Editor-in-Chief Harry J. Levinson gives an overview of the dynamic state of lithography and potential new directions the technology could take.

A prescription for obtaining the three-dimensional irradiance of any periodic pattern is given and applied to obtaining the irradiance and the depth of focus (DOF) for lines-and-spaces patterns. As is well known, the DOF for such patterns can be made large by restricting the angular spread of the illuminating light. At a fixed k₁, the so-called tip-to-tip distance between two opposing line-ends can be made smaller if the numerical aperture of the imaging optics is increased.

Yan Borodovsky, a retired Intel Senior Fellow, reflects on his years working at Intel during the tenure of Gordon Moore, who passed away on March 24, 2023.

Guest editors Gian Francesco Lorusso and Chris A. Mack introduce the Special Section on Advances in E-Beam Metrology.

Extreme ultraviolet lithography has advanced microfabrication of semiconductor devices toward the sub-10-nm generation. In this situation, stochastic defects increase and hence process evaluation requires an entire wafer inspection at high speed. To satisfy this requirement, a large field of view (FoV) inspection with low-resolution enables us to inspect an entire wafer within an acceptable time because the throughput of e-beam inspection depends on imaging resolution. However, low-resolution images are difficult to inspect at high precision using conventional methods because of a smaller photographed defect size and worse signal-to-noise ratio. Moreover, deformation caused by the manufacturing process and larger distortion caused by large FoV result in false detections when we apply die-to-database (D2DB) inspection. To solve these issues, we propose trainable D2DB inspection, which predicts a pixel-value distribution of normal images from a corresponding design layout. The proposed method is robust to low-resolution images because it considers noise and acceptable deformation as variance of the learned distribution. In addition, by introducing a model to predict a misalignment between a design layout and inspection image, trainable D2DB becomes robust to image distortion. Experiments show that trainable D2DB can perform high-precision inspection on images with large noise and image distortion.

Things are drastically changing in the field of metrology. The main reason for that is related to the daunting specification requirements for metrology imposed by high numerical aperture extreme ultraviolet lithography (high NA EUVL). We observe a variety of generation e-beam tools proliferating in imec unique ecosystem, from in-line transmission electron microscope (TEM) to voltage contrast (VC) overlay tools, from die to database (D2DB) large area scanning electron microscope (SEM) to high-voltage SEM, from artificial intelligence (AI)-based inspection tools to massive data acquisition e-beam system. We are facing a renaissance of e-beam metrology. We are going to describe the challenges as well as the latest evolutionary developments of e-beam metrology in the semiconductor industry.

The shrinkage of circuit patterns for improvement of the semiconductor device performance has reduced the tolerances in production. To fit in the tolerances, technologies for improving the uniformity of three-dimensional (3D) shapes of circuit patterns inter- or intra- wafers has been developed. Then, we developed a method for quantifying variations in 3D shapes by critical-dimension scanning electron microscopy (CD-SEM), which can measure widths of circuit patterns with high sensitivity. Since variations in the SEM-image signal are caused by 3D-shape variations, in the method, multiple feature values representing the signal detect shape variations. To compare the effect of the variation in each feature value on the shape variation, the amount of variation in the feature values was normalized by local variations in a reference image. Evaluation on etched wafers showed that several features exhibited independent variation trends that were larger than the local fluctuation. Cross-sectional verification confirmed that one of the feature values correlated with the width at the middle height that cannot be seen in top-view, and variations of 0.24 nm can be detected. It is expected that adjusting processing conditions based on this variation trend will efficiently improve the uniformity of the 3D shape.

We present an automated application for defect detection and classification from ZEISS multibeam scanning electron microscope (MultiSEM^®) images, based on machine learning (ML) technology. We acquire MultiSEM images of a semiconductor wafer suited for process window characterization at the imec iN5 logic node and use a dedicated application to train ML models for defect detection and classification. We show the user flow for training and execution, and the resulting capture and nuisance rates. Due to straightforward parallelization, the application is designed for the large amounts of data generated rapidly by the MultiSEM.

An ultra-fast image simulation algorithm is proposed. The new algorithm uses full fast-Fourier-transform (FFT) to calculate the aerial image intensity. The wavelength, 193 nm, was scaled to a number of powers of 2, through scaling the mask with a scaling factor derived from the discrete Fourier transform (FT) format. The mask can then be transformed to the diffraction spectrum in terms of spatial frequency using the FFT algorithm. Similarly, this mask diffraction spectrum can be inverse transformed to the aerial-image by using the inverse-FFT algorithm. The image is finally scaled back to the original image amplitude of the original wavelength and squared to the image intensity. Comparing to the original FT, there is a 4000 × to 5000 × computation speed improvement with only about 3% intensity deviation. This algorithm provides an efficient engine for lithography optimization.

Low-n masks have gained strong interest due to their potential to simultaneously improve dose and imaging contrast for dense clips. We have previously presented that for the imaging of isolated features mask bias, assist features are crucial to minimize the focus range through pitch. In this paper, we elaborate on aberration sensitivity for different mask-absorber types. We observe that even aberration sensitivities can change significantly by changing the mask-absorber type for the same use case. We show that even aberration sensitivity and best-focus shifts are coupled and that they can also be solved together by applying mask and target bias and/or assist features. Finally, we show how assist-feature position optimization can reduce the impact of odd aberrations on two-bar features.

Formulation optimization plays an important role in the research and development of chemically amplified resist (CAR). However, the CAR profile after development process is influenced by multiple resist parameters and process conditions, so it is hard to determine the optimal CAR formulation in the multivariate problem. An optimization method for the CAR formulation is developed. The simple random sampling is applied to each CAR parameter’s value range independently, and the combinations of these samples from different parameters are used in the simulation of lithography profiles. Kernel density estimation is applied to analyze the simulation results. Then the CAR formulation is optimized based on the probability density distribution from the analysis results. The verification results show that the proposed optimization method can greatly improve the stability of the CAR formulation and thus generating acceptable critical features’ sizes of the CAR profile.

This erratum adds an omitted affiliation.