A novel technique for projection-type electron-beam lithography

One of the challenges of nanotechnology is to fabricate nanometer-scale features of uniform size and shape, and with acceptable speed. Such materials will have a wide range of applications, for example, in developing the future generation of electronic, optical, magnetic, and biological devices, due to their novel and significantly improved physical, chemical, and biological properties. Most top-down approaches to making nanoscale features attempted to date, such as electron-beam lithography and probe tip-based lithography, show good controllability but unsatisfactory throughput. In contrast, bottom-up methods, such as gas-phase and liquid-phase synthesis of quantum dots and quantum wires, pose challenges in controlling size and shape. Here, we propose a novel technique for producing various nanostructures based on crystalline lattice images obtained from high-resolution transmission electron microscopy (HRTEM).

Figure 1. HRTEM image of a silicon (Si) [110] crystallographic zone axis.

Since the invention of TEM by Max Knoll and Ernst Ruska in 1931,¹ the resolution of the microscope has undergone drastic improvement, currently extending down to tenths of nanometers. In the 1950s, James Menter² managed to obtain lattice-fringe images using phase-contrast interference signals between the transmitted beam and diffracted beam. As a result of these advances, imaging the crystalline lattice of many different materials to investigate their atomic structures has become common practice. For instance, Figure 1 shows a high-resolution phase-contrast image of silicon observed at the [110] crystallographic zone axis. The dots represent double columns of silicon atoms separated by 0.33nm. Such atomic-scale images, achieved using HRTEM, may well constitute the ultimate in what humans can currently observe in terms of size. The next question is whether these images can be used as a template to actually make nanoscale dot features. This is precisely the problem we set out to solve.

In HRTEM, the image signal is first formed at the image plane of the objective lens and then magnified further using a series of intermediate and projection lenses to reveal the small-size features. Typically, observing lattice-fringe images requires magnifying up to several hundred thousand times. In our case, however, we are interested using the image to make nanoscale patterns. Consequently, the magnification needed is somewhere between several ten to several hundred times, obtainable in one of the conjugated image planes situated below the objective lens.

Figure 2. Atomic image projection electron-beam lithography hardware. (a) Design diagram. (b) Ray diagram.

To verify our concept experimentally, we modified a field-emission TEM with a 200kV accelerating voltage.³ Figure 2 shows the hardware design diagram with the corresponding ray diagram. The sample stage is installed at the image plane of the objective lens, called a lithography plane, and the patterning lens is inserted between the objective lens and sample stage to control the magnification of the image signal at the sample. The patterning magnification can be varied from 50 to 300 times by adjusting the current of the patterning lens.

Figure 3. (a) Dot and line array patterns from a silicon mask and (b) various patterns obtained from a β-Si₃N₄mask.

Figure 3 shows some of the nanoscale patterns fabricated on a silicon substrate using single-crystalline silicon and polycrystalline beta-silicon nitride (β-Si₃N₄) as mask materials. The mask sample was made using a conventional TEM sampling method of dimpling and ion-beam milling. Hydrogen silsesquioxane is used as an electron-beam resist. After developing the resist, we etched the silicon substrate using hydrogen chloride reactive-ion etching. Note that the high-resolution lattice image we observed in TEM is a 2D projection image of atom arrangements in three dimensions. Accordingly, one can obtain various patterns simply by changing the observation zone axis of a given material. For instance, Figure 3(a) shows the nanoscale dot (observed at exactly the [110] zone axis) and line features (observed by slightly tilting the sample from the [110] zone axis) obtained in a silicon sample. Note, too, that there are an enormous variety of crystalline materials in nature that produce many different kinds of high-resolution images. To illustrate this phenomenon, we show just a few of the patterns obtained from crystalline β-Si₃N₄ used as a mask material.

We have demonstrated that our method does not produce only nanoscale dots and lines. Much more complicated structures can be fabricated using the crystalline lattice images observed with HRTEM. The question now is what kinds of properties these nanostructure patterns will reveal. Our group is actively working on identifying novel functionalities from the patterns.