New Method Accelerates Development of Radiation-Hardened Chips to Meet Long-Term Satellite Operation Needs

Scientists at the Pacific Northwest National Laboratory (PNNL) are collaborating with autonomous artificial intelligence to develop lighter and more durable satellites. These scientists are leveraging superintelligence (SI) and automation technologies to accelerate the development of radiation-hardened materials designed to withstand high-radiation environments, such as space and nuclear reactors. Failures caused by radiation in these components can delay or cancel missions, making radiation hardening essential for operations lasting up to ten years. Erin Barker, a scientist at PNNL, stated, "Space is a high-energy environment, with heavy ions and other particles constantly bombarding our equipment."

The emergence of autonomous workflows comes at a time when we need better radiation testing technologies.

Researchers are investigating layered aluminum gallium nitride and gallium nitride chips that power satellites, communications, radar, and power systems. Before microchips are sent into space, they are produced in a clean facility where even a tiny speck of dust can lead to immediate failure. Once launched into space, these fragile microchips enter a hazardous environment filled with continuous, invisible radiation. Galactic cosmic rays, heavy ions, and near-light-speed electrons permeate the universe. Exposure to adverse conditions can lead to memory corruption in satellites, causing spacecraft to veer off course or completely terminate expensive meteorological or communication missions.

New Technology Accelerates Development of Radiation-Hardened Chips

Testing materials' performance in space environments has long been a slow and cumbersome process. To prevent these orbital disasters, PNNL researchers are changing the way we protect space electronic devices. This new research accelerates the development of radiation-hardened microchips using a multi-agent artificial intelligence system called SciLink. This intelligent software system can assess microscopic radiation damage in just a few hours, replacing the weeks of analysis previously required by human teams. While physical shielding or backup systems can protect electronic devices, the added weight increases launch costs and reduces overall mission efficiency.

Therefore, enabling the microchips themselves to possess radiation resistance may be a more practical solution.

How the SciLink System Works

To achieve this goal, advanced chips made from aluminum gallium nitride and gallium nitride need to be tested under extreme simulated conditions. PNNL researchers utilized ion beam experiments at the University of Tennessee to simulate years of cosmic radiation in just a few hours, generating large and complex datasets. Typically, integrating data from instruments such as atom probe tomography and scanning transmission electron microscopy takes weeks. SciLink organizes its artificial intelligence agents into a corporate hierarchy led by three "project managers," who oversee experimental analysis, planning, and computational simulations.

During interactive meetings, these AI managers discuss hypotheses, synthesize multi-instrument data, evaluate damage mechanisms, and propose follow-up experimental suggestions. Jenna Pope, a scientist at PNNL, stated, "The system allows us to consider all measurements, piece together what happened, and understand the causes of damage."

The AI also assigns a numerical confidence score to each conclusion. In recent tests, SciLink reported that heavy ions dominated the damage to the primary microchips, achieving a confidence score of 90%.

Meanwhile, for defects caused by protons, SciLink's confidence score was reported as 60%, indicating to researchers where further investigation is needed. Despite the rapid pace of autonomous workflows, experts emphasize that humans still play a crucial role in designing experiments, performing calculations, and carefully checking errors identified by the AI. Barker even stressed, "The intelligent agents are a tremendous help to us, but you can't just turn off your brain."

Applications of Artificial Intelligence in Radiation-Hardened Material Development

PNNL's research demonstrates the potential of artificial intelligence in scientific research, particularly in the development of radiation-hardened materials. As space missions increase the demand for the durability of electronic devices, traditional testing methods can no longer meet efficiency requirements. The introduction of the SciLink system not only accelerates the speed of data analysis but also enhances the understanding of radiation damage. The success of this new approach may drive future applications in other fields, showcasing the importance and potential of artificial intelligence in scientific research.