Continued advancements in the architecture of 3D packaging have increased the challenges in fault isolation and failure analysis (FA), often requiring complex correlative workflows and multiple inference-based methods before targeted root cause analysis (RCA) can be performed. Furthermore, 3D package components such as through-silicon-vias (TSVs) and micro-bumps require sub-surface structural characterization and metrology to aid in process monitoring and development throughout fabrication and integration. Package road-mapping has also called for increased die stacking with decreased pitch, TSV size, and die thickness, and thus requires increased accuracy and precision of various stateof- the-art analytical techniques in the near future. Physical failure analysis (PFA), process monitoring, and process development will therefore depend on reliable, high-resolution data directly measured at the region of interest (ROI) to meet the complexity and scaling challenges. This paper explores the successful application of plasma-FIB (PFIB)/SEM techniques in 2D and 3D regimes and introduces diagonal serial sectioning at package scales as a novel approach for PFA and metrology. Both 2D and 3D analysis will be demonstrated in a high bandwidth memory (HBM) package case-study which can be applied more broadly in 3D packaging.

TEM automation is dedicated to providing high-volume, fast and precise TEM data to enable semiconductor manufacturers to develop and control fabrication processes. It automates TEM operation and measurement procedures, and minimizes the requirements of user training. Traditionally, recipes are required for each specific structure, and Automatic TEM Imaging and Metrology had individual recipes for each structure, respectively and separately. TEM professionals review structures and TEM images and manually assign imaging and metrology recipes one by one. After metrology, it is tedious to manually check mis-measurement of hundreds of TEM images and Critical Dimensions (CDs). In this study, An Intelligent TEM automation process was developed by machine learning process that could “fully” automatically conduct TEM Analysis from Imaging to Metrology starting right after TEM holder insertion without human intervention. It is not limited to automatically prepared TEM samples and also works for manually prepared samples. TEM Auto Metrology is carried out automatically right after Auto Imaging in the background. After that, Auto capturing of mis-measurements can be carried out automatically to catch problematic metrology images and CDs as an invaluable complement. As a result, the whole workflow was streamlined, with better efficiency and accuracy achieved.

The operation of modern semiconductor components often relies on nanoscale electronic features emerging from complicated device architectures with finely tuned composition. While the physical structure of these devices may be straightforward to image, the resulting electronic characteristics are invisible to most high-resolution imaging techniques. Here we present electron beam-induced (EBIC) imaging in the scanning transmission electron microscope (STEM) as a high-resolution imaging technique with electronic-based contrast for characterizing complex semiconductor devices. Here, as an example case, we discuss the preparation and imaging of a STEM EBIC-compatible cross section extracted from a commercial AlGaAs high electron-mobility transistor (HEMT). The device exhibits low surface leakage, as measured via electrical testing and STEM EBIC conductivity contrast. The EBIC signal in the active layer of the device is mostly confined to the InGaAs channel, indicating that the electronic structure is largely preserved following sample preparation.

Focused-Ion Beam Scanning Electron Microscopy (FIB-SEM) tomography is a high resolution three-dimensional (3D) imaging method with applications in failure analysis and metrology of semiconductor devices. For the smallest logic and memory structures currently in use, it requires single-digit nanometer 3D resolution. In this resolution range, avoiding distortion artifacts in the data becomes crucial. We present examples and discuss ways to reduce the likelihood of such artifacts during the data acquisition, as well as how to mitigate them in post-processing, and therefore increase the data quality.

Failure analysis of small contamination at the surface and sub-surface interface represents a major set of common microelectronics and semiconductor issues. The application of O-PTIR spectroscopy analyses provides flexibility to sample preparation and improves sensitivity to very small levels of contamination even below <1 micron in layers or particles on or just below the surface. The detection of this contamination can be limited if only bright field imaging is used to contrast the region of interest (ROI) and the surrounding structure. Adding fluorescence microscopy is an additional imaging technique that adds another layer of chemical specificity and provides locations of unseen ROI’s for additional IR and Raman spectral analysis.

An approach to overcome barriers to practical Compressed Sensing (CS) implementation in serial scanning electron microscopes (SEM) or scanning transmission electron microscopes (STEM) is presented which integrates scan generator hardware specifically developed for CS, a novel and generalized CS sparse sampling strategy, and an ultra-fast reconstruction method, to form a complete CS system for 2D or 3D scanning probe microscopy. The system is capable of producing a wide variety of highly random sparse sampling scan patterns with any fractional degree of sparsity from 0- 99.9% while not requiring fast beam blanking. Reconstructing a 2kx2k or 4kx4k image requires ~150-300ms. The ultra-fast reconstruction means it is possible to view a dynamic reduced raster reconstructed image based upon a fractional real-time dose. This CS platform provides a framework to explore a rich environment of use cases in CS electron microscopy that benefit from the combination of faster acquisition and reduced probe interaction.