The impact of poly-removal techniques on thin thermal oxide property in poly-buffer LOCOS technology

The authors used both oxidation and dry-etching methods for removing the sandwiched polysilicon used as a stress-relief film in the poly-buffer local oxidation of silicon (LOCOS) process. The impact on thin thermal oxide (125 AA) grown after PBL-LOCOS was examined by using both time-zero and time-dependent dielectric-breakdown I-V techniques. The results of the time-zero breakdown tests show no difference in oxide-breakdown strength with respect to both methods employed in the poly-removal sequence. A very tight distribution in time-zero breakdown-field in the vicinity of 10 MV/cm was obtained for both oxidation and dry-etching cases. The high-voltage time-dependent dielectric-breakdown tests, carried out at oxide fields of 11-12 MV/cm, however, detected a marked difference in the endurance of high oxide field before the catastrophic damage occurs. In the case of samples using dry-etching poly, the mean-time-before-failure (MTBF) is approximately 6 s in comparison to 200 s measured in the samples in which the poly-film was removed by oxidation. The degradation of thin thermal oxide with respect to high-voltage time-dependent dielectric-breakdown endurance in the case of dry etching is attributed to the plasma damage and/or ion contamination such as carbon/fluorine during the poly dry etching.<
>