Effects of process parameter distributions and ion strike locations on SEU cross-section data (CMOS SRAMs)

Abstract

The effect of statistical parameter distributions and stochastic ion strike locations in hardened memory arrays on observed SEU cross-section data is discussed. Application of numerical analysis to the parasitic bipolar gain distributions in SIMOX SRAMs and the introduction of an effective critical charge based on sensitivities to ion strike locations, explains and quantifies the non-saturating behavior of measured cross-section curves in SO1 and other hardened memories.

Proton SEU cross sections derived from heavy-ion test data

“…This was noted by Petersen [6] and later by Barak et al [7] when discussing the cross section associated with the charge-collection depth exceeding a specified value. Petersen has also used other terminology, a charge-collection gain [8], following Massengill et alwho reported that some devices exhibit a parasitic bipolar gain that varies as ion hit location is varied [9]. The first physical postulate stated in the previous subsection implies that the charge-collection depth at a given lateral location is the integral of along a perpendicular line through the device at that lateral location.…”

“…The solid angle integral of the above equation produces a heavy-ion SEU rate in an isotropic environment (with the integral LET flux proportional to ) on the right side. The result is day-cm cm coul-m (9) where is defined by Heavy-ion rate for a device having a normal-incident cross section and produced by an integral LET flux given by m -sec-ster MeV-cm mg . The constant in the flux was selected so that inconvenient constants do not appear in (9).…”

“…The result is day-cm cm coul-m (9) where is defined by Heavy-ion rate for a device having a normal-incident cross section and produced by an integral LET flux given by m -sec-ster MeV-cm mg . The constant in the flux was selected so that inconvenient constants do not appear in (9). The flux will be called the hypothetical flux.…”

“…This difference will produce different SEU rates in most heavy-ion environments. However, (9) implies that, for this special and hypothetical heavy-ion environment, different devices having the same normal-incident data also have the same heavy-ion SEU rate, even though they may have different directional cross sections at other angles.…”

“…This model does not satisfy the first postulate in Section II, because different values of are associated with different RPPs. However, each RPP satisfies the postulate, so (5), (9), and (10) apply to the individual RPPs. Summing these results over the RPPs produces the same results (5), (9), and (10) but with the cross sections interpreted as device cross sections.…”