Forgot your password?

Gallium Arsenide Semiconductors on the Horizon 119

Posted by michael
from the miracle-tonic dept.
Masem writes: "According to this Chicago Tribune article, Motorola has developed a cheaper solution for putting gallium arsenide on top of silicon in order to allow for better chip designs with speeds nearly 40 times what silicon only chips would allow. While it was well known that gallium arsenide addition was favorable, it was also very expensive; Motorola's new process (covered by 200+ patents) should keep the chip prices low when these new designs are released in 2 years." The AP says they've applied for 270 patents.
This discussion has been archived. No new comments can be posted.

Gallium Arsenide Semiconductors on the Horizon

Comments Filter:
  • What great news ! (Score:2, Interesting)

    by The_Jazzman (45650)
    There are two schools of thought when it comes to computers and the world around us.

    One might say that computers and ourselves are becoming too involved with each other, us being dependant on the computers.

    The other says that each technological breakthrough is a good thing, advancing us to a greater extent each time.

    I subscribe to the latter view.

    Taking in this point, cheaper chips are something that we should really be striving to produce. If we could come up with microchips so cheap that they cost fractions of pennies yet had the processing power of, I don't know, an Atari ST (8Mhz IIRC) then think of the places we could put them - and cheaply !

    For example, The London Underground 'tube' network in England is currently trialing a new ticketing system whereby rather than having a cardboard ticket with a magnetic stripe down one side, they issue tickets which have so-called 'smart chips' inside them.

    The flipside is good for LU - think how much extra effort it would be to forge a ticket.
    For the everyday train user it makes life just that bit easier. No more scrabbling around for your ticket, as long as it's somewhere on your person you'll able to walk straight through the ticket barrier without having to even think about it.
    • The ultimate goal will be for pc's to be a commodity like ballpoint pens are, where if you need a new one you go and fetch it from the stationary cupboard. ;-)

      Actually this is probably more a requirement for wireless computer terminals in various form factors being like that, so you have multiple ways in to your PC, some of which are almost disposable.
    • The flipside is good for LU - think how much extra effort it would be to forge a ticket.

      Untrue. It would make it even easier (and cheaper). Instead of requiring a relatively expensive gadget to read and write magstripes from a PC, the only thing a guy now needs to forge a ticket is a common circuit board with a low-profile PIC, flashed with appropriate code to emulate a real card. While currently not too popular, some crafty freaks have done this with satellite receivers and those famous 'H' cards.

      If this whole "Smart Card" craze spreads to more uses, then today's bleeding-edge hardware crackers will be tomorrow's mainstream neighborhood pirates. Just like Pay-TV blackboxes were "the shiznit" fifteen years ago.
    • Even after reading this a few times, I'm still not sure how you went from an article on Gallium Arsenide chips to:

      "If we could come up with microchips so cheap that they cost fractions of pennies yet had the processing power of, I don't know, an Atari ST (8Mhz IIRC) then think of the places we could put them - and cheaply !"

      And then it was moderated up as "interesting"? How about "offtopic?" Bizarre.
  • Now where's my 2GHz G4!?
  • The articles I've seen on this aren't very detailed as to how the technology actually increases speed, but I would have to guess that the new material allows for less logical gate latency which then allows for faster chip operation. Anyone know if I'm close?
    • the carrier mobility is much much greater in GaAs than in Si.
      however CMOS process is not economically feasible because GaAs has no good clean native oxide to isolate the gate. thus GaAs is usually used in bipolar circuits. maybe the IBM technology has something to do with this?
      silicon epi on a GaAs trench or somesuch?
      i haven't read the article... i'm guessing
      • Having worked in manufacturing at a epitaxial silicon wafer plant, I remember the horror stories from the first time they tried to do GaAs wafers. The bottom line was that the yield just wasn't there. Too many defects showed up due to handling problems which made the process economically unfeasable. They were brittle as hell to the point where the end effectors were causing problems. I'm guessing thats what the 200+ patents are for. They must believe they have a process that can work, but it will be a while until a factory proves it.
  • This is excellent news!

