Thermomechanical distortions of electron-beam projection lithographic mask systems during exposure

All Next-Generation-Lithography (NGL) technologies will be faced with meeting the stringent error budgets for the sub-90-nm regime. One of the major contributors to overlay error for NGL masks is the thermomechanical distortions induced during exposure. The focus of this research is to identify the thermal and structural response of the electron-beam projection lithography (EPL) mask systems under typical exposure conditions. Three-dimensional finite element models were used to simulate the response of the EPL masks.; Submodels with unpatterned membranes were employed to predict the thermal and structural response of the 100-mm EPL test mask under typical exposure conditions. Simple analytical solutions were also applied to verify the finite element data. Results were characterized by very localized transient temperature rises and thermal distortions. The techniques developed for these submodels served as the basis for subsequent simulations of the full EPL patterned mask during electron-beam exposure.; Larger test models were developed to investigate the thermomechanical response of the patterned mask membranes. Due to the relative size and density of the pattern features, equivalent modeling techniques were employed for computational expedience. Equivalent thermal properties such as thermal conductivity and thermal emissivity were calculated for perforated membranes as functions of void fraction. Equivalent mechanical material properties were also included in the simulations. Thermal distortions were affected by various pattern layouts, as well as the positions of the subfields. Stitching errors during the exposure process were also quantified.; For the 200-mm EPL prototype mask, half-symmetry models were constructed. Support conditions corresponding to the electrostatic chuck (with four symmetrically-located pad regions) developed by Nikon were also included for more accurate out-of-plane distortion predictions. Simulations using different test areas showed how image placement errors were affected by the location of the subfields.