{"title":"利用回归和神经网络模型确定晶体硅受控激光热裂解参数","authors":"Yu. V. Nikitjuk, A. N. Serdyukov","doi":"10.1134/s1063774523600679","DOIUrl":null,"url":null,"abstract":"<h3 data-test=\"abstract-sub-heading\">Abstract</h3><p>Regression and neural network models of controlled laser thermal cleavage of crystalline silicon have been built on the basis of the results of the finite element calculation obtained in a numerical experiment with the central composite design. The analysis of thermoelastic fields has been carried out for the cases of exposure to laser radiation with wavelengths of 1.06 and 0.808 μm in six anisotropy versions. The processing rate, silicon wafer thickness, and laser beam parameters have been used as variable factors. An artificial neural network architecture providing the best prognosis of thermoelastic stresses and temperatures in the laser processing zone has been established. The neural network and regression models have been compared. The neural network models are found to be moreaccurate as compared with the regression ones.</p>","PeriodicalId":527,"journal":{"name":"Crystallography Reports","volume":null,"pages":null},"PeriodicalIF":0.6000,"publicationDate":"2024-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Determination of the Parameters of Controlled Laser Thermal Cleavage of Crystalline Silicon Using Regression and Neural Network Models\",\"authors\":\"Yu. V. Nikitjuk, A. N. Serdyukov\",\"doi\":\"10.1134/s1063774523600679\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<h3 data-test=\\\"abstract-sub-heading\\\">Abstract</h3><p>Regression and neural network models of controlled laser thermal cleavage of crystalline silicon have been built on the basis of the results of the finite element calculation obtained in a numerical experiment with the central composite design. The analysis of thermoelastic fields has been carried out for the cases of exposure to laser radiation with wavelengths of 1.06 and 0.808 μm in six anisotropy versions. The processing rate, silicon wafer thickness, and laser beam parameters have been used as variable factors. An artificial neural network architecture providing the best prognosis of thermoelastic stresses and temperatures in the laser processing zone has been established. The neural network and regression models have been compared. The neural network models are found to be moreaccurate as compared with the regression ones.</p>\",\"PeriodicalId\":527,\"journal\":{\"name\":\"Crystallography Reports\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.6000,\"publicationDate\":\"2024-02-06\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Crystallography Reports\",\"FirstCategoryId\":\"88\",\"ListUrlMain\":\"https://doi.org/10.1134/s1063774523600679\",\"RegionNum\":4,\"RegionCategory\":\"材料科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"CRYSTALLOGRAPHY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Crystallography Reports","FirstCategoryId":"88","ListUrlMain":"https://doi.org/10.1134/s1063774523600679","RegionNum":4,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"CRYSTALLOGRAPHY","Score":null,"Total":0}
Determination of the Parameters of Controlled Laser Thermal Cleavage of Crystalline Silicon Using Regression and Neural Network Models
Abstract
Regression and neural network models of controlled laser thermal cleavage of crystalline silicon have been built on the basis of the results of the finite element calculation obtained in a numerical experiment with the central composite design. The analysis of thermoelastic fields has been carried out for the cases of exposure to laser radiation with wavelengths of 1.06 and 0.808 μm in six anisotropy versions. The processing rate, silicon wafer thickness, and laser beam parameters have been used as variable factors. An artificial neural network architecture providing the best prognosis of thermoelastic stresses and temperatures in the laser processing zone has been established. The neural network and regression models have been compared. The neural network models are found to be moreaccurate as compared with the regression ones.
期刊介绍:
Crystallography Reports is a journal that publishes original articles short communications, and reviews on various aspects of crystallography: diffraction and scattering of X-rays, electrons, and neutrons, determination of crystal structure of inorganic and organic substances, including proteins and other biological substances; UV-VIS and IR spectroscopy; growth, imperfect structure and physical properties of crystals; thin films, liquid crystals, nanomaterials, partially disordered systems, and the methods of studies.