Slashdot Mirror
Isotropic Silicon?
jmichaelg asks: "The Register ran this story on isotropic silicon. Among the claims for isotropic silicon is that with a more homogenous lattice, heat is more evenly distributed. That means processors can reach higher speeds simply by being manufactured using isotropic silicon wafers. There's a sidebar at The Register about 1.7 Ghz Athlons manufactured with this stuff. According to the manufacturer's FAQ, isotropic silicon is produced chemically. Is this possible? I thought isotopes were chemically identical and differed only in the number of neutrons they have. Is this real or is it marketing froth?"
You can model a crystal as a bunch of balls (nuclei) held apart by springs (bonds). Heat travels through the crystal as sound waves (phonons), which cause the balls to wiggle around against the springs. If all the balls and springs are the same, the interaction with each plane in turn is identical and the phonon keeps going in the same direction. A phonon which travels to the end of the crystal escapes. If some of the balls are heavier than others, they resonate differently and tend to scatter the phonon instead of allowing it to go in one direction. A phonon which keeps getting scattered is a lot less likely to get to the end of the crystal than one which does not, so heavier balls (different isotopes) decrease the thermal conductivity.
"
/ \ ASCII ribbon against e-mail
\ / in HTML and M$ proprietary formats.
X
/ \
Time is Nature's way of keeping everything from happening at once... the bitch.
Call me silly, but why bother going through tons of very hard processies (I belive isotopes can be derived/speperated through chemical processies..but its HARD!)just to keep on using silicon? There are far better semiconductors out there. Gold for instance, when put in the proper geomeric shapes (at molecular levels) becomes a semiconductor with many more logic levels the silicon. Silicon has two. Gold has the possibility of having 7 or more. (lookup nanoscale physics if your interested, I only have very curssory knowedge of the subject)
Why not spend the reasearch money on developing processors with other better semiconducts is all i'm saying? Instead of putting more and more into an almost tapped out tech (like how we still do research into gasoline internal combustion engines, when we should be doing research on power cells and other eletric vehicles)
You are only young once, but you can stay immature indefinitely.
There is a difference between isotopic and isotropic. Isotopes are variants of an atom. If I remember wel being isotropic is a property of the crystal lattice atoms are placed in. It means some other property is independent of its relation to the direction of the crystal axes. Isotropic means "the same in all directions", I believe.
Sigh
I went to a talk a year or so about this. (Although I can't swear to get all the details right.) There was an american researcher who said he was approached by some russian scientists if having isotopicly pure Si would be usefull. Apparently after the USSR disolved it was decided that they didn't want all the people who knew how to separate isotopes to be unemployed and the machinery to appear on ebay so they went down the periodic table producing isotopicly pure elements.
Because it is so hard to do, isotopiclly pure elements are very valuable, but it is also very hard to find a buyer, so the Russians were nearly giving it away to researchers.
As to whether this can actually be used to make faster Athlons though, I agree I am in the "ait and see" catagory.
may you run like a vicuna
This technology uses high-power frequency-selective energy sources, like LASER and MASER, in which the previously soviet union was the undisputed leader. Imagine that.
By making all the atoms in a crystalline latice the same mass, their positions become very regular, and this supposedly helps heat transfer.
*whup* "Get along, little electrons. Heeyah!"
I am concerned about the origins of the technology and the ability of any process to generate it "scaling up" to the kinds of quantities required for a chip fab.
I'm also highly skeptical that a firm in the former USSR will be able to pull it off - from an investment standpoint, it's a quagmire of accounting practices that border on the fraudulent, and the rule of law has yet to be established.
Given the other uses for isotope-refining techniques, I'd expect some concerns from a proliferation standpoint. Yes, Si ought be a helluvalot easier to separate than the heavy elements, but any technology that can produce large quantities of isotopically-pure Si could likewise be adapted to produce large quantities of isotopically-pure lighter elements.
Finally, as much as we hate IP lawyers around here, there's the matter of patents. Who holds the patent on the gadget? What's to prevent others from using it? (This matters because the spin of the story is that this technology will benefit AMD more than INTC within the next year or two, arguably with an eye to predicting stock prices of both, or with regards to the stock of the Si-producing company.)
So some scientists have a neat gadget, likely a spinoff of former USSR nuke work. Cool, and it may spur interesting further research. But I don't see it having a major impact in chipmaking technology in the immediate future.
It'd be great if it's (a) true, (b) scales up, and (c) presents no proliferation risk. But all three of those things have to be true for it to matter. Put me in the "wait and see" category.
Your definitions are correct, but the story (despite what the Slashdot headline says) is actually about the "isotopic" (no "r") variety.
Supposedly, pure Si-28 has 60% higher thermal conductivity than natural silicon (92% Si-28, 5% Si-29, 3% Si-30). Dunno if it's true or not, but it's a neat idea.
Cool beanz!
