|
As critical dimensions for devices continue to shrink, there is concern over the possible resist sidewall contributions to the critical dimension error budget. Because top surface roughness is substantially easier to measure than the sidewall roughness, it is the purpose of this paper to correlate top surface and sidewall roughness to the processing parameters of dose and development conditions that effect the overall roughness and to explore some possible reasons for the differences in the top surface and sidewall roughness. Initial atomic force microscopy results on the resist top surface indicate that there is a general correlation between the top surface roughness and the processing conditions of dose and development. The sidewall roughness results, however, indicate that the sidewall roughness is relatively independent of the dose and development conditions for the negative-tone, chemically- amplified resist, Shipley SAL 605. The root mean square roughness (Rrms) for the resist sidewalls was on the order of 5.2 +/- 0.5 nm for X-ray exposure. The top surface roughness for the resist at optimized lithographic conditions of 80 mJ/cm2 developed with 0.254 N tetramethylammonium hydroxide was 7.2 +/- 1 nm. These studies, looking at the effects of dose, have shown that increasing the dose decreases the top surface roughness. The extent of the linking reaction, as measured by FTIR, has been compared to the roughness of the resist for samples that have the same approximate linking but have had radically different dose and thermal histories. These preliminary results indicate that there is a general correlation between the extent of linking and the roughness. Samples exposed to a very high dose (650 mJ/cm2) but subjected to short post-exposure bake times (4.1 sec at 108 degree(s)) show similar roughness to samples exposed with lower doses (150 mJ/cm2) but longer PEB times (40 sec at 108 degree(s)). The development conditions provide another major contributing factor in the top surface roughness. Decreasing the developer concentration decreases the top surface roughness of the resist. Adding particular quaternary ammonium salts to the developer decreases the surface roughness and slows the dissolution rate. The goal of these efforts with developer additives was to find the appropriate processing conditions that would yield surface roughness below 3 nm for 100 nm lines. This paper will also explore possible explanations for the effect of developer conditions on the observed roughness in light of current dissolution theories.
|
