Conventional vs. model-based measurement of patterned line widths from scanning electron microscopy profiles.

Area of Science:

• Materials Science
• Nanotechnology
• Metrology

Background:

• Scanning electron microscopy (SEM) is crucial for nanometer-scale metrology.
• Conventional width measurement techniques rely on arbitrary thresholds, leading to inaccuracies.
• Precise dimensional control is vital for semiconductor devices.

Purpose of the Study:

• To quantify the errors in conventional SEM width measurements.
• To evaluate the accuracy of a model-based library (MBL) method for nanoscale feature sizing.
• To compare the conventional threshold method with the MBL approach.

Main Methods:

• Simulated secondary electron signals from silicon line shapes with varying dimensions.
• Employed Monte Carlo simulations of electron transport using four inelastic scattering models.
• Determined feature widths using both conventional thresholding and the MBL method.

Main Results:

• Conventional SEM width measurements can have errors of several nanometers.
• The MBL method, using a reference model (FPA + Auger), showed less than 1 nm variation between models of differing accuracy.
• Errors from conventional methods are unacceptable for modern semiconductor industry requirements.

Conclusions:

• Conventional SEM width measurements are inadequate for nanoscale applications.
• The model-based library (MBL) method provides superior accuracy for nanometer-scale feature metrology.
• MBL measurements are preferable for critical dimensions in the nanoscale domain.