Photonics-Based Technologies Improve Semiconductor Yield

Vijay Sankaran, Sematech

The silicon wafer, the basic unit of material in the semiconductor industry, encounters a range of processing techniques, such as crystal growth, ion implantation, film deposition, lithography, etch, metal deposition and polish. Multiple wafers are subjected to more than 500 intricate steps in today's chip-making process. When one step drifts out of acceptable limits, it generates defects that may affect the downstream process. By the time a wafer reaches the end of the line, it may have cost as much as $3000 to produce.
Considering that manufacturers produce as many as 5000 wafers a week at each fabrication facility, finding defects near the end of the process can amount to significant revenue loss. By detecting process drifts rapidly, the source of the problem can be identified and corrected before substantial waste is incurred. Effective inspection and metrology practices can improve overall yield.
These techniques fall under the domain of statistical process control and are associated with control of lot-to-lot variations. In situ sensors and controllers enable a closed-loop interactive method where the within-wafer or wafer-to-wafer processing variations can be monitored closely. Umpteen photonics-based approaches are used to improve yield, but the industry must overcome further technical challenges to extend their applicability.