随着微电子器件集成度增加,由入射离子在器件灵敏区内引起的δ电子分布对器件单粒子效应的影响越来越显著;尤其是它极易引发多位翻转,严重影响设计加固的有效性。首先利用蒙特卡罗软件包Geant4模拟得到重离子在器件灵敏区内产生的δ电子分布,分析得出以下规律:入射离子单核能越高,其产生δ电子分布的径向范围越大;单核能相同的不同种离子,原子序数越大其产生的δ电子密度越大。其次,通过模拟一款45nm静态随机存储器的单粒子翻转效应,说明δ电子和灵敏区分布共同影响器件的多位翻转。当器件灵敏区间距一定时,多位翻转率随入射离子能量的升高先上升后下降;在多位翻转率峰值和布喇格峰之间,多位翻转率随入射离子线性能量传输(LET)值的升高而降低,在该区域两侧多位翻转率随离子LET值的升高而升高。
As the size of the transistor sensitive volume is decreased with the technology generation, the radial distribution of δ-electrons around the ion path has become more and more important to SEE(Single Event Effects) in semiconductor devices. Because it tends to, by causing MBU(Multiple-bit Upsets), invalidate error correcting codes which allow the device to work properly even when errors occur. In this work, Geant4 toolkit was used to simulate the radial ionization profile of heavy ions with different parameters. From the simulation, the certain rules were deduced: the higher the ion energy per nucleon, the wider the radial ionization track; for an identical energy per nucleon, the heavier the ion, the higher the electron density in track core. Then by simulating SEU(Single Event Upsets) of a 45 nm static random access memory (SRAM), effects of ion track structure on its MBU were illustrated. The maximum value of the MBU probability is determined not only by the structure of the device but also by the distribution of the δ-electrons generated by the incident ion. For ion energies between the top of the curve and the Bragg Peak, the MBU probability of the device decreases with the increasing LET values. For other ion energies, the probability increases as LET increasing.