Every so often a nanotechnology discussion pops up on the SFN forums (one of several recurring themes), and invariably there is a comment or two in the direction of pop-sci topics like nanobots, and the implication that nanofabrication is a future technology. While the length scales have gotten smaller and manipulation techniques have certainly gotten better over time, nanofabrication has been around for a while.
The photos shown below the fold are from some grad school work I did at the Cornell NanoScale Science & Technology Facility in the early 90’s (back then there was no network of such labs, it was the National Nanofabrication Facility. Notice that they’ve been in existence for more than 30 years.) I was fabricating some transmission gratings for an atom-optics experiment to show atomic interference. The structures (grating wire and gap) were each about 125 nanometers across. The basic fabrication process was this:
A silicon wafer has silicon nitride deposited on it. You’d deposit a photoresist layer made of polymethylmethacrylate (PMMA) and expose it using UV light with a mask in place, marking the areas that you wanted to etch, and then develop the resist to remove those areas. A reactive ion etch removed the nitride layer, and then you dumped it in a hot bath of KOH to etch the silicon wafer.
Silicon in KOH doesn’t etch straight in — it goes much faster along the crystal axes, so the etch looked like an inverted pyramid. It also meant you could put in “break lines” in the mask that weren’t wide enough to allow the etch to go all the way through, but would be weaker and let you break the wafer up into smaller pieces after you were done. This shows the directional etch.
When that’s done you’re left with your silicon nitride layer on the front of the wafer, and the back part has a lot of little windows that are exposed, so there’s only the nitride in these regions. Now you can make the gratings themselves.
Again you deposit some PMMA, and you put a layer of gold over that, and put the wafers in a n electron-beam machine that will write the grating pattern and expose the PMMA. The gold gives a conductive path for the electrons so they don’t collect and distort the beam. Then you remove the exposed areas with another reactive ion etch. RIE is directional; you accelerate the ions in an electric field, so they tend to etch along that direction. The reactive ions react (duh) with the exposed surface and then float off as neutral compounds, exposing the next layer of atoms.
The narrow strips are the gratings, ~125 nm wide, with the wider bands left as support.
Another neat thing: a vernier written on the edges of the device to measure how well the fabrication went. The area that the electron-beam writer could cover was small, so you had to do the whole pattern in small steps. By writing a vernier on adjacent sections at the edge of each grating, you could see how well the whole pattern was “stitched” together. The vernier scale is 10 nm per division.