Silicon Carbide Chips Can Go To Hell

IEEE Spectrum had an interesting read about circuits using silicon carbide as a substrate. [Alan Mantooth] and colleagues say that circuits based on this or some other rugged technology will be necessary for missions to Venus, which they liken to hell. That might seem like hyperbole, but at about 460C with an atmosphere full of sulphuric acid, maybe it isn’t such a stretch. When the Soviets sent Venera 13 to Venus, it was able to send data for just over two hours before it was gone. You’d hope 40 years later we could do better.

Silicon carbide is a semiconductor made with an even mix of silicon and carbon. The resulting components can operate for at least a year at 500C. This high-temperature operation has earned them a place in solar energy and other demanding applications.  [Alan], with the University of Arkansas along with colleagues from the KTH Royal Insitute of Technology in Stockholm are building test circuits aimed at developing high-temperature radios for use in environments like the one found on Venus.

The Vulcan II chip they are testing has multiple test circuits including several types of analog to digital converters, an op amp, logic gates, and DC to DC converter and — perhaps in an effort to keep Hackaday commenters happy — a 555 timer. Between the two Vulcan chips, the team claims to have 40 silicon carbide circuits worked out.

The material has a much higher breakdown voltage compared to silicon so you can improve your voltage handling or make smaller devices that have better characteristics such as on resistance due to their smaller size. The thermal conductivity is also much better. But the real key, according to the post, is that silicon carbide has low carrier concentrations at room temperature. This doesn’t sound like a great thing, but if you are interested in high-temperature operation, it is. Silicon devices stop working at high temperatures because the extra energy floods the device and essentially turns it on all the time. That happens at 250 to 300C for silicon, but with the carbon added, devices can switch reliably past 800C.

Silicon carbide itself isn’t new. It was an accidental discovery in 1895 when a chemist tried to make artificial diamonds. By 1906 it was used to detect radio waves —  the carborundum detector — which may make it the earliest semiconductor device, although silicon detectors were the first to be commercially produced.

We’ve looked at Silicon Carbide in electric vehicles. We’ve also seen it used in a non-semiconductor way when melting glass.

75 thoughts on “Silicon Carbide Chips Can Go To Hell

      1. Diamonds are cheap to buy or easy to make in a lab. The expensive part is the serial number that De Beers etches in it for the diamonds destined for the jewelry market (for the purposes of tracking them from the mine and for theft detection).

      2. The Space Industry has an entirely different view about costs than the Earthbound consumer industries. They also have a track history of non-retrievable, multi-billion dollar assets.

        Space (and Military) projects have traditionally been the earliest adopter of expensive new semiconductor technologies – the Apollo Guidance Computer being one example.

        If a 500 degree C semiconductor is the only thing that will reliably work on Venus, then there’s no point wasting money sending anything else all that way just to fry and fail within a couple of hours.

        Complete new electronic packaging and construction techniques will be needed. Soldered joints and FR4 boards will be out of the question. Novel heat dissipation, thermal insulation and cooling methods will also be required

        It’s not just semiconductors that struggle at 500C in an atmosphere of sulphuric acid. Many other common space craft construction materials, including plastics, composites, many metals etc are all going to have a bad day out.

        1. The average surface temperature on Venus is 737 K (464 °C, 867 °F), so it would need to be a Lead free solder that was used :) – Lead’s melting point is 600.61 K ​(327.46 °C, ​621.43 °F). Oh and it would also need to be Tin free as well – Tin’s melting point is 505.08 K ​(231.93 °C, ​449.47 °F).

          All the Sulfuric Acid is in the atmosphere since it boils at 610 K (337°C, 639 °F).

          So every material that can be corroded by Sulfuric acid has to be taken off the table.

          I’m not picturing what will probably be chosen (ceramic wheel), I’m picturing the pimp-est rover ever built and sent to another planet with hollow 3D printed Platinum wheels, or maybe some gold alloy :)

      1. Bose’s 1901 patent mentions galena and a half dozen other materials as detection elements. SiC is not mentioned. Bose and others did research on rectifying devices and materials before then.

    1. LEDs need to be made from 1. semiconductor with direct bandgap 2. with bandgap tuned to energy level of emitted photons, so no you cant just make them from any random semiconductor

      1. But you can have them emit UV and convert that to visible light of whatever color you want using a phosphor. That’s how blue and white LEDs have worked for years, and it’s even a common way for red LEDs to work now.

