Radiation, Temperature, Power Challenges For Chips In Space

Mission-critical hardware used in space is not supposed to fail at all, because lives may be lost in addition to resources, availability, performance, and budgets. For space applications, failure can occur due to a range of factors, including the weather on the day of launch, human error, environmental conditions, unexpected or unknown hazards and degradation of parts to chemical factors, aging, and radiation.

Radiation is the biggest concern in terms of reliability and safety for space chips, especially when compared to automotive, as it doesn’t exist to the same extent on the terrestrial level.

“A problem that you see on the ground might be accelerated at altitude, and when you go to space, there’s no atmosphere anymore, so you are exposed to all the particles that fly around there. Those particles literally have been flying around sometimes for billions of years, and then they hit your semiconductors,” said Helmut Puchner, vice president, fellow of aerospace and defense at Infineon Technologies. “The sun determines a lot of our radiation environment. And the further you go out, the worse it gets.”

When a cosmic ray or a very high energetic particle slams into normal silicon, it knocks off electrons. These flow around like they were current, and as a result the transistor sees it as incoming current. “Of course that wasn’t supposed to be there, and radiation can really disrupt the silicon chip,” noted Marc Swinnen, director of product marketing at Ansys.

Space chips may be hit with 5,000 particles every 40 nanoseconds, and it will keep on happening, according to Varadan Veeravalli, principal functional safety engineer at Imagination Technologies. “Some particles don’t have any impact at all in the system, but some particles can amplify itself and cause a single-event latch-up [SEL], which is created by a particle strike, and there is no recovery from that. An SEL is more like a permanent fault that is a continuous displacement of current and nothing can be done about it. For this reason, you should make it a radiation-hardened device, such that SELs do not happen to it, and even the total ionization dosage should not have any impact on it. Displacement of ions should not happen, because these things can destroy the device, and it would not serve the purpose. That’s why there is a focus on having radiation hardened standards, libraries, layouts, and fabrication.”

Others agree. “It comes down to putting one’s finger on the map of the technical requirements, and saying this microelectronics system must be able to perform correctly and functionally in this type of hostile environment, and then all of the engineering resources dial in,” explained Scott Best, senior director for silicon security products at Rambus. “Depending on where you put your finger on that map, it is more expensive, requires more testing, requires more rad hard techniques — both rad hard by design techniques and rad hard by manufacturing techniques. So different techniques can be used to minimize the effect of single-event upsets due to a malicious, hostile environment like space or a strategic nuclear event. Then the system can be designed, and the cost can be factored in.”

A worst-case scenario involves nuclear missiles. “The missile, when it’s flying, cannot come down because somebody detonated it already,” said Puchner. “It has to operate through that event. That’s the highest level [of radiation], because in a nuclear event, you have a fast burst of neutrons followed by a strong pulse of X-rays. You have to survive that.”

As for satellites, lower-orbit constellations typically do not receive a lot of dangerous radiation, unless they pass over the poles or off the coast of Brazil, because of the South Atlantic Anomaly. But geosynchronous satellites, 35,000 kilometers out from Earth, have to survive all the brunt of the radiation from the sun and space.

Rad hard solutions include gallium arsenide (GaAs). “It’s much more resistant to these effects,” said Ansys’ Swinnen. “Also, there are design techniques that allow for redundancy, so if a particle hits and flips a transistor, there are others in place. There are also specific processes typically used by the military and aerospace that are designed to be radiation hardened and much less sensitive. In a sense, they’re traditional chip designs, except you have different technologies and different design rules that have to be adhered to.”

One challenge for engineering teams is that it is difficult to create a rad hard version of every type of processor or memory.

“If you have to build all the variations of a rad hard memory, it’s going to be expensive,” said Charlie Schadewitz, vice president for aerospace and defense at Cadence. “So you do one of two things. You either throw a lot of money at it and build a lot of variations, or you limit the size of the different variations of memory you can use in your rad hard design.”

