深次微米元件中熱載子效應所引發之氧化層可靠度

Abstract

熱載子效應所引發的氧化層可靠度問題已在深次微米元件技術領域中引起廣泛討論。本篇論文將針對五項熱載子效應的議題進行研究。 首先，吾人利用一套暫態次臨界電流技術研究氧化層缺陷之充放電機制。此量測技術包括兩個交互時相；一為氧化層缺陷放電時相，一為次臨界電流量測時相。根據相關理論，吾人推導出氧化層缺陷密度與臨界電流之解析模式。此外，此技術利用氧化層缺陷與表面缺陷放電時間常數顯著的差異將兩者效應分離。藉由改變氧化層缺陷放電時相之閘極電壓與次臨界電流量測時相之汲極電壓，吾人可分析不同加壓(stress)條件下氧化層缺陷之電場效應與空間分布。根據吾人研究結果顯示熱電洞所產生的氧化層缺陷數量最大而空間上分布最窄，同時其氧化層缺陷放電之時間常數最短。 其次，吾人研究並觀察n型金氧半場效電晶體中，熱載子效應造成的汲極漏電流特性退化與氧化層的關係。根據吾人研究的結果發現，薄氧化層元件汲極漏電流退化主要是由表面缺陷所造成，而在較厚氧化層元件中汲極漏電流呈現兩段式的退化，前段的指數退化是由於表面缺陷產生而後段加速退化則是由於氧化層內部缺陷產生。 另一方面，吾人觀察到淺溝隔絕(STI) p型金氧半場效電晶體中，最大閘極電流加壓將會造成次臨界電流的突起(hump)。此效應起因於在淺溝隔絕邊緣的氧化層有較高的捕獲電子機率。實驗的結果顯示此突起效應和通道長度有很強的關聯但和通道寬度卻無關，因此吾人將提出一定性模式解釋此一現象。此外，在比較不同熱載子退化機制後吾人發現此電流突起將會是未來淺溝隔絕p型金氧半場效電晶體元件尺寸縮小主要的限制。 接下來，吾人發現元件在動態臨界電壓場效電晶體(DTMOS)操作模式下會有增強熱載子退化，而此現象無法用傳統熱載子理論解釋。因此吾人提出一歐傑再結合(Auger recombination)增強電子能量機制。吾人利用熱載子激光與熱載子閘極電流入射實驗來證明通道中電子能量因正偏基極引發電洞入射產生歐傑再結合效應而增加。根據研究結果顯示，在此操作模式下汲極電流與汲極雜訊(noise)退化較傳統熱載子操作下更嚴重。不同於傳統熱載子所造成的元件退化，歐傑再結合所造成元件退化和溫度成正相關，因此對於在高溫操作的先進元件會有更嚴重的可靠性問題。同時吾人也觀察在超薄氧化層元件中因價帶電子穿隧(valence-band electron tunneling)所引發的歐傑再結合增強熱載子退化，此增強退化現象和基極電壓成正相關，因此對於SOI和DTMOS元件將會造成嚴重的可靠性問題。 最後，吾人利用歐傑再結合增強電子能量機制發展出一套新式快閃式記憶體寫入方法。根據相關理論，歐傑再結合效應所強化之閘極入射電流與溫度成正相關，因此此方法適用於高溫操作。同時由於通道電子擁有較高的能量，此方法將可適用於低電壓操作。

This dissertation addresses the issues related to hot carrier effects in CMOS devices. First, an oxide trap characterization technique by measuring a subthreshold current transient is developed. This technique consists of two alternating phases, an oxide charge detrapping phase and a subthreshold current measurement phase. An analytical model relating a subthreshold current transient to oxide charge tunnel detrapping is derived. By taking advantage of a large difference between interface trap and oxide trap time-constants, this transient technique allows the characterization of oxide traps separately in the presence of interface traps. Oxide traps created by three different stress methods, channel Fowler-Nordheim (F-N) stress, hot electron stress and hot hole stress, are characterized. By varying the gate bias in the detrapping phase and the drain bias in the measurement phase, the field dependence of oxide charge detrapping and the spatial distribution of oxide traps in the channel direction can be obtained. Our results show that (1) the subthreshold current transient follows a power-law time-dependence at a small charge detrapping field, (2) while the hot hole stress generated oxide traps have a largest density, their spatial distribution in the channel is narrowest as compared to the other two stresses, and (3) the hot hole stress created oxide charges exhibits a shortest effective detrapping time-constant. Next, hot carrier stress-induced drain leakage current degradation in both nMOSFET's and pMOSFET's is investigated and characterized. Both interface trap and oxide charge effects are considered. In nMOSFET's, the dependence of drain leakage current on oxide thickness is characterized. Results in our study show that the mechanism of hot carrier stress-induced drain leakage current in thin-gate oxide and thick-gate oxide nMOSFET' is markedly different. In ultra-thin gate oxide nMOSFET's, drain leakage current degradation is attributed mostly to interface trap creation, while in relatively thick oxide devices, the drain leakage current exhibits two-stage degradation, a power law degradation rate in the initial stage due to interface trap generation, followed by accelerated degradation in the second stage caused by oxide charge creation. Furthermore, short gate length pMOSET's with STI structure exhibit a hot electron stress-induced subthreshold current hump. This hump effect shows a strong dependence on gate length and is independent of gate width. Enhanced electron trapping efficiency at the edge of STI is found to be the cause of this phenomenon. A qualitative model is proposed to explain this phenomenon. The subthreshold current hump can increase drain leakage current significantly and imposes a limiting factor in device hot carrier lifetime in short gate length STI pMOSFET's. Finally, enhanced hot carrier degradation in DTMOS operation is observed. The enhanced degradation cannot be simply explained by conventional hot carrier theory. Instead, an Auger recombination assisted electron energy gain mechanism is proposed to explain this phenomenon. In order to further confirm this theory, hot electron gate injection current and hot carrier light emission characterization is performed, which provides evidence that the high-energy tail of channel electrons is enhanced by the application of a positive substrate bias. The drain current and flicker noise degradations are about ten times more serious in the DTMOS than in the conventional MOSFET's. As opposed to the conventional hot carrier degradation, the Auger enhanced degradation exhibits positive temperature dependence. Besides, Auger recombination enhanced hot carrier degradation with stress Vg in the valence-band tunneling regime is also observed. Our result shows that the valence-band tunneling enhanced degradation, as opposed to maximum Ib stress induced degradation, exhibits positive dependence on substrate bias. This phenomenon may cause a severe reliability issue in floating substrate devices (SOI) or positively biased substrate devices (DTMOS). On the other hand, a new hot electron programming technique by taking advantage of the Auger recombination assisted hot electron injection is proposed. This method has been shown to have better programming characteristics and excellent temperature stability.