    Speed in future communication equipment WILL depend on gallium arsenide (GaAs), after all. It's well known that electrons travel five to six times faster in GaAs than silicon. I can't wait to see how fast Motorola's new chips become with this addition.
    • I agree think of the amazing new pda's that they could put out with a new generation of GaAs dragonball/coldfire processor not to mention what it could do for the G4 and beyond. Maybe this will be the boost that the faltering semiconductor industry needs to get it back on its feet.
    • That's true, but the holes in GaAs are actually slower: mobility of holes in Si is about 450 cm^2/V*s and about 400 in GaAs (both numbers for intrinsic material).
      • Um, this may be an ignorant question(IANAEE), but how do the holes and the electrons have different speeds? Aren't holes defined as the lack of electrons, and would not a moving electron leave a hole in it's place at exactly the same rate as which it travels?
        • That's not an ignorant question; actually, it's a really good question...My best attempt to explain this without using the word "degeneracy": The limiting factor on electron velocity is the scattering between the conduction electrons (the ones that are basically free). A free electron is a free electron, so when they collide and scatter, that's that. However, holes are what's left behind whenever electrons are freed; the electron that jumps in the direction counter to that of the hole isn't really free in the same sense as a conduction electron, and it has to go to different levels closer to the nucleus. So in addition to "hole-hole" collision scattering, analagous to the aforementioned electron-electron collision scattering, there's this confusion as to where the hole is supposed to come from, and energy is lost in making these "inter-level" transitions. Hope this makes sense; it's not a great explanation.
          • It does make sense, thanks a lot. Degeneration does seem to be the right word though. Inter level transitions would have to take some energy, leading to differing hole/electron effects as holes and electrons propagate to different states? Is that it?
            • The fact that at the top of the valence band there exists degeneracy (i.e. equal energy) between different hole types; i.e. the energy to change types at zero momentum becomes zero so there is all kinds of intermingling when holes accelerate to nonzero velocity again.

              Sometimes weird things happen like in lead sulfide (PbS) where the hole mobility is greater than the electron mobility, but this is because the electron mobility is really low, not vice versa.

  • Great, now maybe the G4 really will be as fast as the P4, or even faster!
  • Kudoes, but (Score:3, Insightful)

    by 4of12 (97621) on Tuesday September 04, 2001 @10:00AM (#2251129) Homepage Journal

    Am I the only one that finds it just a little bit of stretch to talk about about fantastic technology that helps to make GaAs cheaper for real life applications on the one hand -- and then mention 200+ patents on the other hand?

    I know, I know, that the hope of financial gain provides the dollars for this kind of research, but let's be real: it won't be that cheap.

    • Also remember the other, very important aspect of patents. Yes, they have exclusive rights to use a particular process for the life of the patent, but they also have to release the details of that process. This means that other companies or individuals with interest and a bit of money can study the process and look for ways to improve the process (thus changing it, allowing them to patent _their_ process) and hastening the development of the technology. Granted, it doesn't always, or even usually work this way, but the idea is sound, if it were applied the way that it is supposed to be. The patent promotes further study of the technology so that cheaper GaAs semiconductors may sooner become a reality.
    • Re:Kudoes, but (Score:1, Insightful)

      by Anonymous Coward
      Patents are not that evil. It's the way they are used that is.

      As long as all they want is recognition for what they have accomplished and to recoup their (presumably large) investment in developing the technology by licensing it under acceptable terms, I have no objection.

      Patents are evil when they are used to prevent competition, and software patents are almost always about this. Hardware patents are more often used for licensing. AMD is using copper wiring in the Athlons, although there is an IBM patent covering it. But meanwhile IBM goes forward and is going SOI now and low K dielectric next. Yes, this gives the patent holder some advantage, but only in the short term, which is still fair.

