Photon Emission Microscopy (PEM) is a popular technique for microelectronics failure analysis by detecting the photon emission from a defective circuit, when a failing device is electrically exercised at certain voltage. The photon emission contains physical location information, photon emission spectral information and photon emission intensity information. People often use the physical location information to localize a defective circuit and guide the follow-up physical failure analysis to find the defects. However, this procedure does not always work. Sometimes, it shows no defect found (NDF). In this paper, we propose a new computer vision-based analysis of the photon emission intensity for identifying the root cause of the excessively high IDDQ at elevated Vdds. The procedure includes collecting photon emissions at different Vdds and a follow-up photon emission intensity analysis with computer vision techniques. The procedure was applied on a case of microprocessor chip. After analyzing the dependencies of photon emission intensity on Vdd for 4 types of circuits, it was concluded that the SRAM circuit leakage is the root cause of the excessively high IDDQ at elevated Vdd.

In this paper, we discuss the use of spontaneous photon emission microscopy (PEM) for observing filaments formed in HfO 2 resistive random access memory (ReRAM) cells. The setup employs a CCD and an InGaAs camera, revealing photon emissions in both forward ( set ) and reverse ( reset ) bias conditions. Photon emission intensity is modeled using an electric-field equation and inter-filament distance and density are determined assuming a uniform spatial distribution. The paper also discusses the use of high frame rate and prolonged photon emission measurements to assess lifetime and reliability and explains how single filament fluctuations and multiple filaments in a single cell were observed for the first time.

In this paper, we present the first prototype of a Scanning Time-Resolved Emission (STRE) system consisting of a high-sensitivity, low-noise, and low-jitter single-point Superconducting Single-Photon Detector (SSPD) combined with a specialized scanning head of a Laser Scanning Microscope (LSM). This idea was first proposed in late 2006 [1] but required the right combination of detector, customization, and collaboration with a tool vendor to get to fruition. It should be understood that this is still a prototype system under development and significant improvements in acquisition time, resolutions, and performance are expected in the near future. In this paper, we will also present the first preliminary results acquired using a test chip fabricated in 32 nm SOI.

In this paper, an automated contactless defect analysis technique using Computer Vision (CV) algorithms is presented. The proposed method includes closed-loop control of optical tools for automated image collection, as well as advanced image analysis methods to improve image quality and detect potential defects. As an example, the technique was successfully used to identify delamination defects along the perimeter of a large test chip.

In this paper, we present a technique for device temperature measurement using spontaneous near infrared (NIR) emission from an Integrated Circuit (IC). By leveraging modeling and data analysis, time-integrated emission measurements are used to estimate the temperature increase due to switching activity inside the channel of CMOS transistors. The non-invasive nature of the technique allows one to reliably monitor the temperature of any device on-chip without the need for circuit modifications or dedicated on-chip sensors and with a higher spatial resolution than thermal cameras. This method has important applications for modeling heat dissipation during early process development, localizing hot spots, calibrating on-chip sensors, etc. In this paper, temperature is estimated by fitting empirical emission data to an emission model that can be solved for device channel temperature.

In this work, we demonstrate the effectiveness of deconvolution algorithms in improving the spatial resolution of time-integrated emission images from integrated circuits. A mathematical model of the Point Spread Function (PSF) encompassing both the optical system and the imaging detector properties is used for the deconvolution process. Tuning of the PSF parameters is achieved through the minimization of dedicated cost functions that optimize image resolution while suppressing artifacts in the deconvoluted images. The optimized PSF is then used in both the Lucy-Richardson (L-R) and blind deconvolution algorithms. Results from 32 nm and 14 nm SOI devices show that the deconvolution process significantly improves spatial resolution of time-integrated emission images, pushing their resolution beyond the diffraction limit of Solid Immersion Lenses (SILs).

In this paper, the development of advanced emission data analysis methodologies for IC debugging and characterization is discussed. Techniques for automated layout to emission registration and data segmentations are proposed and demonstrated using both 22 nm and 14 nm SOI test chips. In particular, gate level registration accuracy is leveraged to compare the emission of different types of gates and quickly create variability maps automatically.

In this paper, we discuss a set of techniques and analysis methodologies for the reverse engineering and functionality extraction of complex mixed-signal ICs with a special focus for security applications. Front and back side reflected light pattern images at different magnifications are used to identify circuit blocks. Time-integrated and time-resolved photon emission data is used to identify gate logic states, sequences of events, and specific functional activity. Backscattered electron and scanning transmission electron images mosaics are used to reverse engineer individual gates and observe local interconnects. Thermal imaging is used to aid in the functional block identification and analog gates analysis. Different advanced methodologies for tool automation, focusing, mapping, and image processing are also discussed in the context of our proposed electro-optical tester based technique.

In this paper, we present a novel system and method for the automated mapping of pattern and spontaneous photon emission from very large areas of VLSI circuit using Solid Immersion Lens (SIL). To the best of our knowledge, this is the first time that such a technique has been developed and demonstrated on a real chip. The system being presented includes an automation software Application Programming Interface (API) to control the microscope used to acquire the images, an acquisition software that allows to automatically navigate the chip, move (hop) the SIL to the desired location, focus the image after the SIL landing, register the acquired images, and stitch them together to create a high resolution mosaic. In this paper, we will present, for the first time, a real life example involving thousands of images acquired from a 90 nm bulk technology test chip that were used to create a mosaic of more than 25 x 25 images covering a total area of approximately 400 x 400 μm2.

