Effects of wafer noise on the detection of 20-nm defects using optical volumetric inspection

Barnes, Brian M.; Goasmat, Francois; Sohn, Martin Y.; Zhou, Hui; Vladár, András; Silver, Richard M.

doi:10.1117/1.jmm.14.1.014001

Cited by 12 publications

(8 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As the Authors have reported both simulated 19,31 and experimental 32,33 image data from defective and nominal semiconductor patterns previously, the limited number of images informed the decision to assess these loss functions using labeled ML image data sets. First, results reported should be wholly reproducible which requires open access image sets.…”

Section: Data Sets and Neural Networkmentioning

confidence: 99%

Addressing misclassification costs in machine learning through asymmetric loss functions

Barnes

Henn

2023

Metrology, Inspection, and Process Control XXXVII

View full text Add to dashboard Cite

Patterning defect metrology requires data interpretation with classification, each well-suited to machine learning (ML). Defect classification however has notable misclassification costs; mislabeling a defect as nominal has greater impact than the converse. Though quantified costs are not publicly available, total economic misclassification cost (total cost) is optimized across orders-of-magnitude variation in cost ratio C and classification threshold 0.01 <τ< 0.99. Convolutional neural networks are trained using the intrinsically weighted and scaled asymmetric focal losses (AFL, sAFL) with hyperparameter γ with weighted and unweighted binary cross-entropy (wBCE, BCE) functions trained for comparisons. Optimal functions and conditions are identified for reducing total cost. For reproducibility, publicly available ML data sets are surrogates for industrial imaging data. For these data the sAFL mimimizes total cost at τ = 0.5, C ≥ 16. The AFL reduces total cost at 0.1 ≤ τ< 0.5, C > 128. Asymmetric loss functions lower total cost versus wBCE by 15 % to 40 % for 0.2 <τ< 0.5, C< 64. Total economic misclassification cost can be tailored using asymmetric focal losses. Estimations are presented to allow the extension of reported trends to industrial applications with strong class imbalances between defect-indicative and nominal-indicative data.

show abstract

Section: Data Sets and Neural Networkmentioning

confidence: 99%

Addressing misclassification costs in machine learning through asymmetric loss functions

Barnes

Henn

2023

Metrology, Inspection, and Process Control XXXVII

View full text Add to dashboard Cite

show abstract

“…The simulations were performed using a well-verified [28][29][30] in-house implementation of the finite-difference time-domain [31] (FDTD) method to model the electromagnetic field scattered from the patterned layout and its defects. The incident angle of the illumination is chosen to be normal to the substrate for clarity; prior simulation results indicate that the defect detection often varies when using oblique illumination [29,32]. The linear polarization basis within the simulation is defined with respect to the long axis of the nominal pattern.…”

Section: Simulation Detailsmentioning

confidence: 99%

“…In addition to measurement noise, "wafer noise" due to process variations is included [38][39][40][41][42]. Line edge roughness (LER) is known to be present in every lithographically manufactured device, either reducing the signal or increasing the noise [43].…”

Section: Simulation Detailsmentioning

confidence: 99%

Data-driven approaches to optical patterned defect detection

2019

Self Cite

View full text Add to dashboard Cite

Computer vision and classification methods have become increasingly wide-spread in recent years due to ever-increasing access to computation power. Advances in semiconductor devices are the basis for this growth, but few publications have probed the benefits of data-driven methods for improving a critical component of semiconductor manufacturing, the detection and inspection of defects for such devices. As defects become smaller, intensity threshold-based approaches eventually fail to adequately discern differences between faulty and non-faulty structures. To overcome these challenges we present machine learning methods including convolutional neural networks (CNN) applied to image-based defect detection. These images are formed from the simulated scattering of realistic geometries with and without key defects while also taking into account line edge roughness (LER). LER is a known and challenging problem in fabrication as it yields additional scattering that further complicates defect inspection. Simulating images of an intentional defect array, a CNN approach is applied to extend detectability and enhance classification to these defects, even those that are more than 20 times smaller than the inspection wavelength.

show abstract

“…Experimental methods for making full use of this light field have included angle-resolved imaging in a high-magnification platform [5,16-18] as well as the acquisition of focus-resolved images for defect metrology [16,19,20] and dimensional metrology [14,21]. …”

Section: Scatterfield Microscopymentioning

confidence: 99%

Enabling quantitative optical imaging for in-die-capable critical dimension targets

et al. 2016

View full text Add to dashboard Cite

Dimensional scaling trends will eventually bring semiconductor critical dimensions (CDs) down to only a few atoms in width. New optical techniques are required to address the measurement and variability for these CDs using sufficiently small in-die metrology targets. Recently, Qin et al. [Light Sci Appl, 5, e16038 (2016)] demonstrated quantitative model-based measurements of finite sets of lines with features as small as 16 nm using 450 nm wavelength light. This paper uses simulation studies, augmented with experiments at 193 nm wavelength, to adapt and optimize the finite sets of features that work as in-die-capable metrology targets with minimal increases in parametric uncertainty. A finite element based solver for time-harmonic Maxwell's equations yields two- and three-dimensional simulations of the electromagnetic scattering for optimizing the design of such targets as functions of reduced line lengths, fewer number of lines, fewer focal positions, smaller critical dimensions, and shorter illumination wavelength. Metrology targets that exceeded performance requirements are as short as 3 μm for 193 nm light, feature as few as eight lines, and are extensible to sub-10 nm CDs. Target areas measured at 193 nm can be fifteen times smaller in area than current state-of-the-art scatterometry targets described in the literature. This new methodology is demonstrated to be a promising alternative for optical model-based in-die CD metrology.

show abstract

Effects of wafer noise on the detection of 20-nm defects using optical volumetric inspection

Cited by 12 publications

References 12 publications

Addressing misclassification costs in machine learning through asymmetric loss functions

Addressing misclassification costs in machine learning through asymmetric loss functions

Data-driven approaches to optical patterned defect detection

Enabling quantitative optical imaging for in-die-capable critical dimension targets

Contact Info

Product

Resources

About