Slashdot Mirror
Physicists Create a Working Transistor From a Single Atom
stupendou writes "Australian and American physicists have built a working transistor from a single phosphorus atom embedded in a silicon crystal. The group of physicists, based at the University of New South Wales and Purdue University, said they had laid the groundwork for a futuristic quantum computer that might one day function in a nanoscale world and would be orders of magnitude smaller and quicker than today's silicon-based machines."
...it will slip between the fibers on your pocket, fall on the floor, get vacuumed up and get accidentally thrown away.
The future is here.
There's no -1 for "I don't get it."
With transistors that small, how would you harden a microchip against radiation? Would the extra redundancy not make it worthwhile. That is to say, is there an optimal compromise between transistor size and resources consumed through redundancy allocation?
Life is not for the lazy.
We were making single atom transistors ten years ago, but it was hit or miss whether the atom would end up in the right place.
Today, we can place the atom with high precision, in silicon, so that the devices can be made reliably.
Ten years from now, who's to say we won't be able to mass produce them?
Pretty much - that's how transistors work. The phosphorous has a extra electron (compared to the silicon) and the combination forms an extrinsic semiconductor, which you then use to make junctions and transistors and diodes etc.
Just having the phosphorus atom isolated doesn't do much for you, so I think the article is referring to "silicon based computers of today" without really thinking about it properly - you still need to dope it to make it useful for making computer chips, despite it already being an intrinsic semiconductor.
They're stopping at 22/7 sales.
I told them they were being irrational, but there's no stopping them.
I hate being a nay-sayer, but the NYT article is making quite a spectacle about this whole thing. What the group has truly done is demonstrate a novel method for placing a single phosphorus atom within silicon and proceeded to measure the semiconducting properties of the resultant device with quite good precision. Because the doping is the result of a single atom, they can resolve more than just "on" and "off", and in fact can read three states from it, so it gets its quantum computing title.
As a materials scientist, I'm worried that they don't show any long-term data and all their results appear (from my not-so-thorough reading of the originating Nature Nanotechnology report) to be based on a single device. How repeatable is this result and how consistent are the signals across multiple devices? How far will the phosphorus atom diffuse over the lifetime of the device? Or even over the first few hours of its operation at room temperature? How closely can these devices be placed to each other on the silicon chip without getting cross-interference or depriving the dopant of its discrete quantum states? The dopants in a normal device aren't too terribly close to each other. And finally, how big must the surrounding structure be?
Don't get me wrong, this is excellent science and well deserving of its publication in such a prestigious journal, but the spectacle that the NYT is creating around this and the dreams of such a tiny device is a bit premature.
That would have been well over 30 years ago, since 1500nm was reached in 1982 and 800nm in 1989.
The process size is virtually a straight line on a log10 scale. Going on the last 40 years we'll be at 1nm by 2030. Its an order of magnitude every 10 - 15 years
As someone who's been routinely getting "-1, Overrated" on many of my posts for about a year, I most say: Do shut up already.
In the time it takes to downmod someone, a few people have seen the opposing post, and likely agreed, or at least posted something in response that's likely to generate more interest in the original. With the high volume of traffic Slashdot gets, even 20 accounts isn't enough to obliterate any opinion to a reasonable degree. One particularly controversial post of mine managed to get every single moderation, before ending up at "+4, Interesting". I had over a dozen "flamebait", "troll", and "overrated" mods.
Mod gaming is a known problem. Slashdot's system is still above average in my opinion, and has the benefit of enough wide participation (and light enough consequences) that it doesn't matter. Sure, it's disheartening to see one of my deeply-thought-out statements misunderstood, but it's Slashdot. It's not like anything said here has a high probability of drastically changing the world.
You do not have a moral or legal right to do absolutely anything you want.
That was my point - back then creating working dies at 22 nm, which is as good as we can do right now really, would have been laughed at by some. "That's only 100 or so atoms! Good luck!"
The team doing this has demonstrated that they can be much more accurate with single atom placement than in the past, so I don't doubt we'll be building at the single atom scale in mass production eventually, and probably within my lifetime easily.
I want steam based computing. Big things lots of spinning wheels and whistles.
Down with this mamby-pamby micro electronics.
They make a transistor from multiple atoms, all of them silicon but one, which is phosphorus. That is NOT a transistor made from a single atom (as the title suggests). Great advance, in any case, but misleading title.
Today, we can place the atom with high precision, in silicon, so that the devices can be made reliably.
Cornell demonstrated a single atom transistor nearly 10 years ago, and today we are still pretty much at the level of demonstrating / playing / investigating.
Ten years from now, who's to say we won't be able to mass produce them?
It is a pretty big jump from building a single demonstration / proof of concept device and connecting it and integrating it into a design that works reliably at speed. IBM seems to be getting some interesting results with a single atom DRAM, but that is still way closer to a laboratory curiosity than an option for shipping silicon.
But that is just the Fab side of things. To actually design and build chips with this sort of technology is almost certainly going to require some serious upgrades to EDA tools.
much of left-wing thought is a kind of playing with fire by people who don't even know that fire is hot - George Orwell
First, some background: Most agree that Moore’s law, which has held firm, will meet its demise in a matter of decades. This will likely signal the end of the silicon era. The basic problem is the limitation of the ultraviolet process by which a hundred million or more transistors are etched onto increasingly smaller silicon wafers. But another problem is perhaps more daunting: When computing is reduced to smaller and smaller quantum scales (currently, the chip inside your computer can be 5 or so atoms across), one runs into the Heisenberg Uncertainly Principle; it simply becomes impossible to tell exactly where an electron is, so there is leakage. In other words, using quantum computers, given contemporary materials and knowledge, 2+2 might eventually end up being 4, but there might need to be built in recursion and tautological algorithms. Computation using atoms has already been done, as pointed out by another poster. Think it will be a while before we see them at Best Buy. Also, it still seems like silicon based technology
That's what I said... that we've been able to build these things for ten years. As the article explains, the big difference here is the precision of the placement of the atom, making the devices much more manufacturable (though not on a mass-scale, of course).
And yes, there are other steps involved in making actual devices. But we don't have to work in a single pipeline. As the process engineers get closer to making this sort of thing mass producible, the software engineers will be simultaneously upgrading the EDA tools, and the design engineers will be thinking of ways to use this new device. It'll go into high price, low yielding devices at first. Probably military tech, or cutting edge instruments for physicists. Those pilot projects will be used to the design tools, tune the process, and maximize the yield.
It'll be quite some time before they reach consumer electronics, if they ever do, but I wouldn't toss them aside as non-manufacturable.
Ten years from now, who's to say we won't be able to mass produce them?
A little known fact about Moores law. People usually don't know this, but Moores law is actually an inverted bell curve, so a few years from now, circuits will actually start to grow bigger and bigger every year. In the future we will have computers as big as mt to perform the simplest tasks. Unfortunately the bottom of this bell curve occurs at the same time as the end of the Mayan calendar, so not to many people will be around to worry about it.
If my comment didn't sound as good in your head as it did in mine, then I guess we all know who's to blame
>Subatomic
No. The limit is a single atom. Not unless someone comes up with a way of making a transistor out of free quarks. We'd have to have some sort of breakthrough in physics to do that and that's not even on the horizon yet.
-theoretical ---we are not even here yet.
-empirical
-demo devices
-prototype devices
-production/commercial devices
--
BMO