ABSTRACT
Optical systems that perform on-wafer measurement of particles have become one of the most important process control tools available to semiconductor man ufacturers and the IC industry. They give the process owner information neces sary to determine whether or not a specific process is under control, and so are ultimately significant contributors to yield-increasing strategies. It follows that the functional requirements of these systems depend on which process in the production cycle they support. Thus, during process development, a system with high throughput is not as important as a system with high-detection sensitivity and classification capability. On the other hand, a system employed to monitor a production line must have a throughput compatible with that line, without sacrificing sensitivity to those defects that are yield limiting [1]. The manner in which this trade-off between speed and sensitivity is accomplished is what dis tinguishes different commercial inspection systems from one another. However, regardless of the individual approaches, all such systems are based on the same physical principles. They are all concerned with the detection and classification of wavelength-or subwavelength-sized defects using light scattering in an envi ronment that includes a large scattering surface (the wafer) and possibly other light-scattering features intentionally incorporated onto the wafer surface (e.g., circuit patterns). In this chapter we focus on the physical foundations underlying the operation and design of a generic on-wafer inspection system.
