Slashdot Mirror
First Superconducting Transistor Created
holy_calamity writes "New Scientist reports that the first working superconducting transistor has been created, by researchers at the University of Geneva. Field effect transistors with zero electrical resistance would allow much faster operations. Only drawback is they need to be supercooled, something that may be addressed by improving the materials used."
"Only drawback is they need to be supercooled, something that may be addressed by improving the materials used." - that last part is a bit of an understatement. We're still decades (centuries?) away from room temperature superconductors.
To make laws that man cannot, and will not obey, serves to bring all law into contempt.
--E.C. Stanton
At 0.3 kelvin - just above absolute zero - these electrons flow without resistance and so create a superconductor.
So my stock fan won't quite cut it this time?
HAL9000 singing that song popped into my head after reading that headline.
Perhaps this discovery is just one more step in the direction of a singing homicidal AI computer.
Daisy, Daisy
Give me your answer do
I'm half crazy
all for the love of you...
but will it blend?
Josephson Junction has been used for switching in superconductors since I was a kid.
http://en.wikipedia.org/wiki/Josephson_effect
We have no idea how far away we are. We don't fully get it and are pretty much trying substances at random. We might figure out something that works next year or never. It's not something you can predict with any accuracy.
It is no longer uncommon to be uncommon.
As far as I know, the first superconducting transistor was reported in 2006:
cond-mat/0601434
Wow! thats super c... uggh... forget it
love is just extroverted narcissism
Wish I had mod points
...are hard at work trying to figure out how to run one of these things in an aquarium.
Have gnu, will travel.
If we use our best insulators like aerogel or some vacuum flask? Can it be done?
...about liquid-cooled laptops? I've a feeling someone's about to make a joke about that and this story...
Operation Guillotine is in effect.
It could be expected that half-assed semiconductor technology will be at some stage replaced by fullconductors. But with superconductors? I never expected that!
Everyone had that dream in college. Build a liquid nitrogen cooled computer at 4Ghz in your dorm room while you could still use all the electricity you wanted. Even 10 years later that's still an untouchable speed for consumers.
Interesting enough, superconductors only have zero resistance at DC, when you switch them you start getting skin effects, effectively giving rise to an effective resistance.
This impedance is a function of frequency and actually increases faster for superconductors than for regular conductors. I used to have a good graph showing where the crossover was between copper and some high temperature superconductor (YBCO I think). If memory serves it was around 10-100GHz, any GIS experts care to take a crack at finding it?
...motherboards' supercooling systems.
Speed isn't only determined by on-state resistance. Capacitance & inductance matter too and will be the limiting factors for a theoretical transistor that's 0 resistance on and infinite resistance off. Such a theoretical transistor won't dissipate heat, so it won't get hot. However, heat will be dissipated somewhere else because current still must flow from high potential to low potential. Furthermore, transition times aren't arbitrarily fast, and during the transition, the transistor will dissipate resistive power; this could be a big problem for systems cooled below 4 K.
"Only drawback is they need to be supercooled, something that may be addressed by improving the materials used." - that last part is a bit of an understatement.
Is an understatement from the New Sensationalist (as it should properly be called) an oxymoron?
The New Sensationalist runs a story every couple of weeks about how some new breakthrough will revolutionize something or other in the next two years. Has anyone gone through their predictions like we do with psychics to see what their actual hit rate really is?
Prime numbers are exactly what Alan Greenspan says they are -S. Minsky
We'd love to get our hands on some superconducting FETs. The ones I'm designing around right now have 5 milliohms Rds, and they're *still* getting so hot we have to solder big heat sinks onto the backsides of them.
But this just shifts the problem to the gate drive, because during any finite time period between 'off' and 'on' the FET acts like a big power resistor and heats up. Even if people ever make these so they're superconducting at room temp, they'll still heat up when in the active region. (Or we'd need to develop drivers that could produce instantaneous off/on transition times.) So we'd need ones that could remain superconductive in well over room-temp transients. If you have a superconducting FET that suddenly stops superconducting because of a temperature peak, it'll vaporize just about instantaneously. These would be an exciting gamble.
Nostalgia's not what it used to be.
Minor detail indeed.
/.'ers for that matter.
That's like saying, "I have a cancer cure pill that works 100% percent of the time and costs mere pennies per pill, with no patents! Oh, one minor hitch, my revolutionary "cyanide pill" tends to kill the host, but we're optimistic on a workaround!"
Sounds like something you'd say to investors to raise capital, not to peer scientists, or know-it-all
Slashdot "libertarians": Small government for me, big government for those I disagree with. -1, I disagree with you
It seems that it only takes one person that is able to produce the desired result repeatedly. After that people stop arguing about if it can be done, and if so how to do it and actually start working on the problem. There are tons of examples of this throughout history. Now that someone has actually made a superconductor, I would be willing to bet that sometime in the next 15 or so years we will see them in use. Perhaps widely in use.
