https://pureadmin.qub.ac.uk/ws/files/136557002/fsiv_ieea_final_accepted_paper.pdf

An enhanced variable selection and Isolation Forest based methodology for anomaly detection with OES data

Modeling of manufacturing processes is important because it enables manufacturers to understand the process behavior and determine the optimum operating conditions of the process for a high yield, low cost and robust operation. However, existing techniques in modeling manufacturing processes cannot address the whole common issues in developing models for manufacturing processes: a) manufacturing processes are usually nonlinear in nature; b) a small amount of experimental data is only available for developing manufacturing process models; c) outliers often exist in experimental data; d) explicit models in a polynomial form are often preferred by manufacturing process engineers; and e) models with satisfactory prediction accuracy are required. In this paper, a modeling algorithm, namely, the particle swarm optimization-based fuzzy regression (PSO-FR) approach, is proposed to generate fuzzy nonlinear regression models, which seek to address all of the common issues in developing models for manufacturing processes. The PSO-FR first employs the operations of particle swarm optimization to generate the structures of the process models in nonlinear polynomial form, and then it employs a fuzzy coefficient generator to identify outliers in the original experimental data. Fuzzy coefficients of the process models are determined by the fuzzy coefficient generator in which the experimental data excluding the outliers is used. The effectiveness of the PSO-FR approach is evaluated by modeling the manufacturing process liquid epoxy molding process which is a commonly used technology for microchip encapsulation in electronic packaging. Results were compared with those based on the commonly used modeling methods. It was found that PSO-FR can achieve better goodness-of-fitness than other methods. Also, the prediction accuracy of the model developed based on the PSO-FR is better than the other methods.

Modeling of a Liquid Epoxy Molding Process Using a Particle Swarm Optimization-Based Fuzzy Regression Approach

In semiconductor manufacturing processes, virtual metrology (VM) has been investigated as a promising tool to predict important characteristics of products. Although partial least squares (PLS) is a well-known modeling technique that can cope with collinearity and therefore applied to construction of VM, its prediction performance deteriorates due to changes in process characteristics. In particular, maintenance of equipment strongly affects the process characteristics and the prediction performance. In this paper, VM was developed by using locally weighted PLS (LW-PLS), which is a type of just-in-time modeling technique, and it was used to predict the etching conversion differential of an actual dry etching process. The industrial application results have shown that the developed VM based on LW-PLS is superior to the conventional VM based on the sequential update model and artificial neural network model. In particular, it has been confirmed that the LW-PLS-based VM can keep its high prediction performance even after the maintenance, i.e., replacement of parts.

/pdf/adaptive-virtual-metrology-design-for-semiconductor-dry-1or2mvyeyu.pdf

Adaptive Virtual Metrology Design for Semiconductor Dry Etching Process Through Locally Weighted Partial Least Squares

A Convolutional Autoencoder Approach for Feature Extraction in Virtual Metrology

Virtual metrology (VM) is the estimation of metrology variables that may be expensive or difficult to measure using readily available process information. This paper investigates the application of global and local VM schemes to a data set recorded from an industrial plasma etch chamber. Windowed VM models are shown to be the most accurate local VM scheme, capable of producing useful estimates of plasma etch rates over multiple chamber maintenance events and many thousands of wafers. Partial least-squares regression, artificial neural networks, and Gaussian process regression are investigated as candidate modeling techniques, with windowed Gaussian process regression models providing the most accurate results for the data set investigated.

/pdf/global-and-local-virtual-metrology-models-for-a-plasma-etch-3kul7jmnq6.pdf

Global and Local Virtual Metrology Models for a Plasma Etch Process

The objective of this paper is to present the utilization of information produced during plasma etching for the prediction of etch bias. A plasma etching process typically relies on the concentration of electrically activated chemical species in a reaction chambers over time, depending on chamber pressure, gas flow rate, power level, and other chamber and wafer properties. Plasma properties, as well as equipment factors, are complex and vary over time. In this paper, we will use various statistical techniques to address challenges due to the nature of plasma data: high dimensionality, colinearity, parameter interactions and nonlinearities, variation of data structure due to equipment condition changing over time, etc. The emphasis will be data integrity, variable selection, accommodation for process dynamics, and model-building methods. Different techniques will be evaluated with an industrial dataset.

Virtual Metrology Modeling for Plasma Etch Operations

The objective of this work is the creation of predictive models that can forecast the electrical or physical
parameters of wafers using data collected from the relevant processing tools. In this way, direct measurements from the
wafer can be minimized or eliminated altogether, hence the term "virtual" metrology. Challenges include the selection of
the appropriate process step to monitor, the pre-treatment of the raw data, and the deployment of a Virtual Metrology
Model (VMM) that can track a manufacturing process as it ages. A key step in any VM application is dimensionality
reduction, i.e. ensuring that the proper subset of predictors is included in the model. In this paper, a software tool
developed with MATLAB is demonstrated for interactive data prescreening and selection. This is combined with a
variety of automated statistical techniques. These include step-wise regression and genetic selection in conjunction with
linear modeling such as Principal Component Regression (PCR) and Partial Least Squares (PLS). Modeling results
based on industrial datasets are used to demonstrate the effectiveness of these methods.

Dekong Zeng

Papers

Virtual Metrology Modeling for Plasma Etch Operations

Dimensionality reduction methods in virtual metrology