Researchers reveal chips that commit 'circuit suicide' — self-destruction and counterfeit protection in one

As part of IEEE Spectrum's coverage of security advances made at their International Solid-State Circuits Conference (ISSCC) held last week, a team from the University of Vermont showed off a way to manufacture chips (like CPUs) that self-destruct when compromised. This doubles as both a security measure and an anti-counterfeiting measure, which is useful for vendors who wish to protect their designs.

The mechanism works by using physically unclonable functions (PUFs), which can create unique fingerprints for every chip. The PUF, as implemented by the University of Vermont team debuting this self-destruct mechanism, also comes with an extra bonus of two built-in methods of "circuit suicide."

Both methods used pumped up voltages to the lines connecting to the encryption key. In the first method, this causes electromigration, a phenomenon where electricity blows metal atoms out of place, which causes voids and open circuits. The second method instead causes short-circuiting by subjecting transistors meant to operate at under 1V to 2.5V, which causes such rapid time-dependent dielectric breakdown (transistor aging) that the device is killed.

Image 1 of 14

The team from the University of Vermont was assisted by Marvell Technology and led by Eric Hunt-Schroeder, who works at both. The inspiration for this stems from reports that researchers were able to clone PUFs based on SRAM by using an electron microscope. According to Hunt-Schroeder, this method ensures that "When you're done with a part, it's destroyed in a way that renders it useless." The demonstration was conducted with a 3nm test chip.

If self-destruction tech sounds familiar to you, it could be because we've recently covered two different self-destructing USB drives: one from Russian high-tech conglomerate Rostec, and an indie "Ovrdrive" with a unique unlock mechanism that needs modification to enable the self-destruct function.

Other security advances discussed during IEEE Spectrum's blog post include a probe-attack alarm and a built-in electromagnetic signal obfuscation method that prevents EM scanners from cracking keys even after 40 million attempts. Without this protection, the same scanner managed it in 500 tries.