High-temperature semiconductors -- mission to Venus

[lpetrich]

I was motivated to research this by discovering plans for a Venus rover: Windsurfing on a Wicked World | NASA That planet has the problem of a surface temperature of about 450 C and pressure around 92 bar. Its atmosphere is mostly CO2 with a few percent N2 and much less of various other gases.

To date, the champion survivor on that planet is Venera 13 at 127 minutes. Toleration of Venus surface temperature will be necessary for doing better than that. On the mechanical side, that is not much a problem. That temperature is rather mild by some industrial standards, and I easily found several lubricants that can work in much higher temperatures. For electricity storage, sodium–sulfur batteries run at 300 - 350 C, so a 450-C battery may be feasible. Ferromagnetic materials are likely not a problem. They lose their ferromagnetism at above their Curie temperatures, but some common ferromagnetic materials have Curie temperatures well above Venus's surface temperature.

The big weak spot is semiconductor components, and that's what I'm asking about. Design of high-temperature electronics is not very advanced, as far as I can tell (High-Temperature Electronics at Analog Devices, Extreme-Temperature Electronics (Tutorial - Part 1)). Some commercially-available components are rated for temperatures of 200 C or even 300 C, but they look like rather simple ones. There are various things that one can do to improve temperature tolerance, like trench isolation and silicon-on-insulator. Looking at alternative materials, silicon carbide goes up to over 600 C, though only simple components have been demonstrated with it, as far as I've been able to find out.

Fun fact: a favorite way of torture-testing electronic components is to run them at high temperatures. That makes them fail much faster.

Is anyone here familiar with the state of the art in such components?

 

[Hesch]

Extreme-Temperature Electronics (Tutorial - Part 1)

As stated in the link, diamond is the superior material, though very expensive.

But have you thought of simply cooling the "electronic box"? I don't think that thermal isolation is a solution since the electronic circuits will produce heat.

 

[mfb]

Discrete components don't look so hard (there are commercial SiC devices), but I guess such a probe should have some computer inside, and I am not aware of integrated circuits on SiC-basis.

Cooling sounds like the easiest option.

 

[lpetrich]

As stated in the link, diamond is the superior material, though very expensive.

One can do chemical-vapor-deposition diamond films. Is that very expensive?

But have you thought of simply cooling the "electronic box"? I don't think that thermal isolation is a solution since the electronic circuits will produce heat.

Except that the whole purpose of high-temperature semiconductors is to get away from having to cool them. Doing so requires extra apparatus and it is impractical in some environments, like deep oil wells and Venus's surface. For an oil well, it would have to fit into a borehole, a further constraint, though Venus's surface does not impose that constraint. Venus's surface does impose another constraint: inaccessibility for repairs.

 

[Jeff Rosenbury]

Diamond is likely the "best" material because it has a very high thermal conductivity (~5 times that of copper). Thermal conductivity is less useful when the background heat is high. It may not be the best choice for this application.

Keeping power use low is critical. That means simple controls with less computational power onboard the craft.

Vacuum tubes still work for power amplification. They could be used for communications or in other places where power is needed. They could be kept outside a cooled electronics compartment.

 

[lpetrich]

I've done some research and here's what I found:

Silicon Carbide Logic Circuits Work at Blistering Temperatures - IEEE Spectrum

Researchers at Case Western Reserve University have pushed heat-resistant integrated logic circuits to a record 550 °C, a notch hotter than in NASA’s tests of integrated circuits.

Conventional silicon circuits break down beyond 350 °C, but the ability to get sensor information from high-temperature environments, such as inside a jet engine or in a deep oil well, could improve efficiency and save millions of dollars—not to mention provide capabilities for space missions to extreme locations like Venus.

Silicon carbide is good because it has a larger band gap between valence and conduction electrons than silicon does. That means that it takes more heat to kick electrons up into the conduction band, making it too easy for currents to flow. But it's rather difficult to make good insulating layers for CMOS transistors, now often used in IC's, so the Case Western team is using JFET's instead.

Philip Neudeck has a team at NASA working on this problem, and they've gotten chips that can run for thousands of hours around 500 °C. Their first chips had only 10 transistors on a chip, and they are planning on designs with over 300 transistors.

Having working transistors does not mean the end of the problems to be overcome. Connecting the transistors in the chip, packaging the chip, and connecting the chip to other components.

http://www.raytheon.co.uk/rtnwcm/groups/rsl/documents/content/rsl_semi_published_hiten2011.pdf -- including working on the insulator layers in the transistors.

