to save people poking around (because the linked article is quite appalling vague) - what you might call the "active part" of the gate is a single atom. that is surrounded by input, output, and gate connections and insulation that are all multiple-atoms in size. so the single, central phosphorous atom controls the flow from source to drain, depending on the gate voltage.
But as has already been mentioned, the biggest practical hurdle is temperature. This single atom device is measured at milliKelvin temperatures, and fundamentally could not operate at room temperature. It still has important consequences, but building computer chips is not an immediate one.
Making it work at room temperature is probably a greater long term issue. As the temperature goes up the system has a higher probability of accessing states that will make it not function properly (i.e. your logic gates become probabilistic.)
That's great and all that but you can't scale up production at that scale (imagine even building an opamp with it) and it doesn't mention anywhere how long the gate actually lasted. Thermal effects at that scale would probably destroy it pretty quickly.
I think this is more about proof of concept than a presentation of its readiness for production at a large scale or at any scale. Give it 5 more years and maybe we'll be closer to seeing it more frequently in the technologies we use.
"Does this say something fundamental about transistors or something fundamental about atoms?"
Neither. It just says that someone has gotten better at using a hammer to guide an egg, into a frying pan, without breaking it. And they performed that trick while wearing thick gloves.
"The central achievement of the latest work is to use the STM–hydrogen-resist lithography approach to position a single phosphorus atom between source and drain contacts and two (more distant) gate electrodes."
Also, I take strong issue with the term "single-atom transistor". I would expect to hear that from a crank working out of his secret basement lab. You cannot build an electronic device out of a single atom, and obviously, they used many.
Lastly: "Single-atom transistors represent the ultimate limit in solid-state device miniaturization." That's patently false. Thousands of atoms were used here, so there is likely room for improvement.
http://nature.com/nnano//journal/vaop/ncurrent/pdf/nnano.201... (Transistors arrive at the atomic limit)(A single-atom transistor has been made by positioning a phosphorus atom between metallic electrodes, also made
of phosphorus, on a silicon surface)(2012-FEB-19)
It's an issue of definitions. Of course you cannot build an electronic device using only a single atom, because you need leads, interconnects, power supplies, measurement equipment etc.
So when you talk about a single atom transistor, what's meant is that the critical, behaviour determining region of the device is a single atom. If you took that single atom away, you would not have a transistor any more.
In much the same way we talk about 32nm transistors in computer chips for example - it's understood that this number refers to the gate length, not your whole processor.
"It's an issue of definitions. If you took that single atom away, you would not have a transistor any more."
No, you would still have a transistor, albeit a non-functioning one, just like you would still have a car if you removed the engine. I appreciate your effort to explain, but your explanation is worded for people similar to those who work in your lab. Definitions should be correct, especially when they are put forth by the NY Times, a news source for the general public. People have already had their expectations raised over "jetpacks for everyone any day now" stories, so it's rather irresponsible to define something new with a name which is obviously false, and which tends to overrepresent, by far, the achievement that the name defines. I actually work with nanotech, but the average reader of the NY Times has little-to-no knowledge of "leads", "interconnects", "power supplies", "measurement equipment", "critical, behaviour-determining regions", "gate length", or even: "transistor".
http://www.nature.com/nnano//journal/vaop/ncurrent/pdf/nnano...
http://www.nature.com/nnano/journal/vaop/ncurrent/full/nnano...