A method to accelerate creation of plasma etch recipes using physics and Bayesian statistics

Chopra, Meghali; Verma, Rahul; Lane, Austin P.; Willson, C. Grant; Bonnecaze, Roger T.

doi:10.1117/12.2263507

Cited by 4 publications

(4 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Materials integration has also shown promising results in semiconductor manufacturing. Chopra et al developed a software tool for creating plasma etch recipes based on physical models and Bayesian inference (8) and applied it to prediction of experimental results (9). They showed that the etching rate of SiO2 with CF4/Ar gasses could be predicted.…”

Section: Introductionmentioning

confidence: 99%

Reactive Force-Field Molecular Dynamics Study of SiGe Thin Film Growth in Plasma Enhanced Chemical Vapor Deposition Processes

Uene

Mabuchi²,

Zaitsu

et al. 2020

ECS Trans.

View full text Add to dashboard Cite

The SiGe thin films, such as hydrogenated amorphous SiGe (a-SiGe:H) and hydrogenated microcrystalline SiGe (μc-SiGe:H), are known as a potential material. In order to form promising SiGe thin films using chemical vapor deposition, relationship between materials/process and structure/composition should be clarified at the atomic level. We analyzed the influence of the substrate temperature and the ratio of SiH3 and GeH3, which are considered to be the dominant gaseous species on the deposition, on the crystallinity and atomic content in SiGe thin films by reactive force-field molecular dynamics simulations. The crystallinity increases as the substrate temperature increases and reaches its minimum value when the ratio of SiH3 and GeH3 is approximately 0.5. The hydrogen content decreased as the substrate temperature increases, while silicon and germanium content tended to increase as substrate temperature increase. These results were in good agreement with relevant studies.

show abstract

Section: Introductionmentioning

confidence: 99%

Reactive Force-Field Molecular Dynamics Study of SiGe Thin Film Growth in Plasma Enhanced Chemical Vapor Deposition Processes

Uene

Mabuchi²,

Zaitsu

et al. 2020

ECS Trans.

View full text Add to dashboard Cite

show abstract

“…Bouaziz et al try to predict the equipment health factor of semiconductor systems [42]. Khakifirooz et al use Bayesian inference for yield enhancement and industry 4.0 [43], and Chopra et al use Bayesian statistics to calibrate the physical model parameters of plasmon dry etching models [44], [45].…”

Section: Introductionmentioning

confidence: 99%

Physics-Prior Bayesian Neural Networks in Semiconductor Processing

et al. 2019

View full text Add to dashboard Cite

With the fast scaling-down and evolution of integrated circuit (IC) manufacturing technology, the fabrication process becomes highly complex, and the experimental cost of the processes is significantly elevated. Therefore, in many cases, it is very costly to obtain a sufficient amount of experimental data.To develop an efficient method to predict the results of semiconductor experiments with a small amount of known data, we use a novel method based on Bayesian framework with the prior distribution constructed by technology computer-aided-design (TCAD) physical models. This method combines the advantages of statistical models and physical models in the aspect that TCAD can provide visionary guidance on an experiment when a limited amount of experimental data is available, and a machine learning model can account for subtle anomalous effects. Specifically, we use aspect ratio dependent etching (ARDE) phenomenon as an example and use variational inference with Kullback-Leibler divergence minimization to achieve the approximation to the posterior distribution. The relation between etching process input parameters and etching depth is learned using the Bayesian neural network with TCAD priors. Using this method with 35 neurons per hidden layer, mean square error (MSE) in the test set is reduced from 0.2896 to 0.0175, 0.058 to 0.0183, 0.0563 to 0.0188, 0.058 to 0.019 for partition =10, 20, 30, 40, respectively, reference to the baseline BNN where a regular normal distribution prior with zero mean and unity standard deviation N(0,1) is used.INDEX TERMS Artificial intelligence, manufacturing, physics, Bayesian methods, intelligent manufacturing systems.

show abstract

“…This hampers the growth of semiconductor manufacturing due to the reduction in business profits. To address this, methodologies for process optimization were well studied [6]- [8]. Chopra et al developed a software tool This work is licensed under a Creative Commons Attribution 4.0 License.…”

Section: Introductionmentioning

confidence: 99%

“…Chopra et al developed a software tool This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see http://creativecommons.org/licenses/by/4.0/ for creating plasma etch recipes based on physical models and Bayesian inference [6] and applied it to prediction of the etching results [7]. They showed that the etching rate of SiO 2 with CF 4 /Ar gasses could be predicted (R 2 = 0.63).…”

Section: Introductionmentioning

confidence: 99%

Materials Informatics for Process and Material Co-Optimization

Tomita

Sato

Yoshii

et al. 2019

IEEE Trans. Semicond. Manufact.

View full text Add to dashboard Cite

In semiconductor manufacturing, fabrication processes and their materials should be properly co-optimized to achieve required processing results within reasonable development duration and acceptable cost. Unfortunately, it is a very time-consuming procedure, because the number of possible combinations of process/material candidates is very large. Here, we develop a methodology for co-optimization of processes and their materials. We successfully constructed a prediction model for dry-etching of high-k materials (R 2 = 0.65). Also, it was proven that considering both the materials and processes is needed for accurate prediction of etching rates. By trying only <0.00001% of all possible process/material candidates with this model and Bayesian optimization, we can find new combinations of gasses and their processes for more than 100 times higher etching rates than that with a traditional gas/process condition. Furthermore, we discussed that accurate prediction can be made by using a combination of the Bayesian optimization with LASSO and materials knowledge from related scientific papers. Future work will focus on validating the versatility of our methodology by applying it to other development items.

show abstract

A method to accelerate creation of plasma etch recipes using physics and Bayesian statistics

Cited by 4 publications

References 0 publications

Reactive Force-Field Molecular Dynamics Study of SiGe Thin Film Growth in Plasma Enhanced Chemical Vapor Deposition Processes

Reactive Force-Field Molecular Dynamics Study of SiGe Thin Film Growth in Plasma Enhanced Chemical Vapor Deposition Processes

Physics-Prior Bayesian Neural Networks in Semiconductor Processing

Materials Informatics for Process and Material Co-Optimization

Contact Info

Product

Resources

About