    • Patents work pretty good in the real world. The problems with them relate to the patent office, and how they are awarded, overly broadly and ignoring prior art. They work far better than copyright law does.

      As to the licensing of the patents, a maximally high cost per unit is not always the best way to go. There are many patented products that have withered on the vine, due to poor marketing of the licenses, but that isn't a fault of the patent process. Remember, that unlike copyrights in the Disney age, patent protection eventually expires.

    • he was just speaking of making them :

      a) worthwhile to produce on a performance gain v. cost increase comparison.

      b) being affordable once produced.

      i know it's hard to fit a single thought in when you're busy trying to get first post, but c'mon..
  • GaAs (Score:1, Funny)

    by Anonymous Coward
    Hmm, I remember an article in Scientific American at least ten years ago which contained the quote
    Gallium Arsenide, the technology of the future.
    Always has been, always will be.

    This still seems to be true.
  • Motorola's new process (covered by 200+ patents)

    hmm... which means we'll be forced to buy these from Motorola for a few years...
  • I don't think so.

    A major limiting factor for CPU design today is wire delay. Electricity runs over silicon with third speed of light (I think, something in that range, anyway), so you can't speed that up more than 3 times (and even that is highly unlikely). If the gate delay being reduced by 40 times, we won't get chips that are 3 times faster, using the same design, IMHO.

    Though this would be quite an improvement :)

    • this has nothing to do with GaAs tech.
      the wire delay is all above the device layer in the interconnects. it's because the devices are so dense we need more and more metalization layers and chip real estate balloons so the manhattan delay skyrockets. this is why it's important to go to 3d. :-)

      GaAs has a higher carrier mobility than Si, so they are able to switch faster. it says nothing of the circuit being able to switch with it (due to capacitance, resistance, etc).
      but as to wiring delay, IBM switched to copper, which helps. they also are using low-K dielectric(s) which lowers the parasitic cap.
      and they have even gone to using metal gates rather than poly-silicon.
  • Convex, anyone? (Score:2, Informative)

    by Apotsy (84148)
    Gallium arsenide chips have been around for a long time, but as the article says, they are limited to niche applications due to cost. Still, there was one company which actually shipped a mainframe built on GaAs chips -- Convex. It's actually kind of hard to find info on them these days (they were swallowed up by HP in the mid 90s), but this EE Times article [] has a bit of info.
    • Terra (before they bought Cray and it's name from SGI) made their first shipped computer with GaAs.

      Heres a link l

      • Ah, I should have said "at least one". But if I recall correctly, Convex was the first, were they not?
      • Re:Convex, anyone? (Score:2, Informative)

        by morcheeba (260908)
        Ah! That explains it. I knew Seymour Cray was working on a GaAs computer -- he had renounced silicon and was assembling some of the world's best GaAs equipment. So, that was bought by Tera []

        Here's the interesting part of my post: Tera replaced the 24 GaAs chips for one CMOS chip. Here's their blurb from the website []:

        Early MTA systems had been built using Gallium Arsenide (GaAs) technology for all logic design. Today, GaAs parts are predominantly used in cellular phones, not high performance computers. As a result of the semiconductor market's focus on CMOS technology for computer systems, there is little support for GaAs technology.

        Cray's transition to using CMOS technology in the MTA will occur in stages.

        In the first stage, a single CMOS MTA "Torrent" microprocessor replaces 24 GaAs ASICs that had represented 16 different ASIC designs. Torrent chips support up to 128 virtual processors, or threads, and will run at least as fast as today's MTA processors. A Torrent chip requires 50 watts of power compared with 1,000 watts for the GaAs design. The Torrent processor board requires 1,025 connections versus 14,400 connections on the GaAs board.