Infineon’s Puchner agrees. “Every kind of technology has challenges, so it depends on the components you’re looking at, but typically there’s only a small subset of electronic components that are designed to withstand that radiation. Everything can be done, but currently most of this activity is funded by the U.S. government since it requires a complete infrastructure. It’s not only that you need to build it in the fab. You need to build it with libraries with logic library elements, with IP that must be proven first, then developed. The prime contractor that’s been doing this for the last 20 or 30 years is IBM. Then it became Global Foundries. They have contracts with the U.S. government and with third-party prime contractors that create those libraries, and those libraries can then be used by certain projects to design those products.”

The economics of radiation hardened by design products is challenged by a limited customer base.

“You’re sometimes dealing with technology that’s not the latest on the market, because it takes a long time to create these libraries, to calibrate everything, to verify they’re really radiation hardened,” said Puchner. “Then, when you put everything together, there’s still no guarantee that the chip is radiation hardened. You have to test the chip and debug the chip and make sure it’s working. This is a very expensive endeavor. That’s why nuclear missiles cost a lot of money for the taxpayer — to the tune of approximately $30 trillion over 10 or 20 years. So to design a radiation hardened by design NOR flash, you almost need to have government funding to do it. Our 512 Mbit NOR flash for space applications was funded by the U.S. government.”

Not all technologies are susceptible to radiation. Some technologies are inherently rad hard because they do not depend on electric charge. This includes optical waveguides, phase change memories, MRAMs, and other memory technologies that store a state of matter rather than a charge.

Challenges around plasma, heating, cooling, and aging
There are other idiosyncrasies with chips in space. Computational fluid dynamics (CFD), for example, doesn’t necessarily apply because there are no fluids in space. They either sublime or freeze.

“What you do have are plasmas,” said Ansys’ Swinnen. “With the high voltages and high temperatures, especially during re-entry or going through radiation, you have plasma effects. Plasmas are very high temperature materials, to the point where the electrons are ripped off the atoms and they’re all just free-moving charged ions, very highly energetic. It’s what you find inside a neon tube. The gas there is ionized and glows. We use a tool for plasma etching in the manufacturing of chips and other industrial uses of plasma. It’s really the fourth state of fluid.”

Traditionally, NASA and other space agencies use the highest-level reliability components, which appear in the Defense Logistics Agency-certified Qualified Manufacturers List.

“Those components are tested, screened and inspected to a level that’s way beyond what anybody would ever do in automotive,” said Puchner. “Most are built in ceramic packages. They’re bulkier and heavier, but can withstand temperature and stress better, because ceramic has no issue up to 350 degrees, and especially when it comes to temp cycling in space.”

In ceramic packages the silicon die sits on an epoxy, which glues the parts to the package. Bond wires are then attached to the die and hold it in place.

Chips designed for space also have different thermal conductivity, or superconductivity, for low temperatures and for high temperatures, said Imagination’s Veeravalli. “Here, metallization ceramic composites are used. A lot of these things are being done just to protect against faults that are not controlled. In space, there is no one there to tell the system what it should do. It has to adapt itself. So they need to take all of these factors into consideration and build the system accordingly, and also ensure that the system is able to recover by itself. The only way that can be done is ensure that it’s fail-operational. If any fault happens, it still has to work in spite of the faults.”

In space, temperatures can reach approximately -200 degrees in the shade of the moon, but rise when the sun is unobstructed, causing the silicon to expand. “It’s not bounded by anything,” said Imagination’s Veeravalli. “If you use a plastic component, then the mold compound, which is plastic surrounding the silicon, has a different expansion coefficient and creates some stress. I’m not saying that plastic is less reliable — they are flying more and more plastic now in space — but ceramic definitely has better performance. There’s no risk there. With plastic, you always have risks if you don’t design your chip correctly.”

Temperature also impacts aging, and this can happen at a greater extent when compared to automotive, depending on where the auto components are used.

“If you’re in the vehicle hoods, you’re talking extended temperatures right up to 130° to 140° C, depending on how close they are to a combustion engine, so that temperature is always a big factor,” said Infineon’s Puchner. “As for the use case, automotive components are not used continuously. On average, people use their cars maybe 4,000 hours per year. Satellites usually are used 24/7, depending what function they have, and they are rarely switched off. In case of a solar event that is predictable, when there are solar flares hitting the earth, they preemptively switch it off to avoid any damage. But that’s very rare. Usually, geosynchronous satellites that are used for communication or for critical government infrastructure, aka missile detection, have to be up all the time, and they’re designed for minimum lifetimes of 15 years but typically exceed that.”

