Slashdot Mirror
Ohm's Law Survives To the Atomic Level
Hugh Pickens writes "Moore's Law, the cornerstone of the semiconductor industry, may get a reprieve from its predicted demise. As wires shrink to just nanometers in diameter, their resistivity tends to grow exponentially, curbing their usefulness as current carriers. But now a team of researchers has shown that it is possible to fabricate low-resistivity nanowires at the smallest scales imaginable by stringing together individual atoms in silicon as small as four atoms (about 1.5 nanometers) wide and a single atom tall. The secret is to introduce phosphorus along that line because each phosphorus atom donates an electron to the silicon crystal, which promotes electrical conduction. They then encase the nanowires entirely in silicon, which makes the conduction electrons more immune to outside influence. By embedding phosphorus atoms within a silicon crystal with an average spacing of less than 1 nanometer, the team achieved a diameter-independent resistivity, which demonstrates ohmic scaling to the atomic limit. 'That moves the wires away from the surfaces and away from other interfaces,' says physicist says Michelle Simmons. 'That allows the electron to stay conducting and not get caught up in other interfaces.' The wires have the carrying capacity of copper, indicating that the technique might help microchips continue their steady shrinkage over time and may even extend the life of Moore's Law. 'Fundamentally, we have shown that we can maintain low resistivities in doped silicon wires down to the atomic scale,' says Simmons, adding that it may not be ready for production now, but, 'who knows 20 years from now?'"
If the atomic resistance gather together at ohm's law, will they occupy it?
When the foot seeks the place of the head, the line is crossed. Know your place. Keep your place. Be a shoe.
At first I was thinking they meant Moore's Law and somebody had found a way to make really tiny ultra fast processors. Moore, Ohm, Watt... learned all of their laws in the same class in high school. I really need to take up drinking coffee in the morning.
... it scale and can you produce it cheaply?
Since the popular definition of Moore's Law is exponential growth in any tech-related field, I'd say approximately never..
Live today, because you never know what tomorrow brings
They're different laws about different things, they just happen to relate in this instance.
+1 IDisagreeSoHeMustBeATrollOrAnAstroturferOrAShill
So the Star Trek prediction where computers of the future seem to be full of brightly lit crystals may have been accurate?
At what point will we stop hearing about it?
When you stop reading a site dedicated to geeks, computer professionals and computer enthusiasts.
W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
My only issue with Moore's law is that it's a "Law", when really it's more of a guideline. If it was truly a law, then semiconductors would half in size naturally over time, without any research or development involved. Plus I believe that the "Law" gets regularly adjusted as the trend declines.
+1 IDisagreeSoHeMustBeATrollOrAnAstroturferOrAShill
Of course it's valid. It's a law! Not some phony-baloney "theory" like evolution or gravity.
Hate to have to solder one :-)
--an unbreakable toy is useful for breaking other toys--
TFA says that the wires were deposited lithographically (the technique currently used to make chips) and then the phosphorus was deposited. So this, in theory, could be done cheaply.
However, TFA also mentions low temperature. It doesn't measure exactly what temperature, but processors are not usually operated at low temperatures. If this is a "liquid nitrogen cold" temperature, then this could very well be useless on a grand scale. But if the effect survives to room temperature (or higher), then this could have a huge impact.
Just a first order approximation would show that these wires are about 5 times smaller than the current 22nm state-of-the-art. In two dimensions, that means roughly a 2500% increase in density, enough to keep Moore's law alive and well for some time to come.
Great warrior...hrmph! Wars not make one great.
The resistance of interconnects grows polynomially, not exponentially, as they decrease in size.
It's an important difference. As sizes get small enough, we start to see stochastic effects, but we're not there yet.
Don't take life too seriously; it isn't permanent.
Likely because I just quit smoking and are somewhat grumpy, but I am tired of hearing about Moore's Law. Maybe those in the semiconductor industry care about it, but I, and those I work with certainly don't. At what point will we stop hearing about it? /rant
(Thank you for your patience. Now where are the damn pretzels?)
The most important part of Moore's Law was it essentially saying that your new toy will be far better than your old one before it even breaks. When the rate of doubling gets closer to 10 years, buying a new computer isn't going to be so much as the new toy is faster but rather the old toy broke. Once that driving force is over with, electronic companies will be talking about other ways to produce money in more mundane ways.
Seeing that the doubling power in Moore's Law is seen in almost all technological progression, you're going to hear about it a lot more. Probably about twice as much every year or so.
Tic-Tac-Toe, Global Thermonuclear War, and relationships all have the same winning move.
Coles Law is my favorite.
Yummy.
No brain, no pain.
Moore pics please.
My fault I guess, I was still under the impression it still uses the 'processor speed doubles every 18 months' definition (which isn't really exponential, but I will go with the flow).
Groan.
"Law" is being used ironically. It's really not even a guideline, just one man's observation. It has become self-fulfilling, since companies plan their products around it - so it will continue until no longer possible IMHO.
