Effects of Thermal Neutron Irradiation on a Self-Refresh DRAM

Abstract

In this study, static and dynamic test methods were used to define the response of a self-refresh DRAM under thermal neutron irradiation. The neutron-induced failures were investigated and characterized by event cross-sections, soft-error rate and bitmaps evaluations, leading to an identification of permanent and temporarily stuck cells, block errors, and single-bit upsets.

Investigating the Impact of Radiation-Induced Soft Errors on the Reliability of Approximate Computing Systems

“…Based on the characterization tests, we identified three types of errors. The results are in line with the ones that we presented in [22], but with a soft-error rate a little bit higher since that work concerned thermal neutrons (implying lower energy). The first failure mode consists of SBUs.…”

“…To characterize the relative radiation response of the hardware implementations carrying the AxC techniques on the applications, we target the irradiation on the HyperRAM memory. This memory was already characterized by the authors under thermal neutrons and presented different types of errors, such as Single Bit Upsets (SBUs), stuck-at bits, and block errors that affect up to 2048 sequential memory addresses [22].…”

“…In the static test mode, a known data pattern is written in the memory before the irradiation. Then the device is exposed to the beam for a specific fluence; when the beam is stopped, a read-back operation is performed to identify corrupted bits [22], [27], [28].…”

A Model-Based Framework to Assess the Reliability of Safety-Critical Applications

“…Based on the radiation-induced soft errors, the model-based fault injection is proposed as a case study, where it is shown the main ideas to enable the injection of realistic fault models during the simulation or emulation of a target application. In order to define the scope, the next steps are based on neutron irradiation on a COTS self-refresh DRAM memory (results can be explored in more details in [22]), and the reliability assessment is based on the injection of realistic faults extract from this test campaign in applications that may use this memory to run. Then, starting from the radiation test campaigns, based on the results presented in [22], it was defined the cross sections for three different types of errors that were identified: SBUs, stuck-at bits, and block errors.…”

“…In order to define the scope, the next steps are based on neutron irradiation on a COTS self-refresh DRAM memory (results can be explored in more details in [22]), and the reliability assessment is based on the injection of realistic faults extract from this test campaign in applications that may use this memory to run. Then, starting from the radiation test campaigns, based on the results presented in [22], it was defined the cross sections for three different types of errors that were identified: SBUs, stuck-at bits, and block errors. From the estimated cross sections, we can define the E-SER (Execution Soft Error Rate), which provides the expected amount of events within a run of the defined application.…”