|M.Sc Student||Kouniavsky Alexander|
|Subject||Performance Improvement of Optical Instrumentation by|
Control of Illumination and Diffraction Orders -
Theory and Experiment
|Department||Department of Physics||Supervisors||Dr. Joel Seligson|
|Dr. Vladimir Levinski|
|Full Thesis text|
Optical microlithography is one of the key technologies for the production for integrated circuits. One of the sources for imperfections in the microlithography process is the misregistration (overlay error) between two subsequent layers. Today’s lithographic technology permits minimum feature sizes (design rule) of 45 nm, which again requires overlay control down to 15 nanometers, and sub-nanometer performance of the overlay metrology equipment.
The main metric, which describes the optical performance of the overlay metrology tool, is so-called tool induced shift (TIS). TIS is defined as half of the sum of a given overlay measured at 00 and 1800 wafer orientations. TIS can be interpreted as an error in the overlay measurement due to tool asymmetry, and it is common practice to correct the measured overlay by subtracting TIS from it.
An additional metric, derived from TIS, is TIS-variability. It is generally defined as three times the standard deviation of TIS, measured across a given wafer.
TIS-variability is one of the main contributors to measurement uncertainty, as it sets a lower limit to possible TIS-correction. The traditional way to reduce TIS-variability is to improve the optical quality of the metrology tool, and to provide more symmetrical illumination. In this way, the capabilities of automated overlay metrology tools have been steadily improved and single-nanometer levels in metrology accuracy are currently achieved. Any further improvements using these methods, however, involve serious technical challenges and rapidly increasing costs.
The current research deals with TIS-variability reduction using an alternative approach, which does not require any optical quality improvement. It is based on the two-beam imaging principle and includes two different optical architectures.
We have analyzed two-beam imaging architectures as applied to overlay metrology. The analysis is based on theoretical calculations, on simulations using a full Maxwell-equation solver, and on a set of experiments. We have demonstrated the capacity of the two-beam imaging approach to reduce TIS-variability by an order of magnitude.