Acoustic imaging of bump defects in flip-chip devices using split spectrum analysis

Scanning acoustic microcopy (SAM) method with inverted inspection direction has been successfully develop and evaluate the quality of flip chip underfill and interconnect bonds in manufacturing of microelectronic components. Acoustic microscopes utilize high frequency ultrasound transmitting through the silicon chip backside in one scan to access and examine the internal structures in optically opaque materials. These non-destructive methods relatively enable the defect localization which leads microstructural examinations involving destructive analysis sample preparation. For the justification of accuracy of SAM method analysis, a destructive cross-sectioned and mechanical lapping physical analysis were performed for preparing the samples and examine by optical microscope and Scanning Electron Microscope (SEM) for defect verification. Cases studies have been demonstrated that the capabilities of conventional SAM inspection and advantages over other analysis method.

Automated inspection and classification of flip-chip-contacts using scanning acoustic microscopy

Defect inspection of solder bumps using the scanning acoustic microscopy and fuzzy SVM algorithm

Ultrasonic tissue characterization of collagen in lipid-rich plaques in apoE-deficient mice

Scanning acoustic microscopy (SAM) of Haversian bone at high frequency clearly shows differences in the brightness (gray) levels of individual osteons, which were observed to correspond to the differences in reflection coefficients measured with the SAM technique: the darker the bony regions on the SAM micrographs, the lower the SAM reflection coefficient values, and vice versa. We studied the correlation between differences in brightness and the related reflection coefficients of osteons on undecalcified transverse sections of goat tibial diaphysis using a 400-MHz burst mode lens on the Olympus UH3 SAM. Results showed that there was a strong linear correlation between brightness intensities and the corresponding reflection coefficients, with a correlation coefficient r = 0.99 ( P << 0.001) using gray level 210 as cutoff. This result suggested that the image brightness intensity of osteonic bone could be measured retrospectively to provide information on the stiffness of the corresponding bone matrix.

Variations of microstructure, mineral density and tissue elasticity in B6/C3H mice

Renal fibrosis is strongly correlated with renal functional outcomes. Therefore, this is a significant finding in determining renal prognosis. There are various reports on the imaging evaluation of renal fibrosis, but these are not well established. Scanning acoustic microscopy (SAM) uses ultra-high-frequency ultrasound to visualize tissues in just over a minute. SAM can simultaneously measure acoustic data such as speed of sound (SOS). SOS indicates the elasticity (stiffness) of a material. In this study, we aimed to compare and evaluate SAM acoustic intensity images and SOS data with light microscopy images of renal lesions, especially renal fibrosis. Renal specimens containing fibrosis were selected. The acoustic intensity images were compared to PAS-stained images. SOS data of the tubulointerstitium were compared with Masson's trichrome (MT)-stained images. The blue intensity of MT staining, which indicates fibrosis, was numerically valued using image-processing software. Furthermore, the correlations between it and the SOS values were evaluated. The acoustic intensity images suggested tubular atrophy and interstitial expansion in the same areas as in the PAS staining. SOS values of interstitial expansion with fibrosis were higher than normal area, interstitial expansion without fibrosis. A weak positive correlation was observed between the SOS values and the blue intensity of MT staining. SOS data can be used to evaluate renal fibrosis. The combination of SOS data and MT-stained images enables a more detailed evaluation of renal fibrosis. This study can contribute to the evaluation of renal fibrosis and has potential clinical applications in the future.

Scanning Acoustic Microscopy (SAM) is a critical nondestructive technique for detecting buried defects in semiconductor devices [1]. However, interpreting the complex acoustic data requires expert knowledge, especially as microelectronic structures grow increasingly intricate. This study presents a novel, unsupervised approach for defect detection in SAM data using an autoencoder neural network trained exclusively on intact reference samples. The method exploits reconstruction error as an indicator of signal anomalies, identifying potential defects without prior knowledge of their characteristics or the need for defective training samples. The resulting similarity parameter facilitates intuitive visualization of outlier regions through overlay on high-resolution SAM images. This enhances defect detection sensitivity and operator support, making SAM more accessible for less experienced users and highly promising for automation in production environments. While the method effectively highlights signal deviations, it may also flag non-defect-related anomalies such as sample tilting. Nonetheless, it significantly advances automated and operator-independent semiconductor failure analysis by enabling reliable detection of structural anomalies through learned representations of intact signal patterns.

The Discrete Fourier Transform (DFT), which can be calculated efficiently by the Fast Fourier Transform (FFT), is one of the most commonly used tools for frequency estimation of a multi-frequency exponential signal. However, the performance of the FFT is inherently affected by spectral leakage when non-coherent sampling occurs. The Chirp-Z Transform (CZT), which can refine the spectrum to the narrowband range of the source signal accurately, can be generalized to a special form of DFT. This paper proposes a CZT based algorithm which can estimate frequencies of the multi-frequency exponential signal accurately. In our method, the CZT bin, as a nonlinear equation, is firstly used to describe the correspondence among the frequencies, amplitudes, phases and the observed spectrum. Then, the above nonlinear equation is rewritten as a linear equation with unknown frequencies. We further extend the proposed method to the Discrete-time Fourier Transform (DTFT), zero-padding FFT and DFT, which are the special cases of the CZT. We also give the derivation procedure of theoretical variance of our method. In our method, the spectrum leakage is fully considered as useful information to establish the linear equation set. Therefore, it can estimate the frequencies of multi-frequency exponential signal with high accuracy. Simulation results verify the performance of the proposed method.

