Size matters: defect detectability in reticle and wafer inspection including advanced aerial image simulation for defect printability

Abstract

The lithography world is in transition from I-line to DUV for 0.25 micrometer and below technologies. However, no matter what printing wavelength we are using, reticle inspection is still off-wavelength for the foreseeable future. Due to the extreme push for resolution, k<SUB>1</SUB>-factor of 0.4 or even below is being used in manufacturing today. At the same time, the requirements for Usable Depth-Of-Focus (UDOF) are further put under pressure due to the inverse effect that pushing NA has on UDOF. To achieve maximum flexibility, steppers today can accommodate software programmable variable NA and Sigma settings. Resolution enhancement techniques [RET] have been tested and subsequently implemented into manufacturing processes. Lithography at 248 nm or even 193 nm will not achieve the desired resolution and process latitude without using RET. OPC [optical proximity correction] has become a mainstay in 0.25 micrometer designs. With all of these advancements, we also face more and more problems, especially in the area of reticle defect detection and printability prediction. Conventional rules of 1/4 of linesize equals minimum defect specification are no longer applicable. Minimum linewidth variations, for instance, can have a detrimental impact on device performance. Defect printability for optical proximity corrected (OPC) reticles was found context dependent. Reticle qualification needs to have an additional dimension added: on-line defect printability prediction. The ability to characterize the impact of a defect on a given feature, especially in an OPC design, will become an essential tool for mask makers and fabs alike, to evaluate defects and defect repair impact on critical device performance of a particular reticle. In this study, a special reticle design was used to investigate defect size, location and permutation, to evaluate: (1) the defect capture in an advanced reticle inspection system, (2) the defect printability prediction using a sophisticated wafer image simulation software package, (3) the comparison of inspected [reticle and wafer] vs. predicted wafer image, including DR-SEM capability, (4) the true CD impact of a given defect on LW performance using advanced CD-SEM measurements, and (5) the defect capture sensitivity of repaired reticle defects vs. their printability. A test plan was developed to study reticle defect detection, repair, printability prediction and actual wafer print. With the help of MicroUnity, a test vehicle was developed, that would allow for simultaneous inspection of no- OPC, serif-OPC and scattering-bar-OPC in the same inspection path, which then would be incorporated into the reticle in two manners: once without any 'decoration,' then 'decorated' with many different types of pre-programmed defects. In order to be able to also inspect the reticle in die-to-die and do some repair testing on it, the fields were duplicated, and also written at different address units of 0.08 and 0.04 micrometer.