Selective Chemical Mechanical Polishing of Silicon Dioxide over Silicon Nitride for Shallow Trench Isolation Using Ceria Slurries

Abstract

For the shallow trench isolation process, in which a thin silicon nitride film acts as a polish stop layer, a polishing slurry with a high oxide to nitride removal rate selectivity is required to remove the overburden oxide. In this work, we show that using ceria-based slurries and only 0.05% of any one of three different cyclic amine additives, namely, pyridine HCl, piperazine, and imidazole, very low nitride (<2 to 3 nm/min) and high oxide (∼350 nm/min) removal rates can be obtained for blanket films. The role of these additives in the slurry toward achieving the desired removal rate selectivities of silicon dioxide over silicon nitride is discussed considering pyridine HCl as an example, using zeta potential, contact angle, adsorption isotherm, and thermogravimetric analysis data.

Silicon Nitride Film Removal During Chemical Mechanical Polishing Using Ceria-Based Dispersions

“…The absence of a peak shift suggests that no complex was formed, presumably since there was no suboxide on the nitride particle surface to react with the Ce 3þ species. The additive adsorption on the oxide-stripped silicon nitride particles, confirmed by zeta potential, thermogravimetric analysis and adsorption isotherms data in our earlier publications, 6,8,9 must have prevented the hydrolysis and regrowth of the suboxide. Thus in the presence of each of these additives, once the native suboxide is removed by the abrasives, the material removal during polishing stops since the hydrolysis is blocked by the additives and the ceria abrasives do not react with the additive covered nitride surface.…”

“…The nitride RRs increased from 2 nm/min at pH 2 to 60 nm/min at pH 3 and remained >100 nm/min throughout the pH range 4-11, similar to our earlier data. 6,8 Here, we discern the reasons for the observed pH dependence of the nitride RRs. Figure 1b (published earlier in Ref.…”

“…6,8,9 The ceria dispersions were each centrifuged at 2700 rpm for 3 h and decanted. The sediment was redispersed in deionized water, adjusted to the initial pH and again centrifuged.…”

“…6,8,9 These dispersions were also used to obtain the nitride RRs shown in fourth column in Tables I and II, which are substantially different from those in columns 2, except for 2 wt % ornithine or lysine, PVP, and PAD. It is worth mentioning that these differences are not due to changes in particle sizes since there was no significant change in the particle size distribution after centrifugation, filtration and redispersion.…”

“…9 and 10) are new results that were obtained using the G & P polisher in our laboratory and the others were taken from Refs. 5,6,8, 9 and 11. It can be seen that the dispersions containing amines and amino acids suppress the silicon nitride film RRs to <2 nm/min (Refs.…”