Object-oriented information models, tools, and methodologies for semiconductor device design

Recent progress in simulation technologies has opened new possibilities for semiconductor device and process design. Instead of relying strictly on physical experiments in a fabrication facility, designers can now conduct computer-simulated experiments in a "virtual factory" to verify and refine their designs. Use of a virtual factory that models equipment, process, device, circuit, and factory behavior can reduce the number of expensive, slow-turnaround "real" experiments. Unfortunately, wide acceptance of virtual factories is hindered by difficulties in using the simulation environment.; This work creates a device design system called the Simulation Experiment Workbench (SEWB) to support the rapid prototyping of semiconductor devices in the virtual factory. Device, test, simulation experiment, and device behavioral model libraries are built following an object-oriented information modeling methodology to organize and share device design experiences. Information components common to both virtual and physical factories are represented. Design libraries encourage design reuse as the device design libraries provide subsequent technology development generations with past device designs, recommended tests, and parameter extraction methodologies.; The SEWB environment enables the device designer to specify, execute, and analyze simulation experiments seemlessly. Device structures can be borrowed from the device library and graphically modified. Variability in a device structure is described using parametric operators in the graphical editor. Simulation experiments are created with the application of experimental design methodologies on the device objects. Test sequences composed from the Test Library are then added to the simulation experiments. A Task Manager automatically generates the input files, discretizes the device structure for each simulation run, and distributes the runs on a distributed workstation environment. Upon completion of a simulation experiment, design goal parameters are extracted and analyzed using regression analysis and visualization.; SEWB was used to explore the design space of a MOS transistor and trench structure. Different substrate configurations are also examined for their impact on substrate noise. These examples show that rapid prototyping of devices is possible when design libraries and integrated applications are provided to the user. SEWB allows the user to concentrate on the key design issues and minimize the number of mechanical, time-consuming, and error-prone tasks.