Silicon chips get speed boost with a lead start
By Ross Miller 
posted July 20th 2010 8:46AM
posted July 20th 2010 8:46AM
In tennis, the materials of the tennis court affect the performance of the ball. Such is the case, on a much, much smaller scale, for electron movement across circuitry. Silicon chips give resistance that lowers the speed limit, while atom-thick sheets of carbon (a.k.a. graphene) have a special property whereby free electrons are almost weightless and can travel up to 0.003 times the speed of light -- sounds great, but it's hard to produce in bulk. Cut to Han Woong Yeom and Pohang University of Science and Technology in South Korea. His team has added a thin layer lead on a silicon chip, lowering the electron mass (and thus proportionally raising its speed) to 1/20th compared to standard silicon. Still a ways to go for graphene speeds -- by a factor of three, according to Yeom -- but it's also more likely to mass production.
Could of swore you meant the element Pb
@Rohan
that's what I thought before reading the article
@Rohan
Indeed, Pb is what the article talks about. You can find it out by reading both the "via" and the "source" links.
Wait, so 20 times faster than silicon.
So, were talking about 60 gigahertz processors here....
Damn it! These pants were new!!!
@Firewave
Well, i think they mean the actual speed signals can travel between points; not necessarily the clock speed.
@paul34
precisely. faster moving electrons does not equal faster oscillating transistors.
however it should eventually equate to higher performance.
@paul34
Keep in mind the speed of a signal is MUCH faster than the speed of the electrons. I believe they are talking about the speed of the electrons, not the signal, so the potential for a faster signal speed is greatly increased.
@devastator
But to devastator's point, there is still the issue of transistor performance. You can move electrons through the material as fast as you want, but the transistors will still be a limiting factor in computational speed.
@benblake112 The transistors will be made from graphene. That's the point. A small strip of graphene where the chirality of the carbon atoms is n=m causes it to behave like a semi-conductor. Placing an orthogonal electric field allows you to control the rate of electrons flowing through it.
The problem with graphene and carbon nano-tubes is that while they're fast at conducting electrons while in the "on" state, they have a terrible "off" state compared to silicon. It makes it less suitable for digital chips in that respect since it has huge power implications. Makes for great analog circuits though.
@Firewave No, now we will up the speed reference like Megahertz, Gigahertz will be less used
Now we use .06 terahertz
First to 1terahertz intel or amd??
it also means less heat
less heat = more cores
i wonder how this affect intel's road map if amd used this new tech
intel: oh shiz amd is using lead what should we do
intel ceo:umm sue
intel:great idea
@owned66 That does sound a lot like Intel...
@timw4mail
Actually, AMD and Intel have a very confidential but very real agreement concerning the sharing of trade secrets and patents. Haven't you ever wondered how AMD CPUs got SSEx and how Intel CPUs got x86-64 (hope you didn't think 'Intel 64/em64t' was Intel's design). There are multitudes of other areas where they have shared design elements.
This agreement was almost outed recently due to specific wording that allowed AMD to use the x86 license (yes, even the basic x86 license is part of this) only if they were a fap (ie they fabricated their own CPUs). Since AMD has spun off GlobalFoundries into it's own entity, they are now legally considered a chip foundry. Lucky for AMD, Intel didn't push this issue; probably due to the risk of monopoly charges in the EU and the risk of this controversial agreement of making it into public hands (it's controversial because Intel and AMD clearly want to keep others, such as nvidia, out of their segment).
I think this should lead to better power/performance ratios, but beyond that I'm not sure.
@timw4mail
Very clever, lol.
Silicon works not because it has higher resistance, but because its semiconductor properties allow it to be "doped" i.e. electrically charged.
Could someone who actually knows this stuff elaborate what this means for the performance of a chip in the end?
The only related thing I've read once was about how chip designers have to optimize the layout of a chip so that the signal actually reaches different parts in time. But would a faster signal with the same clock speed actually lead to an increase in performance?
@astrath No but faster signaling means you can increase clockspeed for the same layout. Let's say you had your cache placed on the left side and the data fetch placed in the middle. If the travel time for a data read takes less time, then you can clock the chip faster.
Nvm. I see what the author was trying to do now.
I'm sorry, but this article looks like it was thrown into google translator and never proofread:
"His team has added a thin layer (OF) lead on a silicon chip"
"Still a ways to go for graphene speeds ... but it's also more likely to (lead to?) mass production."
so 20 times faster than silicon. android free apps
"0.003 times the speed of light"
just over 2 million miles an hour.
@Cluelesskid
you beat me to it...
.003 speed of light = 588 miles per second = 2 116 800 mph
damn that's fast
@sc0rch3d just ~10 times faster than electron speed in more conventional materials.
Uh, I may be fairly ignorant of such things, but I thought that the mass of an electron was constant, and no alteration of the surroundings could possibly alter it.
@mccand You are right. The actual, free-space rest mass of an electron is constant. Generally, a materials scientist talking about the subject matter of this article would refer to the electron's "effective mass" in the material, which is defined as follows.
When you look at the behavior of an electron in a perfectly periodic materials (that is, where the atoms are arranged in a regular, repeating pattern as they are in a silicon crystal), it turns out that the electron can be treated as though it is a free electron but with a different, generally lower, mass. This is a simplification that applies only for lower energies. If you accelerate the electron to very high speeds, the assumption does break down. The electron actually reaches a maximum velocity and then starts slowing down again, eventually reaching zero net motion again at the top of its energy band (albeit at a higher potential energy).
See Bloch's Theorem for more info.
@NameIsDavid
I learn something new every day . . .
Thanks!
The paper is ab initio simulation!!! It means nothing. There are thousands of such papers out there, which promise huge performance increase. Until someone really makes it it should be noticed by media...
Speed, too much!
I hate to burst your bubbles, but I'm fairly sure this technology won't take off due to fact that lead is getting banned from pretty much everything. I'm fairly sure the RoHS directive is going to make this impossible to sell in the EU and california. See the following snippet from wikipedia:
"On July 1, 2006 the European Union Waste Electrical and Electronic Equipment Directive (WEEE) and Restriction of Hazardous Substances Directive (RoHS) came into effect prohibiting the intentional addition of lead to most consumer electronics produced in the EU. California recently adopted a RoHS law[8] and China has a version as well."
But can it return Roddick`s serve ?
@bufbarnaby
+ 1