Pattern noise in electron beam resists: PMMA, KRS-XE, TOK, HSQ

Abstract

The variation in the printing of nominally identical contacts with electron-beam exposure is used to quantitatively determine the statistical variation in chemically amplified resists (KRS-XE with and without top coat, TOK) and non-chemically-amplified resists (PMMA and HSQ). Uniform 17×23 arrays of 24 and 32nm contacts were exposed at fixed doses with a 100keV electron beam. By looking at data observed from top view scanning electron microscopy images, a normal distribution was fitted to the fraction of contacts that printed versus dose to determine the standard deviation of the distribution relative to the dose at which 50% of the contacts printed. The top coat on KRS-XE increased contact uniformity and reduced the required dose. Quantitative analysis shows that PMMA contained as much noise as the chemically amplified resist systems, KRS-XE and TOK. Except HSQ, this normalized standard deviation ranged from 0.16 to 0.21 which is indicative that the contact hole printing process may be dominated by less than 40 events. HSQ exhibited lower standard deviation values, corresponding to over 1000 effective events.