        Somewhere else on their page they say the system is "Water cooled at 4KW per processor". So, even with the reduced-power CMOS, they are putting out a lot of heat!
        • Ah! That explains it. I knew Seymour Cray was working on a GaAs computer -- he had renounced silicon and was assembling some of the world's best GaAs equipment. So, that was bought by Tera []

          Actually, if memory serves, Tera finished their first MTA computer before they purchased Cray. Seymour, no doubt, had plans of his own before he died, but Tera effort was home-grown. When I was researching Tera as a stock, Tera wanted to switch to CMOS for the cost savings involved. The heat was a problem, but the costs of GaAs was the primary motivation to switch according to their fileings with the SEC.

  • So, these crystals can emit light as well as semi-conduct. It sounds to me like this could possibly help keep processors cool. That is, if there were somewhere for the light to go...

    Hmm... I'm thinking some colored filters and a rotating mainboard and we have ourselves some disco-qauke III action!

    • In order for the light-emitting property to provide cooling to the processor, GaAs would have to be able to convert heat energy directly into light energy. Sorry, that's not how it works.

      I believe GaAs works just like light-emitting diodes; LEDs don't get hot because electrical energy is converted directly to light, but at the same time there is no "cooling" of the LED taking place either. I don't know of anything that can convert heat directly into light. (Unless you count blackbody radiation given off at optical wavelenghts by very hot objects -- let's hope your CPU never gets that hot!)

    • Sorry. (In,Ga)(As,P) LEDs can emit in the visible range, but not GaAs.
  • Back when I was a hardware guy, GaAs was a last
    resort. A GaAs PAL device was 30% faster than
    anything else, but it was also expensive, flakey,
    hot, only available from one manufacturer, and
    suffered chronic yield problems. I saw more than one product suffer in the market because of problems acquiring the single GaAs device that it used.

    It looks like they're going to fix the expensive
    issue; I hope that the other problems are addressed
    as well.
    • yeild problems can almost purely be addressed by the wafer size.
      i don't know when you were a "hardware guy" but GaAs technology is quite mature and this will help the yeild as well.

      one big problem i still see is the power consumption which is what makes them hot. no cmos = power sucking monsters. can't build CMOS with out a good oxide/body interface.

    • Correct me if I'm wrong, but compounds containing Arsenic are poisonous to human beings (and probably plants and animals, too.) Isn't this a classic case of a gilded technological breakthrough where the deadly environmental effects are not realized until almost too late? I mean, with the vast amount of computer hardware being used in the world today, there's bound to be problems in a few years...
  • III-V in the UK (Score:1, Interesting)

    by Anonymous Coward
    For GaAs (III-V) research in the UK, check out the EPSRC UK central facility webpage [] at the Electrical Engineering department [] of Sheffield University.
  • How will this help the rest of us, the masses that don't use motorola technology? (especially when you consider the patent stuff) From what I can see, if motorola can make cheaper chips, those that don't want to use their technology will have to pay a higher premium for their chips until someone comes up with yet another (patented?) method to make things cheaper/faster.
    • You save billions -- since you don't have to do the R&D, hire all the talent, build the factories and labs. You also have zero risk -- it's not you fighting for your life against Intel.

      As for embedded processors -- you won't suffer, because Motorola accounts for a monstrous amount of those. Chances are you've used 15 motorola products today already.

      As for computer chips -- Motorola has been ahead of Intel for years. No secret that risc is better than cisc. Intel chips are faster, but they're the size of a shoe-box, they sound like DC 10's, and you can saute mushrooms with their waste heat. Motorola chips are small, fanless, and you can just about run them off a clock battery. If you haven't switched already, then it can't be bothering you too much.
    • They're licensing the tech to other corps. like Intel.

      Motorola only has an interest in underpricing Intel and AMD for this tech if it intends to take those two on in the desktop proc. market. I doubt Motorola has a desire to do so. It has quite a nice, niche market now. Competition is expensive. ;)

  • Of course, that's probably the switching speed of GA transistors, not overall performance gains.

