Presentation slides for the ISTFA 2024 Tutorial session “Machine Learning in Semiconductor Failure Analysis: Techniques and Case Studies.”

This work demonstrates the capability of E-beam probing, combined with optical techniques, to effectively monitor the activity of the IC structures and extract the signals from a 16nm technology device through the silicon backside. We conducted optical probing to localize the area of interest on the device, where we aimed the E-beam probing to gather the signal. Once the target was located, a trench down to the STI level was opened on the device. This enables the use of E-beam probing, which has a much higher resolution than the optical methods.

In this paper, the failure analysis of InGaAs/GaAs-on-Si nanoridge laser diodes using the electron beam based nano-probing technique is presented. These III-V laser devices are fabricated using the nano-ridge engineering approach where the misfit dislocations generated during the growth of InGaAs/GaAs layers on silicon substrate are confined away from the active region. It is observed that the applied electrical stress causes degradation of electrical properties of the laser devices. We demonstrate the application of the electron beam induced current (EBIC) technique for failure analysis of nano-ridge lasers. This high-resolution technique helps to visualize the local distribution of the electric field in a nano-ridge p-i-n diode. The EBIC signal from the reference (electrically unstressed) device and the electrically stressed device is compared and hence can be used to identify the defective region. Furthermore, in-situ electrical stress experiments are performed for systematic analysis of the impact of electrical stress on the EBIC results.

Non-destructive inspection and analysis techniques are crucial for quality assessment and defect analysis in various industries. They enable for screening and monitoring of parts and products without alteration or impact, facilitating the exploration of material interactions and defect formation. With increasing complexity in microelectronic technologies, high reliability, robustness and thus, successful failure analysis is essential. Machine learning (ML) approaches have been developed and evaluated for the analysis of acoustic echo signals and time-resolved thermal responses for assessing their ability for defect detection. In the present paper different ML architectures were evaluated, including 1D and 2D convolutional neural networks (CNNs) after transforming time-domain data into the spectra-land wavelet domains. Results showed that 2D CNN with wavelet domain representation performed best, however at the expense of additional computational effort. Furthermore, ML-based analysis was explored for lock-in thermography to detect and locate defects in the axial dimension based on thermal emissions. While promising, further research is needed to fully realize its potential.

Contactless probing methods through the chip backside have been demonstrated to be powerful attack techniques in the field of electronic security. However, these attacks typically require the adversary to run the circuit under specific conditions, such as enforcing the switching of gates or registers with certain frequencies or repeating measurements over multiple executions to achieve an acceptable signal-to-noise ratio (SNR). Fulfilling such requirements may not always be feasible due to challenges such as low-frequency switching or inaccessibility of the control signals. In this work, we assess these requirements for contactless electron- and photon-based probing attacks by performing extensive experiments. Our findings demonstrate that E-beam probing, in particular, has the potential to outperform optical methods in scenarios involving static or low-frequency circuit activities.

Localizing security-relevant hard blocks on modern System-on-Chips (SoCs) for physical attacks, such as sidechannel analysis and fault attacks, has become increasingly time-consuming due to ever-increasing chip-area and - complexity. While this development increases the effort and reverse engineering cost, it is not sufficient to withstand resolute attackers. This paper explores the application of camera-based lock-in thermography (LIT), a nondestructive testing method, for identifying and localizing security hard blocks on integrated circuits. We use a synchronous signal to periodically activate security-related functions in the firmware, which causes periodic temperature changes in the activated die areas that we detect and localize via an infra-red camera. Using this method, we demonstrate the precise detection and localization of security-related hard blocks at the die level on a modern SoC.

Presentation slides for the ISTFA 2023 Tutorial session “SEM Based EBIC, EBAC, and E-Beam Probing Techniques.”

Presentation slides for the ISTFA 2023 Tutorial session “Machine Learning Based Data and Signal Analysis Methods for the Application in Failure Analysis (2023 Update).”

The paper presents the approach of enhancing time-domain signal analysis using machine learning techniques for analyzing acoustic echo signals and the subsequent derivation of condition-related class assignments for failure analysis. The examples provided here include two types of flip-chips with defects intentionally induced by thermal stressing. Besides investigating the general applicability and the benefit of the approach the current study also investigated the applicability of different deep learning model-architectures and compared their performances, accuracies, and robustness with respect to external impacts such as noise, jitter or physical defocusing. For independent verification selected defects which have either been identified by an experienced operator or the ML algorithm or both, have been further analyzed and validated by FIB/SEM cross sectional analysis.

This presentation is an introduction to machine learning techniques and their application in semiconductor failure analysis. The presentation compares and contrasts supervised, unsupervised, and reinforcement learning methods, particularly for neural networks, and lays out the steps of a typical machine learning workflow, including the assessment of data quality. It also presents case studies in which machine learning is used to detect and classify circuit board defects and analyze scanning acoustic microscopy (SAM) data for blind source separation.

