Hardware Trojan Detection Potential and Limits with the Quantum Diamond Microscope

Abstract

Assesses whether a hardware Trojan implanted in a fabricated integrated circuit can be detected by magnetic-field imaging with a quantum diamond microscope, under a threat model of a compromised foundry that produced a subset of implanted chips. Its central argument is that test patterns need not activate the trigger, because the current drawn by a Trojan's trigger-monitoring activity is itself detectable if the measurement is sensitive enough, and it projects that the instrument's sensitivity at moderate measurement times is adequate across a wide range of process sizes. The authors state three limits: the sensitivities assume a stand-off of about 10 micrometres, which in practice means a decapsulated chip; an environmental noise floor prevents detection without electromagnetic shielding; and the capability is a projection from physical limits rather than an end-to-end demonstration.