Highly spatial-resolved chemical metrology on latent resist images

Patterning photoresist with extreme control over dose and placement is the first crucial step in semiconductor manufacturing. But, how to accurately measure the activation of modern complex resists components at sufficient spatial resolution? No exposed nanometre-scale resist pattern is sufficiently sturdy to unaltered withstand inspection by intense photon or electron beams, not even after processing and development. This paper presents experimental proof that Infra-Red Atomic Force Microscopy (IR-AFM) is sufficiently sensitive and gentle to chemically record the vulnerable-yet-valuable lithographic patterns in a chemically amplified resist after exposure, prior to development. Accordingly, IR-AFM metrology provides the long-sought-for insights in changes in the chemical and spatial distribution per component in a latent resist image, both directly after exposure as well as during processing. With these to-be-gained understandings, a disruptive acceleration of resist design and processing is expected.

Predictive Analytics for Equipment Maintenance Operations

Predictive maintenance (PdM) is a key application of data analytics in semiconductor manufacturing. The optimization of equipment performance has been found to deliver significant revenue benefits, especially in the wafer fabrication process. This chapter addresses two main research objectives: first, to investigate the particular challenges and opportunities of implementing PdM for wafer fabrication equipment and, second, to identify the implications of PdM on key performance indicators in the wafer fabrication process. The research methodology is based on a detailed case study of a wafer fabrication facility and expert interviews. The findings indicate the potential benefits of PdM beyond improving equipment maintenance operations, and the chapter concludes that the quality of analytics models for PdM in wafer fabrication is critical, but this depends on challenging data preparation processes, per machine type. Without valid predictions, decision-making ability and benefits delivery will be limited.

No Permits, No Fabs: The Importance of Regulatory Reform for Semiconductor Manufacturing

Congress has advanced legislation to appropriate $52 billion in funding for the CHIPS for America Act, which aims to increase semiconductor manufacturing and supply chain resilience in the United States. But more can be done to improve the resiliency of U.S. access to microelectronics beyond manufacturing incentives. This report outlines infrastructure investments and regulatory reforms that could make the United States a more attractive place to build new chipmaking capacity and ensure continued U.S. access to key inputs for semiconductor manufacturing.

A Deep Convolutional Neural Network-Based Multi-Class Image Classification for Automatic Wafer Map Failure Recognition in Semiconductor Manufacturing

Wafer maps provide engineers with important information about the root causes of failures during the semiconductor manufacturing process. Through the efficient recognition of the wafer map failure pattern type, the semiconductor manufacturing process and its product performance can be improved, as well as reducing the product cost. Therefore, this paper proposes an accurate model for the automatic recognition of wafer map failure types using a deep learning-based convolutional neural network (DCNN). For this experiment, we use WM811K, which is an open-source real-time wafer map dataset containing wafer map images of nine failure classes. Our research contents can be briefly summarized as follows. First, we use random sampling to extract 500 images from each class of the original image dataset. Then we propose a deep convolutional neural network model to generate a multi-class classification model. Lastly, we evaluate the performance of the proposed prediction model and compare it with three other popular machine learning-based models—logistic regression, random forest, and gradient boosted decision trees—and several well-known deep learning models—VGGNet, ResNet, and EfficientNet. Consequently, the comprehensive analysis showed that the performance of the proposed DCNN model outperformed those of other popular machine learning and deep learning-based prediction models.