Comparison of measured sidewall roughness for positive-tone chemically amplified resists exposed by X-ray lithography

As critical dimensions for resist features shrink, resist roughness on the sidewall may contribute relatively more to the correspondingly smaller CD error budget. Thus, some photoresists may be more suitable than others for smaller dimensions. This paper compares and contrasts the sidewall roughness values measured by atomic force microscopy of two positive-tone, chemically amplified resists used in X-ray lithography. APEX-E demonstrated a sidewall roughness of 3.1 nm RMS on nested lines that was invariant to the dose. The resist UV5 had a sidewall roughness of around 2.7 nm RMS and was slightly dose dependent with sidewall roughness decreasing with increasing dose. Sidewall roughness for isolated lines printed in APEX-E demonstrated a dose dependence. When this resist was exposed with 60 mJ/cm(2) the roughness was 4.7 nm RMS and decreased to 3.2 nm RMS with exposure of 100 mJ/cm(2). The normal lithographic dose (as measured to the mask) for this resist is 80 mJ/cm(2). UV5 sidewall roughness, however, showed no difference in roughness between nested and isolated lines. The normal lithographic dose (as measured to the mask) for UV5 is 150 mJ/cm(2). The sidewall roughness for APEX-E showed depth-dependence with the resist closest to the substrate smoother than the resist near the top of the resist line. UV5 also showed depth-dependent roughness, but only near the top 120 nm of the resist. APEX-E sidewall roughness showed less that 0.5 nm RMS change with respect to the FEB temperature within a workable process latitude of 4 degrees. The UV5 sidewall roughness, however, decreased by 0.5 nm with an increase in FEB temperature of 5 degrees above the normal FEB temperature of 135 degrees. Other processing variables shown to affect the sidewall roughness of the resist include the gap between the wafer and the mask. Increasing the gap generally resulted in an increase in sidewall roughness. By comparing simulated aerial images to sidewall roughness measurements made at various gaps, we conclude that there are other factors besides the quality of the aerial image as controlled by the gap that contribute to the overall sidewall roughness. Also, differences in the aerial images do not explain the observed difference in sidewall roughness behavior for nested and isolated lines in APEX-E.