Caused by its superior performance regarding bandwidth, power loss and form factors compound semiconductors like GaN or SiC recently moved into the interest of the semiconductor industry, particularly for applications in the high power segment. These improved performance parameters go along with increased power densities and thus, require a specifically optimized thermal concept of the packaging. Ag-sintering as die-attach process has been widely accepted as a promising alternative to soldering techniques with respect to thermal conductance. However, the thermal and the mechanical behavior of the sinter layer is influenced by its composition on the micro -and the macroscopic scale with respect to the presence of voids, inclusions and delamination but also the fractional content of porosity. Scanning acoustic microscopy (SAM) uses acoustic waves that interact with the mechanical properties of the sample materials and is thus, widely established for the non-destructive detection of voids, inclusions and delamination in microelectronic components. The current paper investigates the potential of SAM as a method for the assessment and characterization of the porosity in sintered Ag layers targeting the die-attach technology for novel power semiconductor devices.

Scanning acoustic microscopy (SAM) was used to evaluate the degradation of adhesion on a blown up sample, i.e. a Si-beam coated only on the passivated side with the molding compound under consideration. Such a bending beam sample offers two advantages for evaluation by SAM over conventionally plastic mounted chips: (1) for the acoustic waves the sample is flat, which helps to eliminate or reduce the influence of chip topography, and (2) the bending stress tends to open any partial delamination, giving better defined conditions for SAM examination. From the SAM analysis of the bending beam sample a very low stress molding compound, which had caused an unexpectedly high failure rate, was found to show poor adhesion after thermal cycling in delamination near the corners of the sample. No delaminations could be observed by means of SAM at molded product test chips, however. Thus, it is concluded that poor adhesion of the molding compound which caused delaminations in the bending beam's corner regions resulted in an increased chip failure rate. >

Delaminations of the embedded passive components are often associated with reliability problems. Understanding the delamination can help us quickly identify the failure cause at an early stage of the failure analysis. In this work, defects of buried capacitors have been detected by use of scanning acoustic microscopy (SAM). The results show that SAM can be a powerful analytical tool for the nondestructive evaluation of delaminations in buried passive components. T-scan is an effective method for determining the existence of the internal defects, and TAMI-scan is shown to be capable of detecting the location and size of the delamination and void.

Traditional failure analysis (FA) methods face growing challenges due to the increasing miniaturization and complexity of modern semiconductor devices. In particular, the detection of cracks in the extreme low K layers of flip-chips, utilized in advanced packaging architectures at the back end of line (BEOL), is challenging. To improve detectability of these defects, we propose an AI-based image enhancement workflow, specifically tailored to improve scanning acoustic microscopy (SAM) images. By applying a Deep CNN with Skip Connections and Network in Network (DCSCN) model to lowresolution C-scan SAM images, we are able to improve perceptual quality of images. The approach boosts resolution dependent analysis like AI-based object detection and classification as well as decreases significantly SAM scanning time. Our work demonstrates that AI-based image enhancement can significantly improve FA workflows, offering a robust and efficient solution for modern semiconductor analysis.

In this work, C-Scan Acoustic Scanning Microscopy (ASM) is used to map the defects of three SiC samples. The acoustic images indicate that numerous defects with different shapes and area sexist in the wafers. Some of the defects have areas of more than 100,000 μm2. The number of defects ranges from 1 to 50 defects/wafer. Defect mapping is essential for defect repairing or avoidance. This work shows that ASM can locate the precise positions of the crystallographic defects, which enables defects repair and yield enhancement.

Scanning acoustic microscopy (SAM) is a label-free imaging technique used in biomedical imaging, non-destructive testing, and material research to visualize surface and sub-surface structures. In ultrasonic imaging, noises in images can reduce contrast, edge and texture details, and resolution, negatively impacting post-processing algorithms. To reduce the noises in the scanned image, we have employed a 4D block-matching (BM4D) filter that can be used to denoise acoustic volumetric signals. BM4D filter utilizes the transform domain filtering technique with hard thresholding and Wiener filtering stages. The proposed algorithm produces the most suitable denoised output compared to other conventional filtering methods (Gaussian filter, median filter, and Wiener filter) when applied to noisy images. The output from the BM4D-filtered images was compared to the noise level with different conventional filters. Filtered images were qualitatively analyzed using metrics such as structural similarity index matrix (SSIM) and peak signal-to-noise ratio (PSNR). The combined qualitative and quantitative analysis demonstrates that the BM4D technique is the most suitable method for denoising acoustic imaging from the SAM. The proposed block matching filter opens a new avenue in the field of acoustic or photoacoustic image denoising, particularly in scenarios with poor signal-to-noise ratios.