薄閘極氧化層在奈米尺度應力下之Weibull分佈

The Weibulll Distribution of Thin Gate Oxide Subjected to Nano-scaled Electrical Stress

李志鵬(Chih-Peng Lee)

指導教授 ： 吳幼麟

國立暨南國際大學/科技學院/電機工程學系/碩士(2007年)

摘要

本篇論文結合傳導式原子力顯微鏡(C-AFM)的微觀量測能力與半導體參數分析儀（HP4156C）的強大電性量測能力，研究薄閘極氧化層在奈米尺度應力作用下的行為，並且將所得的樣品之崩潰數據以Weibull distribution方式表示。傳導式原子力顯微鏡的導電探針與裸露的氧化層直接接觸，取代了傳統金氧半電容的金屬閘極，探針與氧化層接觸面積極小，故其所量測到的將是氧化層本質的退化與崩潰。經由結合半導體參數分析儀，我們則可以對樣品施加定電壓應力(CVS)與定電流應力(CCS)，並量測應力施加當中樣品電流或電壓對時間的特性曲線，由此可將所得的數據統計之後繪製成Weibull distribution，並且與傳統量測所得的金氧半電容可靠度相互比較。

　　由所得到的Weibull distribution，我們可以知道氧化層崩潰所需要的電荷(charge to breakdown, QBD）。一般而言，QBD的大小與所使用之金氧半元件的閘極面積大小相關，閘極的面積愈大，在應力施加時愈容易遇到氧化層中之weak point而提早崩潰，故會有著較小的QBD。氧化層崩潰為一種相當局部性的行為，其發生區域的面積約為100nm2，目前文獻所記載的相關研究中並無小於10-7cm2之閘極測試面積，因此傳統之金氧半量測只能提供閘極面積下方的氧化層平均崩潰資訊。本實驗中，當作閘極的傳導式原子力顯微鏡之探針面積約為3x10-12 cm2，由於探針與樣品之接觸面積非常接近於單一個崩潰點的大小，所以展現出來的將會是氧化層本質的退化與崩潰。

　　基本上，氧化層厚度則會影響Weibull distribution中的β值。在薄氧化層中，β值有隨著氧化厚度減少而減少的趨勢。Weibull distribution的β值表示了樣品崩潰時間分散的情形，其值越大表示崩潰時間越集中。β值與氧化層厚度是呈現線性的關係，當氧化層愈厚，電荷注入氧化層造成缺陷進而形成漏電路徑，使得在某一時間發生崩潰的機率愈集中；氧化層愈薄則相反。

本論文首次以C-AFM結合半導體參數分析儀針對厚度為3 nm及5 nm之氧化層施加CVS及CCS並觀測其崩潰。我們成功的將相關的量測面積推至10-12 cm2的範圍並繪製出QBD及time-to-breakdown (tBD)的Weibull plot。我們發現 3 nm的氧化層均有較 5 nm氧化層為佳的QBD及tBD，所量得的?珥?也與傳統量測符合。

關鍵字

傳導式原子力顯微鏡 ； Weibull分佈 ； 定電壓應力 ； 定電流應力

並列摘要

In this thesis, we used conductive atomic force microscopy (C-AFM) in conjunction with semiconductor parameter analyzer (HP4156C) to study the degradation and breakdown behavior of thin gate oxide under nano-scaled stress with different oxide thickness at various stress conditions. Weibull distribution was used in this work to analyze the oxide breakdown events statistically. The C-AFM conductive tip acting as the metal gate electrode in a traditional metal oxide semiconductor (MOS) capacitor for electrical measurements was placed in contact with the bare oxide surface directly in this work. The contact area between the tip and the sample surface is small enough that the breakdown events we measured can be considered as the intrinsic degradation and breakdown behavior of the oxide. By using the semiconductor parameter analyzer we are able to apply either constant voltage stress (CVS) or constant current stress (CCS) to the samples as well as measure the current/voltage versus time characteristics during stress. From the experimental data, we can then analyze the data statistically and make the Weibull distribution plot. In this work, we compared the Weibull distribution obtained by using C-AFM connected with Agilent 4156C with those obtained by using the traditional MOS capacitors measurement.

Charge-to-breakdown (QBD) of the oxides can be determined from the Weibull distribution. In general, the amount of QBD that an oxide can sustain depends on the gate area of the MOS capacitors for test. The increase of the gate area of MOS capacitor enhances the possibility of the weak points of the oxide being met during stress. This would make the oxide breakdown earlier and smaller QBD be obtained. Basically, the breakdown and degradation characteristics of thin oxide are a highly localized phenomenon, typically in a range of hundreds of nm2. However, the gate areas of conventional MOS capacitors used for breakdown test are rarely less than 10-7cm2. Therefore, breakdown test using conventional MOS capacitors only gives average oxide breakdown information under the gate area. In this thesis, the contact area between the conductive tip and the oxide surface is around 3x10-12cm2, which is extremely small so that it can be considered as a single breakdown point when oxide breakdown occurs. Therefore, the Weibull plot in this work depicts the behavior of intrinsic oxide degradation and breakdown.

Basically, the shape parameter β in the Weibull distribution is dependent on the oxide thickness measured. For thin oxides, the β value decreases as the oxide thickness decreases. The β value indicates the scatter of the time-to-breakdown data, which has a linear relationship with the oxide thickness. When the oxide is thicker, the oxide breakdown probability becomes more concentrated due to the charge injected into the oxide during stress and formation of a leakage path in the oxide.

In this thesis, we demonstrated, for the first time, the breakdown characteristics of 3 nm- and 5 nm-thick silicon dioxide by applying CVS and CCS through C-AFM in connection with Agilent 4156C. We successfully pushed the measuring limit of breakdown test area to around 10-12 cm2 and plotted the Weibull distribution of QBD and tBD. From the Weibull plot, we found that the 3 nm samples exhibited better QBD and tBD than the 5 nm samples. We also found that the ?? values of the Weibull plots agreed well with those obtained in the conventional breakdown tests.

並列關鍵字

conductive atomic force microscopy ； Weibull distribution ； constant voltage stress ； constant current stress

參考文獻

[1] International Technology Roadmap for Semiconductors 2006.

[2] Toshiyuki Sakura, H. Utsunomiya, Y. Kamakura, and K. Raniguchi, “A Detailed Study of Soft-and- pre-soft-breakdowns in Small Geometry MOS Structures” IEDM 183-186, 1998.

[3] Andreas Martin, P. O’Sullivan and A. Mathewson, “Dielectric Reliability Measurement Methods: A Review” Microelectron. Reliab., Vol. 38, No. 1, pp.37-72, 1998.

[4] Heiji Watanabe and Toshio Baba, “Characterization of local dielectric breakdown in ultrathin SiO2 films using scanning tunneling microscopy and spectroscopy.” J. Appl. Phys. Vol.85 pp.6704,1999.

[5] B. Kaczer and J.P. Peiz, “Ballistic-electron emission microscopy studies of charge trapping in SiO2.” J. Vac. Sci. Technol. Vol.14 No.4, 1996.