Building transistors today is done with lithography, which is a "top-down'' process that uses patterning to create the complex layers that make up the transistor structure. It’s a bit like exposing a negative on photographic paper to get the pattern you want and then using this pattern as a template to place each material—metal, insulator or semiconductor—in exactly the right location. This process has worked successfully since the 1950s. But as we get to ever-smaller dimensions, new approaches to building nano-scale devices will be required. At IBM’s T.J. Watson Research Center, we use a technique called self-assembly to grow and directly control nanostructures that could one day form parts of integrated circuits. Self-assembly looks at a ``bottom-up’’ approach that builds nanostructures in a way that is dictated by physics rather than by an imposed pattern. In some ways it’s like farming, in that you plant seeds to grow a crop, and then support the growth with the right conditions to get the result you want.
Exploring self-assembly doesn’t mean we are ready to throw away today’s approach; instead, we want to use top-down strategies that we have already learned over many years, and combine them with new tricks that use self-assembly. Think of it as water splashing onto a pane of glass. It spontaneously forms little hemispheres because of surface tension. But the positions and sizes of the droplets are random. Now imagine there is a scratch on the glass. Water droplets form on the scratch, because it is a good, low energy place for the water molecules to stick. We have now combined self-assembly (make a hemispherical droplet on this surface) with an imposed pattern (make a droplet on this part of the surface by using carefully placed scratches.) The result is that we can build more complicated patterns. Flexible, customized patterns—like this water example, but on the nano-scale—help us build integrated circuits. The more precisely we can direct this self-assembly, the more versatility we can achieve.