Holography is a technique used to generate images that look like 3D objects. These images (also called holograms) give a perception of depth through interaction of waves of light beams. Over the years people have used holography in numerous fields, from making art and museum exhibits to environments of augmented reality (AR). In our project we will use holography to do lithography, the process of making semiconductor devices used as sensors, computing chips and microbatteries. Our attempt will be the first of its kind in the world.
Traditionally, to generate holograms, we first record interference patterns using lasers from an object on a recording device; then we reconstruct its image by interfering the pattern on the device with another beam (called the reconstruction beam). Computer generated holography removes the recording step by back-calculating interference patterns on the recording device from our required final hologram. We will use this method to make holographic patterns for lithography. The recording devices we will use are called spatial light modulators (SLMs); these are commonly used in simpler holographic devices, such as cinema projectors.
For lithography we first coat smooth wafers of semiconductors (usually silicon) with a layer of a compound (called a photoresist); then we expose it to beams of high energy light in certain regions. These compounds react with this light to form patterns across the layer. This step is repeated multiple times for different layers, thus making complex sensors and chips. These patterns are usually exposed on wafers by focussing broad beams over patterned masks. For state-of-the-art semiconductors, these masks can produce features as small as 13.5 nm and each of these mask patterns can cost up to $250,000 each! Hence, this process of optical lithography with masks is viable only to make chips that sell in large commercial quantities.
To make devices in fewer numbers, usually for research or custom industrial applications, we often use maskless lithography techniques. These techniques are very slow because they usually scan the wafer surface with individual probes to produce patterns. Additionally, these machines can also be very expensive. For instance, an electron beam lithography (EBL) can be almost 10 million times slower than optical lithography with masks and can cost over $1 million for each machine. More affordable techniques, such as direct laser write (DLW) also suffer from the problem of being extremely slow for widespread use. To improve on this, we will make a machine to do holographic multibeam interference lithography (HMBIL).
Interference patterns from multiple beams have already been used for niche lithography applications in the recent past. However, this type of lithography is restricted to repeating patterns. Instead, by using SLMs to produce interference of holograms, we will extend this functionality to any custom non-repeating pattern. HMBIL promises to be faster and more affordable than EBL or DLW by several orders of magnitude. Additionally, it also promises a larger depth of focus and fewer diffraction effects in comparison to optical lithography with masks. In the scope of this project, we will use near-UV light to make features of up to 100 nm in size. Once this technique gets working, it should be possible to make smaller features by using commercial UV light sources of smaller wavelengths.
NanoDTC PhD Student, c2021