--Mike--
This is isotropic, though, which is different. When something is isotropic it has physical properties, such as conductivity and elasticity, that are the same regardless of the direction of measurement.
"The price of freedom is eternal vigilance." - Thomas Jefferson
So, some crazy wording going on here. Think Carbon 12 vs. Carbon 14, both have percentages in carbon based life forms that allow for Carbon 14 dating to be applied... now slide down a row in the periodic table, and we have Silicon. Both are have four valence electrons, but silicon is our favorite for making semiconductor devices. Now in order to make semiconductors, you need to have "pure" silicon in the first place, which you can then add dopants to to make it an n-type (negative charge carriers) or p-type (positive charge carriers) semiconductor. This electronics grade silicon is elementally pure, but as this article suggests, isotopically not... (some atoms weigh more, extra neutrons!!!)
Now isotopically pure silicon requires that you separate out all of the heavier silicon atoms in your batch, so you only have the "perfect" 14 protons, 14 neutrons and 14 electrons. As one could guess from some general knowledge of thermodynamics, and crystal lattice structure, an isotopically pure crystal would have a natural tendency to shake in a purer way (hence have temperature) without little heavier marbles, (the Si with more than 14 neutrons) becoming centers (defects) holding more heat ( 1/2(mass)(velocity^2) <- mass gets higher, more ability to hold higher than the average kinetic energy and mess up the nearest neighbor atoms and melt the little solder droplets or copper channels or whatever they're afraid of).
So to review.
Perhaps most interesting is the cost of this... Any way I can think of to split off the heavy atoms is not cheap, so hence the isotopically/isotropically pure silicon boules -> wafers get more expensive...
hopefully I haven't put in extra r's where they shouldn't be and confused people more... aaaargh... stupid words.
Fun isotropy fact: Tungsten is the most isotropic metal around, all of it's properties are the same in every direction!... w00t. (it is also highly resistive, but has a reeeeeally high melting point)
Looks like college is paying off... ... err...
Is it just me, or does this seem to be a very bad idea? After all, wouldn't misanthropic silicon just refuse to carry electric current altogether? This kinda negates the whole point, doesn't it?
:)
Oh wait, it's not misanthropic either... don't you hate people who can't fucking read?
To the editors: your English is as bad as your Perl. Please go back to grade school.
From Websters:
/"I-s&-'tä-pik, -'tO-/ adjective
/-'tä-pi-k(&-)lE, -'tO-/ adverb
/I-'sä-tr&-pE/ noun
Main Entry: isotope
Pronunciation: 'I-s&-"tOp
Function: noun
Etymology: is- + Greek topos place
Date: 1913
1 : any of two or more species of atoms of a chemical element with the same atomic number and nearly identical chemical behavior but with differing atomic mass or mass number and different physical properties
2 : NUCLIDE
- isotopic
- isotopically
Main Entry: isotropic
Pronunciation: "I-s&-'trO-pik, -'trä-
Function: adjective
Etymology: International Scientific Vocabulary
Date: circa 1860
: exhibiting properties (as velocity of light
transmission) with the same values when measured
along axes in all directions
- isotropy
maken
A side note that in the future carbon nano-tubes might be used to shift heat.
They act as thermal supercondutors by directing thermal noise as a one dimensional acoustic wave.
I'm on a really slow link so you will have to hunt your own links but any nono site should carry details.
(note: electrical superconductors are lousy at thermal transmission)
cya, Andrew....
This is my sig, exciting huh!
It's not an isotope issue. The number of neutrons in the atom would not change how the material conducted heat, or any other property aside from total mass. It's the isotRopic qualities of the crystals that matter. It's a plain old silicon atom, just crystallized in a different manner.
Um. Okay. What are you smoking?
Basically, the greater the distance the phonons (heat quanta) can travel before bouncing off a site in the lattice that is different from most of the others, the faster heat can travel from one side (the hot side) of the chip to the other (the cooled side). With so many scattering centers around in native Si, it is far from a straight shot, so the phonons bounce from site to site, only slowly diffusing from hotter to cooler regions. Semiconductor-grade Si has usually been so highly purified and so carefully crystallized that there are essentially no lattice defects other than the minority isotopes. Eliminating these can improve cooling dramatically, although there is still phonon-phonon scattering to slow things down.
I thought isotopes were chemically identical and differed only in the number of neutrons they have. Is this real or is it marketing froth?"
Contrary to popular belief, isotopes do play a role in chemistry. Anyone who's taken organic chemistry can tell you about the deuterium isotope effect, which says that deuterium-carbon bonds are harder to break than carbon-hydrogen bonds. This can then influence certain elimination reactions in which otherwise there would be a mixture of products (were there 2 C-H bonds) into only one product (leaving the C-D bond alone).
"
/ \ ASCII ribbon against e-mail
\ / in HTML and M$ proprietary formats.
X
/ \
Time is Nature's way of keeping everything from happening at once... the bitch.