In the application of the grating, it is necessary to quickly obtain the measurement results of the structural parameters, and the parameters of the measured grating are usually reversed by means of scatterometry. We propose a neural network-based grating parameter optimization model. By inversely calculating the diffraction efficiency measurement results, the structural parameters of the grating can be quickly obtained. Applying the model in the experiment, the relative error of the groove depth of the transmission grating is 0.23%, the relative error of the duty ratio is 0.92%, the relative error of the groove depth of the reflection grating is 0.91%, and the relative error of the duty ratio is 2.15%. Using the neural network tool to measure the grating structure parameters, the measurement results can be obtained quickly and accurately. |
ACCESS THE FULL ARTICLE
No SPIE Account? Create one
Diffraction gratings
Diffraction
Neural networks
Neurons
Reverse modeling
Error analysis
Optical engineering