摻雜之硫屬合金及其奈米複合薄膜應用於相變化記憶體之研究

Abstract

相變化記憶體（Phase-change Random Access Memory，PRAM）被譽為下一世代的電子式記憶媒體。本研究以自組之即時電性量測系統（In-situ Electrical Property Measurement）及動/靜態電壓–電流量測系統（Dynamic/Static I–V Measurement System）探討應用於相變化記憶體之摻雜鉬（Molybdenum，Mo）、氮（Nitrogen，N）與鈰（Cerium，Ce）之鍺銻鍗（Ge2Sb2Te5，GST）及銀銦銻鍗（AgInSbTe，AIST）-二氧化矽（SiO2）奈米複合（AIST-SiO2 Nanocomposite）薄膜之相變化動力學和相變化記憶體元件之電壓–電流特性曲線。 本研究的第一部分以自組之電性量測系統、X光繞射分析（X-ray Diffraction，XRD）及電子顯微鏡（Transmission Electron Microscopy，TEM）測量摻雜N與Mo對GST薄膜之電性與相變化行為的影響。研究結果發現Mo摻雜明顯地降低了GST非晶態電阻率，而N摻雜則提升了GST非晶態與晶態的電阻率。XRD和TEM分析結果顯示元素摻雜穩定了GST的非晶相，同時抑制了GST薄膜晶粒的成長。此亦符合Kissinger理論計算結果，摻雜並導致相變化之再結晶溫度（Recrystallization Temperature，Tc）與活化能（Activation Energy，Ea）的提升；將所得之結果代入各種展延效應模式（Percolation Models）和Johnson-Mehl-Avrami（JMA）理論，其結果顯示相變化過程中受到GST層中異質成核效應的影響將會在空氣和試片介面開始發生成核，同時層狀方式沿著垂直試片方向往內部結晶。 在Ce摻雜的研究方面， XRD顯示摻雜能穩定非晶態GST以及抑制再結晶後六方晶（Hexagonal）GST相之形成。TEM之觀察顯示，Ce摻雜會使結晶態GST之晶粒細化，元素分布（Element Mapping）則發現Ce在GST中均呈勻分布，故Ce原子係以固溶態摻雜於GST中，此亦符合等升溫實驗發現Tc與Ea隨著Ce摻雜濃度增加而上升之結果。Ce摻雜之重要特徵為不會使非晶態GST之電阻率下降，結晶態GST之電阻率僅微幅上升，故不會改變非晶態與多晶態GST之電阻比值（R-Ratio □ 105），尤其，有助於維持訊號之對比清晰度，與一般金屬元素摻雜造成非晶態GST電阻特性降低之行為迥異。恆溫實驗配合JMA理論探討摻雜Ce之GST薄膜之相變化機制，發現摻雜使相變化維度下降，推測其為異質成核（Heterogeneous Nucleation）效應所致，但Ce摻雜大幅升高成長活化能而使恆溫相變化活化能（Appropriate Activation Energy，□H）升高。資料保存時間（Retention Time）之分析顯示發現Ce摻雜濃度越高，資料保存效果愈佳。PRAM元件之應用發現臨界轉換電壓（Threshold Voltage，Vth）雖隨Ce之摻雜濃度升高而上升，但Ce摻雜之GST薄膜確實可應用於PRAM元件之製作。 本研究的第二部分探討應用於相變化記憶體之AIST與AIST-SiO2奈米複合薄膜之相變化動力學和微結構。在即時電性量測系統實驗中發現SiO2的添加提升了奈米複合薄膜的Tc。XRD和TEM的結果顯示奈米複合薄膜中有晶粒細化的現象，此亦提昇了藉由Kissinger理論所分析出來的相變化Ea值。Tc和Ea的升高主要是因為AIST晶粒的細化，此意味著分散在基材中SiO2抑制了AIST再結晶時的晶粒成長。JMA分析顯示在奈米複合薄膜中的Avrami指數呈現下降的趨勢，此顯示分散在AIST中的SiO2顆粒增強了在相變化過程中異質成核的效應。PRAM元件靜態I–V特性曲線和動態反轉行為均證明AIST和其奈米複合薄膜應用於PRAM元件的可行性。

