Slashdot Mirror
Dawn Of The Diamond Age?
Wiesel Werkstätte pointed out this article in Nature about recent advances in the use of thin diamond films as semiconductors, a tantalizing possibility which has been thwarted thus far by the overlapping, misaligned structures left by the process of deposting diamond in a film. From the article: "Matthias Schreck and his colleagues from the University of Augsburg have found a way to eliminate the grain boundaries. They have not removed misalignments entirely, but they have restricted them to narrow bands that no longer isolate one crystalline region from its neighbours."
The resistance of metals rises with temperature. The resistance of insulators drops. There is a potential problem there, but maybe you'd have to worry about keeping the wires or SiO2 insulation from melting first.
Incidentally, semiconductor resistance usually drops with temperature -- so you on-state works better, but your off-state leaks more. Apparently diamond, if it can be formed into transistors, would stretch the point where it becomes too leaky to work out considerably.
What I would be concerned about is this: what is the forward drop across a diode junction? This determines the minimum voltage where any semiconductor will work. For Germanium it was a fraction of a volt, but Ge was too leaky at room temperature. For Silicon, it's 0.7V, and leakage is reasonable up to about 150C. Diamond is next on the periodic table, with much higher temperature, so I expect the diode drop will be several volts. If there isn't some work-around for that, you can forget about gigahertz diamond circuits. They might be great for the high-voltage, high-powered stuff like controlling CRT's or the AC end of a power supply.
All the talk I've heard about diamond IC's is for very specialized applications like radiation-hardened chips (EMP's from a tactical nuke won't fry the diamonds themselves, though the rest of the circuit may be in more trouble) or very high-temp applications where for some reason you can't put the logic somewhere else.
But I thought the major problem with diamond chips wasn't fault lines so much as the fact that you can't p-type dope diamonds; for reasons still unknown they simply boot out any such dopants. Which would make it kinda tricky to do anything useful as far as diodes or transistors. Are these people claiming any sort of innovation in that direction? Or is this result more useful for using diamond as a structural material?
It doesn't look too incredibly exciting. First of all, the process by which the diamond is made is substantially different and more flawed than silicon wafers (annealing liquid silicon verses condensation of gasious carbon). They say they've merely discovered a method that reduces condensation disturbances to a usable state.. But when we're designing gate-widths that are only a couple dozen atoms wide, these disturbances will probably be monumental.
:) Another theoretical advantage is the diamond structure should be denser than silicon, which could mean smaller absolute minimum gate lenghts.
The main advantage of these are high temperature (500C instead of 150C with Silicon). One theoretical advantage is running higher voltages at hotter temperatures with less breakdown. So you could over-clock these babby's.
One issue will have to be the metalergic process by which copper attaches to the carbon. It took hundreds of small miracles to find the right intermediate layers to get copper to stick to silicon; will it take just as long for carbon?
Don't know if better or worse, just know that it'll be different. And from the looks of it, more expensive
-Michael
Except of course diamond is far too brittle, we know of no way to weld or bond it and then there's that fire issue...
I don't read ACs: If a post isn't worth so much as a nom de plume to its author then I wont bother either.
DeBeer's main fear at the moment is not diamond films in processors, but those artifical diamonds made in Russia with huge presses.
They spend a fortune just to make test equipment available that can decide wether a diamond is articifical or not.
Does this mean I can get a primer for my daughter? (would you order that from fatbrain?) But just keep in mind now for all the rest of the problems just let Judge Fang spend a few hours with them that should take care of all their problems. This is OT but then again so is this entire comment how good is Zodiac?
Cypherpunks: Civil Liberty Through Complex Mathematics. Those who live by the sword die by the arrow.
Also it's worth noting that diamond stops being a semiconductor at 500C. I hate to dust off ye ol wayback machine (I never know if it will work properly) but I seem to recollect that Diamond became usefull around 220C. Obviously the performance of a 300C chip would suffer somewhat due to more electrons being scattered by the high temperatures, but it's not so much their preformance, but their useability that is the issue. Besides wouldn't we all want to see ads for Cray HVAC/supercomputers?
Warning! Suspect Anecdote Follows, Use Caution.
Once upon a time I was making amorphous substoichiometric thin films of tungstun trioxide (WO3). The Air Force was interested in this (though not me specifically) because W03 could be made into WS2 which was a nice semiconductor, well at least the band gap was attractive. It was a little higher temperature than Si. What the hell would they want with that? Well they wanted to use it to replace the mechanical devices that read the fuel pressure in the tanks of their fighter (probably all planes). I know for commercial planes 1 lb of weight translates to an additional $20k of annual operating cost (for a fighter plane I'd assume it's many times greater). But not only would such devices be much lighter, they'd be much smaller, and more reliable. It's the boon of solid state physics, as one makes things smaller, nearly every property improves. There were some other characteristics of WS2's structure that made it handy from a manufacturing stand point, but it was the band gap (higher temperature operation) that made it useful. I certainly see how the air force might want high performance fighters that had more room for accessories, while being more reliable and cheaper to fly. Something to consider.
--Jimmy has fancy plans; and pants to match.
