Normal Vector Coupled Wave Analysis Application In Microelectronic Structure Analysis Of Cd

Gratings are common structures in microelectronics and integrated optics. Themeasurement to the topography of diffraction gratings has been developed for a longtime in some foreign countries, but it still needs great improvement until now.Optical CD metrology is widely used in semiconductor manufacturing processmonitoring and control for many years. Compared with other CD measurement methods(such as CD-SEM,XSEM,AFM,etc.),here are many advantages for OCD: low costs andhigh yield, small footprint, making it to the main technology of the CD integratedmetrology (IM).With good accuracy,OCD can measure the surface of a complex structure withdifferent refractive index. These advantages make it to be a popular candidatetechnology of equipment and process control. In the past decade, OCD has used forphotoresist trim and gate etch process control, and process monitoring of some complexmemory structure, and IM lithography process control. As the IC industry is progressingtowards32nm technology, Optical CD metrology will play an important role to helpaddress many challenges in process control.OCD metrology measure a fine structure in a way that it contrast the optical signalmeasured with the simulation results of a pre-built structure. The pre-built modelsimulation is based on accurate diffraction theory. With some given structuralparameters,the optical response, usually the0th diffraction rate, can be found bycomplicated calculations. Keep changing the structural parameters, consistent with thesimulation data and measured data, to find the parameters of the target model. Thisprocess is commonly referred to as the optimization analysis.RCWA is the key of simulation. With a perfect performance in one-dimensiongratings，in multidimensional gratings it need many diffraction orders，and convergeslowly. This is because RCWA not fully considering the electric field on the differentdirections of continuity. The Normalvector-RCWA(NV-RCWA) has solved this problemby constructing a normalvector field of the topography and improved the RCWA. Thispaper will examine the realization of the NV-RCWA, and performance comparison with the RCWA, finally proposed a simple NV-RCWA method.The key of an optimization method is to quickly locate the nearest minima pointfrom the initial search point. For the range parameters of OCD, the optimization methodshould be extended to the multidimensional. Levenberg-Marquardt optimization methodis a more suitable method.