Future high temperature applications for SiC integrated circuits - Zetterling - 2012 - physica status solidi (c) - Wiley Online Library -- also mentions lower on-resistance and possibly being radiation-hardened.

 

[marcusl]

JFET's have much higher power draw and dissipation than CMOS so I expect that to be a significant issue.

 

[Svein]

I seem to remember that Honeywell has a high-temperature process. We did something with them back in 2003 and I remember that they had no standard circuit, they just processed circuits designed in VHDL.

 

[Hesch]

What about using the good old radio tubes. You can save a lot of energy on Venus, not using so much power to keep them heated up.

 

[rbelli1]

I wonder if and how small a tube integrated circuit could be made with modern MEMS processes.

BoB

 

[lpetrich]

Glenn Research Center at Lewis Field Development of High Temperature Silicon Carbide (SiC) Electronics for Intelligent Engine Systems -- for jet engines.

Seems to be proposing:

Integration of sensor technology with high temperature wireless communications and energy harvesting to enable a smart systems operable at high temperatures.

High-temperature wireless communications based on SiC electronics and rugged RF passive components.

Energy harvesting systems focusing thermo-electric and photo-voltaic materials for generation of power for remote sensors.

High-temperature radio equipment and photovoltaic cells -- what a Venus rover will need. This presentation also mentions SiC capacitive pressure sensors.

Stability goals:
For testing in a jet engine on the ground: >1000 hr
For most aviation and space applications: >100,000 hr

Also mentions tests of high-temperature-tolerant chip packaging.

http://www.grc.nasa.gov/WWW/cdtb/aboutus/workshop2012/Presentations/Session%203.%20Distributed%20Engine%20Control/DEC_04_Beheim.pdf
It covers much of the ground of the previous one, but with additional numbers. Like these temperature upper limits:

Ordinary silicon chips: 150 C
Silicon-on-insulator chips: 300 C
SOI only for low-power applications; SiC better for high-power ones

The authors concede that JFET's are limited by their high power consumption, about 1 milliwatt per gate -- can't get very many of them onto a chip.

But SiC JFET's can function all the way down to -125 C. Good for space travel.

Also some stuff on multilevel interconnects -- integrated-circuit internal "wires".

Plans: 4-bit A/D and D/A converters, 2*2 static RAM, op amp, ring oscillator, binary AM radio transmitter

 

[mfb]

Philip Neudeck has a team at NASA working on this problem, and they've gotten chips that can run for thousands of hours around 500 °C. Their first chips had only 10 transistors on a chip, and they are planning on designs with over 300 transistors.

Time to dig out the old 8-bit CPU designs? ;)

The potential applications on Earth help, that gives a good reason to spend a significant amount of money on research.

 

[Svein]

The potential applications on Earth

Some of which are:

• Instrumentation of very deep bore holes
• Instrumentation of semi-active volcanoes

 

[Jeff Rosenbury]

I didn't notice whether the temperature of 500ºC was the oven or the junction temperature. Junction temperature seems more likely. Given that jFETs are relative power hogs, 500ºC will severely limit the power available in a passively cooled system with an ambient temperature of 462ºC. And since 462ºC is the mean temperature, the probe will likely need to shut down during temperature spikes. Or perhaps we can limit the probe to the icy wastes of Venus where the temperature plummets to 380ºC. (On a mountaintop.)

Plus there's the need to protect against hot acid. While such protective coatings exist, most of them seem to be poor thermal conductors. Perhaps graphene will work though. Does anyone know how resistant graphene is to acid?

 

[lpetrich]

Venus's surface likely does not vary much in temperature, to within altitude variation. It has pretty much the same temperature in daytime and nighttime, for instance. Venus has a rather long day-night cycle: about 116.75 Earth days. So it's almost 59 Earth days of daytime and 59 Earth days of nighttime.

Atmosphere of Venus - Wikipedia, The Environment of Venus

The highest mountains on Venus are the Maxwell Montes at about 10 km, higher than Mt. Everest. The temperature declines to about 385 C and the pressure to about 47 bar. That's all that one gets for landing on top of Venus's highest mountains.

Venus's sulfuric acid is in its clouds. So it's mainly a problem of one wants to go ballooning there. It decomposes at about 300 C to sulfur trioxide and water, and sulfur trioxide decomposes in turn to sulfur dioxide and oxygen. So on the surface, it's SO2 that one has to worry about. Hot concentrated sulfuric acid can react with carbon (Sulfuric acid - Wikipedia), so it may be able to corrode graphene. There are at least some materials that resist sulfuric acid, however, as ought to be evident from the continued structural integrity of lead-acid batteries and laborarory glassware.

