تجزیه و تحلیل حساسیت مقاومت پارامتر بر اساس شبیه سازی درجه بندی شده برای درک محدودیت های مقاومت در لیتوگرافی نسل بعدی

Simulations for predicting resist effects in the sub 50 nm resolution regime are strongly requested today, as well as for improvement of present resolution and CD control. Therefore this letter reports about a simulative resist parameter sensitivity analysis with help of calibrated resist models, based on Sigma-C’s SOLID software. Target of the study was to learn about the impact of resist parameters on practical resolution limits and to derive specific process and material change proposals. After resist model calibration for 90 nm design rules, process window, mask error enhancement factor, linearity, line end shortening etc. were investigated. The main influencing resist parameters were determined with two independent methodologies: Single-and multiparameter variation, which showed good agreement. Further, the sensitivity analysis was expanded to feature sizes down to 20 nm halfpitch. To decouple all optical influences from the resist, ideal rectangular aerial images were generated and used for simulation. The simulation reveals that an ArF resist might be capable of 40 nm resolution with sufficient exposure dose latitude, comparable to today’s 90 nm design rule. On the other hand even optimized exposure tools can’t provide such ideal rectangular aerial images and there is no commercial resist known today that shows a process window at 40 nm resolution today. Therefore, the main key resist parameters, which are responsible for resolution enhancement, were identified out of this simulation study and proposals for improved processes are given.

Today’s lithography roadmap targets design rules down to 32 nm halfpitch and below. To resolve these small structures there is still no consensus whether 193 nm immersion or EUV lithography will be used. Beside challenges regarding light source, optics, mask, immersion liquid and defectivity it is also still unclear, whether the resist material will be able to meet resolution requirements [1]. The best reported resolution of dense lines/spaces is about 40 nm using EUV or water based 193 nm immersion lithography. In both cases, exposures were carried out under reasonable photospeed conditions below 20 mJ/cm2[2]. Additionally, it has to be considered that all experimental work on resist and process screening has to be performed on the few, worldwide available 193i and EUV exposure tools. To effectively plan today’s resist experiments, it is recommended to evaluate material effects in the sub 50 nm resolution regime also by simulations. Subsequently, simulated results can be verified using appropriate experiments. Therefore, this letter reports a simulative resist parameter sensitivity analysis evaluating important lithography effects. Simulations were performed using calibrated models at a feature size of 90 nm. In a second step, the investigations were expanded down to minimal feature sizes of 20 nm with help of idealized rectangular aerial images. For all simulations, Sigma C’s SOLID and process window analyzer (PWA) software was used.

Resist models (193 nm) were generated by calibration of 90 and 70 nm printing results. A resist parameter sensitivity study for 90 nm simulations was carried out to identify key resist parameters and their impact on printing behavior. Comparison of single parameter sensitivity analysis showed good agreement with the multiparameter approach of Tollkuehn et al. A resolution study down to 20 nm lines/spaces using digital aerial images revealed further insight. Comparison of two resists from the same vendor showed that a large process window is possible with long acid diffusion, fast deprotection rate is and moderate acid generation efficiency. The key parameters have been translated into resist-chemical terms and proposals for process improvement have been given. Modern 193 nm resists offer sufficient exposure dose latitude for 40 nm resolution, if enough aerial image contrast can be provided by the optical system. Future work will focus on the resist line edge roughness issue. New stochastic simulation methods will therefore be used to describe the impact of resist polymer properties and shot noise effects.