Methods for Advanced Multi-Objective Optimization for eDFY of complex Nanoscale Circuits

Cost control, production efficiency, cycle time and yield, besides perfomance aspects, are the  critical quality benchmark for nano- electronics productions. A problem in the production of nanoscaled circuits is the fact that the minituarisation of the device sizes yields a disproportionate dependence of the circuit behavior on the production tolerances. For instance, phenomena like edges or surfaces roughness play an important role. As a result, the figures of merit of a circuit, such as performance and power, have become extremely sensitive to hard controllable statistical process variations (PV) during the chip production.

To ensure stable manufacturability and secure high manufacturing yield, it is mandatory to manage complete design flows and to link traditional methods for design with Technology CAD models. In this context, multi-objective optimization algorithms and statistical analysis are essential on device and behavioral levels to secure high yielding by modeling the impact of inevitable process variations and doping fluctuations on IC performance.

The project is a cooperation between the electronics manufacturer STMicroelectronics, the EDA-tool provider MunEDA, the university of Rome and the Fraunhofer ITWM. Thereby the aim is to exchange and combine the mathematical know how on multi-objective optimization, symbolic techniques and numerical statistical simulation on one side, the industrial design experience, real test cases availability and “Electronic Design Automation” (EDA) software modeling skills on the other.

The scope of the research activity will be to create PV-aware and PV-robust circuit design techniques, tools and models in the frame of the analogue and mixed-signal circuit industrial design.
Thereby the focus lies on the generation of fast simulatable behavioral models under parameter variations that are necessary for the usage of iterative multidimensional optimization methods. Here a combination of response-surface-models and symbolic model order reduction methods will be considered.