Follow Slashdot blog updates by subscribing to our blog RSS feed


Forgot your password?
NASA Space Hardware Science

NASA Building Massively Heat-Resistant Chips 172

coondoggie writes "NASA researchers have designed and built a new circuit chip that can take the heat of a blast furnace and keep on performing. Silicon carbide (SiC) chips can operate at 600 degrees Celsius or 1,112 degrees Fahrenheit where conventional silicon-based electronics — limited to about 350 C — would fail. The new silicon carbide differential amplifier integrated circuit chip may provide benefits to anything requiring long-lasting electronic circuits in very hot environments such as jets, spacecraft, and industrial machinery. In particular, NASA said SiC applications will include energy storage, renewable energy, nuclear power, and electrical drives."
This discussion has been archived. No new comments can be posted.

NASA Building Massively Heat-Resistant Chips

Comments Filter:
  • Great idea (Score:4, Funny)

    by Anonymous Coward on Tuesday September 11, 2007 @11:49PM (#20566717)
    This gives an update for my macbook pro.
  • Too Bad (Score:5, Funny)

    by Kryptonian Jor-El ( 970056 ) on Tuesday September 11, 2007 @11:50PM (#20566721)
    Its too bad, we could have used this when the Pentium 4 Prescott came out...
  • As I read all I could think about was...overclocking
    • Re: (Score:2, Insightful)

      by Falstius ( 963333 )
      Unfortunately, the metal interconnects would probably melt. They're probably doing all of the interconnects with doped carbide, making this chip very slow (or power hungry). Its a shame when your CPU starts dripping metal from running folding@home.
      • no interconnects. this is not a processor. it's a chip. likely a single transistor the size of the silicon in your CPU. wirebonding typically platinum for very high temp packages. al-al or au-au if not quite as severe. other options available below that.
        • it's a chip. likely a single transistor

          Help me out here - what part of "differential amplifier" in the summary is ambiguous?

  • by SpiffyMarc ( 590301 ) on Tuesday September 11, 2007 @11:52PM (#20566749)
    is hundreds of champagne corks popping simultaneously at the AMD campus.
    • is hundreds of champagne corks popping simultaneously at the AMD campus.

      That's 1,112 degrees Fahrenheit not Celsius. You're still gonna need one hell of a heat sink.

  • ...or did he come back? I forget...last one I saw was him dropping into a blast furnace.
  • by lordofthechia ( 598872 ) on Tuesday September 11, 2007 @11:57PM (#20566795)
    In case you didn't know, Apollo has been developing a system to grow diamond wafers through CVD (Carbon Vapor Deposition) for you guessed it, semiconductor use.

    Anyway SiC is used in jewelry [] too (obviously with the same properites), just never realized that it could be used to make microelectronic devices like this. Heh, my wife's engagement ring just got way cooler.
    • Re: (Score:3, Informative)

      Hmm... somehow lost my link to Apollo, well here [] it is.
    • Re: (Score:3, Informative)

      by dido ( 9125 )

      Silicon carbide is really hard stuff. It's another name for the industrial abrasive carborundum [] and it's generally harder than sapphire (9 on the Mohs scale) but slightly softer than diamond (10 on the Mohs scale).

    • Re: (Score:3, Interesting)

      by dstiggy ( 1145347 )
      They're not the only company doing this. Diamonex [] a subsidiary of Morgan Advanced Ceramics has been making diamond on silicon for years. I should know as I have a part of a wafer sitting on my desk at home. As for the people worried about heat dissipation, these things move heat amazingly well (better than copper). I've taken the wafer and on edge it will cut through an ice cube like a knife through warm butter. Unfortunately it will turn your fingers numb in about 3 seconds to
    • Zardoz! (Score:4, Funny)

      by crovira ( 10242 ) on Wednesday September 12, 2007 @08:47AM (#20570487) Homepage
      Remember the 'McGuffin' in Zardoz?

      It was a diamond based processor.

      In fact it was a diamond based, optical processor...

