X-Ray Nanolithography

Project Staff: Dr. David C. Carter, James M. Daley, Michael H. Lim, Euclid E. Moon, and Professor Henry I. Smith

Sponsors : Defense Advanced Research Projects Agency/ U.S. Navy — Naval Air Systems Command, N00019-98-K-0110

For several years, we have been developing the tools and methods of x-ray nanolithography. We have explored the theoretical and practical limitations, and endeavored to make its various components (e.g. mask-making, resists, electroplating, sources, alignment, etc.) reliable and "user-friendly." Because of the critical importance of x-ray mask technology, we discuss this in a separate section.

Our sources for x-ray nanolithography are simple, low-cost electron-bombardment targets. We utilize the L line of copper at l = 1.32 nm. The sources are separated by a 1.5 m m-thick SiN_x vacuum window from a helium-filled exposure chamber. In the future, we hope to replace the Cu_L sources with a higher-flux plasma-focus source.

In earlier research, we showed that for wavelengths longer than 0.8 nm, the important limit on resolution is diffraction in the gap between mask and substrate. With a Cu_L source, a 50 nm feature must be exposed at a mask-to-substrate gap of less than about 4 mm in order to maintain good process latitude. A 25 nm feature would require a gap of 1 mm. For very small features, we eliminate the gap and use contact between the substrate and the flexible membrane mask. This technique has enabled us to replicate features as small as 25 nm in a practical, reproducible way. Figure 3 shows scanning electron micrographs of device patterns with feature sizes less than 40 nm. The x-ray mask is shown on top and the lifted-off pattern is on the bottom.

To create the x-ray masks, the pattern is first written by electron-beam lithography onto an x-ray "mother" mask, using either our in-house e-beam system or a collaboration with the Naval Research Laboratory in Washington, DC. The e-beam written pattern is developed, and gold is electroplated into the resist mold. A negative replica, or "daughter" mask is created by exposing with the mother mask using soft-contact x-ray nanolithography. Finally, the daughter mask is exposed onto the device substrate.

Recent work has focused on investigating process latitude at these extremely fine feature sizes. Figure 4 shows how developed linewidth changes for up to 50% overdevelopment (i.e. developing for 50% longer than it takes for the feature to clear) as a function of linewidth. As can be seen from the plot, the measured feature on the substrate remains within a +/- 10% process window (within the accuracy of the measurement) for isolated features as small as 30 nm and for dense features (greater than 1:3 line:space ratio) as small as 45 nm. This data indicates that soft-contact x-ray lithography is extremely robust and offers very wide process latitude.

Figure 3. Scanning electron micrographs of device patterns with feature sizes less than 40 nm achieved by x-ray nanolithography followed by liftoff. The x-ray mask is shown at left and the lifted-off pattern is at right.

Figure 4. Plot of linewidth varation from nominal (i.e. developed for the time required to clear features) for up to 50% overdevelopment. Isolated features stay within a +/-10% process window for features as small as 30 nm. Dense (line:space ratio of 1:3 or greater) features remain in the process window for features as small as 45 nm.