Problem: Low k1 photolithography processes are increasing the complexity of RET applications in nanometer designs. At 45nm, more complex models and through process window correction and verification requirements significantly increase computational burden. Both the lithographic challenges and the computational complexity associated with the 45nm process node create a need for advanced capabilities for computational lithography tools.

Solution: Calibre nmOPC uses dense simulation, process window optimized OPC that includes a hybrid computing platform utilizing co-processor acceleration (with the Cell BE processor), a new compact resist process modeling capability, and design-intent aware correction algorithms. The fully integrated hierarchical geometry engine enables a fully integrated design to mask flow with a unified command language. Calibre nmOPC also handles practical production features such as: OASIS formatting to minimize output file size; new streamlined hierarchical processing to improve run time and file size compared to flat OPC tools, a progress meter and dynamic CPU allocation capability to manage TAT in a production environment. Process variability can have a dramatic effect on yield. This is especially true in the lithographic process where dose and focus variability impacts image fidelity. To reduce the risk of silicon failure and enable acceptable yield under challenging low k1 conditions, the software uses both dense simulation capabilities, which provide 100 percent simulation coverage for the mask layer, and process window correction optimization algorithms to ensure silicon-patterning success, according to the company. It also features multi-layer inputs into the correction algorithm to enable design critical features to be patterned to preserve design intent and parametric yield.

Applications: 45 and 32nm process generations

Platform: The Remote Acceleration Simulation (RAS) CoProcessor architecture enables the option of connecting a Coprocessor Acceleration cluster with an Ethernet connection to an existing compute cluster. Mentor has partnered with Mercury Computer Systems to offer standard Coprocessor Acceleration clusters based on the ‘Cell Broadband Engine' (CBE). The Cell Processor clusters are claimed to accelerate the image processing components of Calibre nmOPC enabling 4 to 10X improvements in run time with little to no increase in general purpose computing requirements over the 65nm node.

Availability: November 2007 onwards.