      Hmm... Things that make you go hmm...

      Oh, for people who don't know, 'McGuffin' was Alfred Hitchcock's name for a central plot device around which everything in the story rotated.

      And for people who don't know who Alfred Hitchcock was, he was a famous movie director.

      Its not easy getting old. There's all this common 'shared reference' shit to worry about losing.
      • More importantly, even though the plot rotated around the MacGuffin (sometimes McGuffin or Maguffin), it has little other relevance to the story.
    • by crgrace ( 220738 )
      In case you didn't know, SiC was actually the first commercially successful semiconductor. It is also known as Carborundum and was the basis of the original low cost "crystal radio sets" that were used by hobbyists and cheapskates to demodulate AM radio. Basically you had a wire called a "cat whisker" and you moved it around the SiC crystal until you made a good rectifying contact. Then, you would adjust the channel with a variable condenser (now called of course a capacitor) and it drove high impedance
  • I've wondered sometimes why people didn't build silicon carbide chips and put them into Venus landers.
    • by Tavor ( 845700 )
      True, it would allow the chip to function longer on descent. Building a vessel to hold it lightweight enough to launch, but heavy enough to survive the massive pressures of Venus' dense atmosphere is still the challenge.

      Can you imagine a Venus lander 'floating' on a super-dense gas/liquid? It's Sci-Fi for the foreseeable future; and, likely forever.
      /OT: Notice to all operators and grammar nazis. This station is operated by a man with a headcold. All scientific and spelling errors are unfortunate.
      • Re: (Score:2, Insightful)

        by Iowan41 ( 1139959 )
        Kinda like flight. Or going faster than 20 miles an hour. A little engineering thinking makes it easy to get around the pressure problem: Let it be "wet" not "dry" No need to maintain 1 bar in a pressure vessel, now is there? Of course, the other components have to handle the heat, too, not just the chips.
      • Re: (Score:2, Informative)

        by ACDChook ( 665413 )

        Can you imagine a Venus lander 'floating' on a super-dense gas/liquid?
        And why would a Venus lander need to do this? Last I heard Venus was no gas giant. The Russian probes all seemed to find a solid surface to land on.
  • by Anonymous Coward on Wednesday September 12, 2007 @12:06AM (#20566869) []

    TFA talked about an analog amplifier. As such, noise is a problem. The higher the temperature a circuit is operated at, the greater the noise. For some low noise applications, it is standard practice to run an amplifier in a liquid nitrogen bath. For most applications, room temperature is ok from a noise standpoint. The temperature TFA talks about would produce about three times the noise of a room temperature circuit. For many applications, that would be way too much.

    For some applications, high temperature operation would be hard to avoid. Landing a probe on Venus comes to mind in that regard. The extra noise induced by temperature should cause lots of engineering misery.
    • hmmm.. too bad the PhD's never thought of that problem. oh, wait. they did. that's why they're using Silicon Carbide.

      Thermal noise is typically related to the random promotion of carriers from the valence band to the conduction band, which gets worse at higher temperatures because the electrons get more energy. THE primary electronic difference between Silicon Carbide and Silicon is that SiC is a wide-bandgap material. It takes a lot more energy to promote carriers than in silicon. Also, SiC has ~0

      • There are amplifiers with Noise Figures of 600C == ~300K to ~900K) Gives an extra four or five dB of thermal noise.

        Unless your a amplifying an extremely weak signal (for example, in an RF receiver), this amount of excess noise is probably irrelevant. Ku band (~14 GHz) Power amplifiers, for example, often have noise figures of 10 to 20 dB, and no one cares - because the signal level is so much higher than the thermal noise floor.

  • by Foktip ( 736679 )
    A Silicon Carbide chip could also be used to cut or grind steel... the manufacturing costs for making a silicon carbide ship must be insane!
    • wow. a 3 inch silicon carbide wafer costs a couple grand. a whole ship? maybe a small sloop or something. something like the titanic would require the GDP of a small country. or not so small.

      but yes. it's expensive. until we figure out the processing control, costs will stay high too. still too expensive for most commercial app's, but once that changes SiC will replace a lot of Si power electronic devices.

