納米電阻式記憶體導體特點的分析方法

Abstract

在此論文中,我們研究在熱力學下,使用 Metropolis Monte-Carlo 模擬切換揮發性金屬 - 絕緣體 - 金屬結構的存儲裝置的導絲。從氧空位的初始分佈缺陷狀態開始到形成傳導長絲（即成形處理），我們在一個氧化物形成層執行傳導長絲特性的全面分析。包含導絲的氧空缺形成與斷裂，我們利用模擬展示了一個完整週期的操作。我們所提供的電阻開關層內，詳細的氧空位，電位和溫度分佈，並獲得導絲動力學更深入的了解。通過實施基於吉布斯自由能的標準方法，我們的方法是基於其公知的熱力學性質，成為調查統計特性需要大量的模擬的一個重要特徵。 我們實現在氧化鉿基電阻開關器件雙極電阻開關行為的分析方法。開關機制的電壓極性已被證明取決於該金屬 - 絕緣體 – 金屬的頂部電極和電阻開關層之間形成的介面層屬性結構。通過增加漂移擴散模型，代表了移動氧和氧得到的界面層結構，我們結合了動態和兩個移動氧和空位動力學的雙極開關行為。我們的結果確認了先前公佈的基於氧化鉿電阻的隨機存取存儲器的開關特性解釋之數據。 我們進一步分析了氧空位隨機初始分佈的缺陷態和導絲性能之間的相關性。通過定義一個無序參數，對應於該初始分佈，我們進行了低和高電阻的傳導長絲狀態（分別是LRS和HRS）之統計分析。採用了大量模擬每個無序值，我們能夠顯示LRS與無序值與無序值的線性關係，而HRS更強烈。

In this thesis, we develop a method to study the conduction filament properties of a resistive switching nonvolatile metal-insulator-metal structured memory device by using a Metropolis Monte Carlo simulation under thermodynamic considerations. Starting from an initial distribution of oxygen vacancies defect states to a formed conduction filament (i.e. forming process) we perform a comprehensive analysis of conduction filament properties in an oxide forming layer. The rupture and formation of the oxygen vacancies consisting conduction filament are simulated to demonstrate a complete cycle of operation. We provide detailed plots of the oxygen vacancies, potential and temperature distributions within the resistive switching layer and gain deeper understanding of the conduction filament kinetics. By implementing our method based on the Gibbs free energy criteria, we obtain reliable resultsfor various materials and structures, based on their well known thermodynamic properties, and efficiency, a critical feature for investigating statistical properties where large number of simulations is required. We implemented our approach in the analysis of bipolar resistive switching behaviors of hafnium oxidebased resistive switching devices. The voltage polarity of the switching mechanism has been demonstrated to be dependent on the top electrode forming material of the metal-insulator-metal structure which in turn determines the interfacial layer properties formed between the top electrode and the resistive switching layer. By adding a drift diffusion model, representing the mobile oxygen species, and oxygen getting interfacial layer structure, we account for bipolar switching behavior incorporating the dynamics and kinetics of both mobile oxygen species and vacancies. Our results confirm previously published data demonstrating the switching characteristics of hafnium oxide based resistive random access memory. We further analyze the correlation between a random initial distribution of oxygen vacancies defect states and the conduction filament properties. By defining a disorder parameter, corresponding to this initial distribution, we perform a statistical analysis of the low and high resistive conduction filament states (LRS and HRS respectively). Using a large number of simulations for each disorder value, we are able to show a linear dependence of the LRS mean value on the disorder while the HRS is more robust