Phase-change random access memory (PRAM) has been widely recognized as the next-generation electronic data storage media. In this study, a self-assembly in-situ electrical property measurement system and dynamic/static I–V measurement system were adopted to study the phase-change kinetics and I–V characterization of Ge2Sb2Te5 (GST) thin films doped with molybdenum (Mo), nitrogen (N) and cerium (Ce) as well as the AgInSbTe (AIST) and AIST-SiO2 nanocomposite thin films. The applicability of these chalcogenide thin films to phase-change random access memory (PRAM) was also evaluated. In the fist part of this study, phase-change behaviors of GST thin films doped with Mo and N were investigated by in-situ electrical property measurement, x-ray diffraction (XRD), and transmission electron microscopy (TEM). It was found that the Mo-doping mainly reduces the resistivity level of amorphous GST while the N-doping raises both the resistivity levels of amorphous and crystalline GST. XRD and TEM analyses indicated that the element doping stabilizes the amorphous state of GST and suppresses the grain growth in GST films. This resulted in the increase of recrysatllization temperature (Tc) and activation energy (Ea) of amorphous-to-crystalline phase transition in GST layers as revealed by the Kissinger’s analysis. The results of data fitting into various percolation models and Johnson-Mehl-Avrami (JMA) theory indicated the heterogeneous feature of phase-transition process in GST layers that the nucleation first occurs at the air/sample interface and the recrystallization front advances into the interior of sample in a layer-by-layer manner along the direction of surface normal. As to Ce doping, XRD showed that Ce doping may stabilize the amorphous GST and inhibit the emergence of hexagonal GST phase after high-temperature annealing. TEM revealed Ce doping causes the grain refinement in GST. The element mapping depicted an almost uniform distribution of Ce in all types of GST films, indicating that Ce atoms reside in GST in solid-solution form. Kissinger’s analysis found that the Tc and the Ea of doped-GST increase with the increase of Ce content. In contrast to other metallic dopants that suppress the resistivity of amorphous GST, a significant finding in this part of study is that the Ce doping does not alter the resistivities of amorphous and crystalline GSTs and hence the resistivity ratio (R-ratio) remains the same at about 105. This greatly benefits the preservation of signal contrast as well as the high-density signal storage. Isothermal experiment in conjunction with JMA analysis revealed that Ce doping suppresses the dimensionality of phase-change process in GST. This is attributed to the heterogeneous nucleation effects occurring during the phase-change process. The retention time analysis found that the retention time increases with the increase of Ce doping amount in GST. In the study of PRAM device applications, it was found that though threshold voltage (Vth) of device containing doped-GST increases with the Ce content, it nevertheless illustrates that the Ce-doped GST films are indeed feasible to PRAM device fabrication. Second part of this study investigates the phase-transition kinetics and microstructures of AIST and AIST-SiO2 nanocomposite applied to PRAMs. In-situ electrical property measurement found that the incorporation of SiO2 escalates the Tc of nanocomposite films. Both XRD and TEM showed the grain refinement in the nanocomposite which, in turn, results in an increases of the Ea of phase transition as indicated by subsequent Kissinger’s analysis. Increase of Tc and Ea in the nanocomposite was ascribed to AIST grain refinement and hindrance to grain growth due to dispersed SiO2 particles in the sample matrix. JMA analysis revealed the decrease of Avrami exponent of nanocomposite, implying that the dispersed SiO2 particles promote the heterogeneous phase transition. Static I-V characteristics and reversible binary switching behavior of PRAM devices not only confirmed the results of microstructure characterizations, but also illustrated the feasibility of AIST and its nanocomposite layer to PRAM fabrication.