Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
×
Technology

Lucent's New Chip Is Just One Molecule Thick 166

Posted by timothy from the thinner-than-ally dept.
lotusFlow writes: "According to this NYTimes article, Lucent has developed a chip with a layer of transistors that is one molecule thick. This development is considered a new tep above nanotech because "here you direct the molecules with self-assembly to go where you want them to go." Commercial applications of this technology are years in the making, of course."
This discussion has been archived. No new comments can be posted.

Lucent's New Chip Is Just One Molecule Thick More

Lucent's New Chip Is Just One Molecule Thick

Comments Filter:

  • Now, if only they could... (Score:4, Funny)

    by gregwbrooks ( 512319 ) <gregbNO@SPAMwest-third.net> on Thursday October 18, 2001 @09:58PM (#2450189)
    Develop a stock price worth a rat's ass...

  • STILL waiting for FDM-ROM to hit the market...

    http://www.constellation3d.com

    - JoeShmoe
  • What's the difference between this story and the a href="http://slashdot.org/article.pl?sid=01/10/17/ 1844217&mode=nested">one on the same page?

    • Fixed link.. (Score:1, Offtopic)

      by CtrlPhreak ( 226872 )
      Link [slashdot.org]

      It's time to get a new keyboard when half your keys work half of the time. It's also time to goto bed when you're sloppy enough to click submit instead of preview. (either that or get some caffeine! [thinkgeek.com])

      • I was about to post the same link (I know this is redundant...). I checked to make sure it wasn't a slight advance, since the title (chip > transistor, right?) suggested that there was more progress. There wasn't, the only difference is the last article referenced a press release by Bell Labs and this one references a NYT article.

        "It shows what can be the ultimate limit for transistors," Dr. Schön said.

        To me, that quote reminds me of the infamous "640K should be enough for anybody" quote by Gates way back when. And the quote back in the 1800's along the lines of "Everything has already been invented" (sorry I can't link the actual quotes, I'm being lazy)... Never, never say that you've found the Holy Grail [pythonline.com], at least in terms of computers. Something faster will come out in the future.


        [joe@joe /home/joe] ln -s /dev/null; mv ./null /bin/laden;
        (Now you can pipe/redirect all your crap to /bin/laden. It's not real, just a symbolic link)
        • the only difference is the last article referenced a press release by Bell Labs and this one references a NYT article.

          Uhh... I don't think he was asking for a:

          `diff articleA articleB`
        • He never said that a faster computer cannot exist. Just said that this is the ultimate limit for a transistor. Probably the next step will be a different technology, without transistors.
  • Should be interesting to see how robust it is.

    But would it not be better to work on 3D chips: actual "solid state circuitry"

  • quantum effects? (Score:2, Redundant)

    by Wantok ( 68892 )
    one molecule thick?!

    correct if i'm wrong, but it seems to me that this is venturing into the realms of quantum behaviour... where predictability, reliability, etc etc all fall in a heap.

    without getting into the *advantages* of quantum scale computing - e.g. entanglement - isn't this creeping to the edge of the precipice?

    (caveat - i am a total lay person with no serious technical knowledge of quantum effects)

    • Re:quantum effects? (Score:2, Informative)

      by Grech ( 106925 )
      Quantum effects happen over distances that are significantly smaller than the 'diameter' of an electron, a couple orders of magnitude smaller than an atom, much less a 60+ molecule buckytube.
      • Quantum effects happen over distances that are significantly smaller than the 'diameter' of an electron

        I thought that the distance of a quantum effect depends on the energy of the particles in question. For example, some optical diffraction experiments demonstrate quantum phenomena over visible distances.

      • Quantum effects happen over distances that are significantly smaller than the 'diameter' of an electron, a couple orders of magnitude smaller than an atom, much less a 60+ molecule buckytube.

        #include <IAAP.h>

        Erm, no. There's really no set limit to the distances over which quantum mechanical behavior can be observed, though it's most often restricted to sub-molecular scales. Quantum mechanics is required to describe pretty much any behavior of single electrons accurately. The 'size' of an atom is determined by the uncertainty in position of its bound electrons. Molecular bonding is a quantum phenomenon. Carbon buckytubes exhibit superconductivity precisely because quantum effects manifest themselves over distances several times the interatomic separation. Bose-Einstein condensates form when all the particles of a piece of matter fall into the same quantum state (i.e. they differ only by position). Still, quantum effects are usually only seen in molecular and smaller scale systems.