This work presents a new photon emission microscopy camera prototype for the acquisition of intrinsic light emitted from VLSI circuits during their normal operation. This novel camera was designed to be sensitive to longer wavelengths in order to maximize the signal intensities from modern VLSI chips which are characterized by a red shift in the intrinsic emission spectrum. In this paper, we will characterize the performance of the camera using 32 nm and 22 nm SOI chips. The novel camera is able to collect emission images with the circuit under test operating at a supply voltage down to 0.5 V, exceeding the performance of a state-of-the-art InGaAs camera.

In this paper, we present the latest results obtained with a 2D Picosecond Imaging Circuit Analysis (PICA) camera with enhanced Near InfraRed (NIR) sensitivity [1] for taking 2D Time Resolved Emission (TRE). We will discuss key applications where the time-resolved imaging capability is very effective in reducing the debug time and improving the interpretation of the failure signatures of several VLSI chips. Besides conventional chip diagnostics, specific focus will be dedicated to new areas of applications, such as security and reverse engineering [2]. We will also discuss spectral analysis and other techniques that can be used to extract valuable information from the PICA dataset.

This work presents a comparison of two generations of Superconducting nanowire Single-Photon Detector (SnSPD) prototypes used for Time-Resolved Emission (TRE) measurements from VLSI chips. The performance of the systems is compared in order to understand the figures of merit that a single-photon detector should have to enable the acquisition of time resolved emission waveforms for ultra-low voltage applications. We will show that measurements down to a new World record low 0.4 V supply voltage were made possible by a careful optimization of the detector front-end electronics. We also characterized the emission from devices with different threshold voltages in order to understand how the emission contributions depend on this parameter and how this affects the resulting waveform SNR.

Transmission Electron Microscopy (TEM) and scanning TEM (STEM) is widely used to acquire ultra high resolution images in different research areas. For some applications, a single TEM/STEM image does not provide enough information for analysis. One example in VLSI circuit failure analysis is the tracking of long interconnection. The capability of creating a large map of high resolution images may enable significant progress in some tasks. However, stitching TEM/STEM images in semiconductor applications is difficult and existing tools are unable to provide usable stitching results for analysis. In this paper, a novel fully automated method for stitching TEM/STEM image mosaics is proposed. The proposed method allows one to reach a global optimal configuration of each image tile so that both missing and false-positive correspondences can be tolerated. The experiment results presented in this paper show that the proposed method is robust and performs well in very challenging situations.

In this paper, we present a Superconducting Nanowire Single Photon Detector (SnSPD) system and its application to ultra low voltage Time-Resolved Emission (TRE) measurements (also known as Picosecond Imaging Circuit Analysis, PICA) of scaled VLSI circuits. The 9 µm-diameter detector is housed in a closed loop cryostat and fiber coupled to an existing Emiscope III tool for collecting spontaneous emission light from the backside of integrated circuits (ICs) down to a world record 0.5 V supply voltage in a few minutes.

We describe a test chip designed and fabricated in 32nm CMOS SOI. The test chip was developed to assist in the characterization and testing of hot electron emission based test systems for both existing and forthcoming technology nodes, and contains circuit structures of increasing density and complexity. We also describe some unique circuit functions that may be of use in other applications

In this paper, we discuss the use of a tester-based methodology to enhance the spatial resolvability and interpretation of time-integrated and time-resolved emission measurements. This technique, first presented at [1] for chip diagnostics and failure localization, is very powerful for extending the capability of modern analytical tools beyond the limits of existing optics and detectors. In particular, we will discuss how the proposed method works and present several test cases for both static and dynamic emission measurements that allow signals from gates 150 nm apart to be resolved.

In this paper, near-infrared photon emission spectroscopy measurements from ring oscillators in 45 nm and 32 nm SOI process technology are compared. Employing a cryogenically cooled camera, the measurements cover a broad spectral range from 1200-2200 nm. Both leakage and switching emission, increase monotonically with the wavelength, suggesting measurements should be made at longer wavelengths than has historically been practiced. The paper discusses the optimum cut-off wavelength for maximum signal-to-noise ratio and the obvious importance of reduced ambient temperature for performing measurements.

In this paper, we present a comprehensive overview of several advanced methods and software solutions that we have developed during the years in support of chip diagnostic and characterization in our lab. These techniques apply to the analytical tools in our lab, such as time-integrated and time-resolved emission microscopes, and are devised to help improve productivity and ease of use. In particular we discuss emission-based auto-focusing for detecting very faint signals, processing large sets of images for high-resolution mapping of very large field of views, careful registration of emission images to circuit layout shapes, and advanced processing of the data for extracting the valuable underlying information. In the past, these techniques have been extensively used in support of many diagnostic and characterization applications, and they have been effective enablers for the development of innovative methods and tools.