    2. They already exist, the Russians made them years back, they are not very bright though. The discovery that SiC could produce light was made a long time ago.

          1. Wish they were still dim… Too many bright LED these days. It is like people still stuck on using “20mA” to drive high efficiency LED. 20mA was the maximum operating current for LED back in thedays when they were dim.

    1. Send a nuclear reactor there, use molten lead for cooling, and just spin a wheel to generate electricity. Should be “simple”, except with the extreme materials. The lead would melt once at the surface, so there is no problem with “kickstarting” the reactor you otherwise have when the coolant metal solidifies.

      Bob’s your uncle.

    2. Regenerative fuel cell? Could use carbon dioxide from the air, splitting it into carbon monoxide and oxygen using a solid oxide electrolyzer, as was recently demonstrated on Mars, and which has to run at a high temperature anyway. I’ve heard of solid oxide fuel cells as well, and I think you can combine/hybridize the two somehow to make a regenerative fuel cell.

    3. That’s already been solved. A lot of one shot military devices (missiles) use thermal batteries ( that are activated by a pyrotechnic charge that heats them to over 500C in order to melt the electrolyte and allow them to work. A few of these would automatically start generating voltage as soon as the temperture got high enough.

      The trick would be to make them rechargeable and figure out how to recharge them on cloud shrouded Venus where 500 C is your cold sink for running a heat engine.

  1. Folks, do your homework:

    “The phenomenon of electroluminescence was discovered in 1907 using silicon carbide and the first commercial LEDs were based on SiC. […]

    LED production soon stopped when a different material, gallium nitride, showed 10–100 times brighter emission. This difference in efficiency is due to the unfavorable indirect bandgap of SiC, whereas GaN has a direct bandgap […]. However, SiC is still one of the important LED components […]”

    So it seems SiC are the oldest LEDs around, but not extremely efficient :-)

    Ah, the ref? Plain old Wikipedia [1].


    1. I bought a SiC blue LED from RadioShack a very long time ago and it had the nicest color. It wasn’t bright (by modern standard), but any blue LED was amazing when all we had was bright red and dim yellow and green before.

      I wish I could find some of those as I remember the color being very pleasing. Thank goodness GaN came around–thank you Isamu Akasaki.

      1. I saw a datasheet for the New! SiC blue LED in the early ’80s, with indicative pricing of 50 major currency units per 5mm device. For some reason schoolboy me didn’t buy any.

          1. The list price for a reel of blue indium-gallium-nitride LEDs at Digi-Key (part number 2460-L196L-QBC-TR-ND) is 4.6 cents per LED, and they typically output 20 mA.

      2. Ahhh, I remember those blue SiC LEDs, they were dim, but emitted a beautiful powder blue colour unlike the harsh blue of modern LEDs. Expensive though, I think I paid £9 (£26, or $32 in todays money).

  2. Once you have a 555, it is time to thinking bigger!
    The real question is, what will the architecture of the CPU that is eventually sent to stay in hell (Venus) ? Irregardless if it is SiC or a Diamond based.

    Will it be a new architecture on a new chip that has not been radiation hardened and “space qualified”* (yet) : RISC-V, ARM, ???
    Or will it be one of the more traditional architectures that has already has a “space flight heritage”*: MIPS (КОМДИВ-32 or Mongoose-V), PowerPC (RAD750) or SPARC (LEON5).

    *space qualified:
    1 Survivability through the launch environment and deployment into space without performance loss,
    2 Material compatibility in space to either not degrade or cause a threat to other systems, and
    3 Achieving full performance over the duration of the space mission (often in excess of ten years).
    (ref )

    1. I used LEON5 for the SPARC, even though it has not been in space, because most search engine would mistake LEON for a film about an assassin (e.g. “Léon/LEON/The Professional”) and not a SPARC CPU. In hindsight I should have used LEON2, but at the back of my mind was that maybe that would return results for the eventual sequel to the film “Mathilda” which will be made when Portman is old enough for the role.

    2. Or are CPU’s days numbered, maybe it will be a future with Intel FPGA’s (There is currently the Altera Stratix V that are used in LEO satellites).
      Or maybe it will be a future AMD FPGA’s (If the deal goes through, there are Xilinx 90nm:Tolerant or 65nm:Hard Virtex and 20 nm:Tolerant RT Kintex UltraScale used in space).

      1. FPGA has huge handicaps on speed, density, chip real estate (cost, complexity) vs a *commodity* CPU doing *general purpose* computing. There is a reason why they have evolved to have specialized hard IP cores.

