Partition response surface modeling
View Patent ↗A group of models are developed to predict printed contour deviations relative to the corresponding layout edges for different classes of layout topologies. A plurality of calibration layouts with topologies belonging to a class of layout topologies are used to generate a model for the class of layout topologies. A standard least square regression is modified for model creation. The model error may be monitored dynamically.
1. A method, comprising:
by a computing device:
receiving layout data for a plurality of calibration layouts, each of the plurality of calibration layouts having a layout topology belonging to a class of layout topologies;
deriving density information for the plurality of calibration layouts based on the layout data;
receiving printed contour information for the plurality of calibration layouts;
deriving contour deviation information for the plurality of calibration layouts based on the printed contour information; and
generating a model for predicting printed contour deviations for the class of layout topologies based on the density information and the contour deviation information, wherein the model is generated using a response surface methodology.
2. The method recited in claim 1 , wherein one or more calibration layouts in the plurality of calibration layouts are generated using a place and route tool.
3. The method recited in claim 1 , wherein one or more calibration layouts in the plurality of calibration layouts are layout portions extracted from one or more design layouts.
4. The method recited in claim 1 , wherein the class of layout topologies is associated with a litho hotspot.
5. The method recited in claim 4 , wherein the litho hotspot is line end-to-shape bridging, convex-to-shape bridging, slot end pinching or neighboring convex pinching.
6. The method recited in claim 1 , wherein the deriving density information includes:
using a grid to divide each of the plurality of calibration layouts into layout regions; and
calculating a density value for each of the layout regions.
7. The method recited in claim 6 , wherein the size of the grid is dependent upon light wavelength.
8. The method recited in claim 1 , wherein the printed contour information is generated through a lithography simulation process.
9. The method recited in claim 8 , wherein the lithography simulation process includes:
applying one or more RET (resolution enhancement techniques) treatments to the plurality of calibration layouts to obtain a plurality of modified calibration layouts; and
performing lithography simulation on the plurality of modified calibration layouts.
10. The method recited in claim 1 , wherein the deriving contour deviation information includes calculating displacement of a printed contour segment relative to a corresponding layout edge segment.
11. The method recited in claim 1 , wherein the generating a model includes using a least square regression technique.
12. The method recited in claim 11 , wherein the least square regression technique includes ridge regression.
13. The method recited in claim 11 , wherein the least square regression technique is modified by assigning weights inversely proportionally on residual values when computing final coefficients.
14. The method recited in claim 11 , further comprising:
monitoring model errors dynamically with an error bound function.
15. The method recited in claim 1 , wherein the response surface methodology uses at least some of the density information for one or more input variables and at least some of the contour deviation information for one or more response variables.
16. A processor-readable storage device storing processor-executable instructions for causing one or more processors to perform a method, the method comprising:
receiving layout data for a plurality of calibration layouts, each of the plurality of calibration layouts having a layout topology belonging to a class of layout topologies;
deriving density information for the plurality of calibration layouts based on the layout data;
receiving printed contour information for the plurality of calibration layouts;
deriving contour deviation information for the plurality of calibration layouts based on the printed contour information; and
generating a model for predicting printed contour deviations for the class of layout topologies based on the density information and the contour deviation information, wherein the model is generated using a response surface methodology.
17. The processor-readable storage device recited in claim 16 , wherein one or more calibration layouts in the plurality of calibration layouts are generated using a place and route tool.
18. The processor-readable storage device recited in claim 16 , wherein one or more calibration layouts in the plurality of calibration layouts are layout portions extracted from one or more design layouts.
19. The processor-readable storage device recited in claim 16 , wherein the class of layout topologies is associated with a litho hotspot.
20. The processor-readable storage device recited in claim 16 , wherein the deriving density information includes:
using a grid to divide each of the plurality of calibration layouts into layout regions; and
calculating a density value for each of the layout regions.
21. The processor-readable storage device recited in claim 16 , wherein the printed contour information is generated through a lithography simulation process.
22. The processor-readable storage device recited in claim 21 , wherein the lithography simulation process includes:
applying one or more RET (resolution enhancement techniques) treatments to the plurality of calibration layouts to obtain a plurality of modified calibration layouts; and
performing lithography simulation on the plurality of modified calibration layouts.
23. The processor-readable storage device recited in claim 16 , wherein the deriving contour deviation information includes calculating displacement of a printed contour segment relative to a corresponding layout edge segment.
24. The processor-readable storage device recited in claim 16 , wherein the generating a model includes using a least square regression technique.
25. The processor-readable storage device recited in claim 24 , wherein the least square regression technique includes ridge regression.
26. The processor-readable storage device recited in claim 24 , wherein the least square regression technique is modified by assigning weights inversely proportionally on residual values when computing final coefficients.
27. The processor-readable storage device recited in claim 24 , further comprising:
monitoring model errors dynamically with an error bound function.
28. The processor-readable storage device recited in claim 16 , wherein the response surface methodology uses at least some of the density information for one or more input variables and at least some of the contour deviation information for one or more response variables.