Slashdot Mirror
18 nanometer transistor
chrisr was the first of many to tell us that less than a week after the BBC reported Bell Labs had developed a 50 nanometer transistor, researchers at the University of California at Berkeley have announced an 18 nanometer transistor. Best of all, the team has decided to not patent the design, hoping it will lead to faster acceptance.
If they don't patent it someone else will.
Under US law, only the originator of an invention can claim a patent. If the originator publicly announces the invention and fails to file (for 1 year? 2 years? I forget...) after publication, no patent can be granted to that particular invention.
Kudos to the UC Berkeley people for their farsighted anti-patent actions in this case.
Screw those Damn Berkeley Hippies! :)
Blar.
No. Patent law is very clear on this. Only the inventor can patent a discovery. There may be issues about who is the real inventory, but evidence (publications, witnessed lab books, etc) are used to resolve that if it comes up.
The only good weather is bad weather.
Pretty ballsy move. I expect some beancounters at the university must be just a weeeee bit choked right now.
--
rickf@transpect.SPAM-B-GONE.net (remove the SPAM-B-GONE bit)
"People will pay big bucks for the luxury of ignorance."
No, really??? That's so small!
I wonder how they're going to build the thing for CPU's. Electron beam litography is nice in a lab, but I'm pretty sure the plants are quite far from using it for mass production. And optical lithography is not an option either as .18 is probably the best you can do with visible light. You would need to go to far ultra-violet spectrum I guess. That's also a place where patents could get in the way: when someone find a way to mass produce that.
Opus: the Swiss army knife of audio codec
What is the current size of transitors in consumer electronics/computers?
Proof that, once again, Berkeley kicks Stanfurd's ass (at least in EECS). Now if we can only do the same thing for our football team... :-)
I can wait to wire a couple of these transistors together an hook up a Beowulf cluster on them.
Hey, do you think we can port linux to one of them(to a single transistor)?
Windows sux! I hope Bill Gates doesn't try to force us to upgrade to these new transistors so we can run Windows 2k.
Do you think the NSA can use these transistors to monitor our emails? BOMB, NUCLEAR, IRAQ, CHINA, ALLAH.
What did I leave out?
How much current can this thing actually switch, and how tricky is it to amplify the signal from such a device? IOW, how long will it take to actually *use* this kind of device?
-jcr
The only title of honor that a tyrant can grant is "Enemy of the State."
Correct me if I am wrong, but given:
A: IBM's technology to fabricate LCD displays onto plastic, giving rise pretty soon to flexable and cheap LCD panels
B: All these new transistor technologies, that allow pixels to be made smaller...
Is it just me, or are we living in a very exciting era? 8-)
-=- SiKnight
"There are no shortcuts to any place worth going."
"Be regular and orderly in your life, so that you may be violent and original in your work." -Flaubert
Sorry, transistor miniaturization has nothing to do with LCD's. What it does allow is basically faster cheaper smaller coolor microprocessors/memory, which should be enough for anyone. Of course, basic research at Berkeley's no substitute for proven mass production techniques, but its a start...
Can you open source hardware? :) Or maybe release it under a license similiar to GPL?
Very cool to see this. It looks like it would be easier to manufacture than the 'vertical transistor' recently announced.
With all of the advancements in transistor size, you've got to think that wiring the little buggers together is going to get real interesting. Smaller transistors mean less voltage, less drive strength, and lower noise margins. Narrower interconnect means higher resistance. Try to pack the metal lines too close together and you've got even more noise problems.
Aren't we already at the point where the interconnect is beginning to dominate silicon real estate?
When guys are patenting obvious, or worse, prior art like "multimedia transmitted over the internet" and actually getting people to pay up, while other guys are increasing the cost-effectiveness of the information infrastructure by, oh, lets say a factor of 10, and can't receive substantial returns in support their talent for future risk-taking innovation -- the patent of invention has gone the way of the patent of nobility: It is obsolete.
What made the patent of nobility obsolete was the corruption of the nobility by politics. What, apparently, has made the patent of invention obsolete is the corruption of invention by legalistics.
We still need nobility. In technological civilization, nobility is in the creative act. The problem is the politicians and lawyers have demonstrated they are, as a cultural phenomenon, hostile to true nobility.
The creative act deserves the respect, reward and protection traditionally reserved for nobles.
Fortunately, creators, themselves, possess great power.
Seastead this.
Actually I reported that a French team made some 20 nanometers transistors a month ago but of course it didn't make it on Slashdot, as of course only the US can make innovations, and anything else doesn't exist.
Some devices that need to lower power consumption, and have very high clock speeds (read Intel and AMD CPUs) are built at 0.18 micron.
The 0.25/0.18 is only an industry reference, and not the actual size of the transistors. Some high end manufacturers (again, Intel and AMD) have parts of their 0.18 products etched at 0.14 and even 0.12 micron right now. Motorola's upcoming 0.18 micron copper process (not the one that G4s are made with, which is an hybrid 0.25/0.18) is able to make transistors as small as 0.10 micron.
