Machine Learning Applied to Electron Beam Lithography to Accelerate Process Optimization of a Contact Hole Layer.

Determining the lithographic process conditions with high-resolution patterning plays a crucial role in accelerating chip manufacturing. However, lithography imaging is an extremely complex nonlinear system, and obtaining suitable process conditions requires extensive experimental attempts. This severely creates a bottleneck in optimizing and controlling the lithographic process conditions. Herein, we report a process optimization solution for a contact layer of metal oxide nanoparticle photoresists by combining electron beam lithography (EBL) experiments with machine learning. In this solution, a long short-term memory (LSTM) network and a support vector machine (SVM) model are used to establish the contact hole imaging and process condition classification models, respectively. By combining SVM with the LSTM network, the process conditions that simultaneously satisfy the requirements of the contact hole width and local critical dimension uniformity tolerance can be screened. The verification results demonstrate that the horizontal and vertical contact widths predicted by the LSTM network are highly consistent with the EBL experimental results, and the classification model shows good accuracy, providing a reference for process optimization of a contact layer.

Machine learning in electron beam lithography to boost photoresist formulation design for high-resolution patterning.

The reduction of the critical dimension (CD) usually improves the resolution of patterns and performance of chips. In chip manufacturing, electron beam lithography (EBL) is a promising technology for preparing sub-10 nm patterns, and its imaging resolution is primarily determined by the photoresist formulation. However, the smaller CDs are mainly achieved by optimizing process conditions, and little attention has been paid to the photoresist formulation optimization. Screening suitable photoresist formulations remains a significant challenge due to the considerable time and high cost. Herein, we report a formulation optimization technique of a metal oxide nanoparticle photoresist that combines EBL experiments with a machine learning long short-term memory (LSTM) network. Using the LSTM network, a CD photoresist evaluation model is established. Leveraging the CD model, a photoresist formulation optimizer is developed with a line width of 26 nm. The verification results demonstrate that the CDs predicted by the LSTM network are basically consistent with the EBL experimental results, and the photoresist formulations that meet the CD requirements can be screened. This work opens up a novel perspective to boost photoresist formulation design for high-resolution patterning with artificial intelligence and provides guidance for EBL experiments.

Impacts of local soil and vehicle NOx emissions on ground-level NO2 concentrations on a university campus in the city of Shenyang, China.

Nitrogen dioxide (NO2) is a ubiquitous atmospheric pollutant, and fossil fuel combustion is generally considered its predominant source in and around urban areas. As the total nitrogen deposition is high over here, soil NOx emissions from urban green space might also be an important local source of ground-level NO2. In this study, Willems badge samplers were employed to monitor the spatial and seasonal variations of 2-week mean atmospheric NO2 concentrations at a height of 1.7 m on an urban campus in Northeast China from November 2020 to December 2021. We found considerable small-scale spatial variations of ground-level NO2 concentrations on the campus during the growing season, with local soil NOx emissions as the main driver. According to its linear correlation with green space coverage, the increment in ground-level NO2 concentration was partitioned into two components, with one ascribed to the local soil source (referred to as NO2-Isoil) and the other the local vehicle source (NO2-Ivehicle). NO2-Isoil generally reached a maximum (as high as 25.6 µg/m3) during early spring, while its ratio to the background value generally reached a maximum (could be >1) during late spring and could reach 0.52 to 0.92 during summer. Therefore, soil NOx emissions were an important source of ground-level NO2 on the campus, with the contribution even higher than those of other anthropogenic sources during late spring. Even with light traffic on the campus, NO2-Ivehicle could reach 0.48 times the background value at a site with high frequencies of warm starts.

Critical dimension prediction of metal oxide nanoparticle photoresists for electron beam lithography using a recurrent neural network.

The critical dimension (CD) of lithographic patterns is the most significant indicator for evaluating the imaging performance of photoresists, and its value is seriously affected by process conditions. However, the lithographic imaging system is highly nonlinear, and extensive exposure experiments are needed to obtain the desired CD. This consumes lots of time, manpower, and cost in screening for optimal process conditions. Here, we report a combined electron beam lithography (EBL) experiment and recurrent neural network (RNN) study on the CDs of metal oxide nanoparticle photoresists, and establish a CD RNN model. Leveraging the RNN model, a process condition filter is developed to screen suitable process conditions. The experimental results demonstrate that the prediction accuracy of the CD model exceeds 93%, and the photoresist patterns under the screened process conditions can satisfy the requirements of a preset CD. This work opens up a novel perspective for accurate EBL process modeling, and provides guidance for EBL experiments.

Process optimization of line patterns in extreme ultraviolet lithography using machine learning and a simulated annealing algorithm.

