David Watts

Enabling Scatterometry as an In-Line Measurement Technique for 32 nm BEOL Application

Conventional metrology tools are unable to precisely monitor some interconnect attributes such as trench sidewall angle either due to limited capability or excessive cycle time. But these attributes have great impact on interconnect performance for 32 nm technology node and beyond. Scatterometry, a non-destructive metrology technique, is proposed to address the shortcomings of current metrology tools while also potentially providing additional measurement capabilities that enable more comprehensive characterization of interconnect attributes. Enabling scatterometry for back-end-of-line metrology at 32 nm technology node is challenged by the inherent complexity of a multilayer film structure. The research reported describes the implementation of scatterometry measurements to explore the advantages of this technique for the 32 nm technology node. The results obtained demonstrate the superiority of scatterometry techniques over conventional semiconductor metrology tools such as throughput, process control capability, precision, and accuracy. The total measurement uncertainty of scatterometry results with tunneling electron microscope and cross-sectional scanning electron microscope results for line height shows 1.92 and 6.46 nm, respectively, which compare favorably to the reference metrology tools. Scatterometry techniques also exhibited impressive potential to estimate end-of-the-line electrical parametric data. Finally, physical dimensions obtained from scatterometry measurements are shown to be comparable to TEM results from product wafers.

Variability Modeling and Process Optimization for the 32 nm BEOL Using In-Line Scatterometry Data

Process variability is a great concern when it comes to the fabrication of nano-scaled devices precisely. The effect of any imprecision can be directly translated into uncertain behavior of the devices. To address the process related issues, it is now essential to identify physical variability properly for a quality end product. If the effects of the variations are not correctly characterized, there is no other way of guaranteeing that the design will meet the specified budgets. This paper describes the essential variability modeling and analyses for the BEOL critical parameters. We used AQUAIA, to model end-of-the-line electrical resistances and capacitances based on 32 nm technology assumptions. By using scatterometry and reference metrology data, we have compared the correlations among the physical in-line measurements and end of the line electrical measurements which eventually address the potential variability issues between them specifically for the 32 nm technology node. It shows good correlation between scatterometry measurement results and results obtained from the AQUAIA simulation. Fitting parameters are generated with the help of AQUAIAs simulation results and physics model. Finally, we have developed a spreadsheet for the RC graph using those fitting parameters to manipulate and optimize the BEOL specification for 32 nm technology. This spreadsheet can be used as a guideline for the process development and control.