The black silicon method. VIII. A study of the performance of etching silicon using SF6/O2-based chemistry with cryogenical wafer cooling and a high density ICP source

Henricus V. Jansen, Meint J. de Boer, H. Wensink, Ben Kloeck, Michael Curt Elwenspoek

    Research output: Contribution to journalArticleAcademicpeer-review

    75 Citations (Scopus)


    This paper presents a study of the performance of current trends in high speed, highly controllable anisotropic plasma etching of silicon for its use in micro- and nano-engineering. Optimisation rules for tuning the equipment are formulated to enable maximum results with respect to etch rate, etch selectivity, and profile control (e.g. anisotropy). The optimisation process uses the black silicon method to allow for an easy to use design-of-experiment method. After optimisation, etch rates of silicon up to 15 μm/min were obtained at a relatively high pressure (4 Pa = 30 m Torr and ICP power (2 kW), while keeping the anisotropy high. At the same time, the erosion rates of thermal silicon dioxide and ordinary photoresist (Shipley S1805) were around 7 nm/min (i.e. selectivities up to 2000). Increasing the pressure to 20 Pa, the selectivity increased to 10,000, although bottling became more pronounced. The high etch rate and high selectivity are especially important in the case of micro-engineering, where wafer through etching with the help of plasmas become a standard in the near future. In the case of nano-engineering, however, profile control is the main concern. To prevent undercut in such cases, in particular, bottling due to a broad ion angular distribution, the pressure should be sufficiently low. The best results were found at pressures below 0.2 Pa = 1.5 mTorr and at a low ICP power of 350 W to prevent a too strong mask erosion caused by the low pressure. The silicon etch rate decreased to 1 μm/min and the erosion rate of the oxide and resist were both approximately 20 nm/min, giving a selectivity of 50. Reproducibility (wafer-to-wafer) and uniformity were also output factors of prime concern. Surprisingly enough, for two different equipment manufacturers the results were almost identical when using the same parameter setting. This indicates that the SF6/O2-based chemistry was optimised rather than the equipment itself.
    Original languageUndefined
    Pages (from-to)769-777
    Number of pages9
    JournalMicroelectronics journal
    Issue number9
    Publication statusPublished - 26 Feb 2001


    • IR-36018
    • METIS-200199
    • EWI-12794

    Cite this