Resolution, line edge/width roughness, and sensitivity (RLS) are critical indicators for evaluating the imaging performance of resists. As the technology node gradually shrinks, stricter indicator control is required for high-resolution imaging. However, current research can improve only part of the RLS indicators of resists for line patterns, and it is difficult to improve the overall imaging performance of resists in extreme ultraviolet lithography. Here, we report a lithographic process optimization system of line patterns, where RLS models are first established by adopting a machine learning method, and then these models are optimized using a simulated annealing algorithm. Finally, the process parameter combination with optimal imaging quality of line patterns can be obtained. This system can control resist RLS indicators, and it exhibits high optimization accuracy, which facilitates the reduction of process optimization time and cost and accelerates the development of the lithography process.

Canny Algorithm Enabling Precise Offline Line Edge Roughness Acquisition in High-Resolution Lithography.

The line edge roughness (LER) is one of the most critical indicators of photoresist imaging performance, and its measurement using a reliable method is of great significance for lithography. However, most studies only investigate photoresist resolution and sensitivity because LER measurements require an expensive and not widely available critical dimension scanning electron microscopy (SEM) technology; thus, the imaging performance of photoresist has not been adequately evaluated. Here, we report an image processing software developed for offline calculation of LER that can analyze lithographic patterns with resolutions up to ∼15 nm. This software can effectively process all graphic files obtained from commonly used SEM machines by utilizing the adjustable double threshold. To realize the effective detection of high-resolution patterns in advanced lithography, we used SEM images generated from extreme ultraviolet and electron beam lithography to develop and validate the software's graphic recognition algorithm. This image processing software can process typical SEM images and produce reliable LER in an efficient and user-friendly manner, constituting a powerful tool for promoting the development of high-performance photoresist materials.

Process optimization of contact hole patterns via a simulated annealing algorithm in extreme ultraviolet lithography.

The critical dimension (CD), roughness, and sensitivity are extremely significant indicators for evaluating the imaging performance of photoresists in extreme ultraviolet lithography. As the CD gradually shrinks, tighter indicator control is required for high fidelity imaging. However, current research primarily focuses on the optimization of one indicator of one-dimensional line patterns, and little attention has been paid to two-dimensional patterns. Here, we report an image quality optimization method of two-dimensional contact holes. This method takes horizontal and vertical contact widths, contact edge roughness, and sensitivity as evaluation indicators, and uses machine learning to establish the corresponding relationship between process parameters and each indicator. Then, the simulated annealing algorithm is applied to search for the optimal process parameters, and finally, a set of process parameters with optimum image quality is obtained. Rigorous imaging results of lithography demonstrate that this method has very high optimization accuracy and can improve the overall performance of the device, dramatically accelerating the development of the lithography process.

Aberration optimization in an extreme ultraviolet lithography projector via a BP neural network and simulated annealing algorithm.

Extreme ultraviolet lithography (EUVL) presents promise for the advanced technology node in the manufacturing of integrated circuits. The imaging performance of EUVL is significantly affected by the aberration of projection optics. To obtain one optimum aberration for different test patterns, an inverse optimization method for aberration is proposed in this paper. The aberration models of three types of test patterns are first established by applying the backpropagation (BP) neural network. Then choosing the common indicators of the lithography process variation band (PVB) and pattern shift (PS) as the objective function, an aberration optimization method based on the algorithm of simulated annealing is proposed. After applying the optimization method, a set of optimized aberrations and the corresponding PVBs and PSs are obtained and analyzed. These results are finally compared with those from rigorous simulations. The comparison results show that zero aberration is non-optimal distribution in EUVL image simulation with mask topography. In addition, the high prediction accuracy and robustness of aberration optimization is also demonstrated from the results.

Analysis and modulation of aberration in an extreme ultraviolet lithography projector via rigorous simulation and a back propagation neural network.

Lens aberration is a critical factor affecting lithography, one that deteriorates the image fidelity and contrast. As the perfect lens does not exist, the aberration control is important for real optical systems, especially for extreme ultraviolet lithography (EUVL). By choosing the process variation band (PVB) and pattern shift (PS) as the lithographic performance indicators, the inverse analysis model for aberration control is proposed in this paper. First, the effects of aberration with 36 Zernike terms on lithography performance are forward analyzed. Using the definitive screening design (DSD) and with the help of statistical analysis methods of analysis of variance and F test, the combined Zernike terms leading to prominent PVB and PS are identified. After giving a brief introduction of backpropagation neural network (BPNN), the aberration control model based on DSD and BPNN is then established. Finally, several examples are analyzed to demonstrate the effectiveness and robustness of the aberration control model. Predicted results show that the optimum distribution of Zernike coefficients given by the aberration model can generate minimum impact on imaging quality, and this impact is very close to that of zero aberration. The results demonstrate that the BPNN-based aberration model has the potential to be an efficient guiding method for controlling the aberration of EUVL in the optical design stage.