TMF Otimization in VGF Crystal Growth of GaAs by Artificial Neural Networks and Gaussian Process Models

0477799 - ÚI 2018 RIV DE eng C - Konferenční příspěvek (zahraniční konf.)
Dropka, N. - Holeňa, Martin - Frank-Rotsch, C.
TMF Otimization in VGF Crystal Growth of GaAs by Artificial Neural Networks and Gaussian Process Models.
Electrotechnologies for Material Processing. Hannover: Vulkan, 2017 - (Baake, E.; Nacke, B.), s. 203-208. ISBN 978-3-80273-095-5.
[International UIE-Congress on Electrotechnologies for Material Processing /18./. Hannover (DE), 06.06.2017-09.06.2017]
Grant CEP: GA ČR GA17-01251S
Institucionální podpora: RVO:67985807
Klíčová slova: crystal growth * travelling magnetic field * artificial neural networks * multilayer perceptron * Gaussian process
Obor OECD: Condensed matter physics (including formerly solid state physics, supercond.)

In Vertical Gradient Freeze growth of GaAs, the solid-liquid interface shape and subsequently the crystal quality can be improved by forced convection via travelling magnetic fields (TMFs). At present, general methodology to identify the relation and optimize magnetic and crystal growth parameters doesn’t exist. In this study, artificial neural networks (ANN) and Gaussian process models (GP) were used to assess the complex nonlinear relationships among the parameters and to optimize TMF for the interface flattening. 2D CFD simulations provided data sets for ANN and GP. The first encouraging results were presented and the strengths and weaknesses of both mathematical methods discussed.
Trvalý link: http://hdl.handle.net/11104/0274019