A radiation-induced single event upset (SEU) in a semiconductor chip is the main cause of nondestructive faults (the so-called soft errors) in microelectronic devices. The terrestrial neutron is the main cause of SEUs on the ground. Because neutron shielding is difficult, estimating the terrestrial soft error rate (SER_GND) is imperative to ensure the reliability of microelectronic devices. Some spallation neutron sources, which reproduce terrestrial neutrons, have been utilized to estimate SER_GND. However, such special neutron sources are limited and cannot meet the global demand for device tests. We developed a new methodology to estimate SER_GND based on simulation coupled with one-time neutron irradiation testing. This methodology can be applied to various neutron sources. In this methodology, the probability of soft errors caused by a neutron, σ(E_n, Q_fit), as a function of the incident neutron energy, E_n, and the noise charge needed to cause the soft error, Q_fit, are derived via simulation (Fig. 1a-1). Single measured data is used to determine Q_fit (Fig. 1a-3). Thereafter, SER_GND can be derived by integrating the product of σ(E_n, Q_fit) obtained via simulation and terrestrial neutron flux, φ(E_n), with respect to E_n (Fig. 1a-4).
 The developed methodology allows the rapid estimation of SER_GND because it required only one-time neutron irradiation (Fig. 1b). Moreover, it is applicable for various neutron sources and thereby contributes to satisfying the global demand for SER_GND estimation. It contributes to enhance the reliability of semiconductor devices necessary to realize the safe and secure super-smart society.

Part of this research was conducted in collaboration with Socionext Inc. and was supported by the JST Program on Open Innovation Platform with Enterprises, Research Institute and Academia (OPERA) (Grant No. JPMJOP1721).