氧空缺和雙極電荷層於多晶矽-氧化矽-氮化矽-氧化矽-矽形式非揮發性記憶體的應用

Abstract

在本篇論文中，我們將由於氫氣退火在二氧化鉿(HfO2)上產生的氧空缺應用到二氧化鉿(HfO2)薄膜和二氧化鉿(HfO2)微晶粒記憶體電容上。我們發現在300度C氫氣退火一小時後，二氧化鉿(HfO2)薄膜記憶體電容表現出最大的遲滯電壓差值(hysteresis)。並且在500度C氫氣退火一小時後，二氧化鉿(HfO2)微晶粒記憶體電容顯示出最大的遲滯電壓差值。我們確認以二氧化鉿(HfO2)薄膜或二氧化鉿(HfO2)微晶粒為電荷捕捉層的SONOS型記憶體結構可以藉由以氫氣退火所產生的氧空缺來增大記憶窗口。

再者，我們將在氧化鋁(Al2O3)和二氧化矽(SiO2)的接面上產生的”本質偶極 ”(intrinsic dipole)的觀念應用到二氧化鉿(HfO2)薄膜和二氧化鉿(HfO2)微晶粒記憶體電容上。我們指出經由在SONOS型記憶體電容中引入一層極薄(約1奈米)的高介電常數材料(HfO2或Al2O3)可以有效的調節金屬電極的功函數。我們發現對於二氧化鉿(HfO2)薄膜和二氧化鉿(HfO2)微晶粒記憶體電容來說，經由在電荷捕捉層和穿遂氧化層的中間引入一層極薄(約1奈米)的氧化鋁(Al2O3)，配合適當的氫氣退火溫度，我們甚至可以得到更大的遲滯電壓差值。

最後，我們提出一種雙極電荷層的新穎SONOS型非揮發性快閃記憶體結構。我們在穿遂氧化層之上引入一層約1奈米的氧化鋁(Al2O3)來造成雙極電荷的產生，並且使得寫入動作較為簡易。因此我們得到較快的寫入速度。更甚的是，這種新穎結構的SONOS型記憶體較傳統SONOS記憶體有更好的資料持久性。因此我們相信具有雙極電荷層的SONOS型快閃記憶體將有機會參與下個世代非揮發性記憶體的應用。

In this thesis, we utilize oxygen vacancies generated by forming gas anneal (FGA) to our HfO2 thin film and nanocrystal memory capacitors. We find out that HfO2 thin film memory capacitors show the highest hysteresis for FGA temperature of 300oC for 1 hour. And HfO2 nanocrystal memory capacitors demonstrate the largest hysteresis for FGA temperature of 500oC for 1 hour. We confirm that it is effective to enlarge the memory window for SONOS-type memory structure with HfO2 thin film or HfO2 nanocrystal trapping layer by the FGA-generated oxygen vacancies.

Next, we adopt the concept of “intrinsic dipole” formed at Al2O3/SiO2 interface to our HfO2 thin film and nanocrystal memory capacitors. We demonstrate that modulating the gate effective work function by incorporating an ultra-thin (~1nm) high-k layer (HfO2 or Al2O3) in our SONOS-like memory capacitor is valid. We show that both for HfO2 thin film and nanocrystal memory capacitor, the hysteresis could further be enhanced by incorporating a ~1nm Al2O3 layer between trapping layer and tunnel oxide with appropriate temperature of FGA.

Finally, we propose a novel nonvolatile SONOS-type flash memory with dipole layer engineering. We incorporate a ~1nm Al2O3 layer upon the tunnel oxide to induce dipole formation and results in easier programming. Thus we get higher programming speed. Furthermore, the novel SONOS-type flash memory exhibits better retention performance than the conventional SONOS memory. Therefore, we believe that SONOS-type flash memory with dipole layer engineering can be a candidate for next-generation nonvolatile memory applications.