Experimental evidence is presented showing obvious azimuthal dependence of single event upsets(SEU) and multiple-bit upset(MBU) patterns in radiation hardened by design(RHBD) and MBU-sensitive static random access mem...Experimental evidence is presented showing obvious azimuthal dependence of single event upsets(SEU) and multiple-bit upset(MBU) patterns in radiation hardened by design(RHBD) and MBU-sensitive static random access memories(SRAMs), due to the anisotropic device layouts. Depending on the test devices, a discrepancy from 24.5% to 50% in the SEU cross sections of dual interlock cell(DICE) SRAMs is shown between two perpendicular ion azimuths under the same tilt angle. Significant angular dependence of the SEU data in this kind of design is also observed, which does not fit the inverse-cosine law in the effective LET method. Ion trajectory-oriented MBU patterns are identified, which is also affected by the topological distribution of sensitive volumes. Due to that the sensitive volumes are periodically isolated by the BL/BLB contacts along the Y-axis direction, double-bit upsets along the X-axis become the predominant configuration under normal incidence.Predominant triple-bit upset and quadruple-bit upset patterns are the same under different ion azimuths(Lshaped and square-shaped configurations, respectively). Those results suggest that traditional RPP/IRPP model should be promoted to consider the azimuthal and angular dependence of single event effects in certain designs.During earth-based evaluation of SEE sensitivity, worst case beam direction, i.e., the worst case response, should be revealed to avoid underestimation of the on-orbit error rate.展开更多
基金Supported by National Natural Science Foundation of China(Nos.11179003,10975164,61204112 and 61204116)China Postdoctoral Science Foundation(No.2014M552170)
文摘Experimental evidence is presented showing obvious azimuthal dependence of single event upsets(SEU) and multiple-bit upset(MBU) patterns in radiation hardened by design(RHBD) and MBU-sensitive static random access memories(SRAMs), due to the anisotropic device layouts. Depending on the test devices, a discrepancy from 24.5% to 50% in the SEU cross sections of dual interlock cell(DICE) SRAMs is shown between two perpendicular ion azimuths under the same tilt angle. Significant angular dependence of the SEU data in this kind of design is also observed, which does not fit the inverse-cosine law in the effective LET method. Ion trajectory-oriented MBU patterns are identified, which is also affected by the topological distribution of sensitive volumes. Due to that the sensitive volumes are periodically isolated by the BL/BLB contacts along the Y-axis direction, double-bit upsets along the X-axis become the predominant configuration under normal incidence.Predominant triple-bit upset and quadruple-bit upset patterns are the same under different ion azimuths(Lshaped and square-shaped configurations, respectively). Those results suggest that traditional RPP/IRPP model should be promoted to consider the azimuthal and angular dependence of single event effects in certain designs.During earth-based evaluation of SEE sensitivity, worst case beam direction, i.e., the worst case response, should be revealed to avoid underestimation of the on-orbit error rate.
