Advancing the automation and chemistry of scanning tunneling microscopes: autonomous atom assembler

This is movement towards the transition stage of molecular nanotechnology. If the goals described here are achieved (automated three dimensional pick and place of atoms) and that is then sped up and made massively parallel that level of technology would be the level of transition stage molecular manufacturing. The article is from late 2003. The researcher Bob Cellota is now the director of the NIST center for nanoscale science and technology

Researchers are advancing the automation and chemistry of scanning tunneling microscopes

Current work and progress:
-Moving the atom tip even closer so that the atom on the tip and the atom on the surface can form a temporary, sort of tunable, temporary chemical bond and then using the forces of that chemical bond, we move atoms around on the surface. Now that’s been done before, by hand to make very beautiful structures on the surface. And what we’re in the middle of is teaching a computer how to do this. So this can autonomously assemble whatever structures we want starting from a random collection of atoms. We just feed it a drawing of what we’d like and it figures out how to move the atoms and in what order and at what speeds and in which ways to create the structures we want. And we call this an autonomous atom assembler. Celotta’s autonomous atom assembler does it’s work in a high vacuum, at low temperatures.
-It’s just beginning to work and we’re making simple structures on the surface of circles, triangles and squares and once we learn the kind of rules it needs to have in order to make these structures — let me give you an example of a rule — if you’re going to move an atom past another atom, you don’t want to get too close because there’ll be a chemical attraction between the two and they’ll form a dimer — a two atom pair. So you have to give it a certain wide berth when you put them together and we also have to make the quickest routes possible and we have to worry about other defects in the surface. So we’re teaching it a whole bunch of rules to follow and then trying to see how well we can construct a surface where it just does all the thinking. And of course, when you try to teach a computer to do something, you have to know all the rules. So it forces it to understand all the kinds of interaction that can happen on the surface and be able to convey that to the computer so it can work correctly.

Looking at the next steps to advancing scanning tunneling microscopes.
-Have different kinds of atoms and keep them straight, so that you can assemble things out of multiple species of atoms on the surface.
-Do things in three dimensions to bring atoms up so you assemble a small cluster of atoms, say a square that was all filled in on one layer and then build something on top of that square.
-Another major step would be to actually understand the interaction between the atom on the tip and the atom on the surface — that is actually understand the forces … and it’s really chemistry on a very fine level
-Reliably figure out how to pick an atom up on the surface and put it on the end of the tip and then also reliably put it back exactly onto the surface where you want it. It’s kind of like a crane on a construction site. So we’re actually now moving the atoms around more like bulldozers where we’re pushing them around on the surface or dragging them around on the surface might be more correct. A future enhancement might be to pick each one up selectively and put it down.