Abstract

Optical methods are predominantly used in the photolithography process for linewidth measurement, mask and wafer alignment, and photoresist image formation. As the feature size required for modern semiconductor devices shrinks, it is becoming more difficult to accurately print and measure these fine features on the wafer surface. Simulation of the image formation can aid us in understanding the usefulness and limitation of these optical methods, so that improvement can be made to achieve better accuracy. Approximate theories such as scalar diffraction theory do not explain many of the phenomena, and sometimes rigorous E&M theories beyond Born and Wolf’s level have to be used. Several rigorous scattering models applied in the photolithography areas are introduced first. Most of these models fall into two categories: time-domain solvers and frequency-domain solvers. Maxwell equations are solved by these models to different degrees of rigor to compromise between accuracy and speed. Hardware constraints may exist because of the different algorithms used for the models. The pros and cons of these models are discussed. Finally, the discrepancies existing in today’s models as well as the future needs for more rigorous electromagnetic simulations are explained. Hopefully the photolithography simulation community can benefit from the people who specialize in the E&M field but do not specifically work in the integrated circuit area.

© 1991 Optical Society of America