Endpoint Detection Strategy in Bosch Process Using PCA and HMM

Bosch process is developed for advanced microstructure devices and etch endpoint detection (EPD) is demanded for 'notching' (as feature profile degradation) or reducing thickness of the underlying stop. One method commonly used to detect plasma process endpoint is utilizing optical emissio...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Jeon, Sung-Ik, Kim, Seung-Gyun, Han, Yi-Seul, Shin, Sung-Hwan, Han, Seung-Soo
Format: Tagungsbericht
Sprache:eng
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Bosch process is developed for advanced microstructure devices and etch endpoint detection (EPD) is demanded for 'notching' (as feature profile degradation) or reducing thickness of the underlying stop. One method commonly used to detect plasma process endpoint is utilizing optical emission spectrometry (OES) sensor. OES analyzes the light emitted from plasma source to draw inferences about the chemical and physical state of the plasma process. In this paper, an endpoint detection algorithm conjunction with Principal Component Analysis (PCA) and extended Hidden Markov Model (eHMM) using OES signal in Bosch process is proposed. PCA is used to reduce dimension of data without information loss and eHMM is applied to correctly detect endpoint. In 120 um TSVs, this work shows excellent performance.
ISSN:1938-5862
1938-6737
DOI:10.1149/1.3694433