No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Package Innovation Roadmap
Abstract
Electronic Device Failure Analysis Technology Roadmap examines the trends shaping the future of semiconductor technology as well as the tools and techniques that will be required to detect, analyze, and remediate failures. The information is organized into ten chapters, seven of which assess the capabilities, limitations, and future needs of fault isolation technology in light of continued transistor scaling, increasing package complexity, and emerging IC architectures. Other chapters assess the tools and techniques used in die-level post-isolation failure analysis and the opportunities and challenges related to 3D packaging technology. The book also includes a chapter summarizing the technology gaps identified in previous chapters.
ResearchGate has not been able to resolve any citations for this publication.
In their daily work, engineers in the Failure Analysis (FA) laboratory generate numerous documents reporting all their tasks, findings, and conclusions regarding every device they are handled. This data stores valuable knowledge for the laboratory that other experts can consult, however, the nature of it, as individual reports reporting concrete devices and their corresponding processes, makes it inefficient to consult for the human experts. In this context, the following paper proposes a Artificial Intelligence solution for the gathering of this FA knowledge stored in the numerous documents generated in the laboratory. Therefore, we have generated a dataset of FA reports along with their corresponding electrical signatures and physical failures in order to train different supervised classifiers. The results show that the models are able of capturing the patterns underlying the different jobs and predict the causes, showing slightly better results for the physical hypotheses.
The development of the femtosecond laser (fs laser) with its ability to provide extremely rapid athermal ablation of materials has initiated a renaissance in materials science. Sample milling rates for the fs laser are orders of magnitude greater than that of traditional focused ion beam (FIB) sources currently used. In combination with minimal surface post-processing requirements, this technology is proving to be a game changer for materials research. The development of a femtosecond laser attached to a focused ion beam scanning electron microscope (LaserFIB) enables numerous new capabilities, including access to deeply buried structures as well as the production of extremely large trenches, cross sections, pillars and TEM H-bars, all while preserving microstructure and avoiding or reducing FIB polishing. Several high impact applications are now possible due to this technology in the fields of crystallography, electronics, mechanical engineering, battery research and materials sample preparation. This review article summarizes the current opportunities for this new technology focusing on the materials science megatrends of engineering materials, energy materials and electronics.
As a heterogeneous integration technology, the chiplet-based design technology integrates multiple heterogeneous dies of diverse functional circuit blocks into a single chip by using advanced packaging technology, which is a promising way to tackle the failure of Moore’s law and Dennard scaling. Currently, as process nodes move forward, dramatically rising cost, design cycle, and complexity are driving industry to focus on the chiplets. Chiplets allows IC designers to merge dies fabricated at different process nodes and reuse them in different projects, which helps to reduce the cost during design and improve yield. In this review, we look back at the industry’s efforts over the past decade and summary the concepts and techniques associated with chiplets. In the end, a discussion and conclusion will be given to forecast the future of chiplets.
In recent years, X-rays with different radiological devices have been frequently used. X-rays are produced in vacuum tubes by bombarding a metal target with high-speed electrons whereas radiological images occur after the resulting radiation passes through the patient’s body onto a photographic plate or a digital recorder to produce a radiological image. In radiology, devices that use X-rays can be classified as radiography, fluoroscopy, mammography and CT. In this study, we present a general approach for the simulation of X-ray spectra emitted from targets bombarded with electron beams for different energy ranges which are widely used in different radiological devices. The electron and photon transport is simulated by using the SpekCalc GUI code (version 1.1). In this study, we calculated the total photon spectra of the Tungsten (W) target for 150 keV and 12 c anode angle and certain energies for required energy values which are necessary for a successful radiodiagnostic process. We also performed simulations with the same parameters by adding the filter material into the X-ray tube. We achieved a good agreement between the clinical data and the computational method. The results showed that the computational method is in accordance with the manufacturer parameters.
Metal-based thermocompression bonding enables the creation of hermetic seals formed at relatively low processing temperatures and occupying a small portion of the device area. In the current study we have investigated the application of scanning acoustic microscopy (SAM) for assessing the quality of metal thermocompression bonds, both by evaluating its capabilities of localizing areas of poor bonding, and by finding defects in the integrity of the bond seal. Wafer laminates containing a test vehicle of sealing frames with pre-defined defects in the bond metal layer were sealed by Au–Au and Al–Al thermocompression bonding. Employing SAM, an area of five chips of poor bonding was identified non-destructively on the Al–Al laminate. Line defects of width 3.6 µm and point defects of diameter 22.4 µm have also been identified by SAM. The dicing yield for sealing frames was above 96 % for all frames of widths 100–400 µm and for both bond metal systems. The average bond strength was
31.5 ± 11.9 MPa for Al–Al thermocompression bonds and 37.3 ± 9.7 MPa for Au–Au thermocompression bonds. Scanning acoustic microscopy operates non-destructively and proved to be an extremely useful tool complementing current state-of-the-art methods for bond quality assessment.