    Does anyone know what percentage of time in the typical processor is spent waiting for transistors to switch versus simple speed-of-light propagation delays, or any other bottlenecks that this doesn't cover? In other words, how much bottom-line clock speed increase would this be likely to give?

    • Re:"40 times faster" (Score:3, Informative)

      by SysKoll (48967)

      Excellent point. The propagation delays are now about 50-70% of the clock cycle of a modern digital chip at the current speeds of several hundred MHz.

      So any improvement of the semiconductor commutation speed is just a "nice to have" technology these days. Think of it. Assume that your chip spends 70% of its time waiting for signal propagation. Even if you suddenly get your transistors to switch instantly (that is, infinitely fast), you'll only increase the speed of a cycle by 100/70 = 1.43, or 43%. And then no more improvements.

      That's why the biggest performance increases will now come from breakthrough in signal propagation speed: Copper wires, low-K dielectric, and more layers for denser circuits.

      -- SysKoll
      • I guess the next logical question is how fast do signals currently propagate as compared to the speed of light?

        • Re:"40 times faster" (Score:3, Interesting)

          by SysKoll (48967)

          It's a good question, and the answer is "roughly c/3". But it's not the whole picture.

          Signal propagation in chips is not limited just by the speed of light. You have leaks due to line capacitance, which also induces coupling (crosstalk) between adjacent lines. If you send a nice square pulse on a 3-mm long straight metal line crossing half a chip (I've seen it!) you'll get an ugly, slow-rising pulse full of parasites picked by crosstalk on its way. And, oh, it will also bounce so badly that you better be prepared to sustain NEGATIVE voltages.

          Want more fun? Get a 500-MHz signal on a metal line, and have the line do a sharp 90-degree turn. Everything then happens as if most electrons you send miss the turn and keep moving on their trajectory as bullets from a railgun. Not only will your signal be badly attenuated, it will also induce a crazy crosstalk in anything near that 90-degree corner.

          See why chip designers become crazy? Sometimes you wonder how something as simple as an electron can be such a devious little bastard. :-)

          • Whee!

            Hehe. At work we recently had some serious XT/coupling problems after moving to .13um. Fun stuff to debug. Nothing beats having certain opcodes only work after a nop and lacking a good explanation as to why (why, dear god, why? :)

            Anyway, OT, you sound like a chip-designer (I'm a firmware guy, but end up helping our LSI guys when we get new chips), so I was wondering if you had any tools to recommend for finding these vicious bastard couplings? We have one, but it blows, and with a 10-million gate chip its difficult for the layout engineers to catch everything the tool missed.

  • Wrong Answer (Score:1, Insightful)

    GaAs chips running up to 40x faster than straight silicon? What good is a CPU that can run at 30-40GHz to a computer that still uses slow IDE hard drives?

    Perhaps the computer makers should push for faster, cheaper disk/memory tech instead of ever-faster CPUs.

    Cheap SCSI, anyone?
    • And if it does, then you don't have enough RAM and you're using your HD for virtual memory.

      It also sounds like you need to check out advances in other areas besides chips - such as serial ATA (600MB per second is promised)

    • I'm tired of all these Victorian storage devices with their mechanical pieces whirling like some kind of gyroscope. I want solid state storage! Everything else in a computer is solid state; how come my storage is primitive?
    • I agree with Processor speeds as they are the average user and even the power user arent using the added speed. Our delay is coming from our media storage. Consider the fact that most memory and processor run instructions and do look ups in the nano-second range and then step back and look at average hard drives that run in the millisecond. You are taling a factor of a 1000. Computers today are still limited by the time it takes to pull information from the harddrive. Memory can only take you so far. We need to get out media storage up in par with the rest of our computer technology
  • Sorry Folks. (Score:3, Interesting)

    by clark625 (308380) <> on Tuesday September 04, 2001 @11:14AM (#2251361) Homepage

    Nothing to see here. Move along, please.

