Virtual metrology for plasma etch using tool variables

IEEE Trans. Semicond. Manufact.

MacGearailt

2012

Self Cite

Abstract-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.Index Terms-Gaussian process regression, local modeling, neural network applications, plasma etch, virtual metrology (VM).

Section: A Regional Pm Cycle Modelssupporting

confidence: 76%

Global and Local Virtual Metrology Models for a Plasma Etch Process

IEEE Trans. Semicond. Manufact.

MacGearailt

2012

Self Cite

“…Endpoint detection is a somewhat special case of virtual metrology, since it focuses more on event, rather than continuous variable, estimation. By way of example, the paper by Lynn et al [27] provides an example of a range of both statistical and neural network modelling paradigms applied for VM to tool variables, with the use of stepwise regression for variable selection.…”

Section: B Virtual Metrologymentioning

confidence: 99%

Estimation and Control in Semiconductor Etch: Practice and Possibilities

IEEE Trans. Semicond. Manufact.

Bacelli

et al. 2010

Abstract-Semiconductor wafer etching is, to a large extent, an open-loop process with little direct feedback control. Most silicon chip manufacturers rely on the rigorous adherence to a "recipe" for the various etch processes, which have been built up based on considerable historical experience. However, residue buildup and difficulties in achieving consistent preventative maintenance operations lead to drifts and step changes in process characteristics. This paper examines the particular technical difficulties encountered in achieving consistency in the etching of semiconductor wafers and documents the range of estimation and control techniques currently available to address these difficulties. An important feature of such an assessment is the range of measurement options available if closed-loop control is to be achieved.Index Terms-Closed-loop control, plasma etch, run-to-run control, semiconductor wafer, state estimation, virtual metrology.

“…The physical properties of the chamber are changed by chemicals deposited on the chamber walls after each wafer etched, and this ultimately affects the etch response. A sample variable that exhibits PM shifts and process drift is shown in Figure 1 Previous work has examined the disaggregation of datasets to combat the effect of PM events [9] and to identify key process variables from datasets [10]. This paper focuses on the VM of etch rate using sliding-window modelling methods to Normalised variable value from etch process that exhibits shifts due to maintenance events and drift due to chamber conditioning.…”

Section: Introductionmentioning

confidence: 99%

Weighted windowed PLS models for virtual metrology of an industrial plasma etch process

2010 IEEE International Conference on Industrial Technology

MacGearailt

2010

Self Cite

Abstract-Virtual metrology is the prediction of metrology variables using easily accessible process variables and mathematical models. Because metrology variables in semiconductor manufacture can be expensive and time consuming to measure, virtual metrology is beneficial as it reduces cost and throughput time. This work proposes a virtual metrology scheme that uses sliding-window models to virtually measure etch rates in an industrial plasma etch process. The windowed models use partial least squares (PLS) regression and a sample weighting scheme to combat the effects of both process drifts due to machine conditioning and process shifts due to maintenance events. An industrial data set is examined and the weighted windowed PLS models outperform global models and non-weighted windowed models.