Moore never adjusted his "law", though other Intel executives sometimes refer to the time as 18 months instead of 2 years.
W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
The misrepresentation of what Moore actually said is the bigger problem for me. All this "exponential growth" crap isn't part of Moore's law. Nor is it about any other technology. It would be used a lot less if the "geeks" around here would keep Moore's Law to what Moore really said.
I'm not surprised in some ways though. A couple of years ago there was an IEEE podcast of some professor giving a lecture about the near-future problems in the EE field in relation to processing power. Even there he had someone in the audience who fumbled with Moore's Law. If we can't get the correct definitions out of someone who was either a student or a member of the profession to get it right how is it we can expect the wanna-be geek population to get it right?
Now we have to worry about shrinkage? Maybe the microchip was in the pool
I don't know the meaning of the word 'don't' - J
I think everyone understands Moore's law is not a law in the scientific sense, but rather a Coloquialism like "Murphy's law".
It also somewhat useful. It lets us make some basic assumptions like, ok I have W data today, the volume grows at rate X/year, it takes Y machines to handle that today, and based on Y doubling in capacity every 18 months or so I will need Z machines for the future state. Can I continue to scale like this?
Is it accurate, precise, or grounded in solid facts no but its still a nice rule of thumb permits some crude planning around a future with many unknowns.
Repeal the 17th Amendment TODAY! Also Please Read http://www.gnu.org/philosophy/right-to-read.html
exponential growth
y=x(1+r)^t
where x is the starting value, r is the rate of growth (doubling = 100% = 1) and t is a discrete interval (1 = 18 months, 2 = 36 months, etc)
given a starting value of 1000, after 18 months, it doubles.
y = 1000(1+1)^1 = 1000(2) = 2000
after 36 months, it doubles again
y = 1000(1+1)^2 = 1000(4) = 4000
after 54 months, it doubles again
y = 1000(1+1)^3 = 1000(8) = 8000
after 72 months, it doubles again
y = 1000(1+1)^4 = 1000(16) = 16000
this is a straight out of the textbook definition of exponential growth. derp.
The wires are composed of doped silicon, and features of doped silicon are at least several atoms big. It may be made of bunch of atoms of dopants, but they are embebed on a crystal dozens of atoms wide. Also, the wires ccertanly an't work without those dozens of atoms, and another wire can't be as close to share some of those atoms without being connected. For all practical porposes, the wire is dozens of atoms wide.
Why can't /. just anounce a semiconductor breakthrough for what it is? "Smaler wire made of silicon" would make it, for exemple.
And, by the way, Ohm's law holds at the atomic level as well as it holds for big conductors. People learned that by studying organic conductors ages ago. The problem is how to make silicon work the same way. That is what TFA seems to be about (don't really know because it is behind a pay wall).
Rethinking email
It sounds like they've created nano-scale insulated wire, kinda like myelin-coated nerve fibers.
Tiller's Rule: Never use a word in written form that you've only heard and never read. You will end up looking foolish.
A law in this usage of the word is neither a guideline nor a legality; it is an observed fact that is assumed to be true, like the 1st law of thermodynamics.
I got it just before I searched for Cole's Law on Wiki. Groan.
Then I decided to search for it anyway on the off hand chance there actually was a Cole's Law, and it just redirects me to coleslaw. Seems even Wiki is in on the joke!
(Something tells me this may be an old gem somewhere. If the internet is good for something it's recycling old jokes on new audiences...)
My fault I guess, I was still under the impression it still uses the 'processor speed doubles every 18 months' definition (which isn't really exponential, but I will go with the flow).
US scholar Albert Bartlett pointed out the difficulty to grasp ramifications of exponential growth, stating: "The greatest shortcoming of the human race is our inability to understand the exponential function."
Don't beat yourself up too much, it's apparently REALLY hard to understand.
http://www.accountkiller.com/removal-requested
Isn't the first law of Thermodynamics that you don't talk about Thermodynamics?
+1 IDisagreeSoHeMustBeATrollOrAnAstroturferOrAShill
More like a "rule of thumb" then.
I come here for the love
Well, fair enough, but what about Slashdot?
"which makes the conduction electrons more immune to outside influence" How can something be more or less immune than another?
The simple translation of this article is:
"We made really bad nanowires."
All that's necessary to demonstrate this effect is to create a system with enough defects and scattering (aka doping) to make scattering based resistance much larger than quantum resistance. This isn't something I thought was still under debate.
So what is exponential growth then? Given you just dismissed the textbook definition as not being "really exponential".
no, it is just an approximation that is useful. Some materials have negative resistance, others complicated nonlinear functions. ohm's law is not universally true. Many other so-called "laws" are this way,Boyle's, Hooke's, Charle's., etc. Just useful approximations for some common cases with plenty of exceptions and even some things acting the opposite way.
