ICCAD-2003. International Conference on Computer Aided Design (IEEE Cat. No.03CH37486) · 2003 · 66 citations · 17 references
EngineeringDesignManufacturing EngineeringDesign OptimizationManufacturing ProcessesParametric DesignDesign For ManufacturingDevice ModelingFabrication TolerancesIntegrated Circuit DesignManufacturing ConstraintsPhysical DesignDetailed RoutingManufacturability-aware DesignManufacturingPhysical Design ConstraintsPlanarizationProcess CornersProcess Variation AnalysisStatistical Timing Analysis
Ultra-deep submicron manufacturability impacts physical design (PD) through complex layout rules and large guard-bands for process variability; this creates new requirements for new manufacturing-aware PD technologies. The first part of this tutorial reviews PD complications and methodology changes notably in the detailed routing arena - that arise from subwavelength lithography and deep-submicron manufacturing (antennas, metal planarization and mask-wafer mismatch). Process variations and their sources are taxonomized for modeling and simulation. A framework of design for cost and value is described. The second part covers yield-constrained optimizations in PD, especially "beyond corners" approaches that escape today's pessimistic or even incorrect corner-based approaches. Statistical timing and noise analyses enable optimization of parametric yield and reliability. Yield-aware cell libraries and "analog" design rules (as opposed to "digital", 0/1 rules) can help designers explore yield-cost tradeoffs, especially for low-volume parts. We then examine performance impact-limited fill insertion which goes beyond mere capacitance rules. Modeling, objectives, and filling strategies are discussed. Finally, we discuss current and near-term prospects for the overall design-to-manufacturing PD methodology. Key aspects include better integrations with analysis and manufacturing interfaces, as well as cost-benefit tradeoffs for "regular" layout structures that are likely beyond 90 nm, cost optimizations for low-volume production, and the role of robust and/or stochastic optimization in PD.
17