物件式隨機結構搜尋方法之應用

Abstract

隨機結構搜尋已被證明是一種相當強大的方法可用來尋找材料的穩定結構。在高維度的結構空間中，一個全然隨機的取樣固然需要，但卻難免曠日費時，甚至幾乎不可能找到全域的極小值。因此在此方法中，如何利用一些合理的限制產生適當的初始結構成為一個相當重要的問題，例如加上系統對稱性以減少結構空間的獨立維度，或是加上化學鍵結訊息來到達結構空間的低能量區域。在此論文中，我們提出“物件” 的概念，使搜尋複雜結構系統全域能量極小值的過程可以被限制在大幅縮減後的結構空間裡，並得以在合理的時間內應用於現實中。其中，物件的定義為一個或一群原子(例如分子或碳酸根)，帶有由Wyckoff 位置所定義的對稱性；我們檢驗許多的系統並藉此展示物件式隨機結構搜尋法的優勢，包括高壓碳酸鹽(碳酸鈣與碳酸亞鐵)和吸附在石墨烯上的水分子團(一個到四個) 系統。除此之外，為了要驗證物件式隨機結構搜尋法暗示單一水分子在石墨烯上的快速擴散特性，我們更進一步利用了分子動力學模擬來分析其溫度效應下的影響。

Random structure searching has been proved to be a powerful approach to search and find the global minimum and the metastable structures. A true random sampling is in principle needed yet it would be highly time-consuming and/or practically impossible to find the global minimum for the complicated systems in their high-dimensional configuration space. Thus the implementations of reasonable constraints, such as adopting system symmetries to reduce the independent dimension in the structural space and imposing chemical information to reach and relax into low-energy regions, are the most essential issues in the approach. In this thesis, we propose the concept of object which is composed of an atom or a set of atoms (such as molecules or carbonates) carrying symmetry defined by one of the Wyckoff positions of space group and is developed into a method called Random Structure Searching With Object (RSSWO). The method allows the searching of global minimum for a complicated system to be confined in a greatly-reduced structural space and become accessible in practice. We examined several systems, including high-pressure carbonates (CaCO3 and FeCO3) and water clusters (up to four H2O) absorbed on graphene, to demonstrate the power of object concept in random structure searching. Moreover, in order to verify the fast diffusion property of single water molecule absorbed on graphene indicated by RSSWO, a molecular dynamics simulation was performed to illustrate the influence of temperature effect on the system.