Oddly enough the difficulty of using Cu (Copper) on Si (silicon) is somewhat of a special case. Si crystal has the unfortunate (or fortunate depending on your point of view) property of being shaped like Cu sized tunnels. This had been known for ages. Back in the old days researchers would put a Si crystal an a big slab of warm Cu to let the Cu migrate into and throughout the Si. The Cu would zoom through the crystal and home in on the defects, this of course made the defects in Si crystals much easier to study. But having this same thing take place when you're trying to lay down a nice pattern on your precisely doped Si is somewhat less that ideal. Now I'm not a wiz a crystalography, but I can't "see" any orientations of diamond that would be like cu sized tunnels.
--Jimmy has fancy plans; and pants to match.
GaAs (Gallium Arsenide) is alive and well in the semi-con industry. head over to Lucent's Reading or Breinigsville (sp) facilities. Indium Phosphide's even got it's niche. these substrates are used most often in the opto-electronics arena, because of their carrier properties. The reason they never took Silicon on is because of their material properties. they're far too brittle for the traditional semi-con industry. compared to GaAs, you can use Si as a hockey puck. i've held GaAs wafers (2") in my hand and watched them crumble to dust. cause of death? thermal expansion due to the heat of my hand. i can see diamond in use in place of poly-si, but you're right about the cost.
I've seen people researching this for at least
20 years. Radiation hardened military chips
was supposed an application.
In the Periodical Table of the Elements,
carbon is in the same column as Silicon, one above.
Therefore it has properties useful for circuit
design, such as substrates and nano-wires.
(Conversely biochemists have speculated on Si-based life.)
----
Processors currently cost a fortune made out of sand. How much will they cost made from diamond? Also will the current duron/celeron chips be made from cubic zirconium the low cost alternative?
Slashdot: Proof that a million monkeys at a million typewriters can create a masterpiece
In one of Arthur C. Clarkes '2001' series (the end of 2010, I think), the decay of historical monuments has been halted by covering them in a film of diamond atoms.
This idea has stayed with me for years.
*Sigh* imagine Paris, in the springtime, the Tower and Arch sparkling like diamond in the morning sun...
Nothing like a little skewed facts to give businesses a reason to charge more for their products.
-p4
(c) All Rights Released.
(Of course, my knowledge of this stuff runs about as far as yours, so it's possible that this isn't a problem/is unsolvable/is useless. But leave that judgement to the experts.)
How can we continue to believe in a just universe and freedom to eat crackers if we have no ale?
I'll be surprised if they can come out with a practical application for this.
::smile::
They're scientists... I'll be surprised if they CAN'T come up with a practical application for this. Heck... post-it notes were just a cool failure sitting around the lab. Someone will figure out how to build/market something using this tech. I guarentee THAT. How fast... thats another story.
This space for rent. All reasonable inquiries will be entertained at proprietors discretion.
Does that mean I could like have an uptime of like 500 years? ;-)
As if having the ordinary problems of teenage kids wasn't bad enough, now am I going to have to worry about them joining the Drummers, as well as the more mundane problems of alcohol, drugs, smoking, gangs, car accidents, etc.
The living have better things to do than to continue hating the dead.
...but at 500 C, it melted a hole right through my motherboard.
The next Slashdot story will be ready soon, but subscribers can beat the rush and slashdot the links early!
'Diamond chips' would be invaluable in electronic devices exposed to high temperatures. Semiconducting diamond works up to temperatures of 500 C -- silicon devices fail at around 150 C. So not only can you toss the cooling fan, you could use it as a mug warmer too! Seriously though, I wonder if an electronic circuit could work at 500 C. Something I remember from science class about resistance and heat rise together. Considering 0 is freezing and 100 is boiling, we're talking about heat on the magnitude of a small blowtorch. I'll be surprised if they can come out with a practical application for this. --toq
The Russians are doing gem quality diamonds 4 carets every 72 hours. They're also going to do LCD monitors that use almost no power (yet are bright and vivid without backlighting) out of diamond as well.
Simple doping already gives np switching with rough diamond films such as the one shown in the article.
So don't bother buying a diamond engagement ring if you're getting married. In a decade they'll be free with a can of motor oil at Wal-mart.
I've hit Karma 50 and gotten a Score:5, Troll... I win!
So will DeBeer's take charge of all the processors? I can only imagine what prices will be like then.
-- yawn. --
Can you say "skyscraper support beams"? We may not have this before nanotech, but once we have the ability to place carbon atoms exactly where we want to (on a large scale), expect to see 500-story towers that many of its inhabitants would never need leave. I just hope they outlaw dropping things from the roof...
"Feel a glory in so rolling / on the human heart a stone" --E. A. Poe, "The Bells"
...to implode Jupiter into a sun. During the process, millions of metric tons of C60 will be thrown throughout our galaxy, which we can then collect to use for building processors.
Typical short-term solution. Jupiter is a cosmic vacuum cleaner, so it absorbs a lot of comets that would otherwise endanger Earth. If we blow up Jupiter, we'll have to use the carbon to give everyone on Earth a diamond coating so they'd survive an impact event. And we'd have to use lots more on diamond drills so holes could be made to cater to man's procreative/eliminative needs. Then we'd need more carbon to plug the hole that was left so it wouldn't make a draught. Projections indicate we'd run out of carbon in thirty-two years, four months.
So go ahead. Blow up Jupiter. But you'll be condemning humanity to forever going without sex or urination.