    Okay, this just happens to be the research area I work in--and I know full well the problems associated with getting high quality GaAs on Si. It's not nearly as simple as it sounds. So, it appears that Motorola found a "magical" insulating layer to put between the Si substrate and the GaAs layer. Wonderful. But it won't ever be anything but a novelty.

    Here's why: In industry, everything is driven by economic margins. Plus, the pure Si industry is now very mature and they will not simply add new machinery to their processes that screw up their entire production line. That makes sense, really. Why on earth ruin a perfectly great production line just to toy around?

    The other great point is final production cost. There is no way the pure Si industry will adopt a single step that is far costlier than the rest of their production line combined. Then add to the fact that those industries are adverse to any step that may slow down their production runs or cause unnecessary problems.

    Sorry, people. If you want GaAs on Si, there is only one way that it can be made which will result in something the Si industry is not too adverse to. That means epitaxial growth of any buffering layers followed by high quality GaAs growth. The biggest problem that still hasn't been worked out is how does one go about making proper interconnections? Also, the buffering layer can be very conductive--and that is sometimes very hard to control. Motorola has got their heads up where it doesn't belong if they think the world is going to go crazy over this.

    • By this reasoning vacuum tubes would be the logic element in today's computers.
    • Re:Sorry Folks. (Score:2, Insightful)

      by macinslak (41252)
      Do you work for Intel? Had the 'silicon industry' been this averse to changing equipment to accomodate new materials how do you explain the quick proliferation of copper and SOI chips? Every fab takes a risk of screwing their production up with each process shrink, but they have to to stay competitive.
    • All right,

      I'm going to step in as a materials engineer and contradict this. There is already an existing market for chips made from gallium arsenide. GaAs is used to make chips used in solid state lasers (in your CD player), LEDs, cell phones and certain high power applications. GaAs is used when silicon would simply fall apart. (Silicon does not emit light efficienly either)

      These chips are made from GaAs wafers that are 3" in diameter. Current Silicon technology uses 12" wafers. Because of the difference in area, you have about 15x as much space on a state of the art silicon wafer. The reason that they don't work with bigger GaAS wafers is that

      1. GaAs Single Crystals must be grown under extremely high pressure
      2. It is damn near impossible to grow large GaAs single crystals without defects (specifically dislocations and twins)

      So to be able to essentually turn a Si wafer into a GaAs wafer would be a godsend for GaAs processing technology.

      In fact, if Motorola's claims are true, and it will cut the cost of GaAs circuitry by 90%, GaAs could start to become the material of choice for high end applications (after all, there was a brief time period when Cray was making their supercomputers with GaAs, but they just couldn't keep up with the advances in Silicon technology)

      However, I will add one word of caution. There has long been a joke in Materials Science that GaAs is the material of the future, and always will be. YMMV.

      • ". Current Silicon technology uses 12" wafers" Well, sort of. Almost. D1B is Intel's only Fab that's using 12", all the others are 8". Fab22 is designed for 12", but we're still stacking it for 8" and won't make the switch up to 12" for a couple years. AFAIK, TSMC is the only other foundary moving to 12". Everyone else seems to be taking a "wait and see" attitude. We'll get there eventually. Anyway, the point is that even though 12" is possible and even in extremely limited production, it's just that, experimental right now. 8" is standard.
    • Yeah, copper is another case in point. Nice idea, but of course we'll never see it. Oh, wait...
  • GaAs is the semiconductor of the future, and always will be.

    - one of my professors, ca 1983

  • Whether this will take succeed or not, there's no doubt that we will keep wanting faster chips.

    But a funny thing hit me when I read the words "Gallium Arsenide." It reminded me of a lecture in one of my Comp Sci classes, when the prof described the nasty environmental effects of creating these chips in the first place.

