ABSTRACTThe original intent of this study was to compare rapid thermal, thin (80-100Å) gate oxides with standard, furnace-grown, thin gate oxides for endurance. Wafer processing before gate oxide growth was chosen to duplicate processing used ina typical non-volatile memory product. In particular, care was taken to duplipate pre- and post- gate growth processing of field oxide isolated polysilicon capacitors for all wafers in order to eliminate the previous difficulties in comparing oxides when different cleans and processing steps are used.[1] Substrate defects, atypical to this process, were presumably introduced during the initial wafer cleaning and scattered the time-to-breakdown (TTB) values during a constant current stress of these oxides to the point where statistical comparison of TTB averages was dubious. However, for unannealed wafers and for post polysilicon definition heat treatments of 900°C, RTO oxides grown with HCL had the same oxide trapping rate as the furnace oxides grown with TCA and RTO oxides grown in pure O2 had a faster trapping rate. Higher temperature post polysilicon definition heat treatments had different effects. RTO oxides exhibited better yield than the furnace oxides. These results illustrate the differences between RTO and furnace oxidation in the presence of non-ideal wafer substrates.