      • Let's hope they stay high. You are aware that AMD and Intel will immediately stop trying to get faster without increasing the heat if they could, yes? I mean, yes, we'd have CPUs running at approximately 5 times the speed in a year, but wasting about 10 times the energy to do that. 20 times if you factor in the AC energy cost to keep the room at a temperature below 100F.
        • Power consumption and thermal dissipation are two separate problems. You're used to thinking of them as rigidly related, because the thermals for modern CPUs are very rigidly defined, but by conflating the two you're only displaying ignorance.

          With lower need for thermal dissipation at any given power draw, the demand on the cooling system is lower. That means smaller, quieter, and lighter components everywhere a high performance IC is present.
  • 350C for Silicon? (Score:4, Informative)

    by Mateorabi ( 108522 ) on Wednesday September 12, 2007 @12:13AM (#20566937) Homepage
    Summary claims 350C for traditional silicon, but most silicon based transistor manufacturers list only 125C as the maximum junction temperature. (Which makes the package temp max out at 70-85C.) Makes me question how fast and loose the author was playing with the numbers. Article starts with 600C for SiC, but in the same paragraph they are down to 500C for only < 2000hrs of operation? Hype?

    Also, do SiC transitors switch as fast as doped silicon? Otherwise the "make a pentium with it!" ideas might fall flat.

    • Re: (Score:3, Interesting)

      by ApharmdB ( 572578 )
      SiC is meant as a replacement for GaAs in RF circuitry. SiC and GaN are the next generation semiconductors with much, much better thermal properties and max voltage than GaAs. SiC is not meant as a replacement for Si on digital chips as far as I know, but I'm an RF electrical engineer.
  • by pushing-robot ( 1037830 ) on Wednesday September 12, 2007 @12:19AM (#20566997)
    600 degrees Celsius or 1,112 degrees Fahrenheit

    I love those "pull-significant-digits-out-of-my-ass" unit conversions.
    • Re:Sorry, OT... (Score:4, Informative)

      by Dunbal ( 464142 ) on Wednesday September 12, 2007 @12:58AM (#20567365)
      Don't worry, NONE of those digits are significant, since the ACTUAL temperature of a blast furnace is around [] 2000-2300 degrees C. I certainly remember the kilns at the cement plant my dad worked at used to get up to around 1900 C. 600 degrees is nothing as far as furnaces are concerned, although it's more than your oven can do.
  • by Anonymous Coward
    Presumably the chip has to sit on some type of circuit board, connected to other components.

    So it's OK if the chip survives but the rest of the circuit melts?
    • no. see comments above. specifically []
    • Copper? (Score:3, Informative)

      by 2901 ( 676028 )
      The traditional challenge is to get the melting point of solder low enough. The worry about moving to lead-free solder is "how will we keep the melting point down." It is the silicon chip that is the delicate component
      that is damaged by heat.

      Ofcourse the low heat tolerance of silicon chips, by limiting permitted temperatures during manufacturing, also limits required temperatures. No-one requires circuit boards to withstand more heat than the components can take. So some materials that sheltered behind the
  • Venus Lander! (Score:4, Insightful)

    by Derling Whirvish ( 636322 ) on Wednesday September 12, 2007 @12:31AM (#20567127) Journal
    Maybe we can finally get a decent lander or rover on Venus.
    • by vought ( 160908 )
      Maybe we can finally get a decent lander or rover on Venus.

      Maybe we can even get a descent lander.
    • You could instead use more normal components in a 100% sealed container thats vacume isolated and suspended on a magnetic field and uses small a heat generated turbine to make electricity since its so hot.

