Device pattern impact on optical endpoint detection by interferometry for STI CMP

Inhalt:
In this study, the focus is made on interferometry endpoint detection for Shallow Trench Isolation (STI) Chemical Mechanical Planarization (CMP) process done on the first platen of the tool for a three platen-polishing configuration. The interferometry principle is based on light interferences produced by thin transparent and semitransparent layers. Thus the removed thickness can be correlated to the intensity measured over time. The optical signal received during the SiO2 removal is affected by patterned wafers and pattern type. For STI stack, parameters like Gap Fill SiO2 thickness, trench depth, pattern density and surface topology, could be involved in laser light diffraction and scattering. In some cases, these unwanted effects are stronger than the periodical signal of the removed SiO2 layer thus, preventing from endpoint algorithm usage. A correlation analysis has been performed based on experimental data between memory product layouts parameters and the response of raw signal detection. It confirms that the more sizable the Electrically Erasable Programmable Read-Only Memory (E²PROM) coverage is, the weaker and noisier the signal is. However, our experiment also shows that regardless of the SiO2 thickness, a deeper trench will produce a more exploitable signal. In conclusion, perspectives for future experimentations are explored, in order to set up CMP optical endpoint detection by interferometry for a broader type of memory device patterns. Keywords: Planarization, Chemical-mechanical Polishing, optical endpoint, interferometry, STI CMP, device pattern.