    I wonder if, at the same time as we ask what will make for faster and cheaper chips, someone somewhere will start to ask if there's a way to make these manufacturing processes safer.

    The plants that manufacture our computer chips are generally pretty nasty environmental hazards.
  • Though this is very interesting I think that the main use will be to intergrate opto-electronics on the substrate. The problem is that

    A) GaAs has a crumby native oxide

    B) there isn't a very good or simple complimentary process for GaAs. This is murder for power disapation, which is really the main problem in high speed devices.

  • 1) first of all, if the speed (transistor activity that is) is actually 40 times faster, then wouldnt that generate quite a bit more heat? hey, i mean we have enough heating problems in terms of chips anyway, we dont need for every chip to need a water cooled peltier system in order for the damn chip not to overheat! 2)200+ patents. wait...can anyone say monopoly?
    • Well if the transister is switching 40x faster then you should be producing 1/40 of the heat (heat created depends on switching time and voltage - less time + less voltage = less heat.) This of course only counts if you're running at the same MHz and using the same type of material - they're not.

      I would actually also be quite interested in the heat produced. I assume it would be higher - thus, you're not going to see them making any PPC chips out of these. However, the communication industry could benefit greatly from such devices (ie, small, simple, but very very fast.)

      This all looks great but shouldn't effect the computer industry much. However, I'd still like to see Motorola's product catalog in a few years.


      • one other thing: if yuo are running at the same MHz, then what is the point of having the transistors switch faster (except to generate less heat, which would be usefull in things like laptops, where space is critical). They are probably going to run at higher Mhz, and most likely also need more voltage going through the chip: increasing both of yuor (presumably correct) heat factors; hence, more heat.
  • EMP Hardened? (Score:2, Interesting)

    by DaedalusLogic (449896)
    I remember from a materials science course that GaAs semiconductors were more resistant to electromagnetic fields' influence... What kind of applications could these chips be better for than straight silicon? The military has plenty of applications... but what industries could specifically benefit from cheap electromagnetically "rugged" chips?
    • GaAs is actually a very, very poor choice for "rad hard" chips. IDK who gave you your information, but they're pretty much dead wrong. Rad-hard materials are like SiC, GaN, ZnO, diamond...things with bandgaps > 3 eV.
  • by Ted V (67691)
    Hasn't Vitesse [] been doing Galium Arsenide semiconductors for the past 10 years?

  • by matt-fu (96262)
    "Gallium Arsenide Valley". I'm just not seeing it.
    • "Gallium Arsenide Valley". I'm just not seeing it.

      How about "GaAs Valley". Except that it already describes L.A.

