BiLSTMBased Dynamic Prediction Model for Pulling Speed of Czochralski SingleCrystal Furnace

Abstract

With the rapid development of microelectronics science and technology, the quality of ICgrade silicon single crystal directly affects the yield and stability of the performance of semiconductor device production. As the main equipment for the preparation of such materials, the monitoring and maintenance of the working condition of the single crystal furnace are crucial. Bidirectional long shortterm memory (BiLSTM) is an innovative neural network paradigm that is used to predict future occurrences by learning the bidirectional longterm dependencies of timesteps and serial data. This paper built a BiLSTM based model that can dynamically predict the pulling speed of a Czochralski (Cz) singlecrystal furnace by modeling the time series of operational parameters. The BiLSTM model is validated using real data from a silicon singlecrystal factory. It is proven that the model achieved higher accuracy than LSTM, ANN, SVR, and XGBOOST. The experimental results verify the validity of modeling the pulling speed of singlecrystal furnace devices through the BiLSTM model by using the time series of multidimensional parameters. Therefore, the BiLSTM model can serve as a reference for modeling the parameters of such devices.