Computer memory that can endure extreme temperatures has been developed by a team led by the University of Michigan. This new solid-state memory device is capable of functioning at temperatures exceeding 1100°F (600°C), surpassing the surface temperature of Venus and the melting point of lead. The development involved collaboration with Sandia National Laboratory.
“It could enable electronic devices that didn’t exist for high-temperature applications before,” said Yiyang Li, assistant professor of materials science and engineering at U-M and senior corresponding author of the study published in Device, a Cell Press journal.
The device currently holds one bit, comparable to other high-temperature memory demonstrations. According to Li, further development could potentially allow it to store megabytes or gigabytes of data. However, for devices not constantly exposed to extreme heat, new information can only be written above 500°F (250°C). A heater might address this issue for lower temperature operations.
The technology functions by moving negatively charged oxygen atoms instead of electrons. Traditional silicon-based semiconductors begin conducting uncontrollable current levels when heated above 300°F (150°C), which can erase data. In contrast, oxygen ions in this device remain unaffected by high temperatures.
These ions move between two layers—the semiconductor tantalum oxide and metal tantalum—through a solid electrolyte barrier. Three platinum electrodes guide whether oxygen is absorbed into or expelled from the tantalum oxide layer, resembling battery charging processes but used here for data storage.
Depending on its oxygen content, tantalum oxide acts as either an insulator or conductor, allowing it to switch between digital 0s and 1s voltage states. Finer control over the oxygen gradient may enable computing within the memory itself with more than 100 resistance states rather than simple binary ones.
“There’s a lot of interest in using AI to improve monitoring in these extreme settings," said Alec Talin, senior scientist at Sandia National Laboratories and co-author of the study. "In-memory computing chips could help process some of that data before it reaches the AI chips and reduce the device’s overall power use.”
The device retains information states above 1100°F for over 24 hours while offering benefits like lower voltage operation compared to alternatives such as ferroelectric memory. It also provides more analog states for in-memory computing.
This research received funding from several sources including the National Science Foundation and was conducted at facilities within the University of Michigan. A patent application has been filed with plans to commercialize this technology.
Alerts Sign-up