二維離子薄膜之結構相變與電子特性

Abstract

近年來異質磊晶成為製造超大型積體電路的一個重要技術,我們也經常藉由異質磊晶來研究不同材料之間的介面特性，希望藉由深入了解磊晶成長機制，進而能自由的操控薄膜特性。氯化鈉和矽有相近的晶格常數，但在原子間的鍵結本質上有很大的差異，氯和矽的電負度差是1.2，而和鈉的電負度差是2.1，這樣的離子共價介面存在許多有趣的問題:氯化鈉和矽的介面將傾向形成離子鍵或共價鍵?在介面又會存在甚麼樣的物理特性?我們可否人工製造出單原子層的離子薄膜? 在本論文中，我們結合了原子層沉積法與自我限制反應的概念，設計了一個兩階段的磊晶方法來研究這些問題。 首先，我們利用表面矽原子懸鍵的活性製造出一個有序的氯/矽(100)面，然後蒸鍍不同數量的鈉原子在這個表面上，如此便可直接的觀察三種原子之間的交互作用，我們同時利用掃描隧道顯微鏡（STM），X射線光電子能譜（XPS）和第一原理計算（DFT）來研究薄膜成長過程中的表面形貌，電子狀態和界面性質的演變。STM結果顯示，當少量的鈉吸附時表面同時存在三種結構有p(2x2)、單一暗點、和一大塊的暗斑。當鈉覆蓋率達到0.5 ML，表面會形成唯一的p(2x2)有序結構，當覆蓋超過 1 ML時，表面會戲劇性的轉變為大規模且均勻的p(1x1)結構。經由高度及距離量測，我們推斷其應為二維NaCl(100)薄膜。XPS結果則顯示，隨著鈉覆蓋增加，Cl 2p和Na 2p的束縛能同步增加約 1.1 eV。它反映出磊晶過程中產生不同的鍵結形態。當鈉覆蓋率超過1 ML，Si 2p光譜顯示表面矽原子的電子恢復成懸鍵狀態，此時氯化鈉薄膜與基底只存在微弱的凡德瓦作用力。 第一原理計算結果顯示，單一鈉原子喜歡吸附在氯原子之間的洞位，且鈉原子可以很容易地沿二聚體擴散組裝成氯化鈉（111）叢集結構。此結構對應於STM圖中的單一暗點，當叢集密度增加後表面將從氯化鈉（111）轉化為氯化鈉（111）-（100）共面結構，此結構對應於STM圖中的p(2x2)有序結構，當鈉原子覆蓋率超過1 ML，氯、鈉原子將自組裝成氯化鈉（100）薄膜，對應於STM圖中的p(1x1) 有序結構。整個磊晶成長過程遵循Franck-van der Merve成長模式。最後我們計算了該系統一系列的電子特性，包括電子能態密度、空間分布，STM影像模擬以及XPS光譜模擬，模擬結果與我們的實驗相當吻合，它不僅成功解釋了實驗數據，更幫助我們了解深層的物理意義。

Heteroepitaxy has now become an important technique for fabricating very-large scale integrated circuits (VLSI). It also serves as an ideal sample preparation tool for understanding the nature in heterogeneous interface. Furthermore, one hopes to gain sufficient insight to manipulate the thin film structures with desired properties. Silicon and sodium chloride have similar lattice constant but much difference in crystal structure (fcc & diamond). The electronegativity difference between chlorine and silicon is 1.2 and that between chlorine and sodium is 2.1. Many interesting questions can be raised regarding the ionic crystal-covalent crystal interface: What correlations can be drawn about the difference in electronegativity between bonded atoms and their tendency to form ionic or covalent bonds? What kind of physical properties exist in interface? Can we produce a NaCl thin film of only single layer in thickness? In this thesis, we designed a two-step epitaxy to research these issues through a combination of atomic layer deposition (ALD) and self-limited concept. First, we prepared a oriented Cl/Si(100)-(2x1) surface, and then deposit various amount of sodium atoms on the surface. On the sample surface, we can directly observe the interaction between three species of atoms. The evolution of the surface morphology, the electronic state, and the interfacial properties during the film growth are investigated by scanning tunneling microscopy (STM), X-ray photoelectron spectroscopy (XPS) and first principle calculations (DFT). The STM result shows, in low sodium coverage,the surface consists three kind of adsorption structures, the single dim spot, the p(2x2) domain and a domain of dim spots. At intermediate coverage, the surface would transform into unique p(2x2) ordered structure. When the coverage over 1 ML, the surface dramatically transformed into large scale and uniform p(1x1) structure. By a steric measurement, we concluded the structure is a two dimensional NaCl(100) film. The XPS spectra indicate, with increasing of sodium coverage, the Cl2p and Na2p peaks synchronously increase by about 1.1 eV. It suggests significant transition in their bonded electronic state during the Na adsorption process. The Si 2p spectra indicate, when sodium coverage is over 1 ML, the electron state in surface silicon retrieved dangling bond state; namely, the Van der Waals force dominated the interaction between the NaCl thin film and silicon substrate. From first principle calculation, we showed that the single sodium prefer to adsorb on hollow sites between chlorine atoms and that the sodium atom can easily diffuse along the dimer row assembling into NaCl(111) cluster. The structure should correspond to single dim spot that we observed in STM. With the increasing of cluster density, the structure would transform into NaCl(111)-(100) concomitant phase; It just corresponding to p(2x2) ordered structure that we observed in STM. When the sodium coverage over 1 ML，the cluster would entirely self-assemble into NaCl(100) monolayer. We demonstrated that the epitaxy process follows Franck-van der Merve growth mode. Finally, we also simulated a series of electronic properties including density of state (DOS), spatial charge distribution, STM image and XPS spectra simulation. The simulation result showed good agreement with our experiment data; it not only successfully explains the experimental data but also clarifies the underlying physics.