Modeling and random search optimization for the polysilicon CVD reactor

Bangwen Xi; Gang Xiong; Kirill A. Kozin; Chang He; Tariku Sinshaw Tamir; Yonggang Song; Xiong Liu; Zhen Shen

doi:10.1016/j.rico.2023.100320

Modeling and random search optimization for the polysilicon CVD reactor

Bangwen Xi
, Gang Xiong
, Kirill A. Kozin
, Chang He
, Tariku Sinshaw Tamir
, Yonggang Song
, Xiong Liu
, Zhen Shen^*

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

2 Scopus citations

Abstract

Polysilicon is mainly obtained by the Siemens process chemical vapor deposition (CVD) reaction, in which hydrogen gas and trichlorosilane (TCS) are fed into the CVD reactor to produce polysilicon. However, adjusting the feeding parameters in a step-by-step manner according to the actual deposition results is inefficient. In this paper, based on the existing mechanism model of the CVD reactor, we take use of historical data to construct a more accurate simulator. A random search algorithm is used to find good feeding parameters. The simulator can help increase polysilicon productivity while reduce the unit energy consumption.

Original language	English
Article number	100320
Journal	Results in Control and Optimization
Volume	13
DOIs	https://doi.org/10.1016/j.rico.2023.100320
State	Published - Dec 2023
Externally published	Yes

Bibliographical note

Publisher Copyright:
© 2023 The Author(s)

Keywords

CVD reactor
Nonlinear regression fitting
Random search
Simulator

ASJC Scopus subject areas

Control and Systems Engineering
Modeling and Simulation
Control and Optimization
Applied Mathematics
Artificial Intelligence

Access to Document

10.1016/j.rico.2023.100320

Cite this

@article{e854e888b84f4203b7ebb6870d19f06e,

title = "Modeling and random search optimization for the polysilicon CVD reactor",

abstract = "Polysilicon is mainly obtained by the Siemens process chemical vapor deposition (CVD) reaction, in which hydrogen gas and trichlorosilane (TCS) are fed into the CVD reactor to produce polysilicon. However, adjusting the feeding parameters in a step-by-step manner according to the actual deposition results is inefficient. In this paper, based on the existing mechanism model of the CVD reactor, we take use of historical data to construct a more accurate simulator. A random search algorithm is used to find good feeding parameters. The simulator can help increase polysilicon productivity while reduce the unit energy consumption.",

keywords = "CVD reactor, Nonlinear regression fitting, Random search, Simulator",

author = "Bangwen Xi and Gang Xiong and Kozin, \{Kirill A.\} and Chang He and Tamir, \{Tariku Sinshaw\} and Yonggang Song and Xiong Liu and Zhen Shen",

note = "Publisher Copyright: {\textcopyright} 2023 The Author(s)",

year = "2023",

month = dec,

doi = "10.1016/j.rico.2023.100320",

language = "English",

volume = "13",

journal = "Results in Control and Optimization",

issn = "2666-7207",

publisher = "Elsevier B.V.",

}

TY - JOUR

T1 - Modeling and random search optimization for the polysilicon CVD reactor

AU - Xi, Bangwen

AU - Xiong, Gang

AU - Kozin, Kirill A.

AU - He, Chang

AU - Tamir, Tariku Sinshaw

AU - Song, Yonggang

AU - Liu, Xiong

AU - Shen, Zhen

PY - 2023/12

Y1 - 2023/12

N2 - Polysilicon is mainly obtained by the Siemens process chemical vapor deposition (CVD) reaction, in which hydrogen gas and trichlorosilane (TCS) are fed into the CVD reactor to produce polysilicon. However, adjusting the feeding parameters in a step-by-step manner according to the actual deposition results is inefficient. In this paper, based on the existing mechanism model of the CVD reactor, we take use of historical data to construct a more accurate simulator. A random search algorithm is used to find good feeding parameters. The simulator can help increase polysilicon productivity while reduce the unit energy consumption.

AB - Polysilicon is mainly obtained by the Siemens process chemical vapor deposition (CVD) reaction, in which hydrogen gas and trichlorosilane (TCS) are fed into the CVD reactor to produce polysilicon. However, adjusting the feeding parameters in a step-by-step manner according to the actual deposition results is inefficient. In this paper, based on the existing mechanism model of the CVD reactor, we take use of historical data to construct a more accurate simulator. A random search algorithm is used to find good feeding parameters. The simulator can help increase polysilicon productivity while reduce the unit energy consumption.

KW - CVD reactor

KW - Nonlinear regression fitting

KW - Random search

KW - Simulator

UR - https://www.scopus.com/pages/publications/85175311649

U2 - 10.1016/j.rico.2023.100320

DO - 10.1016/j.rico.2023.100320

M3 - Article

AN - SCOPUS:85175311649

SN - 2666-7207

VL - 13

JO - Results in Control and Optimization

JF - Results in Control and Optimization

M1 - 100320

ER -

Modeling and random search optimization for the polysilicon CVD reactor

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this