SEMBA: A SEM based acquisition technique for fast invasive Hardware Trojan detection

Abstract

In this paper, we present how SEMBA, a fast invasive technique for white team Hardware Trojan detection, has been used to differentiate between a maliciously infected integrated circuit and a genuine one. Our methodology is based on the observation of the component's hardware structure and includes the use of wet etching, Scanning Electron Microscopy and Multiple Image Alignment. Once the Integrated Circuits' image have been fully reconstructed, image processing allows to detect the presence of the Hardware Trojan (HT). SEMBA is a fully automated approach with a 100% success rate, detecting any `transistor-size' HTs and requiring `affordable' resources and time.