      Btw, how much of venuss heat is due to presure rather than the greenhouse effect, since only 30% of light/heat gets to the surface it would take a long time to heat up, it still loses
      heat due to normal thermal dynamics . But if venus was at mars distance, how hot would it still be? Try any gas at 90 earth
      • You could instead use more normal components in a 100% sealed container thats vacume isolated and suspended on a magnetic field
        I'm no rocket scientist, but wouldn't that make it kind of hard to get any data back to earth?
        • by suitti ( 447395 )
          Not really. You can have a wire from the inside to the outside attach to an antenna.

          The vacuum chamber is a good insulator, but the computer will be generating heat on the inside. So, it's still best if the computer produces as little heat as possible, and if there is a way to get rid of the heat. I suspect that Venus will cook it eventually, anyway. Plenty can be learned in a few hours, though.
          • Not really. You can have a wire from the inside to the outside attach to an antenna.
            Wouldn't that make it less than 100% sealed? :)
      • by khallow ( 566160 )

        Btw, how much of venuss heat is due to presure rather than the greenhouse effect, since only 30% of light/heat gets to the surface it would take a long time to heat up, it still loses heat due to normal thermal dynamics . But if venus was at mars distance, how hot would it still be? Try any gas at 90 earth here and does it get hot?

        The only energy inputs are the Sun and internal heat from Venus. The former dominates. Pressure is not an energy source so it doesn't have an effect on long term temperature d

  • by chris_sawtell ( 10326 ) on Wednesday September 12, 2007 @12:36AM (#20567195) Journal
    Venus darling, Please don't get alarmed, but those wretched Earthlings have made a super-dooper hot-stuff control whatsit. You'd better watch out because from what I hear that super-hot atmospheric condom of your's isn't going to protect you from frequent and repeated penetration much longer. Sorry to be such a harbinger, but I just thought you ought to know. Haved a chat to Mars, he knows all about what they get up to.
  • It's nice and all but will we be able to build a fanless PC from it? Although I'm not sure how fast the PC would operate if it could be allowed to get that hot?
  • My first questions (Score:3, Insightful)

    by artifex2004 ( 766107 ) on Wednesday September 12, 2007 @01:02AM (#20567403) Journal
    Have they got a carrier or other method of holding it to a circuit board that will stand up to that heat? Speaking of, have they got circuit boards that stand up to that heat? And obviously solder can't be used. So how will they interconnect? Glass fiber may melt at higher temps, but I'll bet the optical properties distort well before then, considering it glows when it gets hot enough. Not to mention they have to make the emitters and receivers withstand that temp as well.
    • Re: (Score:3, Informative)

      by GloomE ( 695185 )
      Not all circuit boards are glass fiber.
      Ceramics are already used where you need precision e.g. wave guides.
    • That was my first thought as well. I think the tech is really better suited for just increasing the overall reliability of spacecraft electronics, since they have to be ridiculously hot/cold tolerant. Typically, it's not in the nature of aerospace guys to shake up an entire mode of engineering on a single discovery; so if anything, this stuff is going to be phased in nice and slow starting with everything staying the same except for the chip substrate (SiC instead of plain Si).

      But in order for this to ope
  • Anywhere you run a PC where it's above room temperature you've got cooling issues - if this gets to a sensible price you could do away with a lot of gadgetry that has to be added to keep things like servers working.

    I'm also thinking of SCADA deployment in dry and dusty places - less parts means more reliability.
    • by B5_geek ( 638928 )
      I thought I should point out that any room that you have a PC in, is at room temperature.

      • I was debating with myself to caveat this situation :-).

