Plasma processing of materials is used extensively in the semiconductor industry. In plasma etching, patterns are defined in silicon wafers by exposing them to an ionized gas through a lithographically defined stencil, or mask. One of the advantages of plasma etching over liquid chemical etching is the directional etching that results from bombardment of the surface by positive ions raining down from the plasma. Because the ion bombardment is normal to the substrate surface, undercutting, or etching of material under the mask, can be avoided. This enables the fabrication of ever-tinier transistors critical to the continuing advancement in integrated circuit performance.
In typical manufacturing plasma processes today, the energy of ions reaching the substrate surface is coarsely controlled bv varying the amplitude of an RF sinusoidal bias voltage applied to the substrate electrode, but the resulting ion energy distribution (IED) at the substrate is generally broad. The energy provided to the substrate surface upon ion impact can enhance chemical reactions via several mechanisms, with significant implications for etched feature profiles and etch selectivity. The proposed activity employs a newly developed technique for significantly reducing the width of the IED at the substrate, with the potential for significantly improving these aspects of plasma processes. Furthermore, because of past difficulties in controlling ion energy at the substrate under realistic process conditions, this method opens up the possibility of examining the role of ion bombardment energy in etching processes for different materials and process gas systems with minimal impact on other process parameters.
This technique for ion energy control replaces the conventional sinusoidal substrate bias voltage waveform with a waveform that produces, on the substrate surface, a constant potential punctuated by periodic voltage spikes. As a result, most ions see a constant voltage drop across the "sheath" that arises between the substrate and plasma, and therefore arrive at the substrate with the same energy. Potential measurements have confirmed the feasibility of this approach. In addition, preliminary etch rate measurements of blanket films show dramatic improvement in etch selectivity for SiO2/Si etching in a fluorocarbon plasma.
This method will be applied to several etching problems, as well as to basic understanding of the role of ion energy in plasma processing. The first step will be to directly measure the IED at the substrate to both confirm the conjecture that this method produces a narrow ion energy distribution, and to verify that average ion energy can be effectively monitored through external voltage measurements. In addition, both etch selectivity and etch feature profile control on patterned silicon wafers by precise tailoring of the IED using this method will be examined. A particular process of interest is etching of silicon dioxide, which continues to pose challenges for plasma processing. Finally, by scanning the ion energy while measuring etch rates, improved understanding of the mechanisms by which ion bombardment affects surface erosion during etching processes is expected. ***