Mask Error Modeling via Aerial Image Sensitivity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional photolithography process models are inaccurate and inefficient in modeling mask errors, leading to reduced efficacy in optical proximity correction and increased time-to-market for integrated circuits, which can result in significant revenue loss.
Innovation Solution
The system models mask errors using aerial image sensitivity by measuring critical dimensions and fitting an uncalibrated process model with edge bias and corner rounding terms, based on coefficients that represent sensitivity to local and long-range pattern densities, allowing for accurate prediction of aerial image intensity and mask error deviations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional photolithography process models are used, then the manufacturing process is simple, but the accuracy of feature size predictions is poor
Solution Approach 1:
The process model is segmented into multiple components: an uncalibrated process model representing the base lithography process, and separate mask error modeling terms (edge bias term and corner rounding term) that account for specific mask errors. This segmentation allows each component to be calibrated independently using measured CD data, improving overall prediction accuracy without overwhelming complexity
Solution Approach 2:
The patent introduces adjustable parameters including edge bias coefficient and corner rounding coefficient that can be calibrated to match measured critical dimensions. These parameters are modified based on aerial image sensitivity calculations, allowing the model to adapt to specific manufacturing conditions and improve prediction accuracy for feature sizes
2Reliability
If conventional process models are used, then the computational process is fast, but the accuracy of optical proximity correction is reduced
Solution Approach 1:
The patent performs preliminary calibration of the process model by fitting it to measured CD data from actual mask layouts before using it for optical proximity correction. This preliminary action establishes accurate edge bias and corner rounding coefficients that improve OPC reliability. The calibrated model is then reused for multiple OPC calculations, reducing overall computational time
Solution Approach 2:
The mask error modeling terms are applied locally at different regions of the mask layout where specific error types occur. Edge bias terms are applied at edges and corner rounding terms at corners, allowing accurate local corrections without requiring complete remodeling of the entire process, thus maintaining computational efficiency
3Manufacturing precision
If accurate process modeling is implemented, then manufacturing precision improves, but computational efficiency decreases
Solution Approach 1:
The patent uses aerial image sensitivity as a computational proxy to model mask errors without requiring exhaustive simulations of every possible mask variation. By calculating sensitivity of the aerial image to small mask modifications, the system efficiently estimates edge bias and corner rounding effects, achieving accurate process modeling at reduced computational cost
Data Source
AI summary
One embodiment of the present invention provides techniques and systems for modeling mask errors based on aerial image sensitivity. During operation, the system can receive an uncalibrated process model which includes a mask error modeling term which is based at least on an aerial image sensitivity to mask modifications which represent mask errors. Next, the system can fit the uncalibrated process model using measured CD data. Note that the mask error modeling term can also be dependent on the local and/or long-range pattern density. In some embodiments, the mask error modeling term can include an edge bias term and a corner rounding term. The edge bias term can be based on the sensitivity of the aerial image intensity to an edge bias, and the corner rounding term can be based on the sensitivity of the aerial image intensity to a corner rounding adjustment.