This paper describes a new method based on Geant4 and Technology Computer Aided Design (TCAD) to assess SEU sensitivity of SRAMs. Combined with space environment models, we present a method of predicting single event upset rates in SRAMs.1. We use TCAD simulations to replace ground simulation experiments to gain SRAM's response to radiation and Heavy-ion induced upsets. Heavy-ion simulation data are fitted by Weibull Function. Then we calculate the energy deposition of secondary particles produced by proton-silicon nuclear reaction, using Geant4 which based on Monte Carlo method. Finally, we can gain proton-induced upsets combined with Weibull Function.2. We simulate charge sharing and charge collection between adjacent devices. Simulation results show that charge sharing is mainly affected by charge diffusion and areas between adjacent devices are more sensitive to MBU than drain area. Charge sharing and MBU can be mitigated by nodal separation.3. We develop multi-layer shielding analysis software and on-orbit upset rates prediction software based on Geant4. Multi-layer shielding analysis software allows the users to define complex materials and record primary particles which pass through the shield. On-orbit upset rates prediction software calculates the energy deposition in sensitive volume, then gains the upset rates combined with Weibull Function which gained by ground experiments or TCAD simulations. |