LITHOGRAPHIC PROCESSING

Perhaps one of the more unusual processing sequences in micromechanics is the selective epitaxial silicon deposition and electrochemical etch technique that Yokogawa Electric Corporation uses to produce its resonating force transducer. This process, which Yokogawa has worked on for several years, is responsible for the resonating pressure transducer that Yokogawa markets. Twenty-thousand units were sold in 1992. Specifications of 0.1 percent accuracy with 26 inch wafer devices with 100:1 turndowns at the same accuracy reflect the high quality of the device. Discussions at Yokogawa about the choice of this particular processing tool for this type of sensor indicate that major issues are known material constants and behavior. The device is formed from single crystal silicon. This concern is shared by Toyota Research, where materials research into films such as silicon carbide is justified with comments on the possible unsatisfactory long-term behavior of polysilicon when used as a sensor construction material. The reader is cautioned: this may or may not be so. There are no hard published data available. The point here is simply that reliability data for polysilicon and many other materials are badly needed to avoid unpleasant surprises in the future. In a sense, possible problems are avoided by restricting material choices to bulk silicon.

One of the major issues in micromechanical processing tools for actuators involves the ability to construct high aspect ratio structures. The best tool for this involves X-ray assisted or LIGA-like processing, which has now achieved structural heights of 1 cm with minimal run-out (Guckel et al. 1994). This processing tool depends on access to a bright X-ray source: an electron storage ring or synchrotron. Since machines of this type are typically large and always expensive, they are normally shared by many users. Access may therefore mean off-site processing. Reaction to X-ray assisted processing falls into two categories: acceptance via participation or avoidance via alternative approaches to high aspect ratio processing. Both situations occur in the United States and in Japan.

Interest in X-ray assisted processing in Japan is high. Nearly every facility that was included in the site visits expressed an interest. Two sites, SORTEC, a national synchrotron facility, and Mitsubishi Electric, with its own superconducting synchrotron, have or are constructing beam lines for deep X-ray lithography. Seiko Instruments has worked with the German company Microparts to produce and test microgears, and indicated continued interest during the site visit. However, there are also problems. Concern was expressed over the lead that Europe and the United States have in this area. The patent issue in particular was mentioned, and the cost of the process was criticized at Toyota. These remarks in turn are balanced by the questions that one senior Japanese researcher posed: Can Japan really afford not to participate?

Technical concerns that the Japanese must consider start with the source. Deep X-ray lithography requires photon energies above 3,000 eV. VLSI X-ray lithography, in which there are significant efforts in Japan, typically uses 200 eV photons. Since the new sources at Mitsubishi Electric and Sumitomo are aimed at VLSI lithography, their outputs are too soft for deep X-ray lithography. It is unclear but doubtful that these compact machines can be furnished with insertion devices to shift their outputs to shorter wavelengths and larger powers. SORTEC, a warm ring at 1 GeV, could be modified with a wigler to make it a very acceptable photon source for X-ray assisted processing. The Photon Factory in Tsukuba has the required characteristics. However, this machine belongs to the Ministry of Education, which makes it a doubtful candidate for a manufacturing effort (Clemens and Hill 1991).

The source issue has taken a somewhat unusual turn in Japan. Representatives of the Fujita Corporation, a large Japanese construction company, visited Madison about one year ago. They indicated that they were interested in exploring the possibility of constructing a high energy synchrotron in Chiba and operating it at a profit with multiple users. The project, which the Fujita Corporation called the Nano Hana Project, is still active and viewed with mild pessimism by some Japanese researchers.

Support for X-ray assisted processing requires thick photoresists and the processing technology that makes them useful. There are no indications that X-ray sensitive polymers for this type of work are of major concern in Japan. However, the results of high aspect ratio photoresist work at SORTEC, which produces strain induced mechanical failures in high resolution patterns with photoresist thicknesses of 5 microns, are applicable to thick photoresists with larger feature sizes and thicknesses, but similar aspect ratios. The SORTEC work is based on a German photoresist that is manufactured by Hoechst.

Electroplating, a major problem for LIGA-like processing, is being used at Professor H. Fujita's laboratory at the University of Tokyo with conventional thick photoresists to produce wobble motors and other actuators (Hirano, Furuhata, and Fujita 1993). There are other activities in Japan that deal with conventional electroplating issues, but there are no known activities that deal with the very special challenges that thick, high aspect ratio, LIGA-like photoresist technologies offer.

Processing procedures that attempt to duplicate results from X-ray assisted processing are normally called "high aspect ratio machining," or HARM. In the United States, emphasis in HARM centers on thick photosensitive polyimide technology as practiced by M. Allen of Georgia Tech. In Japan, Professor Masayoshi Esashi of Tohoku University and NTT researchers have reported on plasma etching of fully imidized polyimide to achieve vertical flanks with structural heights in the range of a few hundred microns (Murakami et al. 1993). Professor Esashi's group uses a nickel mask during plasma etching. The etch gas is oxygen. The reactor uses a cooled substrate support that can be used to produce substrate temperatures as low as 77K during etching. The plasma energy in the parallel plate reactor is reduced by coupling a static magnetic field from external permanent magnets into the discharge. Very high gas flow rates typify the etching procedure. The NTT work is similar to the Tohoku procedures, but utilizes a more automated reactor (Furuya et al. 1993). Both research groups report very nice HARM results that can be used to produce interesting actuators such as the distributed electromechanical actuator, or DEMA, which has been described by Professor Esashi (Yamaguchi et al. 1993).

Toyota research laboratory also reports that it has a HARM process. The details of the Toyota process have not been disclosed.


Published: September 1994; WTEC Hyper-Librarian