The use of optical techniques for attacking integrated circuits (ICs) is increasingly being reported, particularly the nefarious extraction data from embedded SRAM. Such attacks can provide access to highly sensitive information such as encryption keys and bypass various security measures. Attackers usually exploit one of several interactions between light and semiconductors to generate logic-state images that reflect data in memory. Thermal laser stimulation (TLS) and laser probing via electro-optical frequency mapping (EOFM) have been reported in the literature, but photoelectric laser stimulation (PLS) gets little attention. Considering the potential advantages of PLS over other techniques (e.g., less power is required to generate current-voltage changes and the effect can be triggered at shorter wavelengths, which can lead to improved spatial resolution), the authors set out to determine if logic state images can be generated from various types of devices with PLS and assess the strengths and limitations for each case. The results of the investigation are presented in this paper.

Limited spatial resolution and low signal to noise ratio are some of the main challenges in optical signal observation, especially for photon emission microscopy. As dynamic emission signals are generated in a 3D space, the use of the time dimension in addition to space enables a better localization of switching events. It can actually be used to infer information with a precision above the resolution limits of the acquired signals. Taking advantage of this property, we report on a post-acquisition processing scheme to generate emission images with a better image resolution than the initial acquisition.

For VLSI, internal electrical measurements are key steps to solve design debug issues and to perform failure analysis. Due to multiple metal layers, active areas of the chip are only accessible from the backside of the die. The ability of optical contactless techniques to operate through the silicon substrate and the few sample preparation required have widely contributed to promote them as unavoidable tools of the defect localization workflow. Timing issue or unusual consumption can be detected by static and dynamic photon emission analysis. The identification of the emission spots is an essential step of the process. Due to scaling, more and more emission nodes are located within the acquisition area so that large variations of emission intensity can exist. Because of various limitations, former thresholding techniques cannot ensure an exhaustive localization. In this paper, an automated process is reported to locate spots in these complex areas. We will underline the challenge and define application boundaries of this technique.

Time Resolved Imaging (TRI) acquisitions allow precise timing analysis of emission spots. Up to date technologies deeply challenge their isolation by hiding the weak ones, under sizing or over sizing visually detectable emission spots and finally by jeopardizing timing resolution. We report on an algorithm based on 1 and 2D signal processing tools which automates the identification of emission sites and optimizes separation between noise and useful signal, even for weak spots surrounding strong emission areas. The application of the algorithm on several sets of data from different types of devices and their results are also discussed.