corresponding STM images. Fig. 3-3-4 (a) is the STM image of the island in the size of 14 nm2, whereas Fig. 3-3-4 (b) is that of the island in the size of . The. % difference of the two in measured lattice. constants (0.76 nm vs. 0.80 nm) indicates a discrepancy in a degree 68.

(77) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. of the atomic compactness. Further inspecting the images, one observed. another. variation. in the distance between. the. neighboring atomic rows. For the island shown in Fig. 3-3-4 (a), the distances between the rows are measured 0.82 nm, which is a fixed value (see the inset in Fig. 3-3-4 (a), the rows under discussion are labeled as 1, 2, 3 and 4). On the other hands, Fig. 3-3-4 (b) shows the island that the distances are inconsistent: two of which are measured 0.7 nm (see the inset in Fig. 3-3-4 (b), the discussed rows are labeled as 2, 3, 4), and one of them is larger, measuring 0.85 nm (i.e. the distance between the rows labeled as 1 and 2). Fig. 3-3-4 (c) shows an STM image of a relatively big island, which the size amounts to 100 nm2 and the relaxation inspects to be. %.. Studying the inner structure of the island, one observed the fact that the distances between the neighboring rows are fixed (see the inset in Fig. 3-3-4 (c), the discussed rows are labeled as 1, 2, 3, 4, 5, 6, and 7). Having analyzed a number of images, we noticed that unfixed inter-row distances in the same island are typical only for some small islands. In addition, for a particular island, the inter-row distance which is different from the other appears between the row located at the island edge (i.e. the first or the last row) and its. 69.

(78) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. nearest neighbor. Thus, it is reasonable to attribute the above-mentioned phenomena as edge-effect. We propose that the ratio of the island edge length to the island total area directly influences the magnitude of the effect. One observed the deviations in distances between the rows located near the island edge in small islands due to the large edge-to-area ratio. However, we have not found a critical value for the edge-to-area ratio beyond which the effect is not shown. To account the edge-effect, one must realize that whereas free energy of step is believed as a dominant factor in deciding the degree of the atomic compactness in small islands, but the strain also influence the atomic configuration due to the fact that the periodicity of the substrate the island. is different from the one of. . Although the significance of strain energy is. considered minor comparing to step energy, and is restricted to the island edge, we shall propose the magnitude of strain as the other factor that affect the atomic arrangement near island edge. We suggest that the strain was influenced by the way that the island edge is positioned against the background of the substrate surface.. 70.

(79) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. a. b. c. d. Fig. 3-3-5 Schematic diagrams showing several possible locations of Co-. islands on the Ag/Ge(111)-. surface. Filled. circles correspond to Co atoms, while pink solid lines mark the edges of Coindividual. islands. Thin dashed lines separate the unit cells. The black solid lines mark the. boundaries of the. unit cells which are in close vicinity. of the island edges. [240] Fig. 3-3-5 demonstrates several possible arrangements for Co-. islands on the Ag/Ge(111)-. template. For. clarification, the filled circles represent Co atoms, whereas the pink solid line encloses Cothe individual. islands. The thin dashed lines separate unit cells, and black solid lines represent. the approximate boundaries for Co islands in relation to 71.

(80) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. Ag-. unit cells. Studying different cases for the relative. position of the island concerning the Ag/Ge(111)boundaries (see Fig. 3-3-5, only a few is shown), it is to believe that the degree of strain influenced by the island edge varies over a wide range for the whole population of CoAg/Ge(111)-. islands grown on the. surface. One is confirmed that a critical strain. strength for the occurrence of the edge-effect existed. Yet, to further test our proposal, more investigations are needed. We realize that for big islands, with smaller edge-to-area ratio, it is the strain energy contributing primarily, not the edge-effect. We name the objects orderly for those the edge-effect is not exhibited, whereas disorderly for which the edge-effect is shown. In the context, we understand that small islands could be in either ordered or disordered arrangement, whereas big islands commonly show a ordered structure. It is to believe the two types of atomic arrangement are the dominant reason leads to the large differences in the relaxation within the small islands.. 72.

(81) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. For instructive reasons, the island is represented by rings of Co atoms surrounded concentrically the central Co atom in our model illustrated (see Fig. 3-3-6, the central atoms are labeled as A and a, respectively). It is to assume that the atom positions at the geometrical center of the unit cell of the fundamental substrate. Yet, considering Co–Co interaction, the positions of Co atoms besides the central one must shift to a certain degree in terms of the centers of the Ge unit cells. The larger the ring radius measures, the more shifting of the position of Co atoms occurs, with respect to the center of the substrate unit cell. Due to the consecutive shifts, Co atoms located at the last ring are within close vicinity at the edge of the substrate unit cell (see Fig. 3-3-6 (a), which atoms B represents the last ring of the big island). It is to assume that the positions of Co atoms in the ultimate ring are similar between a small island (see Fig. 3-3-6 (b)) and a big island. Thus, we believe that the lattice constant for a smaller island is smaller than that for a bigger island. The fact indicates that the atomic arrangement is more compact in small island than in big island, and it is indeed observed in our STM images.. 73.

(82) 3-3 Strain-relaxation in the structure of Co-2×2 islands on …. a. b. Fig. 3-3-6 Schematic diagrams of big (a) and small (b) Coislands. Balls represent Co atoms against a background of the honeycomb-like unit cells of the underlying Ge substrate surface. Co atoms form rings concentrically surrounding the atom situated in the center of islands (in (a) and (b) marked as A and a, respectively). B and b refer to Co atoms situated in the last rings. [240]. 74.