New Methods for Cell-based OPC

The invention relates to the field of integrated-circuit layout optimization, particularly to optimization of integrated-circuit standard-cell layouts for cell-based optical proximity control.

Optical proximity correction (OPC) has been a key enabler of the aggressive IC technology scaling implicit in Moore 's Law. OPC determines the photomask patterns that enable drawn layout features to be faithfully and accurately reproduced by optical lithography onto the wafer. However, the runtime of model-based OPC tools (i.e., software tools that use optical simulation and geometric operations to determine the photomask pattern for each layout feature) has grown unacceptably long with each successive technology generation, and has emerged as one of the major bottlenecks in the turnaround time for IC data preparation and manufacturing.

The cell-based OPC approach is to run OPC once per each cell definition (i.e., per “cell master”) rather than once per placement or unique instantiation of each cell (i.e., per “cell instance”). In other words, in the cell-based OPC approach, the master cell layouts in the standard-cell library are corrected before placement, and then placement and routing steps of IC design are completed with the corrected master cells. Unfortunately, optical proximity effects in lithography have a certain interaction radius between layout pattern geometries. Since the neighboring environment of a cell in a full-chip layout is completely different from the environment of an isolated cell, the cell-based OPC solution can be incorrect when instantiated in a full-chip layout: as a result, there can be a large difference in feature critical dimension (CD) between cell-based OPC and conventional model-based OPC.

Several solutions to the above short-fall have been proposed in recent literature, however none offer a complete solution. In “Merits of Cellwise Model-Based OPC”, Gupta et al, ( Proc. SPIE Conference on Design and Process Integration for Microelectronic Manufacturing , Vol. 5379, pp. 182-189 (2004)) three kinds of dummy features are devised to emulate – for purposes of calculating the OPC solution – possible different neighboring environments of a cell. The dummy features are inserted at three predetermined areas described as Border Poly, Top-Bottom Poly and Contact Poly before performing OPC. It has been shown that the OPC solution created in the presence of dummy features is not necessarily a correct or accurate solution when the layout is inserted into an arbitrary environment within a standard-cell block. The dummy feature has only limited ability to capture proximity effects from pattern geometries of neighboring cells, and hence CD errors will still remain after cell-based OPC.

In “Exploiting hierarchical structure to enhance cell-based RET with localized OPC reconfiguration.” by Wang et al ( Proc. SPIE Conference on Design and Process Integration for Microelectronic Manufacturing , Vol. 5756, pp. 361-367 (2005)), a method which accounts for OPC re-correction in proximity interaction areas between cells of a standard-cell block is described. The approach is to first perform cell-based OPC, and then to stitch already-corrected cells into the layout, re-correcting the OPC solution within interacting areas at the cell boundaries, to obtain a proximity-corrected final layout. However, the stitching area can increase OPC runtime and degrade the ostensible benefits of cell-based OPC such as cell reuse and cell-based timing analysis.

In, “The novel approach for optical proximity correction using genetic algorithms,” by Matsunawa et al. ( Proc. BACUS Symposium on Photomask Technology, Vol. 5992, pp.54-1 - 54-9 (2005)), a genetic algorithm to correct the stitching area is introduced. However, the method also requires rework of the layout (e.g., GDSII representation) of the OPCed cell design.

The present invention employs the use of an auxiliary pattern (AP) which shields poly patterns at the cell outline from proximity effects and thus minimizes the CD difference between cell-based OPC and conventional model-based OPC. Also devised is a new design methodology to insert at least one vertical AP (“vertical-AP” between horizontally adjacent cells (i.e., neighboring cells in the same cell row) so that cell-based OPC achieves the same accuracy as conventional model-based OPC methods. The AP and postplacement techniques together provide substantial improvements for OPC runtime, cell-based timing characterization, and cell re-spins for ECO (Engineering Change Order). 

The present invention is realized in software and is available for commercial evaluation and licensing.

Laboratory Link: Professor Andrew B. Kahng

Case No: SD2006-215, SD2006-824
Inquiries To:  invent@ucsd.edu