Real-time virtual metrology and control for plasma etch

Abstract

a b s t r a c tPlasma etch is a semiconductor manufacturing process during which material is removed from the surface of semiconducting wafers, typically made of silicon, using gases in plasma form. A host of chemical and electrical complexities make the etch process notoriously difficult to model and troublesome to control. This work demonstrates the use of a real-time model predictive control scheme to control plasma electron density and plasma etch rate in the presence of disturbances to the ground path of the chamber. Virtual metrology (VM) models, using plasma impedance measurements, are used to estimate the plasma electron density and plasma etch rate in real time for control, eliminating the requirement for invasive measurements. The virtual metrology and control schemes exhibit fast set-point tracking and disturbance rejection capabilities. Etch rate can be controlled to within 1% of the desired value. Such control represents a significant improvement over open-loop operation of etch tools, where variances in etch rate of up to 5% can be observed during production processes due to disturbances in tool state and material properties.

Real-time virtual metrology and control of etch rate in an industrial plasma chamber

“…When real metrology from drifting processes is available, the VM model will require refreshing to maintain model currency (as proposed in [4]). A similar refreshing technique may be required for the independent model of the PFC system, as investigated in [12]. Secondly, separate VM models may be required for different tools in fabrication plants due to mismatch between individuatl tools and sensors.…”

“…Finally, Sections V and VI give the control results and paper conclusions, respectively. The results in this paper, along with further results on the real-time control of plasma electron density [11], have been submitted for publication in journal format in [12].…”

In-situ virtual metrology for the silicon-dioxide etch rate by using optical emission spectroscopy data

“…Control schemes in views of real-time control and run-to-run (R2R) control have been suggested. More recently, a publication by the same authors, the estimation of plasma's electron density and etch rate were estimated in real time was focused to achieve a real-time model for a predictive control scheme [12].…”

Growing Structure Multiple Model System for Quality Estimation in Manufacturing Processes

“…Work by Lynn et al[21] examined a weighted PLS implemented within a moving window, determining that weighted updating of a PLS based VM model was slightly more accurate than VM based on a global PLS or using a moving window alone.Besides linear regression models, neural networks are also very commonly used models in VM, primarily due to their well-known ability to approximate non-linear dependencies.Generally speaking, neural networks comprise numerous interconnected computational nodes representing non-linear functions of relevant inputs, and weighted sums of the outputs of those nodes are used to obtain a functional relationship for the desired output. Within VM, two types of neural networks have been studied: BPNN and RBNN.…”