There are always two stages in the development of any new technology: inventing it, and making it useful. The first computer was the size of a room, and slower than a first-grader counting on his fingers. The first telephone transmitted noises so garbled they could barely be understood even when you knew what was being said. We’ve had abilities in nanoscale writing, or “lithography,” for some time, but it was too expensive, too slow, too limited, to be of use in many contexts. Researchers often despaired at using the technique outside of very specific uses, or with huge grants, and many ended up using it simply to write their names as a proof of concept. Now, Boston University researchers say they’ve taken nanoscale lithography to the a new level of usefulness, creating a machine that can lay down previously impossible patterns at the atomic scale, and do it without crippling hassle or expense.
The technique involves heating a writing material in a vacuum so that it enters a gaseous state, and using a simple shutter system to carefully rain the atoms down onto a surface. It’s controlled by a two-plate system of polysilicon plates which slide relative to one another. One has a 20µm (micrometer) hole, large in comparison to the scales the researchers are interested in, and the other an assortment of pinholes as small as 50 nanometers across. When the hole in the top plate aligns with a desired hole on the bottom one, the atoms flow through and settle onto the surface below. Four ultra-sensitive springs can move the bottom plate in any direction the researchers desire.
This technique allows a range of applications that were previously impossible, most notably creating loops or other shapes with holes; to prove the machine’s abilities, the researchers made a series of eights, or infinity symbols. Prior solutions of this type rely on nanoscale stencils that cannot support hollow areas within the shape. The researchers are confident that their machine can reliably lay down single atoms, but experiments proving this capability are still forthcoming. The most important advance, though, is the cost, which is much lower than prior technologies. Team lead David J. Bishop said that after using a bottom plate for experiments, they can “go get another clean one — for a dollar or two.”
That’s a huge step forward from the current industry standard, photolithography, which involves using light to essentially score a surface, then treating the surface to either eat it away or build it up in only those places affected by the light. (See: Seeing double: TSMC adopts new lithography technique to push Moore’s law to 20nm.) This requires that the target be immersed in liquid, along with all sorts of intermediate steps to increase the resolution of the beam, and it’s becoming prohibitively expensive for today’s smallest process nodes (28, 22, 14nm). This new technique is additive only, but drastically cuts both the time and money needed to lay down complex, nano-scale structures.
Studies at the atomic scale are increasingly important, as our computer chips approach that scale and we learn that even fundamental forces seem to act differently down there. It’s not just about the oft-cited weirdness of the quantum world, as certain properties of magnetism and the increasingly important area of superconductivity also seem to work differently. Understanding exactly how these work differently will be a key part of continuing advancement in semiconductors.