High-speed anisotropic etching of quartz using SF 6 /C 4 F 8 /Ar/O 2 based chemistry in inductively coupled plasma reactive ion etching system

Abstract

Etching of quartz and glass for microsystems applications requires optimization of the etch process for high etch rates, high aspect ratios and low rms surface roughness of the etched features. Typically, minimum surface roughness of the etched feature accompanied with maximum etch rate and anisotropy are desired. In this article, we investigate the effect of different gas chemistries on the etch rate and rms surface roughness of the Pyrex(R) 7740 in an inductively coupled plasma reactive ion etching (ICP-RIE) system. The gases considered were SF6 and c-C4F8, with additives gases comprising of O2, Ar, and CH4. A standard factorial design of experiment (DOE) methodology was used for finding the effect of variation of process parameters on the etch rate and rms surface roughness. By use of 2000 W of ICP power, 475 W of substrate power, SF6 flow rate of 5 sccm, Ar flow rate of 50 sccm, substrate holder temperature of 20°C, and distance of substrate holder from ICP source to be 120 mm, we were able to obtain an etch rate of 0.536 μm/min and a rms surface roughness of ~1.97 nm. For an etch process optimized for high etch rate and minimum surface roughness using C4F8/SF6/O2/Ar gases, an etch rate of 0.55 μm/min and a rms surface roughness of ~25 nm was obtained for SF6 flow rate of 5 sccm, C4F8 flow rate of 5 sccm, O2 flow rate of 50 sccm, Ar flow rate of 50 sccm. Keeping all other process parameters the same, increasing the SF6 flow rate to 50 sccm resulted in an etch rate of 0.7 μm/min at an rms surface roughness of ~800 nm whereas increasing the C4F8 flow rate to 50 sccm resulted in an etch rate of 0.67 μm/min at an rms surface roughness of ~450 nm . Addition of CH4 did not contribute significantly to the etch rate while at the same time causing significant increase in the rms surface roughness. Regression or least square fit was used define an arbitrary etch rate number (Wetch) and rms surface roughness number (Wrms). These numbers were calculated by least square fit to the data comprising of ten correlated etch variables and enable quantization of etch parameters in terms of process parameters. The etch numbers defined in this work as function of process parameters present a very useful tool for the optimization, quantification and characterization of the dielectric etch processes developed in this work for MEMS fabrication and packaging applications.