Reactive ion etch modeling using neural networks and simulated annealing

Kim, B.; May, G.S.

doi:10.1109/3476.484199

Cited by 32 publications

(7 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They used the "leave-one-out" cross-validation procedure in developing the model (cf., Cohen [27]). Kim and May [28] discuss a D-optimal experiment to design the network architecture of a neural network plasma etch model, and Kim and May [29] use a modification of the backpropagation algorithm in which the cost function allows the network to slowly degrade or forget information that is no longer needed. This is essentially the same as weight regularization or weight minimization to reduce the network complexity.…”

Section: Whole System Model For Feedback Controlmentioning

confidence: 99%

Pre-production results demonstrating multiple-system models for yield analysis

Rietman¹,

Friedman²,

Lory³

1997

IEEE Trans. Semicond. Manufact.

View full text Add to dashboard Cite

We have assembled an integrated view of the entire via manufacturing process. This integrated study includes five key plasma processes that culminate in the production of vias on CMOS wafers. There are essentially no linear cross-correlations between the processing steps and there are no linear correlations between the individual process steps and the yield for vias, as measured by the resistance between metal-one (M1) and metaltwo (M2).Using a neural network, we demonstrate that the key processing steps to determine the M1M2 resistance are the thick oxide deposition and the anisotropic via etch. Of lesser significance are the etchback planarization, an isotropic etch and plasma enhanced tetra-ethoxy siliane (PETEOS) deposition. Keeping in mind that there are five processing steps, the numerical value of M1M2 resistance can be predicted ahead of time, before completion of all five processes. This prediction can be done to an accuracy of about 1 : By using adaptive neural networks, the intelligent agents can modify their predictive behavior with respect to process changes effected by the engineering staff. Our pre-production demonstration suggests that these programs could be used in feedback and feedforward control for production yield.

show abstract

Section: Whole System Model For Feedback Controlmentioning

confidence: 99%

Pre-production results demonstrating multiple-system models for yield analysis

Rietman¹,

Friedman²,

Lory³

1997

IEEE Trans. Semicond. Manufact.

View full text Add to dashboard Cite

show abstract

“…Neural network combined with statistical experimental design has been used to understand underlying etch mechanisms. 1,5,6 Neural network models once demonstrated improved prediction over statistical regression models in modeling plasma etch processes. 7,8 In the context of oxide film, neural network has only been applied in a CHF 3 reactive ion plasma.…”

Section: Introductionmentioning

confidence: 98%

“…The oxide film was reactive ion etched in CHF 3 /O 2 plasma and qualitatively modeled. 1 Using a high density plasma, the film was also etched in C 4 F 8 /H 2 , 2 C 2 F 6 , 3 and CHF 3 4 plasmas.…”

Section: Introductionmentioning

confidence: 99%

“…7,8 In the context of oxide film, neural network has only been applied in a CHF 3 reactive ion plasma. 1 In this study, neural networks are used to model oxide films etched in a CHF 3 /CF 4 magnetically enhanced reactive ion plasma. A statistical experimental design was employed to systematically characterize relationships between process factors and etch responses.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Modeling oxide etching in a magnetically enhanced reactive ion plasma using neural networks

Kim

Kwon

et al. 2002

Journal of Vacuum Science &Amp; Technology B: Microelectronics and Nanometer Structures Processing, Measurement, and Phenomena

View full text Add to dashboard Cite

Oxide films etched in a CHF3/CF4 plasma were qualitatively modeled using neural networks. The etching was conducted using a magnetically enhanced reactive ion etcher. A statistical 24-1 experimental design plus one center point was used to characterize relationships between process factors and etch responses. The factors that were varied include a radio frequency power, pressure, CHF3 and CF4 flow rates. The resultant nine experiments were used to train neural networks and trained networks were subsequently tested on eight experiments not belonging to the training data. A total of 17 experiments were thus conducted for modeling. Etch responses modeled are etch rate and etch profile. Root-mean-squared prediction errors of optimized models are 111 Å/min and 3.50° for the etch rate and etch profile, respectively. Interaction effects between the factors were examined from the prediction models. Besides the dc bias, the F and CF2 intensities measured with an optical emission spectroscopy were related to the etch responses. Polymer deposition effect was transparent in the etch rate for a variation in CHF3, but little for the one in CF4. Conflicting effects between CHF3 and CF4 were noticed with respect to either etch response. Pressure-induced dc bias played an important role in determining etch responses.

show abstract

“…Each neuron contains the weighted sum of its inputs filtered by a sigmoidal "squashing" function, providing neural networks with the ability to generalize with an added degree of freedom that is not available in statistical regression techniques. Due to the inherent ability to learn complex nonlinear mapping, neural networks have been applied to semiconductor process modeling, and those studies reported excellent results using neural network-based techniques [9].…”

Section: Neural Networkmentioning

confidence: 99%

Neural network modeling of reactive ion etching using principal component analysis of optical emission spectroscopy data

Hong

May

13th Annual IEEE/SEMI Advanced Semiconductor Manufacturing Conference. Advancing the Science and Technology of Semiconductor Ma

Self Cite

View full text Add to dashboard Cite

In this paper, neural networks trained by the error back-propagation algorithm are used to build models of etch rate, uniformity, selectivity and anisotropy as a function of optical emission spectroscopy (OES) data in a reactive ion etching process. The material etched is benzocyclobutene (BCB), a low-k dielectric polymer, which is etched in an SFs and 9 plasma in parallel plate system. Neural network training data are obtained from a multi-way principal component analysis (MPCA)of the OES data. These data are acquired from a z4 factorial experiment designed to characterize etch process variation with controllable input factors consisting of the two gas flows, RF power and chamber pressure. Evaluation of the trained neural networks is performed in terms of root mean square (RMS) error, and less than 3% prediction errors are achieved.

show abstract

Reactive ion etch modeling using neural networks and simulated annealing

Cited by 32 publications

References 13 publications

Pre-production results demonstrating multiple-system models for yield analysis

Pre-production results demonstrating multiple-system models for yield analysis

Modeling oxide etching in a magnetically enhanced reactive ion plasma using neural networks

Neural network modeling of reactive ion etching using principal component analysis of optical emission spectroscopy data

Contact Info

Product

Resources

About