 

[lpetrich]

Time to dig out the old 8-bit CPU designs? ;)

Many embedded systems still use 8-bit chips, so one can use some of those designs. Their main value is their low power consumption -- they don't have as many transistors as 32-bit ones do. That's also why many embedded systems still use 16-bit chips.

 

[marcusl]

There are at least some materials that resist sulfuric acid, however, as ought to be evident from the continued structural integrity of lead-acid batteries

That's a relief. We can build our space vehicle out of lead and just strap 6 or 10 rockets together to hoist it into space

 

[lpetrich]

That's a relief. We can build our space vehicle out of lead and just strap 6 or 10 rockets together to hoist it into space

That is not necessary. http://www.electrical4u.com/construction-of-lead-acid-battery/

The glass, lead lined wood, ebonite, hard rubber of bituminous compound, ceramic materials and molded plastics are having the above mentioned properties, hence the container of lead acid battery is made of either of those materials.

Ebonite is a kind of hard rubber.

Venus's sulfuric acid is only a problem at cloud altitudes, and the temperature there is around Earth temperatures. So coating everything with plastic ought to work. There's even a commercial product that does that: http://www.neptuneresearch.com/product/acid-coat/

 

[Jeff Rosenbury]

Yet plastic is a poor thermal conductor. Thus coating the probe with plastic will limit thermal dissipation.

Perhaps we could place the probe in a shell that breaks open after passing through the acid? The inside could have some antacid to handle any residuals.

 

[Baluncore]

Time to dig out the old 8-bit CPU designs? ;)

Initially it would be better to use redundant serial processors, with only 1bit each. That will be lower power and more fault tolerant, but they only run at the speed of dark. Speed is not important when you are stuck on the surface of Venus because you do not have to run MS Windows. Serial processsors, such as the PDP-8/S were used in the early days of computing for mobile applications in the oil industry. The CPU of the PDP-8/S, had only about 519 logic gates.

 

[mfb]

Perhaps we could place the probe in a shell that breaks open after passing through the acid? The inside could have some antacid to handle any residuals.

That is the standard procedure. You enter the atmosphere with a heat shield but you don't want to land with it. If you can keep it long enough to travel through the clouds, it doubles as acid protection (might need some more components to fully cover the probe from all sides).

Speed is not important when you are stuck on the surface of Venus because you do not have to run MS Windows.

You still run against the time the probe will stay operational on the surface.

 

[lpetrich]

Speed is not important when you are stuck on the surface of Venus because you do not have to run MS Windows.

You still run against the time the probe will stay operational on the surface.

I seriously doubt that running its control logic would be very time-consuming. It does not seem to be for most other spacecraft .

 

[mfb]

It was relevant at the time of the Apollo missions, and they had more than a few hundred transistors. Okay, modern transistors have a higher speed.

 

[Baluncore]

I see no reason why a probe on the surface of Venus should not be designed to last for a year or more. What will a probe on Venus be doing once it is on the surface? Any power supply, instruments and peripherals will also need to be high temperature. Perhaps remote control of a camera of some sort and managing the communications link(s) between surface and an orbiting relay satellite would be the major tasks. Is a Venusian Clarke Orbit possible with such a slow spin so close to the Sun?

 

[mfb]

A proper venus-stationary orbit is not possible, but you can go the L1 or L2 points where you have a long visibility time (~60 days). Or go to a lower orbit and transmit data half of the time. I guess the lower orbit is the better option, the short transmission distance allows to get a reasonable rate.

 

[the_emi_guy]

Use plutonium as a power source, power will be needed for other stuff anyway.
Reserve some of that power to cool the well insulated electronics.

No need to build a crude CPU from a handful of diamond transistors.
You could put dozens of high end risk processors in there for redundancy if you want. Unit will function for as long as the plutonium lasts.

 

[mfb]

RTGs have bad efficiencies even on Mars and in deep space where the cold side is very cold. With a ~500°C "cold" side their power output looks problematic.

 

[meBigGuy]

I think it is wrong to think you need super high temperature transistors technology to build whole computers just because you want to go to Venus. You obviously need high temperature technology, but I think that is going to be mostly mechanical with limited high temperature electronics for that which absolutely needs to be in contact with the environment.

We can put rovers the size of SUV's on other planets,now.

I would think you would build your sophisticated electronics within a cooling system that would make the electronic system a manageable task. The interfaces with the external environment is a challenge.