  • ...but... In the past couple of weeks there was an article in the Wall Street Journal about Motorola seeking to sell its semiconductor business because it just wasn't performing. Now, all of a sudden, that division comes up with a "holy grail" of chip manufacturing. Hmmmmm. Does this mean the executives at MOT are so dumb they didn't know there R&D boys had a world beater coming down the pipe? (OK. OK. They're pretty dumb. But are they THAT dumb.) Or does it mean that the R&D boys are trying to get some extra hype for something that may-be-ok-but-is-not-all-that-great? Far be it from me to think that people may be trying to use publicity as a tool in an internal power struggle. Oh, no. Couldn't possibly happen. I guess I'm just too cynical.
  • There's a similar article in EE times. Sounds like they're using something called a "compliant substrate". The idea is that if the substrate is very thick relative to the film being grown then the tendency is for the film to deform it's lattice to match the substrate. The lattice strain stores energy, and as the film increases in thickness the amount of strain energy per unit volume of film increases. If the mismatch between substrate lattice dimensions and film dimensions is large enough the strain energy per unit volume can become large enough to nucleate dislocations at the interface. These dislocations allow the film to "relax" back to something near it's equilibrium lattice dimensions by periodically deleting or adding atomic planes near the interface. The problem is that these dislocations can thread up into the top of the film (i.e. where the device layers are) and act as non-radiative recombination centers and carrier traps. The dislocations can also jump from one layer to subsequently grown layers. A compliant substrate tries to force the substrate to deform, and thus the strain E in the film never gets high enough to nucleate dislocations. For example, if you make the substrate very thin then as the film grows the substrate will deform to match the equilibrium lattice dimensions of the film rahter than the other way round. Traditionally in Si technology this has been done by ion implanting O2 in a thin layer some small distance below the surface fo the wafer. The wafer is then annealed to let the crystal structure recover from all the damage the ions did to the surface. This leaves a thin layer of "single crystal" silicon floating on a thin layer of glass. At growth temperatures of >1000 C in MOCVD the glass layer is fairly gooey, and the thin silicon layer practically floats on it. So as long as the epi film is thicker than the Si compliant substrate you're golden. But this adds 2 steps to the production run, and ion implanting isn't generally a high throughput process. (i.e. $$$$$$$$$$$$$) Seems like Motorola's trick is to deposite a layer of some oxide with a crystal structure similar to GaAs. They then let oxygen diffuse down into the Si to form glass. So they bipass the implantation step. The intermediate layer probably doesn't match the GaAs exactly anyway, which means you still get dislocations. Alot fo Motorola's research time and patents were probably devoted to converting existing techniques for reducing dislocation density to work with the intermediate layer material. Anyway, I hope this gives you some idea of why I'm kinda skeptical. Old dog, maybe not-so-new tricks. But if Motorola has pulled it off it would be pretty sweet.
  • How come they have so many new inventions for this new chip? What - none of them could have fit on any of their older technology? Yes, I'm sure a lot of effort went into that thing, but to get over 200 new ideas on one product (especially one that is mostly an old idea that's just been too expensive before) seems amazing.
  • by crgrace (220738) on Tuesday September 04, 2001 @02:59PM (#2252425)
    GaAs has been used for chips for years. Cost is of course a problem but there are others that make it very unlikely this will be used in general purpose microprocessors. The first problem is GaAs has a much higher defect density than silicon because it is a superlattice of gallium and arsenic and not a single crystal like silicon. For this reason GaAs chips have MUCH less yield than silicon chips so the number of transistors that can be integrated in GaAs in much less, even if it is put on a silicon substrate.

    The second problem is the lack of a good thermal oxide in the GaAs material system. Silicon uses SiO2 which is an excellent insulator and more importantly has an extremely clean interface with silicon, so there are very few traps at the oxide-si interface. Because GaAs doesn't have a good oxide, MOS field-effect transistors (MOSFETS) are impossible and so digital GaAs chips use MESFETS, which are FETs without the oxide. It turns out the good oxide in silicon makes a lot of things possible that are impossible in GaAs. For example, the si oxide makes for a very high input impedance for Si transistors so they can be used to make dense RAM and very simple registers that rely on a high impedenence node. This structures are not possible in GaAs so more complicated, higher power circuits are required in GaAs to achieve the same functionality.

  • Most of the stuff I've read about GaAs chips indicates they're used in RF signal processing. If Motorola makes them inexpensive, will they also show up as CPUs in desktops, laptops, and PDAs?
  • Is this simple a clock speed multiple? For example, the PowerPC G4 currently tops out at 867 MHz. If it was made with this new gallium technique, would it then be able to be clocked up to 30 GHz?
  • Its important to be able to integrate inductors onto chips in order to reduce component count and to provide a degree of uniformity in consumer designs like Wireless. Unfortunately, having a silicon layer under the GaAs makes for a lousy inductor as it absorbs much of the field. The same problem is true for SiGe. Plain old GaAs is not dead and is getting even more important as lower freq spectrum space is being used up, pushing applications above 4Ghz.

"Text processing has made it possible to right-justify any idea, even one which cannot be justified on any other grounds." -- J. Finnegan, USC.