Spatial frequency analysis of line edge roughness in nine chemically related photoresists.

An experimental study of the line edge roughness (LER) of nine 193 nm photoresist formulations is presented. In these formulations, the same polymer platform is used while the photoacid generator (PAG) properties and base concentration are systematically varied to produce controlled LER in 130 m dense line/space features. SEM images of each resist are recorded using a KLA 8250 XR inspection tool. The SEM images are post processed using software developed at Shipley to extract frequency dependent LER as well as RMS amplitude LER. We present an investigation of the dependence of frequency limits and sensitivity on the magnification level and image quality. The primary source of noise in the LER measurements is found to be image amplitude noise, which makes determination of the line edge more difficult. The noise introduced by the line edge measurement errors is primarily high frequency noise. The top down resist profiles of the different formulations are used to calculate the LER power spectral density functions. While the absolute amplitude of the spectral density functions is different for each resist, all of the plots show a similar functional form. The resists show a maximum amplitude LER near the low frequency limit with an exponential decay at higher frequencies. The log plot of all of the resists show that the LER follows 1/f noise statistics. The dependence of the amplitude of the LER on the aerial image is also demonstrated.