鈦上電極對濺鍍法製備之氧化鋯薄膜於電阻式轉換記憶元件之研究

Abstract

隨著科技越來越進步， 各種消費性電子產品的發展快速，各種產品所需要的記憶體容量越來越大，而其中不需要電源供應仍能儲存資料的非揮發性記憶體，也漸漸受到矚目。其中，電阻式非揮發性記憶體具有高密度、高操作速度、低功率消耗、高耐久性、可微縮化、非破壞性讀取資料、能符合現在CMOS製程以及結構簡單等特性，因此有機會取代flash memory，使其成為次世代非揮發性記憶體的熱門選擇。 在這篇論文中，電阻轉換特性研究是著重在鈦/氧化鋯/鉑的結構，包含製程條件可以分為大兩部份，第一部份是氧化鋯製程溫度的選擇，第二是部分氧化鋯製程結構的改變。 在氧化鋯製程溫度的選擇中，其成長溫度從25oC，100 oC，150 oC，200 oC，250 oC中，而成長厚度均為30奈米，其中200 oC有最好的特性，其電性分析為:持久性可以達到10000次;動態脈衝可以以快速的50奈米寬的脈衝波轉態1000次;非破壞性讀取可以在0.3V下維持106秒而沒有資料的破壞。 接著以最佳的200 oC製程溫度為主，另外作氧化鋯結構的改變，目的為分析轉態機制的探討，分為氧化鋯薄膜厚度的改變以及改變鈦上電極面積兩部分。首先藉由改變厚度，可以探討介面轉態的部份是發生元件的何處，因為鈦上電極的強吸氧性，形成一層介面轉態層，利用這層介面轉態層，跟元件的高良率、穩定的特性，轉態的極性有關，而改變厚度更可以證實利用介面轉態的機制。第二部份為鈦上電極面積的改變，利用改變面積的微影製程，結合改變厚度所得到的結論，來探討導電絲斷裂所形成的電阻大小。而電極面積越小,電阻會越大,在低阻態時電流不會下降，但高阻態時電流會下降，使鈦/氧化鋯/鉑的轉態機制能夠更加明瞭。

Many kinds of consumer electrical commercial products are becoming more and more popular following with the development of the technology. All kinds of products need the memory, especially non-volatile memory, which can store data without power. The resistive switching random access memory (RRAM) is one of the next generational memories that have the chance to become the mainstream. It has the advantages of high cell density, high operation speed, low power consumption, high endurance, lower scale limit, non-destructive readout, and easy processing that can fit in the CMOS process, so it is one of the potential substitutions for flash memories. In this thesis, the resistive switching characteristics are investigated based on the Ti/ZrO2/Pt structure, and the research focus on two parts. First, the process temperature of ZrO2 is changed. Second, the thickness of ZrO2 and the area of the top electrode are changed. In the first part, the process temperature of the ZrO2 is changed to five temperatures: 25oC, 100 oC, 150 oC, 200 oC and 250 oC. The optimal value is 200 oC, so the following research of the mechanism will use this temperature. The performance of 200oC is good. DC sweep cycle times can achieve 10000 times; high speed 50ns pulse width can be switched at least 1000 cycle times; retention test is 106s; and there is no data loss at the nondestructive readout test for over 10000 seconds. In the second part, the purpose is to know the mechanism of the RRAM, so the structure is changed by various thicknesses of the ZrO2 and the various area of the top electrode. First, we can use the various thicknesses of the ZrO2 to find where the switching part in the filament is. Because Ti is a strong absorption metal, so there is an interface layer to be formed. The high yield, steady performance, and the polarities of the switching is related with the interface layer, and we can use the various thicknesses of ZrO2 to improve this mechanism. Second, the area of the top electrode is changed by the lithography. The resistance of the filament can be affected by this factor. At on-state, the current is not related with the area of the top electrode, but at off-state, the current is decreasing following the scale down of the top electrode. It is combined with the various thicknesses of ZrO2, so the mechanism of Ti-ZrO2-Pt is much clearer.