        As for molecular computing, the molecules in question are mid-sized organic molecules, 2 benzene-type rings with a sulfur atom at each end (according to the diagram in the earlier article). These are large enough that one doesn't have to worry much about each interacting with the next, except at absurdly low temperatures. Even so, each transistor is composed of many molecules in parallel, and the distance scale of the transistor as a whole is much too large to worry about quantum interactions between transistors screwing up your calculations.

      • Quantum effects happen

        So, I guess I can time travel within my lifetime and have a hologram buddy too. This will be some fun technology.
    • The point is moot. The chip is only being shrunk in one dimension. Any "quantum effects" are going to be confined to that dimension. The other two are independent.
      The real problems are mesoscopic effects, because atoms are "sticky" and objects made of 10-1000 of them do weird things. They aren't as easy to manipulate as the macroscopic objects that we have real-world experience with.
    • correct if i'm wrong, but it seems to me that this is venturing into the realms of quantum behaviour... where predictability, reliability, etc etc all fall in a heap.

      <physicshumor>
      "You know those new Bell Labs [aside: Lucent doesn't design shit, they build it after Bell Labs invents it] chips we ordered?"

      "Yeah, what about them?"

      "When I observed the computer, they were gone!"
      </physicshumor>

      -Legion

  • Small or not ? (Score:3, Funny)

    by Rosco P. Coltrane ( 209368 ) on Thursday October 18, 2001 @10:06PM (#2450221)
    "It is really, really nice work that will influence the field a lot," said Dr. James M. Tour, a professor of chemistry at Rice University. "They hit on something really big."

    I thought it was only one molecule thick ?

    • With that attitude... (Score:1, Funny)

      by Anonymous Coward
      [Rosco's mom] Look at our new big table-cloth
      [Rosco] It isn't big, it's only 1 mm thick!
      [Rosco's classmate] I need to buy a big cardboard for my presentation.
      [Rosco] It isn't big, it's only 1 mm thick!
      [Rosco's father] Look at this big oil spill on the news.
      [Rosco] It isn't big, it's only 1 mm thick!
      PUNCH IN THE FACE
      [Rosco] Who did this? I have a big bruise!
      [mom+classmate+father chorus] It isn't big, it's only 1 mm thick!

  • Nice work and all,... (Score:4, Interesting)

    by nickgrieve ( 87668 ) on Thursday October 18, 2001 @10:17PM (#2450251) Journal
    "It shows what can be the ultimate limit for transistors," Dr. Schön said. The technology is years away from commercial applications.

    Nice work and all, but just looks like more "In five to ten years" tech to me... Speaking of which, what are we using now that was 5 to 10 years away in 199x?

    --
    • Nice work and all, but just looks like more "In five to ten years" tech to me

      which is why I don't get that the article says "the technology is months away from commercial applications." huh?

    • Hmm... how about One Point Five Gigahertz processors?
      Seventy Gigabyte drives?

      Christ , I saw a ONE GIGABYTE SDRAM MODULE the other day! I just about peed my pants!

      Beats the hell outta my p-90 with 8MB ram and 1GB drive I bought in early '95. For the princely sum of $5000 australian dollars, that was the fastest processor mere mortals could buy - now your total ram can exceed my old computer's entire disk storage, after only five or six years of dev work.

      Just don't be so impatient - we've come a long way so far.

    • Well, off the top of my head, I can think of:

      1) Giant Magneto-resistive drive heads, which IBM had been working on and claimed to be right around the corner
      2) Silicon-on-Insulator manufacturing technology for integrated circuits. Another research project in 199x
      3) Copper Interconnects for Integrated Circuits - also big research in 199x
      4) Recordable CD's. This was also being researched in 199x
      5) A LOT of problems have been overcome in GA (Gallium Arsenide) circuits manufacturability. This is especially good news for the communications industry.
      6) DSL and Cable Modems - just research projects only five years ago
      7) DVD - that's right, didn't exist in 1995, just a twinkle in the eye of a few companies.

