氮化硅对硅和二氧化硅的高选择性蚀刻

A highly selective dry etching process for the removal of silicon nitride (SiN,) layers from siliconand silicon dioxide (SiO,) is described and its mechanism examined. This new process employs aremote O,/N, discharge with much smaller flows of CF, or NF, as a fluorine source as comparedto conventional Si,N, removal processes. Etch rates of SiN, of more than 30 nm/min were achievedfor CF, as a source of fluorine, while simultaneously the etch rate ratio of Si,N, to polycrystallinesilicon was as high as 40. and SiO, was not etched at all. For NF, as a fluorine source, Si,N. etchrates of 50 nm/min were achieved, while the etch rate ratios to polycrystalline silicon and SiO, wereapproximately100 and 70, respectively. In situ ellipsometry shows the formation of anapproximately10-nm-thick reactive layer on top of the polycrystalline silicon. This oxidizedsuppresses etching reactions of the reactive gas phase species with the silicon.

I.INTRODUCTION

The stripping of silicon nitride mask material after thelocal oxidation of silicon (LOCOS) is a possible source ofdevice damage during integrated circuit (IC) fabrication. Thepad oxide can suffer degradation during the overetch. Fur-thermore, etchants can reach the underlying Si substratethrough imperfections in the pad oxide, and etch the sub-strate at a significant rate. This effect leaves behind craters inthe substrate, and is referred to as “pitting.’ Current dryprocesses used for the SiN4 stripping step favor this undes-ired effect, since they typically etch Si much faster thanSiN..l-4 Therefore, a process which etches SiN, selectivelyover both SiO,and Si is desirable.

In previous work by this group~.- and by Blain et al.,’ theetching of SiN, in the afterglow of CF,/0,/N, and NF, /0discharges has been investigated. The flow of the fluorinesource, CF4 or NF3, was kept at constant values for mostexperiments, and O, and N, were added in varying amountsThe etch rate of SiN, was found to be correlated to thedensity of nitric oxide (NO) in a linear relationship for bothgas mixtures. No correlation was observed between theSiN, etch rates and the density of atomic fluorine. The Fatom concentration, determined from the polycrystalline sili-con etch rate26 and mass spectrometry, was found to behigher by at least a factor of 20 than necessary to sustain themeasured Si,N. etch rates. We concluded that the arrival ofNO is the etch rate limiting step in SiN, etching, and Fatoms are available in abundance for SiN, etching.

The etch rates of SiO,(Refs. 2, 5) for the same param-eters were found to be independent of the NO concentration.but followed the F radical density very well. The CF4 basedchemistry produces the desired ratio between the etch ratesof SiO, and Si,N.. With O, and N, added to CF4, ratios of 10or slightly higher could be obtained easily. The etch rate of SiN4 typically was about 30 nm/min, that of SiO2 3 nm/minTypical etch rate values for SiO, and SiN, in the afterglowof NF, /O, discharges were 60 and 80 nm/min, respectively.too high to achieve selective etching of SiN. relative toSiO2.