In high-resolution X-ray computed tomography (CT), also known as 3D X-ray microscopy (XRM), low photon counts can lead to extremely long data acquisition times (in the order of hours). Reducing the number of radiographic projections (Np) acquired for CT reconstruction can be a cost-efficient solution in some cases. But the risk associated with reducing Np, if analytical filteredbackprojection algorithms are used for CT reconstruction, e.g., Feldkamp-Davis-Kress (FDK), is that it may produce a significant loss of image quality. Typical Np thresholds for a faithful 3D image reconstruction, required by the Nyquist-Shannon sampling theorem, are in the order of thousand projection views with modern XRM instruments. It is now well known, however, that deep learning (DL) based algorithms for CT reconstruction can improve the scan time (throughput) and image quality capabilities of XRM. This paper proposes the use of DL-based algorithms as an option for reducing Np, even down to a few hundred projections, without a significant loss of image quality. The integration of DL-based reconstruction techniques into 3D XRM workflows is presented throughout this article. It is shown that 3D XRM data reconstructions produced by DL-based workflows can provide up to 8X and 10X throughput improvement at similar or better image quality compared to standard FDK reconstruction.
Over the past decade, 3D X-ray technique has played a critical role in semiconductor package failure analysis (FA), primarily owing to its non-destructive nature and high resolution capability [1,2]. As novel complex IC packages soar in recent years [3,4], X-ray failure analysis faces increasing challenges in imaging new advanced packages because IC interconnects are more densely packed in larger platforms. It takes several hours to overnight to image fault regions at high resolution or the crucial details of a defect remain undetected. A high-productivity X-ray solution is required to substantially speed up data acquisition while maintaining image quality. In this paper, we propose a new deep learning high-resolution reconstruction (DLHRR) method, capable of speeding up data acquisition by at least a factor of four through the implementation of pretrained neural networks. We will demonstrate that DLHRR extracts signals from low-dose data more efficiently than the conventional Feldkamp-Davis-Kress (FDK) method, which is sensitive to noise and prone to the aliasing image artifacts. Several semiconductor packages and a commercial smartwatch battery module will be analyzed using the proposed technique. Up to 10x scan throughput improvement was demonstrated on a commercial IC package. Without the need of any additional X-ray beam-line hardware, the proposed method can provide a viable and affordable solution to turbocharge X-ray failure 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.
In this work we present a large-volume workflow for fast failure analysis of microelectronic devices that combines a stand-alone ps-laser ablation tool with a SEM/Xe Plasma FIB system. In this synergy, the ps-laser is used to quickly remove large volumes of bulk material while the SEM/Xe Plasma FIB is used for precise end-pointing to the feature of interest and fine surface polishing after laser. The concept of having a stand-alone laser tool obeys the logic of maximizing productivity as both systems can work simultaneously and continuously. As application examples we first present a full workflow to prepare an artefact-free, delamination-free cross-section in an AMOLED mobile display. We also present applications examples that require cm-sized long cuts to cut through whole microelectronic devices, or removal of cubic-mm of material to prepare mm-sized cross-sections in packages. We discuss a way how to implement correlation data across the laser and FIBSEM platforms through SYNOPSYS Avalon SW allowing precise navigation to the area of interest using layout circuit overlays. We also show an example of image bitmap overlay to navigate across platforms and end-pointing.
In this paper we will demonstrate new approaches for failure analysis of memory devices with multiple stacked dies and TSV interconnects. Therefore, TSV specific failure modes are studied on daisy chain test samples. Two analysis flows for defect localization implementing Electron Beam Induced Current (EBAC) imaging and Lock-in-Thermography (LIT) as well as adapted Focused Ion Beam (FIB) preparation and defect characterization by electron microscopy will be discussed. The most challenging failure mode is an electrical short at the TSV sidewall isolation with sub-micrometer dimensions. It is shown that the leakage path to a certain TSV within the stack can firstly be located by applying LIT to a metallographic cross section and secondly pinpointing by FIB/SEM cross-sectioning. In order to evaluate the potential of non-destructive determination of the lateral defect position, as well as the defect depth from only one LIT measurement, 2D thermal simulations of TSV stacks with artificial leakages are performed calculating the phase shift values per die level.
This paper describes an alternative cross-section sample preparation approach using triple ion beam (TIB) milling to prepare flip chip samples. The 3 cases study results clearly demonstrate the capability and advantages of TIB milling for preparing Lead-free and eutectic C4 solder bumps and back-end-of-line (BEOL) defects as compared to conventional mechanical and Focus Ion Beam (FIB) cross-section.
