Metrology related to the multilayer mirrors in an extreme-ultraviolet stepper

Extreme-ultraviolet (EUV) lithography is slated for introduction in the 2012 time frame for volume production of chips with 32-nm design rules. NIST has a program to perform metrology on the multilayer mirrors used to focus the 13.5 nm EUV radiation employed in the stepper. This program includes accurate mirror reflectivity measurements and accelerated lifetime testing. One of the major obstacles to attaining high volume manufacturing is lifetime of the mirrors comprising the projection system. Ongoing endurance testing of Ru-capped multilayer mirrors at the National Institute of Standards and Technology synchrotron facility has revealed that the damage resulting from EUV irradiation does not in general depend on the exposure conditions in an intuitive way. Previous exposures of Ru-capped multilayers to EUV radiation in the presence of water vapor demonstrated that the mirror damage rate actually decreases with increasing water pressure. We will present results of recent exposures showing that the reduction in damage for partial pressures of water up to 5×10 ^-6Torr is not the result of a spatially uniform decrease in reflectivity across the Gaussian intensity distribution of the incident EUV beam. Instead, under a range of water partial pressures, we observe a drop in the damage rate in the center of the exposure spot where the intensity is greatest, while the reflectivity loss in the wings of the intensity distribution appears to be independent of water partial pressure. An investigation is underway to find the cause of the non-Gaussian damage profile. Preliminary results and hypotheses will be discussed. In addition to high-resolution reflectometry of the EUV-exposure sites, the results of surface analysis such as XPS will be presented. Although the observations presented here are based on exposures of Ru-capped samples, unless novel capping layers are similarly characterized, direct application of accelerated testing results could significantly overestimate mirror lifetime in the production environment.