An information-theory and virtual metrology-based approach to run-to-run semiconductor manufacturing control

Virtual Metrology (VM) and soft sensing modules have become popular in the past years and are now widely adopted in semiconductor plants. Nevertheless, few scientific works have so far investigated interactions between VM and Run-to-Run (R2R), the most common control approach in the field. In this paper, a novel strategy aimed at integrating VM and R2R relying on Information Theory measure is presented and motivated. The proposed control method penalizes statistical measurements based on their informative distance from real metrology data. This approach is also able to cope with the virtual loop control, in which the R2R runs for several process iterations without actual measurements, relying only on VM predictions. The results are compared with the current state-of-the-art by means of simulation studies based on realistic assumptions.