        Space industry has money, but low volume and tough requirements. So they shouldn’t be used for generalizing trends. FPGA is a good fit as once you quaified a part, the same phyical part can be programmed to do different things.

        1. There are faults that can develop with chips in the presence of Ionizing radiation.

          Cumulative effects
          Total Ionising Dose Effects
          Displacement Damage Effects
          Single Event Effects (SEE)
          Single Event Latch-UL (SEL)
          Single Event Upset (SEU)
          Single Event Transient (SET)
          Single Hard Errors (SHE)

          There is no solution for the Cumulative effects other than possibly additional external Z grade shielding, which could produce additional Single Event Effects.

          There is a good “public” 4 page summary which includes information on the effects of radiation on electronics from the ESA, search for a pdf file called “The RISC-V Klessydra Orbital Lab project”.

          My thinking is that with enough gate area that SHE failts (which are typically permanent, usually a heavy ion traversing through a active gate causing burn-out of that gate), that some SHE’s at least could be fixed by a reconfiguration of the hardware.

          1. It stripped my formatting!
            Cumulative effects
            —Total Ionising Dose Effects
            —Displacement Damage Effects
            Single Event Effects (SEE)
            —Single Event Latch-UL (SEL)
            —Single Event Upset (SEU)
            —Single Event Transient (SET)
            —Single Hard Errors (SHE)

  3. If this material becomes commonplace for general electronic components and even for computer chips, it would be an interesting way to push back the thermal limits of throttling. Then again, these components will have to be wire-bonded, packaged-up, and connected/soldered on some circuit, and have some power source aswell… The chip may survive high temperatures, but everything around the chip will have to do so too.

  4. Several companies (CREE for one) use Silicon Carbide to make Field Effect transistors. SiC is better suited for high voltages than silicon. SiC FETs are used in large quantities in many off-line applications: Power Plants, Electric Cars, LED light bulbs and even some cell phone chargers.

  5. 480c isn’t even all that hot. Sure it rains gallium but maybe all you really need is a ceramic outer shell and a couple inches of aerogel insulation.

    y’all are just too focused on planet surfaces. There’s a level of venus’ atmosphere where the temperatures are similar to tropical earth regions, the air pressure is about 1, and you are way above the clouds of extreme acids. Admittedly the atmosphere at that level is mostly co2 and the solar albedo will be insane, but the temperature isn’t an issue.

      1. Completely cloudy, but somehow still blindingly bright. Actually it’s possible that the albedo will be worse coming up from the surface than down through the clouds.

        Mars has a relatable landscape wherein the wettest part is dryer than the driest part of earth, and the warmest part has an average temperature lower than the coldest part of earth. And the dust is extremely toxic. And there isn’t enough gravity to prevent your bones losing their calcium. And despite the thin atmosphere, there might not be enough of the right kind of solar radiation to grow plants. The available solar radiation on the equator of mars is about what you get on Devon Island, near greenland, 75 degrees north. Perhaps they could use large lens arrays? They’ll need a lot of insulation or artificial heat to make it through the night without freezing, too.

        Humans could theoretically inhabit dirigibles in the upper atmosphere of venus, sending down probes (or just fractionating columns) to harvest resources from the lower strata.

        You could walk around on a platform in that atmosphere but you couldn’t breathe the atmosphere and you wouldn’t want to expose your skin or eyes to the sun, and you couldn’t expose plants directly to the atmosphere or sunlight either.

    1. maybe all you really need is a ceramic outer shell and a couple inches of aerogel insulation.

      And a heat pump, because aerogel isn’t perfect thermal insulation, and whatever’s inside will generate waste heat. And then you need a way to power the heat pump that doesn’t either heat up the interior excessively or use the inside-vs.-outside temperature difference as its source of power (because that would defeat the purpose). That can be done, but it’s not as easy as just passive insulation (unless you only want your lander to survive for a few hours).

  6. I’m surprised that with all the early discoveries of “solid state” rectification, (as mentioned in previous comments) that radios for decades after used a vacuum tube rectifier. I am aware of the galena crystal used in “crystal sets”, and the “selenium stack” rectifier that came about later (1940s?)

    HaD once mentioned a Russian who also discovered/made a solid state rectifier back in the late 1920s(?) that did not get into general use…

  7. Im pretty sure your processor isn’t going to be the weakest point on a space craft subject to such environments. Unless you’re planning to make the whole craft from silicon carbide?

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.