Aging is therefore calculated into the components themselves. For Defense Logistics-certified components, this means they must demonstrate one critical test that is burns in or accelerates that stress for 4,000 hours at 125° C, assuming an average operating temperature of, say, 70° to 80°. “That gives proof the component can survive those lifetimes,” Puchner said.

Aging has wide-ranging impacts. “This is what we’re doing with the space station, all the experience and learning we’re doing there, because microgravity has an effect on electronics and people, and so does the background radiation of space,” said Todd Tuthill, vice president for aerospace and defense industry at Siemens Digital Industries Software.

Cooling is also dealt with differently in space than on Earth. The problem with cooling is that space is a vacuum, so the classical convection cooling fan doesn’t work. “Cooling is done through heat exchanger radiation, so you can only radiate off heat,” Puchner said. “In the space shuttle, it has big loading docks and big doors open up. You’ve seen an image of the space shuttle in space with the doors open. Why keep the doors open? It’s because they act as radiation shields for the heat exchanger so they can cool the electronics in the head of the space shuttle. So, it’s a little bit different. You need to have a heat exchanger. If your satellite is in a stable orbit, then it’s very predictable, and it’s designed this way to have a control temperature. So, you can control the temperature better, but it’s not free. You have to design a system around it to have a control temperature. If you are sitting in a sensor on the outside of a spacecraft, you are experiencing all the temperature variability and, in general, every semiconductor performs better at cold temperature because the mobility is higher. And if you don’t have a risk condition, then typically it performs better, with less leakage and less power consumption. You can take advantage of that, as well.”

Another concern in space applications is electrostatic discharge.

A particle in space is electric charge — it’s hitting the semiconductor and causes some problems. “For example, on a spacecraft, if there is heavy proton exposure, it can charge up any metal,” Puchner said. “The protons get absorbed by the metal, and the metal will just collect the charge and will charge up. At some point it will discharge somewhere. It would spark somewhere. It’s like when you have a fur and you’re rubbing plastic on it. You’re charging up electrostatic charge. Spacecraft designers have to consider electrons in the same way. There’s a huge electron belt that can hit you, so it’s the same thing. The metal can charge up and cause a discharge event.”

In addition, when chips are getting bombarded with alpha waves and particles, their reuse is completely different. Everything may have to be retested in a different way than before.

Processing, power, and communication trends
Space-targeted technologies have changed significantly over the last 10 to 15 years, particularly processing cores.

“FPGAs were typically 10 years behind. Now we are maybe two years behind,” Puchner observed. “The latest AMD Xilinx FPGA that’s used in space is a very powerful engine, has AI cores, and enables a lot of implementation. There also are some companies flying the latest NVIDIA boxes – not the high-performance one, because you don’t have enough power on a satellite. For the big satellites, maybe we limit it to 3 kilowatts, or something in that range of power, because you cannot make the solar panels infinite. You cannot fly the best, top-notch ground systems and build a data center. You don’t have enough energy. But there are some lower power systems, like the Jetson AGX, that’s actively pursued and flying right now. For any low power AI chip that you hear about in the next couple of years, people will try to fly it and see how it behaves. There are a lot of new space companies. They pick up an architecture and ask, ‘Hey, you want to fly it?’”

Solutions are coming from multiple directions to create more power in space and potentially enable higher performance computing. Siemens’ Tuthill sees many opportunities to attack the problem from multiple sides, creating more power, using less power, and finding alternate ways to put power on orbit in spacecraft.

Solar is one obvious solution. “Maybe part of the answer is to have better battery storage systems to store the power collected, and more efficient solar collectors,” he said. “Then you’ve got things like SMRs, small modular nuclear reactors. To my knowledge, there aren’t any nuclear-powered spacecraft, but I think there’s a real potential for nuclear-powered vehicles in the not-too-distant future.”

The public concern around nuclear stems from past failures, but technology exists today to make those things far more reliable and far safer. “Obviously, we’ve got to figure out what to do with the spent uranium eventually,” he said.

