Measurement of photoresist grating profiles based on multiwavelength scatterometry and artificial neural network

We employed a grating profile measurement method, which is based on the combination of multiwavelength scatterometry and artificial neural network, to determine the critical dimensions of submicrometer-period photoresist gratings with wavy sidewall profiles. Six laser beams in three wavelengths and two orthogonal polarizations were adopted for the scatterometry measurement, and the incident angle of each beam was chosen following principles that we propose for achieving high sensitivity. We measured diffraction efficiencies of a large number of photoresist gratings made on glass substrates and high-reflectivity multilayer substrates coated with a chromium thin-film layer, and determined the grating groove parameters using a neural network model. The experimental results are statistically compared with results extracted from scanning electron micrographs. Good agreements between the indirect, neural network predicted results and the direct, scanning electron microscopy results are obtained.