Correlating low energy neutron SER with broad beam neutron and 200 MeV proton SER for 22nm CMOS Tri-Gate devices

For terrestrial single event upset (SEU) characterization, JEDEC JESD89A requires using either broad beam neutron sources or at least four single neutron/proton particle energy sources. While computing ambient upset rates from broad beam neutron sources, such as the ICE House at Los Alamos Neutron Science Center (LANSCE), is simple and direct, their availability, cost, and particle fluxes are very limited. Several authors in the open literature therefore have considered alternatives, such as high energy (e.g., 200 MeV) proton beams or low energy (e.g., 14MeV) neutron beams. In this paper, we address the questions of accuracy and correlation when these alternative facilities are employed. SEU data were collected on 22nm Tri-Gate bulk CMOS devices at Boeing (BREL), Los Alamos (LANSCE/WNR) and IUCF (RERP and LENS) and the corresponding upset rates are compared.