Single-Atom Transistor

Acid-F1ux writes: "Using a single cobalt atom as a switch, a research team at Cornell University has demonstrated a working transistor only 1.3 nanometers in length. Silicon transistors today are generally more than 100 nm long." We posted a slightly more general article about nanotransistors recently, too; this one concentrates on the Cornell researchers.

Problems to come

[brejc8]

When you get transistors the size of 1nm the time it takes for a clock signal to travel from one side of a 10mm chip to the other gets to millions of gate delays.

Current high speed processors have about 10 gate delays between clocks (and dropping).

The only way to keep devices increasing in speed along with the technology is to move to non clocked systems at least for long distance communication. This can be
done by GALS (Globally Asynchronous Locally Synchronous).

Also at that size transistors become rather unpredictable. Comes take a ps to switch sometimes three times more. You will no longer be able to say this pipeline stage
will be completed by x ps because sometimes it might not be. So margins become so large that most operations take only 50% of the time allocated.

My research is into Delay Insensitive circuit conversion. Basically you know when the pipeline stage is completed. This improves speed and copes with fluctuations in
voltage, heat or lazy transistors.

I tried it out [man.ac.uk] on a MIPS R3000 clone and got 30% speed improvement. I'm hoping for 50%.

"Single atom" is misleading

[Wayne Hoxsie]

After reading the article, it seems that this is a complex of silicon, gold, hydrocarbons, etc. all combined into a system that can manipulate and detect a state change in a single cobalt atom. While a major accomplishment (previously quite a bit more stuff was needed to manipulate and detect single atom state changes) I would hardly call it a single atom transistor.

Re:"Single atom" is misleading

[cmowire]

Single atom transistor is the buzzword. 1.3 nm transistor is the reality. They've got 5 iterations of Moore's law (7-8 years) to perfect it for production for it to be useful.

It's like One Transistor SRAM. The One Transistor is the buzzword, the reality is that it's DRAM that can be indivually accessed like SRAM. People ignore the support logic for the larger goal of making a cute futuristic buzzword out of it.

Re:"Single atom" is misleading

[Alsee]

They've got 5 iterations of Moore's law

Transistor density is based on the SQUARE of the size. I could be off, but it looks like Intel is currently running 130 nm. 1.3 nm allows 10,000 times as many transistors by area. That's 13.3 iterations of Moore's law - almost precisely 20 years.

By that time I'd say there's a good chance quantum computing will make classical computers look pathetic.

I love the future, some day I hope to live there :)

-

Re:"Single atom" is misleading

[cmowire]

Ahh.. Yes, you are absolutely right.

See, I'm still wondering what happens after the point where we make single-atom features. Either there's sub-atomic computing out there, or computers will start to get big and multiprocessor again.

Perhaps we'll be doing sub-atomic quantom computing with teleported laser beams transporting the signal faster than light.

Re:"Single atom" is misleading

[Alsee]

See, I'm still wondering what happens after the point where we make single-atom features.

That's where quantum computing breaks the barrier. All the old rules go out the window. They are almost infinitly parallel for free.

To use chess as an example, to look one move ahead there may be 20 possililites. A normal computer would have to do a 20 units of work. To look 12 moves ahead may be a hundred quadrillion possibilities. A normal computer would have to do a hundred quadrillion units of work. The problem explodes in complexity. No matter how fast the CPU becomes you can't look very many steps.

A quantum computer can see all the possibilities in parallel for free. To look 12 moves ahead only takes 12 units of work. It could look 100 moves ahead with just 100 units of work. The explosion of complexity vanishes.

The jump from normal computers to quantum computers and may be comparable to the jump from slide rules to computers. It's hard to even guess at what will be possible. Assuming we can figure out how to make useful quantum computers of course :) So far everything seems to say we can.

-

Re:"Single atom" is misleading

[cmowire]

See, I'm a little doubtful that we'll make a truly useful quantum computer. Just a gut feel, and it might be becaue I haven't tried to wrap my mind on how one would control a quantum computer.

Doubtlessley, even if we had a quantum computer with single-atom features, we'd find a way to exhaust its capabilities and would still want more power. ;)

irreverently heckling from the back row

[JUSTONEMORELATTE]

I know I shouldn't, but I simply can't help bringing up one of my favorite [strange-matter.com] Strange Matter [strange-matter.com] comics.

I bought one.

[bcrowell]

The article doesn't mention that these are actually available commercially now. I ordered one to try out, and it's sitting here on my desk right now.

Oh...

Wait, where did I put it?

Diamond Age & MS NanOS

[iamcadaver]

I'm re-reading Neil Stephenson's work. In his piece, the nano-mites were around 130 nm, and used rod-logic for computations.

This seems far more likely, because it's under the 'little-yellow-different-better' variety, all the major players will still be in place as it is today.

I SHUDDERED when I followed this line of thought and realized we are more likely to see an MS NanOS within the timeframe of Diamond Age, than the anarchistic/utopia. I can't say I can claim which is better.

Let me off this rock now, please.

Bizarro

[p3d0]

Next, McEuen said, his group planned to try a new molecule that can switch between "1" and "0" by changing shape, rather than by losing or gaining an electron. This physical mechanism, used in place of an electronic switch, could have advantages in economy and efficiency.

I don't know if I have ever seen that sentence before.

First use of this technology...

[ipsuid]

By sensitivity to "environmental" factors, I'm sure they are referring to light.

Requiring cryogenic cooling for operation, it will be a long time before we see these "transistors" in a CPU humming under our desk.

An application we are likely to see for this technology in the near term is supersensitive CCD imagers for use in devices such as the Hubble. Having the switching element consist of a single atom is likely to make each transistor sensitive to single photons of light, with the proper cooling and biasing.

From reading the article, the biggest trouble with these devices is they all share the same gate (the silicon substrate). So for real world use we are still no further ahead... in other words, still dependant on the feature size we can produce using lithography techniques on silicon.