The biggest problem the space industry is facing right now is related to Earth observation, image detection, and bandwidth, because communication is always up and down.

“Our sensors can detect so many images so quickly, and generate gigabits of data, terabits of data, very quickly,” said Puchner. “How do you know which ones to keep and which ones to throw away? Historically, you are collecting the data and sending the images down to earth, and then they get analyzed on earth. You cannot afford that. The downlink is still not fast enough to do that. So we are creating systems now on boards that collect those data, analyze them, and prioritize them. For super-critical data, we send them through the mesh to other stations and then downlink, or direct downlink, or we store them locally and wait for the next downlink.”

In terms of Earth observation, the U.S. Space Development Agency is driving advances through a project for its next-gen missile defense program. “It’s using a communication layer, which is a lower-orbit network, a transmission layer that communicates into the geosynchronous satellite, and a tracking layer that’s really tracking events,” Puchner said. “This is very important for the U.S. government to drive that program, because it will give us a continuous coverage around the planet.”

Another challenge is that the more people become dependent on satellite constellations and connectivity, the less carriers will invest in terrestrial infrastructure, particularly in rural areas.

“What does that mean for 911 emergency situations?” said Puchner. “Keep in mind those constellations are not designed to survive nuclear events in space. They’re not designed to survive mechanical events in space. It’s very tricky, and they’re flying at about 550 kilometers away, or for a different implementation, between 325 to 900,000 kilometers, in that range. There’s a lot of space up there. But any missile from, say, North Korea can reach that altitude easily. So protecting that asset becomes another problem. You can say we have 4,000, what if I lose a couple? But it’s not that easy because there are some events that can take out the whole infrastructure. And then what are you going to do? The financials will talk, and then the best thing is to make sure everybody’s not fighting over that supremacy in space and keep it open. Any event up there, like demo or collision of satellites, creates a lot of debris, and that debris will travel forever. You cannot collect it.”

Space debris
A U.S. intelligence program, SINTRA, tracks space debris to make sure it is not hitting satellites. Meanwhile, Los Alamos National Laboratory developed postage stamp-sized “license plates” to help track and protect satellites in low Earth orbit and a “Spacecraft Speedometer” to predict satellite location without GPS.

Fig. 1: The U.S. Air Force tracks and shares data on over 19,000 orbiting objects. Here, satellites are shown in red, rocket bodies in blue, and other debris in gray. Credit to: James Yoder, Stuff in Space. Source: Los Alamos National Lab

For rockets, nothing should come back to earth but it’s always possible that some debris will, especially with the launchers now re-using stages, Puchner said. “When a satellite deteriorates, it usually blows up in the atmosphere, so there’s nothing left. It’s not like avionics, where you have to demonstrate that if your avionic system crashes somebody can’t reconstruct the image or the content of your memory. It’s very unlikely and impossible.”

Conclusion
Space is considered the final frontier of human exploration, whether that is to find life on other planets, colonize Mars, or find new resources.

In fact, autonomous robots could help with resource mining, data collection, and infrastructure construction on Mars or the moon.

Fig. 2: Advanced AI tools perform tasks that would be too dangerous, costly, or impractical for astronauts. Source: Cadence

Meanwhile, at Rochester Institute of Technology’s Center for Detectors, a NASA-funded project recently achieved a first light image using a single photon imaging detector. The goal is to detect life on other planets, but first they need to prove the new sensors can maintain extreme sensitivity while exposed to radiation in space.

“One of the key concerns is not only that I have to protect the people on the spacecraft that’s going to travel all the way to Mars, but I’ve also got to protect all the electronics,” Siemens’ Tuthill said. “We’re going to put people on Mars, maybe not in my lifetime, but maybe in my grandchildren’s lifetime, and we’ve got a lot to learn about how people and electronics and things survive in the background-radiation environment of space.”

Related Reading
Stakes Are High For Aerospace, Defense IC Designs
Mission-critical applications involve higher levels of abstraction than automotive and often need to meet safety-critical requirements, as well.
Extra Safety Measures Needed For Aerospace ICs
Fail-safe operations and boot load memory are essential for ensuring pricey resources aren’t lost in space.