Single-Atom Transistor

zarsky99 writes: "EETimes is reporting that Japanese researchers are close to creating the world's first single-atom transistor. This could be a boon to power problems and Moore's Law. The article is here, and please enjoy. Now if they could only get a single girl to date me." OK, you take the transistor, I'll take the girl ;) J adds: For those of you graphing Moore's Wacky Law: November1999, 50nm; November1999, 18nm; October2000, 1nm; December2000, 30nm; five days ago, 30nm. We don't make the semiconductors, we just report 'em.

Comments

Single electron, not single atom

[CrazyMadPsychoBandit]

The transistor itself contains several atoms, although it's still MUCH smaller than today's devices. A one-electron difference in charge on the transistor's gate is all that's needed to switch it on/off.

What atom would that be?

[Arandir]

Okay, it's been quite a long time since I took high school chemistry, so maybe something radical changed in the field since then. But I distinctly recall only about 200 to 250 possible elements. Which isotype of which element does this single atom belong to?

Oh hell! This can't possibly be right. Not even Japan can alter the laws of physics. Let me read the article to see what the truth of the matter is... ...yes, Slashdot goofed again (does this surprise anyone). They read the eeTimes' equally innacurate headline and never bothered to read the article. Quoting the real information, we find that:

The transistor Aono is developing makes a switch circuit consisting of ... a 10-atom-diameter cluster of 500 silver atoms that acts as a capacitor...

and

"We can make an atomic switch in a cluster of silver atoms"

Very amazing. But it's not a "Single-Atom Transistor" like Slashdot says. The key component in the transistor may be a single atom, but the transistor itself is not.

Re:Consequences of Moore's Law

[magi]

Speaking of Moore's Law, does this mean that in 18 months we'll have scientists promising half-atom transistors?

Naah, they start with Lawrencium 262, then in a few months they kick out a few neutrons and protons to make Fermium 256 (just to make these calculations easier). Then, with 18 month steps: Xenon 128, Zirconium 64, Sulphur 32, Oxygen 16, Beryllium 8 (9 is more stable though), Helium 4, Hydrogen 2, and finally Hydrogen 1 (ehm...a single proton).

Thus, it takes 12 years before they have to go to subatomics!

pico-wires

[Spy Hunter]

While the single-atom transistor thing is cool, it seems to me that the interesting part of this discovery/invention is the super-tiny wires they have to connect them. Now THAT's cool, and a big problem down there at the nano(pico?)-scale level.

What I want to know is, how will they connect this with normal electronics? They'd probably need 5 or 6 buffers in between to step down the current so as not to fry the tiny wires. Also, wouldn't a chip made with this technology be super-sensitive to interference? If a random cosmic ray hit it, it would probably be fried.

Hope they can solve all the problems. This sounds like really cool technology.

[me@localhost]$ prolog
| ?- god.
! Existence error in god/0

single atom AND / OR gates

[josepha48]

In IEEE there was an article a year ago on single atom AND and OR gates. Basically they were dealing with the spin of the electrons. This would be equivelant to two transistors(maybe). At the time I am not sure if it was theoretical or not, but it does not suprise me. The problem is dealing with an interface to that level of technology. How do you connect a wire to this to actually get the output somewhere. This is the problem they had a year ago. If you have a micro chip that has atom sized transistors, how do you connect the wires to these chips to get the power to. Also with a conventional power supply how do you step down the power to that level without the massive overhead of creating a major conversion mechanism. 220/110 W to ??

Anyway it is a step, now we need the leap. What would really be neat is to see smaller PC parts today. A 2" network card and a 2" modem connecting to an 8"x6" MB would be sweet. Then my pc would be cut down to about 1/3 its current size. A cdrom drive that only 1/2 the cd went it rather than a drawer. Hmmm

I don't want a lot, I just want it all!
Flame away, I have a hose!

I Sure Hope...

[Robotech_Master]

...that after they make these things, they're careful to remember where they put them.
--

What do women and transistors have in common

[Anonymous Coward]

- They are often biased
- They often amplify things out of proportion
- They can switch their minds in an instant
- They are often non-linear
- They consume energy

Consequences of Moore's Law

[DeadMeat (TM)]

Speaking of Moore's Law, does this mean that in 18 months we'll have scientists promising half-atom transistors? Now that'll be really interesting . . .

(Disclaimer: don't bother flame^H^H^H^H^Hcorrecting me about Moore's Law not really being a law. I know that; I'm just joking.)

It won't do any good

[Dr. Prakash Kothari]

I feel that too much emphasis is being placed on processor speed today. An important processor is nice, but we need to take the emphasis off of multi-gigahertz processing, and take a step back to develop other components to the same technological advancement. We're rapidly approching a point where a two, five, or even ten gigahertz processor isn't going to improve performance, because it's no longer the limiting factor. I'd love to have my own processor made from single atom transistors, but first, we need to work on improving bus speeds, memory latency, and hard disk access time. Without all of these factors working in unison at high speeds, the only thing you're going to get from an ultrafast processor is a lighter wallet.

Just look at some of the benchmarks on Tom'sHardware. The majority of them are all dead even after processor speeds hit 8 or 9 hundred Mhz.

Some of the newer Athlon processors have 12x multipliers. That means the processor is working 12 times as fast as the rest of the system. This is wasteful, and you end up with a lot of dead processor cycles because the RAM/system bus can't provide enough data for processing. Manufacturer's need to stop throwing money at superfast processor development, and work on improving system bus speeds, and latency/throughput of RAM.