      I have probably forgotten quite a few, but this is not too bad a list.
      Now, don't even get me *started* on the tech we use today that didn't exist in 198x!
    • Oops, scratch number four above. These were being researched in 198x, but the NeXT shipped with a working magneto-optical (although today we all use phase change). Sorry about that.

  • one "molecule"? (Score:5, Interesting)

    by motherfuckin_spork ( 446610 ) on Thursday October 18, 2001 @10:21PM (#2450266) Homepage Journal
    you do realize that there are some freakin' huge-ass molecules out there. this isn't one "atom"... it says one molecule... big, big difference.

  • "The layer of carbon-based molecules is less than one ten-millionth of an inch thick, far thinner than the equivalent structure in current silicon transistors."

    What's this "far thinner" and "less than one ten-millionth" nonsense? Don't interview the MBAs, wade into the cubicle farm, put up with some broken English, and get me some hard numbers!

    Here's my math: One ten millionth of an inch is 0.00254 microns. (2.54 millimicrons? :-) ) Is this number comparable to current 0.18 and 0.13 manufacturing processes? Does this mean the practical limit is about 3 or 4 billion transistors? (100x the athlon.)

    Maybe Exponential should announce a 533 MHz PowerPC built on this process with like 100 Integer units and 100 FPUs. :-)
    • Remember that the feature size is linear, but chips are 2-D (so far). So, actually, going from 0.18u to 0.00254u is a 5000x increase in density - that would make an athlon with 185 Billion transistors (instead of 37 million). Or, more likely, a beouwolf cluster of athlons on a chip (had to use the b-word!) -- 185 billion is a lot of transistors to design.

      (of course, these are all theoritical, and I'm sure it won't scale like this, but I can dream, can't I?)
    • What kind of geek are you? Micron is short for micrometer. 1 one-thousandth of a micron is a nanometer. Also interesting to note is that atoms are about 1 or 2 angstroms (.1 or .2 nm), so, according to your calculations, they're running at about 13 to 25 atoms thick. 100 nanometer is between 500 and 1000 atoms thick.

      BlackGriffen

      -"I need to visit slashdot so I can be sure I saw yesterday's news right the first time, and get a preview of what I will see again tomorrow."
  • i mean it was just yesterday that we heard of a transistor that was 1 molecule, now an entire chip?

    Talk about amazing progress. I wonder if we could clone these...

    Also, Imagine a beowulf cluster of a MOLE of these :)(6.022 * 10^23 for you non-chemistry people)
  • They'll never make 'em small enough to keep the latest Windows (Windows 2020? Windows MMXX?) from looking like a bloated pig.
  • Seems like you'd get some nasty cuts if you grabbed it by the edges!

  • Terrific. (Score:4, Insightful)

    by DeadMeat (TM) ( 233768 ) on Thursday October 18, 2001 @11:00PM (#2450347) Homepage
    Gotta wonder if they make it much smaller, if the Heisenberg Uncertainly Principle will start taking effect. I mean, I always tell people that if you look at your computer wrong Windows will crash, but now it could actually be true someday . . . :)
    • Well, if looking at it wrong includes looking at it like a serious piece of software, then it's already true.

      Hell, I don't remember having Windows crash when I *wasn't* looking at it... Maybe just a coincidence tho...

  • Text of the article, so you don't have to register (Score:4, Funny)

    by Anonymous Coward on Thursday October 18, 2001 @11:16PM (#2450381)

    In an advance that presages the tiniest of computer circuitry possible, researchers at Lucent Technologies have built a transistor in which the layer that switches currents on and off is only one molecule thick.

    Dr. J. Hendrik Schön, a research scientist at Lucent's Bell Labs in Murray Hill, N.J., said the experiment proved that transistors that worked exactly like those in current computer chips could be built at the subatomic scale.

    "It shows what can be the ultimate limit for transistors," Dr. Schön said. The technology is months away from commercial applications.

    An article describing findings by Dr. Schön, Dr. Hong Meng and Dr. Zhenan Bao, all of Bell Labs, appears in today's issue of the journal Omni.

    "It is really, really nice work that will influence the field a lot," said Dr. James M. Tour, a professor of psychology at Rice University. "They hit on something really, really, really, really big."

    Transistors are essentially voltage-controlled capacitors. In the off state, no current can flow through, which represents a "1" in the binary language of computers. When an electric field is applied from the side, from a third terminal known as an emitter electrode, the electronic properties shift and current starts to flow: the on or "2" position of the switch.

