Ehsan Dehghan-Niri

Directed Energy Deposition (DED) is an additive manufacturing process that enables the fabrication of complex metallic components; however, process induced defects such as porosity remain a critical concern. This study investigates Acoustic Emission (AE) as an in-situ, non-destructive testing (NDT) method for correlating real-time process signals with internal porosity in DED-fabricated stainless-steel specimens. AE signals were continuously acquired during fabrication, while post-process X-ray Computed Tomography (CT) was used to quantify pore size, distribution, and volume fraction. The novelty of this work is the spatial correlation of in-situ AE measurements with CT-identified porosity to identify defect-prone regions during fabrication. The results show that stable AE amplitudes above approximately 55 dB, with fewer high-energy transient events, were associated with improved material consolidation and lower porosity, while fluctuations in AE activity corresponded to regions with increased pore formation. These findings demonstrate the potential of AE monitoring for assessing process stability and defect evolution in DED, providing a practical complement to conventional post-process inspection methods.

Microelectronic packaging is crucial for protecting, powering, and interconnecting semiconductor chips, playing a critical role in the functionality and reliability of electronic devices. With the growth in complexity and miniaturization of these products, the implementation of efficient inspection techniques becomes crucial in preventing failures that may result in device malfunctions. This review paper examines the progress made in utilizing Scanning Acoustic Microscopy (SAM) to assess the structural integrity of microelectronic systems within the broader field of Nondestructive Evaluation/Testing (NDE/T) methods. With an exclusive emphasis on SAM, we point out SAM technological advancements in multi-die stacking, Through Silicon Vias (TSV), and hybrid bonding inspection that improve inspection sensitivity and resolution required to be prepared for upcoming challenges accompanying 3D- and heterogeneous integration architectures. Some of these approaches compromise the depth of inspection for the benefit of lateral resolution, while others do not sacrifice the in-depth range of evaluation. These developments are of the utmost importance in addressing the substantial obstacles associated with examining microelectronic packages, facilitating the early detection of potential failures, and enhancing the reliability and robustness of semiconductor devices. Furthermore, our discussion consists of the fundamental principles and practical approaches of SAM. It also examines recent investigations that integrate SAM with machine learning concepts and the application of deep learning models in order to automate defect detection and characterization, thus substantially augmenting the efficiency of microelectronic package assessments.

Tap testing is an effective way of characterizing material conditions and flaws in various materials, including wood. Given its versatility and widespread usage, wood requires thorough inspection to assess its quality, identify potential defects, and ensure the safety and durability of wooden structures across diverse applications. This technique has the advantage of being simple, efficient and inexpensive. The tap testing method, when performed manually, requires an operator to tap each point of the structure using a hand-held object (e.g., a coin or tap hammer). Consequently, the precision of this test is highly reliant on the inspector's subjective interpretation of the vibrational acoustic response. In order to overcome this drawback, a bio-inspired tap testing approach with augmented objectivity of signal analysis has been proposed. The pioneer tap testing is inspired by an animal named aye-aye recognized for its unique acoustic-based foraging behavior called 'tap-scanning' or 'percussive foraging'. The aye-aye's near-field versatile acoustic sensing capabilities enable it to locate small cavities beneath a tree bark with complex materials. Current work describes a quantitative and instrumented robotic tap test system that creates repeatable mechanical impacts using a biomimetic approach. Two specimens were utilized to validate the effectiveness of this biomimetic approach. One of the specimens possessed identical diameter flat bottom holes but of varying depths, and the other had different diameters at positions of the same thickness from the test surface. Biomimetic tap scanning was applied over the defect-free and damaged areas of the specimens utilizing the 3D printed animal pinna and head in the experimental setup. The findings indicated that the biological characteristics of the animal's external auditory organs including the pinna and ear cannel substantially enhanced the system's sensitivity in detecting artificial defects within wooden blocks. This enhancement was primarily attributed to a notable improvement in the signal-to-noise ratio. Moreover, the outcomes demonstrated that the head and external ear structure exerted a superior discriminating factor for damage detection compared to both the pinna with ear canal configuration and the microphone-only setup within the experimental framework. The underlying cause behind this heightened discriminating factor remains undetermined and warrants further investigation by the research team.