        It think you're forgetting something - not al PCs are used as workstations. Think of ATMs out in hot countries, and not everyone with a small company has the resources to place a server in an airconditioned place either, either for lack of energy or space. And quite a bit of SCADA platforms are out there in the nice hot sun, and I'd feel much happier if they had less parts that could fail. Granted, the PC style stuff is not usually used in ESD chain
  • by NerveGas ( 168686 ) on Wednesday September 12, 2007 @02:33AM (#20567963)

        Intel re-released the Pentium-D line, using this technology.
  • the packaging... (Score:2, Informative)

    by Anonymous Coward
    the limiting factor with Si is infact the packaging AND that is going to be the limiting factor with SiC operating temp as well

    NASA are right in saying that Silicon can operate at 350C but that is the exposed die that isn't on any substrate and using spring-point connections
    Start packaging the thing up and you have the die solder down onto something, solder wires onto the die and it is these things that put the operating temp at 125C

    Semikron have IGBT modules that they say can operate upto a die temp of 175
    • close very close.p> the limiting factor for silicon is loss of junction action as more carriers get thermally promoted to the conduction band. typically right around 200C, the intrinsic carrier concentration overtakes the typical doped carrier density. But, you start getting increased leakage currents and higher current gain well below that. Depending on the type of transistors used, latch-up failure becomes more likely. the prime factor affecting what temp things start going bad is the amount of dop
  • Excellent now we can have a Matryoshka shell of advanced computing equipment orbiting nearer to the sun just like in the (soon to be outdated) scifi novels!
  • I was looking into this in the early 90s (along with the development of diamond as a semiconductor.) We were in fact interested in very robust front ends for things like telecoms systems, that could survive lightning strike or EMP. Here the issue is not to operate at very high temp., but to survive brief excursions to it. The problem with any form of lightning protection is that the sensitive amplifier must be outside the protected area, or how can it get the signal?

    In fact SiC has a long history as a semic

  • by sunspot42 ( 455706 ) on Wednesday September 12, 2007 @04:29AM (#20568671)
    Now hardware capable of running it is finally available, Duke Nukem Forever should be released any day now!
  • In particular, NASA said SiC applications will include energy storage, renewable energy, nuclear power, and electrical drives.

    Yeah right. Everyone knows these are just government funding cover stories for the true purpose: extreme overclocking.
  • 600 C is not even hot compared to a blast furnace

    A blast furnace is a metallurgical furnace generally used to produce iron. It operates between 2000 C to 2300 C (3600F to 4200F). (Irons melting point is 1538 C or 2800 F).

  • by mbone ( 558574 ) on Wednesday September 12, 2007 @08:07AM (#20570079)
    Since Venus's surface temperature is about 460 C, these chips would presumably work on the surface of Venus, which would allow for long duration landers, or even rovers, should we want to do that. I think of Venus, "Earth's evil twin," as being a very interesting planet, but there has never been very much interest in exploring it at NASA.

    The only pictures [] we have of the surface of Venus are from the Venera landers. (These USSR Venus landers [] were all inernally insulated and weren't designed to last on the surface more than about an hour; since the data were relayed from the fly-by bus spacecraft which was only in range for about that duration, there was no point in doing more.)
  • How can we kill the terminators now?

    Thanks a lot, Nasa. You've just doomed us all.
  • As an electronics enthusiast with the exquisite l33t soldering skills of a drunk baboon (and that's on a good day), I for one welcome our new heat-resistant overlords. Er, chips.
  • Hard to believe all the people posting here that think some how this is related to digital logic or CPU chips. Even the summary said "differential amplifier integrated circuit chip" I guess those words go over the heads of most readers. What this really means is that now, with this kind of chip you can have shorter wires leads on sensors even to the point of packaging a sensor and an operational amplifier in the same package. This will go a long way to lower noise in certain measurements.
  • Think, folks, think. How often do you need electronics IN a hot zone?

    What's worked just fine for many decades is to have sensors in the hot zone, ceramic or Teflon-coated wires to a cooler place where you have the electronics.

  • I worked for a materials science company that was trying to manufacture ultra-pure Calcium Fluoride optics for extreme UV lithography. Our vacuum furnaces had a lot of specially-constructed silicon carbide parts which were manufactured using a vacuum deposition process, and were very strong and heat resistant (for the temperatures we were generating in the furnace). My boss, a material scientist of some renown, told me he wanted to eventually get into manufacturing SiC semiconductors because they could be

1 Angstrom: measure of computer anxiety = 1000 nail-bytes