SANTA CLARA, Calif. -- Results from beta tests of a new process control software program from Yield Dynamics Inc. here indicate that it can substantially increase yield and reduce rework, the company said today.
The company's Model Automated Production Applications (MAPA) software has been in beta site evaluations at three separate IC manufacturing facilities. The software is now available commercially.
"MAPA increased overlay-limited yield 8% for a 0.25-micron design-rule microprocessor and reduced overlay-related rework from 10% to 2% in an ASIC process shrink from 0.25-micron to 0.18-micron technology generations," said Terrence Zavecz, Yield Dynamics' vice president of lithography applications. "By simulating extended lot performance, MAPA can improve the overlay error budget by approximately 35 nanometers, enabling current mix-and-match lithography exposure tools to achieve the threshold required to produce 0.15-micron devices," he added.
MAPA gathers data automatically from multiple metrology sources. Raw, modeled and simulation data are stored in a relational database. Data is automatically analyzed using common statistical process control (SPC) techniques as well as statistics derived from tool and process models andsubsequent simulations tied to the derived model structures.
Systematic and random error modeling for overlay and registration information is provided. The APC software and its engineering modules address calibration, modeling and simulation of stepper and scanner lens wavefront analysis, focus uniformity and critical-dimension (CD) performance.
Typically, production data under-samples ("sparse sampling") the amount of data needed to properly characterize the performance of wafer fab lots. Using tool and layer-dependant models and simulations, MAPA extends sparse sample production data to adequately describe lot distributions.
Models provide a base estimate for the systematic and random components of the errors. The components are then simulated into field, wafer and lot-projected array distributions to obtain production performance statistics that are more accurate than any previously available. The accuracy of the technique results in lower rework rates and higher capacity utilization and improves work in process by removing the need to use dedicated exposure tools, explained Zavecz.