This journal paper explores the application of Deep Learning (DL)-based Time-Series Classification (TSC) algorithms in ultrasonic testing for pipeline inspection. The utility of Electromagnetic Acoustic Transducers (EMAT) as a non-contact ultrasonic testing technique for compact robotic platforms is emphasized, prioritizing computational efficiency in defect detection over pinpoint accuracy. To address limited sample availability, the study conducts benchmarking of four methods to enable comparative evaluation of classification times. The core of the DL-based TSC approach involves training DL models using varied proportions (60%, 80%, and 100%) of the available training dataset. This investigation demonstrates the adaptability of DL-enabled anomaly detection with shifting data sizes, showcasing the AI-driven process's robustness in identifying pipeline irregularities. The outcomes underscore the pivotal role of artificial intelligence (AI) in facilitating semi-accurate but swift anomaly detection, thereby streamlining subsequent focused inspections on pipeline areas of concern. By synergistically integrating EMAT technology and DL-driven TSC, this research contributes to enhancing the precision and near real-time inspection capabilities of pipeline assessment. This investigation collectively highlights the potential of DL networks to revolutionize pipeline inspection by rapidly and accurately analyzing ultrasound waveform data.

This final report summarizes the research findings of the current project. This project is a collaboration between New Mexico State University (NMSU) as a lead (Recipient) and Arizona State University (ASU) as a Co-Recipient. Tubular structures are common components of boilers and heat exchangers in power plants. Over time, these components may suffer corrosion, cracks, and stress-corrosion cracks in either the body or the welded connections. A single tube leakage can cause an outage of several weeks. Regular inspection is a key safety factor when ensuring that power plants are maintained in reliable, operational condition. This inspection, however, is challenging, time-consuming, and in many cases impossible due to accessibility issues and safety concerns. Recent developments in robotic-based inspection can offer a great solution. Hard-to-reach places can be inspected without overhauling the unit, saving time and cost. Although several robots have been designed and implemented successfully for inspecting power plant components, in particular tubular structures, their mobility and flexibility are limited. Most of these robots use wheels for mobility which reduces their maneuverability of these robots. Moreover, they usually use magnets to attach to tubes which will not work on tubes with non-ferromagnetic materials. These robots usually carry measurement tools for a particular non-destructive testing (NDT) method such as ultrasound testing (UT) that requires a couplant to perform a point-by-point (scanning) inspection of the tubular structure. In this project, we developed a versatile lizard-inspired tube inspector (LTI) robot with embedded inspection sensing components for tube inspection which removes the need for point-by-point scanning of tube surface for crack and corrosion detection. Inspired by a “lizard”, the novelty of the current project is the integration of couplant-free ultrasound sensing and transmission, advanced data-driven defect detection and imaging, and friction-based mechanical mobility components in a single robot to eliminate a need for smooth surfaces and simple geometry for mobility and scanning. The LTI robot could replace the wheel-based approach with friction-based mobility to significantly increase the flexibility and maneuverability of the robot. The LTI robot can get into a power plant unit, such as a boiler, from a small area allowing it to access a component of interest for inspection (e.g., move on OD, curved and flat surfaces, non-ferromagnetic or ferromagnetic materials, and tubes with rough surfaces and complex geometries). Additionally, an advanced data-driven method using ultrasound data was pioneered to allow the robot to detect defects in the entire area between the robot’s multi-functional mobility system (grippers). Integrating a couplant-free ultrasound sensing in the robot’s grippers as well as using advanced data-driven methods allowed the LTI to detect defects in the entire cross-section of a tube using its grippers when stationary and when mobile.