Lock-In Thermography is an established nondestructive method for analyzing failures in microelectronic devices. In recent years, a major improvement made it possible to acquire time-resolved temperature responses of weak thermal spots, greatly enhancing defect localization in 3D stacked architectures. One limitation, however, is in the method used to determine defect depth, which is based on the numerical estimation of the delay between excitation and thermal response inferred from the value of the lock-in phase. In structures where the region between the origin of the defect and sample surface is partially or fully transparent to infrared signals, interference between radiated and conducted signal components largely falsifies the phase value on which the classical depth estimation relies. In the present study, blind source separation based on independent component analysis was successfully used to separate interfering signal components arising from direct thermal radiation and conduction, resulting in a precise estimation of the defect depth.

This presentation is an introduction to machine learning techniques and their application in semiconductor failure analysis. The presentation compares and contrasts supervised, unsupervised, and reinforcement learning methods, particularly for neural networks, and lays out the steps of a typical machine learning workflow, including the assessment of data quality. It also presents case studies in which machine learning is used to detect and classify circuit board defects and analyze scanning acoustic microscopy (SAM) data for blind source separation.

Lock-in thermography (LIT) has been successfully applied in different excitation and analysis modes including classical LIT, analysis of the time-resolved temperature response (TRTR) upon square wave excitation and TRTR analysis in combination with arbitrary waveform stimulation. The results obtained by both classical square wave- and arbitrary waveform stimulation showed excellent agreement. Phase and amplitudes values extracted by classical LIT analysis and by Fourier analysis of the time resolved temperature response also coincided, as expected from the underlying system theory. In addition to a conceptual test vehicle represented by a point-shaped thermal source, two semiconductor packages with actual defects were studied and the obtained results are presented herein. The benefit of multi-parametric imaging for identification of a defect’s lateral position in the presence of multiple hot spots was also demonstrated. For axial localization, the phase shift values have been extracted as a function of frequency [4]. For comparative validation, LIT analyses were conducted in both square wave and arbitrary waveform excitation using custom designed and sample-specific stimulation signals. In both cases result verification was performed employing X-ray, scanning electron microscopy (SEM) and energy dispersive x-ray (EDX) as complementary techniques.

Signal processing and data interpretation in scanning acoustic microscopy is often challenging and based on the subjective decisions of the operator, making the defect classification results prone to human error. The aim of this work was to combine unsupervised and supervised machine learning techniques for feature extraction and image segmentation that allows automated classification and predictive failure analysis on scanning acoustic microscopy (SAM) data. In the first part, conspicuous signal components of the time-domain echo signals and their weighting matrices are extracted using independent component analysis. The applicability was shown by the assisted separation of signal patterns to intact and defective bumps from a dataset of a CPU-device manufactured in flip-chip technology. The high success-rate was verified by physical cross-sectioning and high-resolution imaging. In the second part, the before mentioned signal separation was employed to generate a labeled dataset for training and finetuning of a classification model based on a one-dimensional convolutional neural network. The learning model was sensitive to critical features of the given task without human intervention for classification between intact bumps, defective bumps and background. This approach was evaluated on two individual test samples that contained multiple defects in the solder bumps and has been verified by physical inspection. The verification of the classification model reached an accuracy of more than 97% and was successfully applied to an unknown sample which demonstrates the high potential of machine learning concepts for further developments in assisted failure analysis.

GHz scanning acoustic microscopy (GHz-SAM) was successfully applied for non-destructive evaluation of the integrity of back end of line (BEOL) stacks located underneath wire-bond pads. The current study investigated two sample types of different IC processes. Realistic bonding defects were artificially induced into samples and the sensitivity of the acoustic GHz-microscope towards defects in BEOL systems was studied. Due to the low penetration depth in the acoustic GHz regime, a specific sample preparation was conducted in order to provide access to the region of interest. However, the preparation stopped several microns above the interfaces of interest, thus avoiding preparation artifacts in the critical region. Cratering related cracks in the bond pads have been imaged clearly by GHz-SAM. The morphology of the visualized defects corresponded well with the results obtained by a chemical cratering test. Moreover, delamination defects at the interface between ball and pad metallization were detected and successfully identified. The current paper demonstrates non-destructive inspection for bond-pad cratering and ball-bond delamination using highly focused acoustic waves in the GHz-band and thus illustrates the analysis of micron-sized defects in BEOL layer structures that are related to wire bonding or test needle imprints.

Tremendous research efforts have been devoted particularly to the development and improvement of through silicon vias (TSV) in order to provide a key enabling technology for vertical system integration. To achieve high processing yield and reliability efficient failure analysis techniques for process control and root cause analysis are required. The current paper presents an advanced approach for non-destructive localization of TSV sidewall defects applying high resolution Lock-in Thermography and Photoemission Microscopy imaging and defocusing series.