Random forest-based robust classification for lithographic hotspot detection

Geometrical positioning surveying-based features for BEOL line-end-pull-back modeling using regression-based machine-learning

“…Several ML applications were proposed to solve manufacturing-related problems in domains such as litho HS detection, 3,[10][11][12][13][14][15][16][17][18][19] optical proximity correction (OPC), [20][21][22] sub-resolution assist features (SRAF) insertion, 23,24 and litho simulation. 25 Litho HS detection is one of the most plentiful domains for ML applications, with a wide variety of related layout representation features.…”

“…Zhang et al 16 represented the layout clips by concentric circle area sampling (CCAS) and proposed an ML model that employs smooth boosting. Dawar et al 17 extracted aerial image and geometrical parameters with a random-forest (RF) classifier to detect litho HS. The extracted aerial image parameters are done with approximations and include: (a) process variation band, (b) normalized image log-slope, and (c) mask error enhancement factor.…”

“…Frequency domain/DCT features 13 are not computationally friendly-layout patterns are clipped and subjected to density-based scanning, then transformed into the frequency domain using DCT. Similarly, aerial image features 17 and PFT features 22 need optical simulations to be calculated, which may be suitable for ML-OPC and litho HS detection applications where features calculation is bounded by the downstream application and embedded within the overall flow. Adaptive-squish-pattern 18,19 is an adaptive-grid bitmap representation for a layout pattern that provides a compact lossless representation of layout patterns.…”

“…Each of these constraints has boundaries and can be constructed in the form of a DT model. Dawar et al 17 charted an example of a DT model for a simple constrained rule of tip-to-tip. We reintroduce this example in Fig.…”

“…This analogy can be extended to GPS features as well, as the features directly represent directional topological and dimensional measurements for a POI and its surrounding context of patterns and can be constructed in the form of a DT model. In addition, DT models have been incorporated with density pixels features, 14 aerial image, and geometrical features 17 for litho HS detection applications, and with CCAS features for SRAF placement. 23 Initially, we selected random forest (RF) 35 and adaptive-boosting RF (AdaBoostRF) 36 ML models from Scikit-learn ensemble regressors.…”