Comparison of Cl2 and F-based dry etching for high aspect ratio Si microstructures etched with an inductively coupled plasma source

Abstract

The differences between Cl2 and F-based dry etching are compared in this article. Inductively coupled plasma sources have been used to generate plasmas using both Cl2 and SF6/C4F8 chemistries. Trenches etched using Cl2 suffered less aspect ratio dependent etching effects because the trenches can be etched at a much lower pressure than with F-based gases. A 1.4 μm wide, 65 μm deep trench can be obtained with an aspect ratio of 46 in 12 h. The average Si etch rate was 90 nm/min and the selectivity to electroplated Ni was 23. The sidewall was vertical and smooth and the trench openings were nearly the same width before and after etching. Adjacent trenches with 0.14 μm mask opening and 2 μm line width were etched using these two etching technologies. With Cl2 etching, a wider 0.25 μm trench opening, due to the mask erosion effect, with a depth of 5.6 μm was obtained in 50 min. However, the 0.33 μm undercut increased the trench opening to 0.8 μm for 10.7 μm deep trenches after the F-based etching for 55 min. The Si etch rate in a large open area using F-based etching was 1818 nm/min, which is much faster than 201 nm/min when Cl2 etching was used. However, the Si etch rate, 112 nm/min for Cl2 and 195 nm/min for F-based gases, was similar when the trench opening was decreased to submicrometer dimensions. This shows that the Cl2 etching provides better dimension and profile control with comparable Si etch rate to F-based etching when etching submicrometer trenches. The loading effect using Cl2 chemistry is less than with F-based etching. The Si etch rate was 1.74 μm/min for ∼100% Si exposed area and 3.68 μm/min when the exposed Si area was ∼0% in F-based etching. Scalloping, which is a periodic undercut near the top of the sidewalls, disappeared when using an electroplated Ni mask. The size and period of the scalloped features decreased as the Si exposed area and etch time increased.