For 28nm NAND flash memory, the aspect ratio of shallow trench isolation (STI) becomes large, and it makes difficult to fill STI structure by conventional oxide film without void formation. Several gap-fill methods such as high density plasma (HDP) and thermal chemical vapor deposition (CVD) are widely accepted, but they are difficult to make gap fill without void because of gap-fill margin. Recently, STI fill has been tried with polysilazane (PSZ) based inorganic spin-on dielectric (SOD) film. PSZ oxide has its good gap-fill ability, low moisture content and high etching resistance. However, PSZ oxide would produce amine etch residue during the NF3 dry etching process because of NH3component in PSZ oxide. For preventing this resudue, it is very important to control process parameters affecting the chemical reaction because the etch rate, and the surface quality depend upon the process reaction. During the STI etching back process, other layer also exposed. So, selectivity between other oxide must be considered. In this study, we have employed plasma dry etching of nitrogen trifluoride (NF3) and H2O mixtures. We studied the effect of the gas chemistry on the etching of oxides. The materials used for these experiments were thermally grown SiO2 (1000Å, 10000Å), PSZ oxide (500Å), TEOS (315Å) and ALD oxide (300Å). Samples of size 25 x 25 mm were cleaned first in diluted SC-1 [NH4OH(25%): H2O2(38%): DIW =1:2:50] at 60°C for 10 min. The etching mechanism of PSZ oxide was investigated in plasma activated NF3/H2O gas. This process generates ammonium hexafluorosilicate ((NH4)2SiF6) as a by-product. By-products were formed on the PSZ surface which was confirmed by observing N–H, NH4+ and SiF6 2-peak in ATR-FTIR spectrum. These peaks are representative of (NH4)2SiF6 powder composition. Etch rates of several SiO2 were reported as a function of H2O ratio, substrate temperature, and pressure. NF3/H2O mixtures were excited using pulsed RF plasma. The NF3 flow rate for all experiments was kept at 150 sccm. The temperatures were varied from 8 to18ºC, the chamber pressures from 7 to 9 Torr., the H2O flow rates from 1500 to 2500 sccm and process times from 20 to 100 sec. Helium was used as a carrier gas. After dry etching, an annealing process was performed. The reaction mechanism was proposed by analyzing the oxide layer surface. We used an ellipsometer to measure the oxide thickness for evaluation of the etch rate. Also, various analyzing tools such as SEM, contact angle analyzer, and FT-IR were used to characterize the surface contaminants. The optimized drying process provides the high selectivity without any residue formation.