Untitled Document

TECHNICAL INSIGHTS ALERT

INSIDE R&D ALERT, Vol. 30, No. 32 / AUGUST 8, 2001

ISSN 0300-757X

Posted with permission of the Editor, Inside R&D, published by Technical Insights, a business unit of Frost & Sullivan.

***************************************************

MICRODISCHARGES ETCH SILICON WITHOUT MASKS

Here's a neat etching trick that you could see in microfluidics devices about as quickly as someone decides to use it. Caltech chemists have come up with a way to use microdischarges to pattern silicon. The discharges are used as stencil masks to pattern bare silicon in CF4-argon chemistry. Using the process to form discharges in multiple hole and line shapes will permit direct pattern transfer in silicon. It gives you an alternative to ultrasonic milling and laser drilling in MEMS fabrication.

The big advantage of the technique, Caltech chemist Konstantinos Giapis tells me, is that it lets you use the same mask again and again to pattern multiple blanket wafers and devices. There is no polymer spinning, no lithography, no solvents, no exposures, no post-etching stripping. You just place the prefabricated mask on the wafer you want to pattern, strike the plasma, and you are done. There is very little power wasted for processing since the plasma is formed and concentrated only where it is needed.

The whole concept should be very attractive to MEMS (micro electromechanical systems) manufacturers. They will use it wherever circular cavities and channels are needed in microfluidics, microreactors, and the like. It will also be used to create plasma cavities for displays.

The attractions go on. Since the mask is flexible, you can pattern curved surfaces such as cylinders or even spheres. It does this--something no other technique can do at all--in a simple and elegant way.

Keep an eye on microdischarge research--a lot of labs are. They are interested in such characteristics of microdischarges as high-pressure operation and intense ultraviolet radiation. In neon discharges, excited ionic states more than 50 eV above the ground state are found. Excimer formation in the discharges suggests a relatively large concentration of high-energy electrons. These electrons could help produce reactive radicals at high pressures, rendering microdischarges suitable for materials processing. Using the plasmas as light sources has also been studied.

The Caltech chemists were interested in another aspect of the microdischarges. Their size makes them capable of maskless pattern transfer. Discharges can be formed in structures as small as 50 micrometers so wafers can be etched directly, eliminating the need for a lithographic step. The ability to form microdischarges in flexible structures will allow patterning of curved surfaces such as cylinders and spheres. Length scales over which plasmas are formed aren't great--they may not be small enough for microelectronics--but the patterning technique could help fabricate MEMS where dimensions are often on the order of 10-500 micrometers.

At Caltech, the stencil masks are two-layer structures made from 100-micrometer-thick copper foils spin coated with polyimide films. Holes and slots are cut mechanically to produce the pattern. The silicon substrate serves as the cathode, giving a much higher etch rate--more than 7 µm/min-- than using a metal electrode between the dielectric and silicon. It has been determined that power requirements to operate the discharge are less than 50 mW to etch a single hole to a depth of 100 micrometers.

So far, Giapis says, they have been able to get straight sidewalls only during early stages of the etching process. When they go deeper the sidewalls become curved, a drawback until they figure out how to tune the process. Because the plasma is formed inside the pattern, Debye shielding limits how small you can go. They figure 1 micrometer will be about it, keeping the process out of microelectronics. They also have more routine work to do: improve the robustness of the mask, improve pattern transfer fidelity, and extend the patterning to materials other than silicon.

The team has filed for patents and is ready to work with companies on development, particularly in microfluidics where the process should fit in immediately.

Details: Konstantinos Giapis, Associate Professor, Chemical Engineering Dept., Spalding Laboratories, MC 210-41, California Institute of Technology, 1200 E. California Blvd., Pasadena, CA 91125. Phone: 626-395-4180. Fax: 626-568-8743. E-mail: giapis@cheme.caltech.edu.

R010168/2080801