Investigation of body-tie effects on ion beam induced charge collection in silicon-on-insulator FETs using the Sandia nuclear microprobe

Abstract

Silicon-on-insulator (SOI) technology exhibits three main advantages over bulk silicon technology for use in radiation environments. (1) SOI devices are immune to latchup, (2) the volume of the sensitive region (body) and hence total charge collection per transient irradiation is much reduced in SOI devices and (3) the insulating layer blocks charge collection from the substrate (i.e., no funneling effect). This effectively raises the single event upset (SEU) threshold for the SOI device. However, despite their small active volume SOI devices are still vulnerable to single event effects (SEE). Inherent in the SOI transistor design is a parasitic npn bipolar junction transistor (BJT), where the source-body-drain acts as an emitter-base-collector BJT. An ion strike to a floating (not referenced to a specific potential) body creates a condition where the excess minority carriers in the drain-body cause the parasitic BJT to turn on and inject more charge into the drain than was deposited in the device by the ion. In extreme cases the floating body effect (FBE) can trigger a high-current state called single-event snapback (SES) where channel conduction is sustained indefinitely through regenerative electron-impact ionization near the drain junction. Tying the body to the source limits the emitter-base current and reduces the sensitivity of the device to single ion strikes. Unfortunately, the body-tie loses effectiveness with distance due to resistivity, and in regions far enough from the tie the BJT is still in effect. Using the Sandia nuclear microprobe we have created charge collection maps on Sandia CMOS6rs SOI FETs of varying channel widths. These devices have body ties at both ends of the channel region. Results clearly demonstrate that distance of the ion strike from the body tie has an inverse effect upon charge collection and SES sensitivity due to the resistivity of the channel. Experimental results compare well with DAVINCI simulations and electrically induced snapback thresholds. In addition, an interesting saturation effect of SES versus the amount of injected charge is observed.