USC Memory Chip Operates for 50 Hours at 700°C, Outlasting Venus Landers
The Chip That Could Survive Venus
Every piece of electronics you have ever owned shares a critical weakness. Push it past roughly 200 degrees Celsius and it begins to fail. Your phone, your car's computer, the satellites orbiting above your head right now — all of them have the same thermal ceiling baked into their design. For decades, that ceiling has been one of the most stubborn walls in engineering. Now, a team at the University of Southern California may just have broken through it.
In a paper published March 26 in the journal Science, researchers led by Professor Joshua Yang report a new type of memory device that kept working reliably at 700 degrees Celsius — hotter than molten lava and hotter than the surface of Venus, which has defeated every lander ever sent there, destroying most electronics within minutes and the sturdiest probes within a few hours. The USC device ran for more than 50 hours at that temperature without needing a refresh, survived more than one billion switching cycles, and operated at 1.5 volts with speeds in the tens of nanoseconds — specs that would be respectable for a chip designed for an air-conditioned server room, let alone something sitting on the surface of another planet. The first author is Jian Zhao.
"We tried to build a different graphene device," Yang said. "To be honest, it was by accident, as most discoveries are."
The mechanism turns out to be elegantly simple. The team built the memristor with tungsten as the top electrode — the element with the highest melting point of any metal — a layer of hafnium oxide ceramic in the middle, and graphene at the base. When the device heats up, tungsten atoms tend to drift and migrate, eventually causing the kind of atomic-scale short circuits that kill conventional electronics. But graphene does something unexpected: its surface chemistry with tungsten acts, as Yang put it, "almost like oil and water." The tungsten atoms that drift toward the graphene surface cannot take hold. The structure holds.
This is not a roadmap claim. The device was actually built and tested. Seven hundred degrees was not the device's limit — it was the limit of the test equipment. The actual failure point was never found.
The research was carried out through the CONCRETE Center at USC, shorthand for Center of Neuromorphic Computing under Extreme Environments, sponsored by the Air Force Office of Scientific Research and the Air Force Research Laboratory. Yang has co-founded a startup, TetraMem, with Qiangfei Xia, Miao Hu, and Ning Ge, to commercialize room-temperature memristor chips for AI computing — the Venus-friendly variant is a separate track, but the underlying physics applies in both directions.
The Venus application is the most striking but not the only one. Space agencies have long needed electronics that can operate above 500 degrees Celsius — the rough surface temperature of Venus — because deep space missions routinely encounter temperatures that exceed what silicon can handle. Jet engines, nuclear reactor instrumentation, oil and gas drilling electronics, and industrial processes all have the same problem. Any environment where heat is a constraint, and where replacing hardware is difficult or impossible, is a candidate.
Yang was direct when asked to characterize the result: "You may call it a revolution. It is the best high-temperature memory ever demonstrated." That is not a quote you see often in academic papers, and it is not the kind of thing researchers say casually when they have something incremental to report.
The gap between a lab demonstration and a mission-ready part is real and should not be dismissed. The device was tested in a controlled environment, not sitting on the Venutian plain under 90 atmospheres of pressure and a corrosive atmosphere. Getting from here to there requires more engineering than the USC team has published so far. But the fundamental physics appears to be solved. The question now is not whether high-temperature memory is possible — it apparently is — but how fast it can be packaged into something that survives actual Venus conditions, not just the thermal part of them.