Sample-to-Sample Variability and Bit Errors Induced by Total Dose in Advanced NAND Flash Memories

Abstract

The variability in the total ionizing dose response of 25-nm single level cell NAND Flash memories from two different lots is studied. More than 1 Terabit of floating gate cells were irradiated with gamma rays and the number of errors was statistically analyzed. The behavior of the two lots is remarkably different in terms of floating gate errors. The statistical parameters such as mean, standard deviation, and shapes of the error distributions were studied. The sample-to-sample statistical distribution of TID errors is not Gaussian and it is attributed to variability in neutral cells, due to phenomena such as random discrete dopant fluctuations. Finally, the impact of scaling on variability is discussed

Single Event Upsets Induced by Direct Ionization from Low-Energy Protons in Floating Gate Cells

“…Different from the doses delivered with low-energy protons at RADEF, for 4-MeV protons and higher energies the doses exceed 1 krad(Si). However, these doses are still too low to induce FG errors in these samples, at least after gamma ray exposure [16], [18]. This opens the door to two possible scenarios: The results of irradiations with protons at even higher energies can be helpful to evaluate the first possibility.…”

Space and terrestrial radiation effects in flash memories

“…The threshold voltage shift induced by TID exposure was shown to change only marginally with technology scaling [24], as the thickness of the tunnel oxide in flash memories can hardly be shrunk below 8-10 nm due to reliability issues (e.g., degradation of retention and endurance properties). However, in some recent devices a dependence on the feature size has been observed [25].…”

“…The TID-induced ΔV th is proportional to the received dose; for enough small doses, the relationship ΔV th versus dose is linear [32]. If the shift is large enough, raw bit errors appear and they monotonically increase for increasing doses ( [25,32,33]). This is depicted in figure 5, for a 25 nm NAND SLC flash memory irradiated unbiased (with shorted pins) with gamma rays at the Co60 ESA/ESTEC gamma facility [25].…”

“…If the shift is large enough, raw bit errors appear and they monotonically increase for increasing doses ( [25,32,33]). This is depicted in figure 5, for a 25 nm NAND SLC flash memory irradiated unbiased (with shorted pins) with gamma rays at the Co60 ESA/ESTEC gamma facility [25]. All these errors were observed in programmed FG cells (level L1), whereas erased cells (L0) showed no fails; this is due to the fact that the neutral distribution is below the reference voltage for these samples, so even though an erased cell is completely discharged by TID, no errors take place.…”

“…One of the issues recently investigated is sample-tosample and lot-to-lot statistical variability in TID response of FG cells [25,34]. Figure 6 plots the number of raw bit errors per chip, before and after gamma irradiation up to two dose levels, 20 krad(Si) and 33 krad(Si), for 20 nominally identical samples (same part number, same lot).…”