Abstract
Optimal performances of integrated optical devices are obtained by the use of an accurate and reliable characterization method. The parameters of interest, i.e., optical indices and thickness of the waveguide structure, are calculated from effective indices by means of an inversion procedure. We demonstrate how an artificial neural network can achieve such a process. The artificial neural network used is a multilayer perceptron. The first result concerns a simulated anisotropic waveguide. The accuracy in the determination of optical indices and waveguide thickness is and , respectively. Then an experimental application on a silica–titania thin film is performed. In addition, effective indices are measured by m-lines spectroscopy. Finally, a comparison with a classical optimization algorithm demonstrates the robustness of the neural method.
© 2007 Optical Society of America
Full Article | PDF ArticleMore Like This
Yann Battie, Stéphane Robert, Issam Gereige, Damien Jamon, and Michel Stchakovsky
Appl. Opt. 48(28) 5318-5323 (2009)
Yuan-sheng Ma, Xu Liu, Pei-fu Gu, and Jin-fa Tang
Appl. Opt. 35(25) 5035-5039 (1996)
Stéphane Robert, Alain Mure Ravaud, Stéphanie Reynaud, Sabine Fourment, Franck Carcenac, and Philippe Arguel
J. Opt. Soc. Am. A 19(12) 2394-2402 (2002)