單電荷與輻射效應在先進CMOS 和SONOS 元件可靠性之統計量測與模式

Abstract

在奈米元件内，由於元件具有強烈的三度空間效應，通道内電場與local strain均具不均勻性， 元件内各個trap可因位置不同而對元件Vt有不同之影響，各個trap亦可因活化能不同或所在位置 電場不同而具有不同之時間常數。相較於大面積元件内元件退化特性是眾多trap之平均結果，在 奈米尺度元件内，每個元件可能僅含有一顆或少數顆trap，此時各個trap之差異性將逐漸顯現， 而導致奈米元件特性退化AVt呈現一較寬的分佈。在實際應用上，通常係AVt分佈之tail部分導 致元件與電路無法正常工作，本計劃的目的即在於研究元件内trap特性之統計分佈並建立元件退 化特性分佈之統計模式。

本計劃將研究VLSI元件三項重要單電荷(或少數顆電荷)效應所造成之可靠性議題，分別為 (a)奈米CMOS元件内AC BTI stress所造成AVt之統計分佈（b)奈米線電晶體内RTN之振幅分佈及 (c)a-粒子入射對於SONOS Vt分佈的影響。在BTI方面，吾人將測量數百顆high-k CMOS介電層 内trapped charge產生或釋放之時間常數，研究其分佈特性並粹取high-k trap活化能分佈及探討活 化能與元件結構，high-k/IL製程及通道材料之關係。並建立在AC BTI stress下元件退化AVt之統 計分佈模式。在RTN方面，將由平面式元件擴展至奈米線元件，利用3D數值模擬及吾人已自行 發展完成之量子線蒙地卡羅模擬研究奈米線元件内RTN之振幅分佈及理論模型。吾人並將以 FinFET近似奈米線元件量測RTN振幅分佈。在輻射線對SONOS可靠性方面，由於SONOS内電 荷儲存方式與理論與傳統快閃式元件不同，吾人將研究a-粒子照射對512Mb SONOS記憶體Vt 分佈之影響，並建立Vt分佈統計模型，此模型將包括儲存電荷因Frenkel-Poole emission而流失及 a-粒子照射兩種機制所造成之Vt distribution tail。

In the past, studies of device reliability in CMOS and flash memory were focused on Vt degradation and responsible degradation mechanisms in large-area devices, which are an average result of many generated trapped charges. As device dimensions reduce to a deca-nanometer scale, a single trapped charge (or a few charges) would cause device failure. A single trapped charge induced AVt varies significantly from a device to a device due to strong three-dimensional electrostatic and percolation effects. Moreover, trapped charge characteristics, such as trap time-constant and activation energy, spread in a wide range due to stochastic process of trap generation and dispersion of activation energy arising from local strain. In these nano-scale transistors, a statistical model of an entire AVt distribution and its stress time evolution are needed to ensure that the tail of AVt distribution does not cross the reliability criteria. In this proposal, we will perform a statistical study on three major reliability topics in VLSI technologies, (a) AC BTI stress induced AVt distribution in decananometer CMOS, (ii) RTN amplitude distribution in nanowire CMOS and SONOS and (iii) impact of a-particle irradiation on SONOS reliability. Hundreds to millions of CMOS and SONOS devices will be measured to characterize trap statistical behavior and trap induced AVt distributions. 3D numerical simulation and our own developed quantum-wire Monte Carlo simulation will be used to analyze RTN in nanowire transistors.

In AC BTI degradation, both BTI stress and BTI relaxation are included. Trap statistical behavior in BTI stress and in relaxation will be characterized from hundreds of high-k CMOS. 3D numerical simulation will be performed to study device edge and 3D percolative effects. Trap activation energy distribution will be extracted and a statistical BTI AVt distribution model will be developed. We will explore the correlation of trap activation energy distribution with HK/IL processing and channel materials (eg. Ge). In respect to RTN amplitude distribution, we will extend from planar FET to nanowire FET. In nanowire structures, 3D percolation effect is no longer a major cause of RTN amplitude distribution. Quantum-wire Monte Carlo simulation will be employed to calculate RTN amplitude distribution. FinFET devices will be used in this project as a substitute of nanowire FET to characterize RTN amplitude distribution. In SONOS reliability, a-particle irradiation is an interesting and new research topic because SONOS and floating gate flash have a quite different charge storage concept. We will characterize Vt distribution in 512Mb SONOS memory after a-particle exposure. A SONOS Vt retention loss model combining a-particle irradiation and stored charge loss through Frenkel-Poole emission will be developed.