Evaluation of the proximity effect and GHOST correction technique for submicron electron beam lithography at 50 and 20 kV

Abstract

The magnitude of proximity effect for submicron electron beam lithography has been experimentally quantified using an electrical proximity effect tester (PET). The PET was delineated in MP 2400-17 resist in a trilevel stack on aluminum, using an AT&T electron beam exposure system. At 50 kV the proximity exposure on a 0.5-μm coded feature under the extreme layout conditions in the PET resulted in linewidth variation >0.25 μm. The effective range of the backscattered electrons was measured to be >10 μm. The GHOST proximity correction technique was used to reduce the magnitude of the linewidth variation. Using a 19-μm defocused beam and a correction dose of 30% of the incident, the magnitude of the linewidth deviation on a 0.5-μm feature was reduced to ±0.05 μm. However, the smallest feature that could be successfully GHOSTed was 0.375 μm at 50 kV. This was attributed to uncorrected electron scatter associated with the forward traveling electrons. Resolution of finer geometries was unpredictable. At 20 kV the resolution limit for successful proximity correction with GHOSTing was 0.5 μm. In conjunction with the experimental results, a computer model was used to confirm the limitations of the GHOSTing technique, as well as to establish the tolerances on the GHOST parameters, correction dose, and defocused beam size. At 50 kV the correction dose can vary from 27% to 33%, and the defocused beam diameter can vary from 15 to 26 μm with only a ±0.05-μm linewidth variation on a 0.5-μm feature.