Hacker News Reader

Comment by iwwr | Hacker News Reader

These chips, at least in IC form, could enable the long-term exploration of Venus from the surface. This is because SiC can work at 400C or higher temperatures.

View on HN

Not just Venus, the whole cost of cooling chips and cost of raw material extra's like heat-sinks could be reduced and that in itself would make a huge saving thru volume. Some IC'c on this type of silicon wouldn't need that extra cooling cost and that is were the big saving will come in, even with the extra cost of the IC's factored in.

That and if you can remove all cooling, not just cost savings, but in some cases being able to remove active cooling and less moving parts is always an engineering win win.

Unless you're changing the underlying currents or resistances, you'll still have heat losses that need to go somewhere.

I don't think it's that a 700C chip doesn't need cooling the way a 90C chip does; it's that the delta between 700C and room temperature means that a much smaller and much more passive cooling apparatus may be sufficient.

As radiative cooling is a temperature to the 4th power (T^4) process, vs. linear (T^1) for conductive and convective cooling, one could certainly increase power density for general electronics, though the danger of setting something on fire, and actually needing higher temperature packaging / soldering / bonding processes might put a crimp on that.

However, space craft would be where being able to run hotter would significantly reduce the size and complexity of cooling systems, as they can only use radiative cooling bleed off the heat they generate into space.

Besides the thermodynamic advantages with conduction / convection / radiation, you can also use boiling points. Strap a pan with water to it. The pan won't heat up past 100 unless all the water has boiled off. You still need to remove the same number of Watts, but this is potentially easier to manage.

Makes sense, though you can get that with contemporary automotive spec (120-150C) parts.

>The pan won't heat up past 100 unless all the water has boiled off

Imagine... a smart rice cooker where the heating element is its own embedded system running some kind of awful javascript-based OS and monitoring stack optimised for maximum waste heat. When the rice is done the system emails you and then issues itself an ACPI shutdown.

I think SiC devices will be a way station on the road to diamond devices for extreme environments. GaN devices will be the next after SiC, then on to diamond.

I've seen a diamond FET operating at 700°C. Devices like that could use the Venusian atmosphere for cooling.

GaN is only really good for applications requiring direct bandgaps like LEDs and extremely high electron mobility like plasma micro-antennas. I believe that SiC will eventually replace the photonic applications with direct-band gap strains and dopants, but not the high-mobility applications.

Normally i would suspect that mixed solid is more sensitive to heat (lower melting point) then a pure material. Why isn't this the case for SiC?

There are many cases where mixed solid materials are stronger than the sum of their parts. Heat tolerance is all about high bond strength between the materials, and strategic mixing of materials can enhance that property or others

Been a while since I was in the chemistry game, but my guess is lattice structure and packing. Pure Si has relatively long bonds compared to carbon, so the intermolecular forces are a bit weaker. Apparently SiC has over 250 polymorphs, but the most common is the alpha.

... ok I'm reading about the material properties of SiC and it's insanely complex so I'm just going handwave and say that α-SiC has tighter packing with stronger bonds because of the alternating Si and C atoms.

Chips aren't a single piece of Si or SiC crystal, but a superbly complicated arrangement of doped regions, insulating, metal and plating layers, so the melting point of the base crystal doesn't matter here (Si melts at ~1500 °C, C depends, and SiC seems to sublime at very high temperatures like carbon, instead of melting - corrections welcome). The limit to sustained high-temperature operation of ICs is due to all of these things starting to diffuse around and getting into places they're not supposed to be.

I’d be more worried about the solder used for the mounting of the chip and other components.

Crystals are conceptually very different from "mixed solids". Eutectics, e.g. solder, count as mixed solids.

Keep in mind that SiC is a substance, so a pure SiC crystal is not a mixed solid.

Mixed solids are more sensitive to temperature variations (there's no change on the melting point), but actual chips are already full of mixed materials, and temperature variation isn't the bottleneck. A SiC chip has actually a bit fewer kinds of materials on the mix, but still, probably not enough to make a difference. (Anyway, we are talking about discrete transistors here, so the material count is exactly the same.)

You are probably thinking about pure substance (the ones that only have a single kind of chemical element). If so, you are completely wrong, pure substances usually have a very low melting point, and very few reach as high temperatures as you can get with composed ones.

SiC is a wide band gap semi conductor so it takes a much higher temperature for the intrinsic carrier concentration to rise to the point where it is no longer a semi conductor and just a conductor - it was 12 years ago since I finished my PhD in SiC and related devices so I can’t explain in much more detail but that’s the crux of it

It does have a lower melting point than diamond. The bond strength is determined by the electro-negativity of the bond pair. In this case, it is between silicon and carbon.

That's awesome, although I was kind of excited to see how far we can get with clockwork.

Wikipedia has the surface temperature of Venus at 464C, would SiC be able to handle that?

Directly probably not but the higher the working temperature of your electronics is the easier it would be to cool it to a working temperature.

That would allow survival for a few hours or days at most I imagine.

I'm sure it's possible to build something that could operate at these kinds of temperatures permanently, but it would probably need an RTG (nuclear battery) and may be hard to test it on earth and also allow for thermal expansion and shrinkage (hard to keep the probe at this temperature for the entirety of travel to Venus). We'd also need to design lubricants for gears, cameras and sensors that can tolerate a wide range of temperatures, etc.

The upside of all this is that there are probably legit industrial use cases here on earth for electronics that are resilient in extreme temperature ranges, and in satellites as well.

I don't see why it would be hard to test it on Earth. Pretty easy to construct a 500C room.

But you're right - just because the silicon can go to 500C doesn't mean there aren't a gazillion other problems. You can't even use normal solder at that temperature.

We have high temp electronics that are used in deep wells, nuclear reactors and scientific instruments… Do they go to 400C+? I don’t know but even on digikey you can find IC’s with operating temperatures above 200C and even 250C fairly easily.

The issue with Venus is really cost vs benefit any probe would have a very short life span as in hours to maybe maybe days.

At the same cost you can send probably multiple missions to Mars that would last months if not years.

We had probes on Venus using 1960’s technology we can do that again. Selling the political oversight and even the public on this however is far harder than designing and building them.

Reply on news.ycombinator.com