Fingerprinting silicon chips just got easier

Single-electron tunneling offers a technique to count the number of electrons confined within in a small space. These electrons exhibit spin-qubit behavior and can be used to investigate Kondo effects and quantum computing. There are many trap sites in devices, and electrons are trapped randomly. Scientists can map these unique characteristics to fingerprint silicon transistors in chips. A new study explores the use of single-electron traps in conventional silicon transistors to identify physically unclonable function (PUF).

The PUF is a unique structure that cannot be duplicated in the material, like the recognizable knot in an old oak tree. These features arise as transistors and circuits are fabricated and offer an inexpensive way to identify a device for security purposes, like protecting personal information.

“Our method can be equipped to many LSI (large-scale integration) chips, because we are using conventional transistors and the detection circuits are considered to have small area in LSI chips,” said Tetsufumi Tanamoto, first author on the study. “When other ordinary security systems are not sufficient, customers can take their chips to some institutes or universities to check their fingerprints.”

Tanamoto et al. examined single-electron tunneling at sufficiently low temperatures (approximately 1.54 Kelvin) and were able to show the characteristics, like current and voltage, could be used as fingerprints of chips using image recognition algorithms (AKAZE, BRISK, and ORB). The team applied Open Source Computer Vision Library image matching software to align edge, corner and flat landmarks to evaluate differences between images.

Future studies plan to explore the clear quantum behavior of electrons caught in the traps as an advanced way to fingerprint a chip.

Source: “Application of single-electron effects to fingerprints of chips using image recognition algorithms,” by T. Tanamoto, Y. Nishi, and K. Ono, Applied Physics Letters (2019). The article can be accessed at https://doi.org/10.1063/1.5100644 .