On the other hand, the nazi answering him in a very thoughtful way did NOT get moderated.
Something is wrong here ...
Just imagine being one of the guys who announced the 50nm vertical transistor a couple of days ago. They must be feeling pretty sick right now...
;o).
Would anybody be willing to fetch the JPEG image of this new device out of their browser cache and stick it up on the web? All the graphics seem to have been removed from the original site (slashdotted I guess
Consciousness is not what it thinks it is
Thought exists only as an abstraction
And in the press release, they indicated that they are looking to reduce the size to one-half of what is is now; and it wasn't in that, "You know, someday we might..." way, it was like a simple statement of fact. Jesus, a 9 nanometer transistor may actually be possible soon; could we actually see working nanotechnology within the next 20 years? Despite the vast number of technological advances that have occured over the past 100 years, the prospect of working nanotech just seems too... William Gibson, if you will. :) Like the man said in The Matrix, "This is a very exciting time!". :)
Deosyne
Surely their decision NOT to patent is merely an open invitation to all those opportunists out there to go ahead and patent the technology?
I have no real idea about what you can and can't patent and maybe other people can't patent it (pls tell me if this is the case) but if they could then the patent would almost surely be saught after by someone else?
In a related story, the New Post reports that the Free Software Foundation has released the text of version one of the Hardware GPL.
Seriously though, while it is an excellent advancement to create a transistor this tiny, I believe the most significant part is that it's not patented. I wish we could see more of this kind of behavior by educational establishments.
-----
Do you even know anything about perl? -- AC Replying to Tom Christiansen post.
You can only patent things to your entity's name that you can be prove were first created in the name of that entity. ie: you can create a patent in your name of something you created, or a company can create a patent of something one of its employees made while on company time. Since Berkely already announced the invention, there is no way some one can insist that they had it first (unless the genuinely did, which must be proved).
-----
Do you even know anything about perl? -- AC Replying to Tom Christiansen post.
Actually, I want the equivalent of a beowolf cluster of these I can put up my nose and instead of using fans to cool it, I can just breathe......
I'm not too surprised that they are not applying for a patent. They are a state university, so the university would hold the rights to the patent. The actual researchers would have their names attached, but they would receive little monetary gain from any licensing (assuming this is the same type of agreement I had to sign at NCSU).
e nsion/policies/patent_policy.html
;~)
In light of the recent discussions on patents and copyrights, I thought many of you might like to see what my alma mater has to say on this matter. The whole document may be accessed at
http://www2.ncsu.edu/ncsu/research_outreach_ext
Anyway, the preamble is a pleasant read for all of those opposed to frivilous/progress-stifling patents applications...
"North Carolina State University is dedicated to teaching, research and extending knowledge to the public.
It is the policy of the University to carry out its scholarly work in an open and free atmosphere and to publish results obtained therefrom freely, limited only by a short time delay in cases in which this is necessary to prepare and file applications. Patentable inventions sometimes arise out of research activities of its faculty, staff and students which are carried out wholly or in part with University facilities. As a public service institution, the University has an interest in assuring the utilization of such inventions for the public good. Protection must be provided for at least some of these inventions through patents and licenses to encourage their development and marketing. Patents and their exploitation, however, represent only a small part of the benefits accruing from either publicly or privately sponsored research.
A portion of the research conducted by the University is supported by government and a portion by private industry. Service to the public, including private industry, is an integral part of the University's mission. As a public institution, the University, in its agreements with private industry or other private organizations, must keep the interests of the general public in view. The rights and privileges set forth in cooperative agreements or contracts, with respect to patents and copyrights developed as a result of research partly or wholly financed by private parties, must be fair and just to the inventor(s), the sponsor and the public. Research should be undertaken by the University under support from private parties only if it is consistent with and complementary to the University's goals and responsibilities to the public."
Although, this document may be copyrighted, I'm not sure
Eric
How many minuses does he get now?
So can anyone answer a couple of questions?
It looks like these suckers will require a couple of poly layers to get the gate to wrap around the channel... will that require any type of mass changes to the fab process (besides going to, oh, I don't know, the x-ray band for the masking)
Since the channel is about 18nm wide, these babies will have a (reletively) massive amount of resitance, as oppsed to the Bell Labs design which has a small gate, but a fairly large channel. Will this effect the charging of the next few gates down the line from them because...
In a nutshell, would someone use these devices in high-performance applications, or would they only be suitable for getting better density on a chip? Oh ya, my knowledge of VLSI sucks, so please be kind with flames.
- Dan
As I understand, the 50 nm transistor was created with a vertical channel, and the 18 nm transistor was created with a gate with a special shape (a fork) in order to pinch more effectively the electron stream, (if I made some mistake feel free to correct me).
:-)
How about a transistor which combines both technologies ?
