含氮氧化層應用於二氧化鉿儲存層快閃記憶體資料儲存時間之改善

Abstract

本論文著重於如何改善快閃記憶體的可靠度問題，如儲存資料的保留時間(retention)。傳統快閃記憶體的穿隧氧化層(tunnel oxide)採用乾氧成長二氧化矽(dry SiO2)。但當氧化層厚度小於7奈米時，氧化層中的缺陷很容易產生漏電流路徑，使得被儲存在捕捉層(trapping layer)的電荷透過穿隧氧化層中的漏電路徑而流失，而造成資料的誤判。因此本論文採用新的含氮氧化層的技術取代傳統的乾氧化層，藉由減少界面狀態及層內的缺陷(interface states and bulk defects)來提升快閃記憶體的可靠度特性。而本論文所提出的新含氮氧化層製程技術是符合互補式金氧半(CMOS)場效電晶體的標準製程，因此對於目前工業界的生產技術是可行的改善方法。 傳統的非揮發性記憶體是採用複晶矽浮停閘(Poly-Silicon Floating Gate)作為電荷載子儲存的單元，而電荷在複晶矽中是均勻分佈的，因此若出現漏電路徑，儲存的電荷將會全部流失。而電荷在氮化矽中是屬於離散式的分佈，因此SONOS的結構被提出而取代傳統的複晶矽浮停閘。但為了提升快閃記憶體的寫入速度及降低操作電壓，本論文採用二氧化鉿層作為電荷捕捉層，稱之為SOHOS結構。我們製作SOHOS憶體元件，最後再做完整的電性測量分析。 由本論文的結果，可知道新的含氮氧化層可完全運用在快閃記憶體元件上，並且可以透過提升穿隧氧化層的品質，而達到改善快閃記憶體的可靠度問題。因此對於未來元件的尺寸微縮及改善特性，新穎的含氮氧化層製程技術是很有潛力及備受期待的。

This study focuses on how to improve the reliabilities of Flash memory, including the data retention. The tunnel oxide of conventional Flash memory is dry SiO2 layer. When the thickness of tunnel oxide layer is thinner than 7nm, the defects of tunnel oxide will form the leakage path easily. The trapped charges in trapping layer leak out through the leakage path and let we read the wrong data information. Therefore, the novel oxynitride process proposed in this study can replace the conventional dry oxide layer and improve the reliabilities of Flash memory by reducing the interface states and bulk defects. Moreover, the novel oxynitride process is compatible with standard CMOS process today and it is practicable improvement in industry manufacturing. The charge storage unit of conventional nonvolatile memory is Poly-Silicon Floating Gate and the charges in poly-Si distribute uniformly. If there were leakage path, the trapping charges would all lead out. While the charges distribute discretely in Si3N4 layer, the SONOS structure was proposed and replaced the conventional poly-Si floating gate. To promote the program speed and lower the operating voltage, the HfO2 layer worked as trapping layer in this study and was called SOHOS structure. We found the better PDA conditions of HfO2 film applied the optimum condition in integrated Flash memory devices. Finally, the complete electrical measurements and analysis were carried out.From the result of this study, we confirm that the novel oxynitride process can be applied in Flash memory cells fully. The oxynitride can promote the reliabilities of Flash memory by improving the quality of tunnel oxide. Therefore, in order to reach the scaling down and modification of memory device in the future, the novel oxynitride process is potential and expectable gradually.