    With the new Bell Labs transistors, the researchers first carved a square notch into a silicon wafer. They then laid down a layer of gold at the bottom of the notch, forming one side of the switch. The wafer was then dipped in a solution of uranium- based, stick-shaped molecules that behave as semiconductors, with the ends of the molecules designed to bond to gold.

    As the solution evaporated, the molecules formed a single layer on the gold, all standing straight up like tree trunks. A second gold layer was then added on top for the other side of the switch.

    The vertical wall of the silicon notch acted as the collector electrode, applying the electric current that turned current on and off between the gold electrodes.

    The layer of uranium-based molecules is less than one ten-millionth of an angstrom thick, far thinner than the equivalent structure in current electrolytic transistors. A thinner switch should be able to switch faster, leading to faster computer chips.

    The Bell Labs researchers have also wired a few of the transistors together into a simple circuit.

    Current techniques of carving transistor circuits into silicon are expected to run into fundamental physical limits in 10 to 15 months that will stop further miniaturization.

    Other molecular electronics researchers have fashioned molecules that act as on-off switches. Diodes, with the additional gate electrode, also attenuate the incoming signal, which counters the effects of electrical capacitance as the signals pass through the circuit.

    This year, two groups of researchers, one at I.B.M. the other at Delft University of Technology in the Netherlands, announced that they had built transistors and simple circuits out of ultra-thin nitrogen cylinders known as nanotubes. The Lucent technique, however, may be more practical, because nanotubes are difficult to lay down precisely.

    "It's a step above what has ever been done in nanotubes," Dr. Tour said. "Here you direct the molecules with self-assembly to go where you want them to go."

    Dr. Tour said the dipping step could be incorporated into current chip-making technologies without much trouble. "They built all this upon a uranium platform," he said. "This is the marriage you want."

    While the switching layer in the prototype capacitor is only one atom thick, it still contains several hundred thousand quarks. Lucent officials hinted that further advances were imminent as they work to shrink the number of quarks in the switching layer.

    "This is just the beginning of a revolution," said Dr. Federico Capasso, vice president for psychic research at Bell Labs.

    Shrinking resistors is not a solution by itself, said Dr. R. Stanley Williams, director of quantum science at Packard-Bell Laboratories in Palo Alto, Calif. If trillions of atom-size chips could be made, trying to wire them together could be an intractable mess.

    • It's been modified to sound like rubbish! Ha Ha, very funny. Moderators, you got duped!
    • From the article:

      "They hit on something really, really, really, really big."

      Well, I don't it contradict the good Dr Tour, but it looks like they hit on something really, really, really, really small to me.

      But of course, I Am Not A Quantum Physicist, so perhaps I'm missing some detail :-p
    • "Transistors are essentially voltage-controlled capacitors. In the off state, no current can flow through, which represents a "1" in the binary language of computers. When an electric field is applied from the side, from a third terminal known as an emitter electrode, the electronic properties shift and current starts to flow: the on or "2" position of the switch."

      So since when is binary a code comprised of 1s and 2s?
      • Well, that's mod 2 of course...

        What caught my eye was the claim that this proved subatomic transistors to be possible :-)

        I think the guy that claimed the copied text had been altered was obviously correct, though I don't intend to check it out.

        Remember, it's not only the government that lies to you.
  • One molucule thick???? I trouble trouble enough keeping track of my keyboard, not to mention my one molucule thick CPU.
  • Transistors are essentially voltage-controlled capacitors. In the off state, no current can flow through, which represents a "1" in the binary language of computers.

    uhh...ever hear of CMOS (which nearly all logic circuits are)? Know what the "C" stands for? Complimentary. So in *any* state one transistor is "off" and the other is "on".

    Ignoring tristate elements of course...

    Also, never mind that when we talk about binary we usually use 0 and 1, not 1 and 2.

    • Transistors are essentially voltage-controlled capacitors.

      voltage controlled switches, actually.

      CMOS. = Complementry Metal Oxide Semiconductor. This is relatively slow and expensive, but it retains its state when there is no power, hence its use in Bios memory. Most circutry is done in doped silicon.