Bell labs originally created the transistor. It was designed to replace the Tubs on the Phone network in an effort to make service more reliable. At first they held the patent and licensed the technology to anyone who wanted it. Although most U.S. companies in general were happy with the vacuum tube. A small Japanese company saw the potential for making a small (And affordable) pocket transistor radio. That company eventually because Sony.
Later, Bell removed all fees for producing the transistor. Although this was mostly a move to reduce Monopoly pressure from the Gov't.
The inventors of the transistor were happy to see they started the computer age. Their only regret was that the small size of pocket radio allowed teenagers to listen to the radio away from adults and thus contributed to the spread of Rock and Roll music.
I think Bell Labs is likely to patent the process. However, I think they will license it for free.
At UC Berkeley, the inventors get 33% of royalties, the department they work in get 33%, and the Regents get 33%.
Wouldn't a better idea be to patent the design and then GPL it, or something similar to that? Remember, Intel isn't making faster chips to make the world a better place, they're making faster chips because they like downloading greenbits from your wallet.
`/\/\
(^.^)
(")(")
not quite an analog pussy, just a cat that plays with vinyl
Gordon's back again!
This stuff looks nice and fine, and the no-patent is a neat decision. But etching a 3-D gate is years-light away from a normal CMOS process. And for the industry, that's the very problem : how to preserve the HUGE investment in manufacturing equipement.
Moreover, I don't really see what's the great edge of this process versus using SOI (silicon on insulator). What's SOI ? Basically, take an average silicon wafer, polish it, grow a thick silicon oxide on one side (the insulator),re-polish, and then grow a very thin (10nm to 1um) pure Si layer (the silicon), cook it (medium rare) and re-re-polish. Therefore, the wafer process is very different from usual stuff. The ion implantation is also very different (no deep diffusion, etc). The rest of manufacturing (active and wiring layers) is completly preserved.
SOI gives a complete control over the depth of the active layer and avoids current leaks under the gate, the very problem addressed by Berkeley invention. You can even tweak the transistors threshold for ultra-low voltage application.
Anyone has a hint about SOI acceptance in the industry and SOI behavior with very thin gates ?
We won't see any great leap in technology *EVER*. I don't care if they could creat 1-atom sized transistors simply by waving a magic wand over some beach sand - the chip industry, the U.S. government, and marketing would ensure that we'll only see incremental improvements over time. Marketing would be very happy to see Moore's Law grind to a halt. And they will do what they can to stop this sort of technology from seeing the light of day until it becomes an incrementally better technology than what we've got. Then, they will sell it to us for a huge number of $$$, and let the price dribble slowly over time. Then, finally, in about 2025, we'll finally be able to buy something using this technology. This is the way it'll play out. Face it.
Gibson's got nothing. More like Engines of Creation.
To the editors: your English is as bad as your Perl. Please go back to grade school.
There are a lot of potential reasons why your contribution didn't get posted. Maybe you sent it to the wrong address. Maybe the email never arrived. Maybe you submitted a bad URL. Maybe the URL was good, but the server was misbehaving when Slashdot checked it, so it seemed bad. Maybe....
But of course there's only one possible explanation for every problem in the world from the French perspective: some evil American conspiracy. But of course.
So-called "pillar" transistors following just this concept have been made in the past, I think by IBM. The real trick would of course be connecting that device cheaply/efficiently to other devices. Actually, the "surrounding" gate idea being pushed here is really not that new and has been accomplished before (and, what do you know, announced at the same conference), also by IBM but with a lateral SOI type of device. Hmm... maybe I should be working for IBM... Anyway, what probably IS new here is the technque for forming the fin gate. I'm willing to guess it was deposited down the side of some sacrificial material (such as silicon dioxide) so that the deposited film thickness would define the gate width (and channel length) instead of being limited by lithography. This would also mean that while this transistor can be made very small, another one could not be placed very close to it because to do so would require another sidewall to be defined by lithographic techniques (which are cannot draw lines that densely, yet).
I think the moderators just happened to hit "1" on their dartboard. Don't let it discourage you from posting similarly in the future. I'm sure I'm not the only one who appreciated the details in your post.
In the US, only one party is legally entitled to file: the inventor. If two or more parties claim to have invented the same thing, the legal issue becomes the collection of evidence regarding who actually invented it first. Once that has been determined, only they can file. If they wait longer than a year to file after they "publicly demonstrate" the invention, even they aren't entitled to file. In that case, nobody is, and the invention is in the public domain, from the perspective of patent rights. The Japanese and Europeans use the "first to file" system. This means that they can spot a US (or any other nation's) invention and file for a patent in their own country if the real inventor hasn't already done so. The purpose of doing so is almost always to extort "licensing" fees from the actual inventor, who is forced to license his own technology if he wants to enter the market. There are a lot of nuances and subtle differences from country to country regarding who is actually allowed to file at all that address some of the problem, so it's not quite as bad as it used to be. However, the fact that Berkeley isn't patenting at all doesn't prevent some Japanese company from patenting, without any need to claim to have invented it first. Presumably, the Europeans could do likewise, though that may no longer be true.