Glad you asked :)
Look at this chart here, from the wiki on Moore's Law -
and compare it to this one here, from the wiki on Exponential Growth (specifically the green line) -
Looks to me like the growth in the Moore's Law chart is linear (in layman's terms = a smooth, steady incline), rather than exponential (again, in layman's terms = a steep increase after a smooth incline).
Does it match the textbook definition? Perhaps not. But it passes the smell test to me.
The chart you referenced has a logarithmic y-axis; a 'straight' line in logarithmic space is not linear; it's exponential. Try plotting the same data with a linear y-axis.
The Y axis on the Moore's law is distorted to make the resulting line linear.
I Cater to the Needs of Stupid People. - from a coffee mug Christmas gift
Except that axis of the Transistor count are already given in exponential order. So, yet it is exponential growth.
It's exponential if your counting in binary.
You don't have to be technically correct when repeating moors law because moors law isn't technical... or even a real law. It's just a guideline the industry grabbed on to as a way to time their delayed release cycle for maximum profitability.
Ah, I hate to be the one to point this out to you, but look at the y-axis scale in the first chart. See? That's what we in the business call a "log scale". The logarithm is the inverse function to the exponential, so when one makes the y-coordinate the log of y, you get a straight line.
If you want to see the algebra, if log y = a t (log is a linear function for some constant a), the e^(log y) = y = e^(a t), the exponential, where the particular log function I'm using (there is one for each possible base) is the natural log. If e makes you uncomfortable, substitute log_10 y = b t 10^(log_10 y) = y = 10^(b t). It's all the same.
You should really work a bit harder on understanding the textbooks and claims like this before ranting. Lots of very smart people have looked at them (and indeed, Moore was a Very Smart Guy) and you have to realistically compute the odds. Which is more likely, a failure of your understanding or that all of these very smart people are wrong? That's not that they can't be wrong -- only a suggestion not to shoot from the hip; do a bit more work -- well, a lot more work -- before claiming that something like this that has been reviewed many times is wrong.
rgb
Even when the experts all agree, they may well be mistaken. --- Bertrand Russell.
Damn, my fault for not reading the axes and just getting a chubby over seeing the charts that matched what I was thinking. Thx everyone for the lesson.
Exactly, the reason we hear about it is that chip companies decided it was a decent way to space out their releases. Develop tech that gives an 8 fold increase in transistors? Don't need to work that to release immediately, focus on tech giving only a two fold increase or watered down step along the path of the 8 fold increase and get that ready for production. That way you already know where to go from there.
"I think everyone understands Moore's law is not a law in the scientific sense, but rather a Coloquialism like "Murphy's law"."
First of all. I'd caution you to show more respect to Mr. Murphy. Second Moore's law became a self fulfilling prophecy the minute the major chip development companies heard it. It's a way of colluding without breaking the law. Got a technology that can give a 10 fold increase, another that could give 2 fold, and another that could give 4 fold... all in the research lab of course so which do you work on making ready for production? If you went straight for the 10 fold you wouldn't be guaranteed to have anything ready to rock in the next 18 months. Thanks to Moores law, your competition feels the same way so while you might compete on your answer to the next 18 month milestone, neither of you are going to skip by much.
Nah, they just won't measure the performance of a computer in terms of chip anymore. There are plenty of specs they can use to convince the average idiot that the new pc is the bigger better one and the old one sux0rs.
1.5 yr is not 1 yr or so. If anything it'd be 2 yrs or so since it rounds up.
It certainly is exponential. Although that's not actually Moore's law, what you stated is exponential.
P = T^2
Where P=Processor power, and T=Time expressed in 18 month units. 2 is the exponent, which makes it "exponential."
"National Security is the chief cause of national insecurity." - Celine's First Law
No that's quadratic.
The form of Moore's law stated was exponential. I went fast, and wrote the formula wrong, it should be:
P = 2^T
i.e. a doubling every 18 months: 1 2 4 8 16
"National Security is the chief cause of national insecurity." - Celine's First Law
The most important part of Moore's Law was it essentially saying that your new toy will be far better than your old one before it even breaks.
I wouldn't put it that way, because then you get into the messy part of whether a 2 GHz computer is twice as good as a 1 GHz computer, not just twice as fast. Honestly, computers have scaled faster than I've been able to scale my use for them. My upgrades are increasingly a matter of luxury rather than need. You can always say you will find new uses, but it's not like people are going to stop listening to music because playing music takes 0.1% CPU power. Or watching HD movies for that matter, now that we have hardware acceleration for that too. Even for games it's diminishing returns as you use more and more processing power getting details right. So even if you keep Moore's "law" going, it becomes less and less significant to people.
Live today, because you never know what tomorrow brings
The most important part of Moore's Law was it essentially saying that your new toy will be far better than your old one before it even breaks. When the rate of doubling gets closer to 10 years, buying a new computer isn't going to be so much as the new toy is faster but rather the old toy broke. Once that driving force is over with, electronic companies will be talking about other ways to produce money in more mundane ways.
Yeah, they'll actually have to make their software run efficiently.