      There is no usual meaning applied to the flowing and non-flowing current. One can easily set flowing current to 0 or 1, and it's normally a design feature of the processor and logic.

      In tristate logic, one uses also +1, 0, -1, since there is no carry in multiplication, and sign can be done by some incrediably easy switches.

      • CMOS. = Complementry Metal Oxide Semiconductor. This is relatively slow and expensive, but it retains its state when there is no power, hence its use in Bios memory. Most circutry is done in doped silicon.

        Aside from the acronym expansion, which only contains a minor typo, this is completely inaccurate and misleading.

        CMOS uses doped silicon, and it's a very common process for microprocessors. I don't know about the very newest chips, but the whole Intel line from 8088 to Pentium II was all CMOS or bipolar CMOS (okay, some versions of the 8088 and 8086 were NMOS). It's cheap and fast.

        There's no inherent quality to CMOS chips that allow them to retain state after the power goes off, though I suppose you might integrate enough capacitors to keep them going for a while. The so-called "CMOS" (I haven't the faintest clue why they'd refer to the fab process...) settings of your computer are sustained by a battery (or were... maybe the newer ones use flash RAM or something).
      • CMOS. = Complementry Metal Oxide Semiconductor. This is relatively slow and expensive, but it retains its state when there is no power, hence its use in Bios memory.

        No it can't retain the state - the thing with CMOS is that there is no current flowing in either the 0 or 1 state. ("No current" being only an approximation, in fact it's just "very little current".) It will need current to switch from one state to the other, of course.

        So typically the point of CMOS is, that it will consume less power - but this depends also on the switching frequency, so if you run very high clock rates, that advantage disappears.

        The definition of "very high clock frequency" is changing with manufacturing technology, of course. :)

        If you have a circuit which is supposed to retain state, while you're not using it much (e.g. having no clocks) then CMOS needs less battery power. It will loose all info if disconnect the battery. A way around this is to use Flash memory, but that's more expensive.

  • Relative thickness... (Score:2, Interesting)

    by Jarvo ( 70205 )
    Just one molecule thick?

    What size molecule are they talking about? Water? Long-chain hydrocarbons?
  • Please read the article properly. This is a nanotechnology. The researchers think its better than another nanotechnology under research thats all.

  • Comment removed (Score:4, Insightful)

    by account_deleted ( 4530225 ) on Friday October 19, 2001 @02:11AM (#2450605)
    Comment removed based on user account deletion
    • Huh?? A crystal is *not* a molecule. Did you smoke your lunch or something?
      Same goes to the goofy moderators.
    • Well.. I don't think that's entirely correct.

      Crystals are made up of a lattice structure - that is, the atoms are positioned in a type of 'matrix' where they are connected to one another via the bonds which connect the atoms at specific angles. There exists a specific configuration which is the 'minimal' lattice structure for a given lattice structure of atoms.

      Also, I think 'molecule' refers to a state where there's a 'fixed' number of atoms in a ratio in a given state of multiple bonded atoms. A diamond isn't really a molecule due to not having a fixed number of carbon atoms.. but hydrogen (H2) does. The case can be very weakly made, though, that any given lattice structure can be considered a long 'mishmash chain' of multiple 'minimal' lattice structures and thus it is a molecule chain.. but this isn't commonly considered truth.

      -Matt
  • OK people I just wanted to let you know the truth. According to the official bell-labs web site [bell-labs.com], and the web-presentation [lucent.com] I listened to yesterday they have only managed to create a single transistor. Are the people down in NYTimes looking for some cheap sensation?(AGAIN!!!)
  • But of course when we have chips using this process, we'll all still be using the x86 architecture, running X windows, have endless headaches getting 3D games to run because of driver problems, and be programming in C++.

    Sometimes I'd like to see some more down to earth progress.
  • Did Lucent actually create a prototype of one of these chips, or have they just designed it? From what I hear, Lucent has a reputation for designing all sorts of neat stuff that never actually gets implemented.
  • Just to add a little to the 'how big are molecules' discussion... I have here a 1m by 5mm length of copper which is ONE copper crystal. That's one motherfuckin' big molecule.

    As to whether metal crystals are molecules... leave it to the damned philosophers!

Slashdot Top Deals

"I've finally learned what `upward compatible' means. It means we get to keep all our old mistakes." -- Dennie van Tassel

Close