Identification of Pulsed-Neutron-Induced Upset Bursts in Static Random Access Memories using Monte-Carlo Simulations

Chao Qi,Wei Chen,Yan Liu,Xiaoming Jin,Shanchao Yang,Xiaoqiang Guo
DOI: https://doi.org/10.1109/RADECS.2017.8696159
2017-01-01
Abstract:In contrast to the accumulation of upsets at a relatively slow rate in static random access memories (SRAMs) caused by neutrons at the steady-state modes of nuclear reactors, pulsed neutrons generated at a pulsed-state mode with a significantly high flux induce a burst of upsets in a short time (<10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ms). Herein, pulsed-neutron experiments are conducted on SRAMs, showing that the upset cross sections differ as the pulsed-state mode changes, indicating the nonlinearity between the upset bits and the neutron fluence, unlike the linear accumulation of single event upsets (SEUs) with increasing neutron fluence at the steady-state modes. To identify the upset pattern of pulsed-neutron-induced upset bursts, the experimental results are compared to Monte-Carlo simulations, which calculate the growing upset bits of different accumulated upset types and simulate the upset accumulation process at steady-state modes. The experimental results exhibit consistent and precise accordance with the simulation results, indicating that the pulsed-neutron induced upset bursts are caused by the same accumulating process as that of SEUs at the steady-state modes. The observed nonlinearity is attributed to the uncertainty of neutron-fluence measurements, uncounted secondary upsets, and some unidentified systematic errors in cross-section calculations. The identification of pulsed-neutron-induced upset bursts with the accumulation of SEUs can convincingly distinguish such upset bursts from dose rate upset (DRU) bursts.
What problem does this paper attempt to address?