Prediction of plasma etch process by using actinometry-based optical emission spectroscopy data and neural network

Kim, Byungwhan; Kwon, Minji

doi:10.1016/j.jmatprotec.2008.06.006

Cited by 8 publications

(5 citation statements)

References 14 publications

(10 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To achieve equilibrium between the tracking error and the chattering, an ASMC based on the bipolar sigmoid function is proposed, 26,27 which can be derived from equation (6).…”

Section: Asmc Based On Bipolar Sigmoid Functionmentioning

confidence: 99%

“…Theorem 3. For the system (2), under the action of the sliding surface (4) and the controller (27), the adaptive law (29) can generate an appropriate control input which can guarantee the system output x to track the desired output x d asymptotically.…”

Section: Asmc Based On Hyperbolic Tangent Functionmentioning

confidence: 99%

“…In the internal controller, h(0) ¼ ['(0), a(0), b(0)] T ¼ [À1, 2, 0] T . The control law(27) and adaptive laws (29) are employed. Experimental results are illustrated in Figure12.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Robust control of the A-axis with friction variation and parameters uncertainty in five-axis CNC machine tools

Zhao

Shi

2014

Proceedings of the Institution of Mechanical Engineers, Part C:

View full text Add to dashboard Cite

A-axis is an essential assembly in the five-axis CNC machine tools, and its positioning precision directly affects the machining accuracy and surface quality of the parts. Considering the influence of parameters perturbation and uncertain cutting force on the control precision of the A-axis, a nonlinear dynamics model of the A-axis system is established, which reveals the relationships among the drive torque, the load torque, the motion direction and the system parameters. Then, two adaptive sliding mode controllers (ASMC) are designed. The first one is based on the bipolar sigmoid function, which can adjust the switching gain and the boundary layer thickness adaptively, then, equilibrium between the tracking error and the chattering can be achieved. The second one is a compound adaptive controller constituted by the traditional sliding mode controller (TSMC) and an internal controller that is based on the hyperbolic tangent function. When the state trajectory comes into the boundary layer, the TSMC will be replaced by the internal controller, thus, the adaptive controller can continuously switch between these two controllers, which can effectively eliminate the high-frequency chattering in the TSMC. Stability of these two controllers is guaranteed by the Lyapunov theory. Experimental results demonstrate the effectiveness and feasibility of the proposed ASMC, which can smooth the input chattering and reduce the tracking error by 16.62% and 21.44%, respectively.

show abstract

“…To achieve equilibrium between the tracking error and the chattering, an ASMC based on the bipolar sigmoid function is proposed, 26,27 which can be derived from equation (6).…”

Section: Asmc Based On Bipolar Sigmoid Functionmentioning

confidence: 99%

Section: Asmc Based On Hyperbolic Tangent Functionmentioning

confidence: 99%

See 1 more Smart Citation

Robust control of the A-axis with friction variation and parameters uncertainty in five-axis CNC machine tools

Zhao

Shi

2014

Proceedings of the Institution of Mechanical Engineers, Part C:

View full text Add to dashboard Cite

show abstract

“…A sizable body of research has been done on artificial neural network (ANN) models used as surrogate models for prediction of macroscopic plasma processing outputs (such as etch rate, deposition rate, etc) from the processing reactor control variables, such as RF power, pressure, or feed gas flows. Examples from plasma etch process modeling and real-time process control include, among others, the extensive work of Kim et al [26][27][28][29], Himmel and May [30], Rietman and Lory [31], Han et al [32], Stokes and May [33], or Tudoroiu et al [34]. The same is true for other areas of plasma processing.…”

Section: Introductionmentioning

confidence: 99%

A regression model for plasma reaction kinetics

Haničinec

Mohr

Tennyson

2023

J. Phys. D: Appl. Phys.

View full text Add to dashboard Cite

Machine learning (ML) is used to provide reactions rates appropriate for models of low temperature plasmaswith a focus on A + B $\rightarrow$ C + D binary chemical reactions. The regression model is trainedon data extracted from the QBD, KIDA, NFRI and UfDA databases. The regression model used a variety of dataon the reactant and product species, some of which also had to be estimated using ML.The final model is a voting regressor comprising three distinct optimized regression models: a support vectorregressor, random forest regressor and a gradient-boosted trees regressor model; this model is made freely available via a github repository.As a use case, the ML results are used to augment the chemistry of a BCl$_3$/H$_2$ gas mixture.

show abstract

“…In such cases, the modeling using artificial neural network (ANN) can be advantageous because ANN has the capability to learn arbitrary nonlinear mapping between sets of input and output parameters. A number of works have been found in the area of electronic packaging which have been modeled various fabrication process using ANN [3], [5]- [6]. However, not many works have been demonstrated the modeling of the fabrication process of polymer optical waveguide.…”

Section: Introductionmentioning

confidence: 99%

Process modeling of polymeric optical waveguide using artificial neural network

Haque

Wahid

Hossain

2012

2012 7th International Conference on Electrical and Computer Engineering

View full text Add to dashboard Cite

Polymeric optical waveguides have recently drawn a great interest in the fabrication of many functional optical devices. Photolithography is the widely used technique for the fabrication of such waveguides. To facilitate low-cost manufacturing and to achieve high reproducibility, determination and control of the process parameter in the different fabrication steps is very important. However, the mathematical modeling of waveguide fabrication process is quite difficult due to complex physical/chemical sub-processes involve in the lithography process. In this work, waveguide fabrication using photolithography has been modeled using artificial neural network which is capable to extract a simple and accurate empirical model of a complex nonlinear process. This modeling scheme predicts the value of the parameters required in the lithography process including spin speed and time, expose dose, develop time, power and pressure, while the geometry of the waveguide that includes film thickness and wafer size is predefined. The modeling results are compared with the experimental results and found a good experiment between them. Such agreement indicates that the modeling approach will be very useful for the process optimization and supervisory process control scheme.

show abstract

Prediction of plasma etch process by using actinometry-based optical emission spectroscopy data and neural network

Cited by 8 publications

References 14 publications

Robust control of the A-axis with friction variation and parameters uncertainty in five-axis CNC machine tools

Robust control of the A-axis with friction variation and parameters uncertainty in five-axis CNC machine tools

A regression model for plasma reaction kinetics

Process modeling of polymeric optical waveguide using artificial neural network

Contact Info

Product

Resources

About