New insights into EUV mask modeling based on at-wavelength reflectance measurements

EUV reflectometry was used to study the multilayer structure of several test masks, on areas with and without TaBN absorber. For interpretation of the data we propose an updated multilayer mask model with oxide layers on- and below the capping layer, and a genetic fitting algorithm was used for layer thickness optimization. The updated model yields improved goodness-of-fit to the EUV reflectance data, and indicates the presence of SiO₂ and RuO₂ layers in the capping layer stack. Moreover, fits to the EUVR data for the absorber pad indicate that the TaBO layer on top of the absorber is significantly thicker than nominal. CS-TEM measurements showed the TaBO to be present on the absorber sidewalls as well. Wafer exposures of dense line features through pitch (P28-P40) were performed using the same set of test reticles. The nominal mask model was found to strongly underpredict pupil induced contrast fading and best focus shifts. The updated mask model predicts a 2x stronger pupil induced contrast fading and 1.5x larger best focus range through pitch, hereby removing the main discrepancies between wafer data and simulations. This result emphasizes the importance of mask characterization measurements for derivation of an accurate mask model for imaging simulations.