Lock-in thermography and magnetic current imaging are emerging as the two image-based fault isolation methods most capable of meeting the challenges of short and open defect localization in thick, opaque assemblies. Such devices are rapidly becoming prevalent as 3D integration begins to ramp up production. This paper expands on previously published work with a qualitative comparison of the techniques on single chip and stacked die packages with known designed-in or FIB-created defects.
The standard Ga focused ion beam (FIB) technology is facing challenges because of a request for large volume removal. This is true in the field of failure analysis. This article presents the first combined tool which can fulfill this requirement. This tool offers the combination of a high resolution scanning electron microscope (SEM) and a high current FIB with Xe plasma ion source. The article focuses on failure analysis examples and discusses the different steps of extra large cross sections (deposition of protective layer, rough milling, and polishing). Several applications of the novel Xe plasma FIB/SEM instrument are shown with respect to the failure analysis. The performance of the instrument is tested and discussed in comparison to gallium liquid metal ion source FIB systems. Results show that the Xe plasma FIB offers much higher milling rate, greatly reducing the time necessary for many failure analysis tasks.
This article describes how Focused Ion Beam (FIB) milling methodology enhances the capability of package-level failure analysis on flip-chip packages by eliminating the artifacts induced by using conventional mechanical techniques. Dual-Beam Focused Ion Beam (DB FIB) cross sections were successful in detecting failure mechanisms related either to the die/C4 bump or package defect inside the organic substrate. This paper outlines detailed sample preparation techniques prior to performing the DB FIB cross-sections, along with case studies of DB FIB cross-sections.
It was already demonstrated, that the method of Lock-in Thermography (LIT) enables 3D localization of thermal active defects, e.g. electrical shorts and resistive opens, on die level and within fully packaged single and multichip devices [1,2]. The depth of a defect can be derived from phase shift measurements of the defective compared to a reference device For a general approach of this method, thermal modeling is used and verified by experimental data to investigate the internal heat propagation under periodic stimulation in correlation to the LIT measuring process. [3]. A basic requirement for the successful application of the method is a precise and reproducible measurement of both the thermal material properties of each material layer and the phase shift between the internal heat excitation and thermal response measured by LIT. Significant influences from the material and measurement setup to the detected phase shift have to be identified and taken into account. However, to identify and distinguish the relevant influences measurements with defined internal heat sources are necessary which are presented in this paper. First, the relationship between geometrical thickness of a material layer and the resulting thermal parameters for both homogeneous and heterogeneous materials are measured and discussed. A new measurement setup generating a defined point heat source will be presented to calibrate the LIT system for quantitative phase shift measurements and to determine the phase shift to thickness parameters of single material layers. In addition the variation of the phase shift caused by the defect geometry and the defect environment will be investigated. Finally, a case study is presented comparing the experimental results to the obtained results from a real stacked die device.
With the growing variety, complexity and market share of 3D packaged devices, package level FA is also facing new challenges and higher demand. This paper presents Lock-In Thermography (LIT) for fully non-destructive 3D defect localization of electrical active defects. After a short introduction of the basic LIT theory, two slightly different approaches of LIT based 3D localization will be discussed based on two case studies. The first approach relies on package internal reference heat sources (e.g. I/O-diodes) on different die levels. The second approach makes use of calibrated 3D simulation software to yield the differentiation between die levels in 8 die μSD technology.
In this paper, we describe the novel technique of using leading edge X-Ray Microscopy (XRM) technology to replace physical cross-sectioning in Failure Analysis (FA) and 3-Dimensional Integrated Circuit (3DIC) process development. Contrary to general consensus that 3D X-Ray is too slow, we explain how XRM can be used to obtain high quality cross-section images within 5-300 min per measurement depending on the physical properties (materials, feature sizes, and outer dimensions) of the sample and the minimum tolerable image quality needed to visualize a defect. The specifics of the inspection technique itself and how X-rays interact with the sample to achieve high-quality images will be discussed and contrasted with conventional 3D MicroCT technology. Furthermore, understanding the effects that imaging parameters, such as voltage, power, exposure time, resolution, number of projections, etc, have in the quality of an image, can help the user reduce the 3D X-Ray inspection time considerably. A TSMC test vehicle package is used to illustrate the effects of inspection time in image quality, and to compare and contrast the quality of an optical image taken from a physical cross-section and a virtual cross-section image taken from an XRM tomography.
Artificial Intelligence vs. Machine Learning vs. Deep Learning
- Oppermann
Progress in Failure Analysis – a Key for Reliable High-quality European Electronics
- Klaus