基于蒙特卡洛和器件仿真的单粒子翻转计算方法

Calculation of single event upset based on Monte Carlo and device simulations

文章提出了一种基于蒙特卡洛和器件仿真的存储器单粒子翻转截面获取方法,可以准确计算存储器单粒子效应,并定位单粒子翻转的灵敏区域.基于该方法,计算了国产静态存储器和现场可编程门阵列(FPGA)存储区的单粒子效应的截面数据,仿真结果和重离子单粒子效应试验结果符合较好.仿真计算揭示了器件单粒子翻转敏感程度与器件n,p截止管区域面积相关的物理机理,并获得了不同线性能量转移(LET)值下单粒子翻转灵敏区域分布.采用蒙特卡洛方法计算了具有相同LET、不同能量的离子径迹分布,结果显示高能离子的电离径迹半径远大于低能离子,而低能离子径迹中心的能量密度却要高约两到三个数量级.随着器件特征尺寸的减小,这种差别的影响将会越来越明显,阈值LET和饱和截面将不能完全描述器件单粒子效应结果.

An extraction method for single event upset cross section based on Monte Carlo code and device simulation is proposed, which can be used to calculate single event effects and sensitive regions in memories accurately. Single event upsetcrosssectionsofdomesticstaticrandomaccessmemory（SRAM）andfieldprogrammaticgatearray（FPGA）devices are calculated, and results agree well with these from heavy ion test. Simulation results reveal the physical mechanism of the relationship between single event upset sensitivity and surface area of off-state NMOSFET and PMOSFET.Sensitive regions of single event upset under different linear energy transfer（LET） values are obtained. The radial ionization profiles of heavy ions with different energy, but the same LET, are also calculated using the Monte Carlo method. The track radius of high-energy ion is significantly larger than that of low-energy ion, while the charge density at the track center of low-energy ion is higher by two or three orders of magnitude. With decreasing technology scaling,the impact of these differences on single event effects will be more pronounced, and the threshold LET and saturated cross-section will not be capable of describing the single event response completely.