If you're using Helium-4 to cool your transistors to 0.3K you actually can't use aerogel, in the superfluid state Helium doesn't have surface tension and just oozes through ridiculously small openings. Do a youtube search for "Liquid Helium II: The Superfluid". It's very cool stuff.
Yes, you can insulate a device, so that in almost all cases (definitely in the case of a fast-switching transistor) the main heat source is the device itself.
Here's a commercial box that cools a 2-inch wafer of high-temperature superconductor to around 80K. This box uses 80 watts including whatever other signal processing stuff is in there.
Another source (Cryogenics 42 (2002) 705-718) says that 1W of cooling power at 4K will cost you 5kW of input power using a straightforward helium compressor. This scales as 1/temperature^2 for higher temperatures, but for lower temperatures you'd switch to a different type of refrigerator.
0.3K refrigerators using helium 3 would not use more than 10kW, but this is already too much for most applications.
So the practical significance of this research is that it may be reproduced with higher temperature materials, not that we will build THz DSPs at 0.3K.
Use of the term "supercooled" in this context is bogus. Something is supercooled if it remains a liquid, even though it should be a solid at those conditions (or it remains a gas where it should be a liquid). If you put a glass of very clean distilled water in a freezer you'll find out that you can cool it down to -7*C or lower without freezing. It will momentarily freeze if you drop a snow flake into it though, or when you hit the glass with a screwdriver.
(For the curious: this is because extremely small crystals and droplets have higher free enthalpy than the bulk phase due to surface effects, so their formation is inhibited.)
This has nothing to do with superconductors, because they are always solids and cannot be supercooled. For superconductors you're looking for "cooled below its critical temperature", but I admit that it doesn't sound as good as "supercooled".
Those who would give up liberty to obtain working drivers, deserve neither liberty nor working drivers.
The only drawback is not just supercooling.
Let's talk fabrication. Anyone know what the yields are on lanthanum aluminate? What are the physical manufacturing chanllenges?
Making one transistor is easy compared to making millions of billions (per die, per wafer, per lot) of them reliably and cost-effectively. That is the major obstacle. Remember how long it took to switch to copper wires in the late 1990's? And that was with billions of dollars invested by many companies, and in a hurry!
It took 40+ years to go from a BJT the size of your fist to a modern CPU. I suspect it will take a decade to do the same with this new technology, assuming there is major capital investment for it. Probably 20 years if one of the big three (IBM, Intel, that company is Southeast Asia that has a the 45nm process who's name I am forgetting) doesn't get involved.
https://www.accountkiller.com/removal-requested
Wait a second... This is does not look like a superconductor... Where are the cooper pairs? Where is the lattice? They are talking about an interface between two crystals, it doesn't seem like a nice regular lattice necessary for the formation of Cooper pairs.
What it looks like is a 2D electron gas cooled so much that the electron mean free path is longer than the size of the crystal. No scattering, thus no resistance but no superconductivity either. Electron gases have been around for a long long time, no novelty here.
Moreover, to settle the ongoing debate about the cost of cooling such a device: liquid Nitrogen is not cold enough to reach 0.3K, for that you need a pumped He3 system. Now the cheap ones go for 100K$ (then add the cost of liquid He4 necessary to keep the system cold).
This looks like a nice device but somebody somewhere clearly overstated its importance!
I say they can do it in 5.
Since reducing the fab size doesn't reduce resistance, it won't speed up a superconductive processor. If someone is willing to put up with a system size of a fridge, or entire building (and keep it cold), it will work just fine, potentially at speeds that are orders of magnitude above modern supercomputing. I very much doubt that this will ever see a consumer desktop, or even many server rooms. But anybody already in the supercomputing business would probably be happy to go through hell to be able to whatever it is their doing faster* at the same price.
*It will still take several minutes to boot because of the slow hard drive of course.
Liberte, Egalite, Fraternite (TM)
Think of power converters. Think of a broadcast TV transmitter.
Many devices use switches to control voltage and current. One of the most common switches and also one of the most efficient is the transistor. Much like a relay it can handle high current and is able to be controlled easily and fast. But the price of this speed is heat. The faster it is switched the hotter it will become. Cooling is only a stop-gap as transistors work much less efficiently when chilled below their operating range. ---------- Bobwilliams Hollywood North Acting & Modeling
You know, it's almost as if you all have never heard of the Cooper pair transistor:
http://www.google.com/search?q=cooper+pair+transistor
You can make a superconductor by supercooling a conductor? Holy crap. What a breakthrough!
Oh, you're not stuck, you're just unable to let go of the onion rings.
Have they tried using amorphous metals with this? I remember hearing years back that amorphous metals were able to superconduct at fairly high temperatures (relatively of course). However, this was years ago, and I don't know exactly the specs on them. Still, I just thought I'd throw it out there.
I am not sure how much energy it takes but I believe one approach to keeping the helium cool is to put some liquid nitrogen cooling around it since liquid nitrogen is fairly cheap.