Advances in digital transformation have enabled the integration of three-dimensional (3D) sensing technologies and numerical analyses in non-destructive testing (NDT). This study presents NDT simulators of ultrasonic and infrared inspections, achieved through the fusion. Consumer-grade devices, including mobile LiDAR sensors and digital cameras, are used to generate high-resolution surface models of inspection targets. We use two primary techniques to capture 3D shapes — light detection and ranging (LiDAR) and photogrammetric range imaging — and generate high-resolution surface models of inspection targets. Both methods first generate point cloud data, which we convert into a voxel. This 3D voxel data is then fed into numerical simulations using structured grids, such as the finite difference and finite element methods, to analyze ultrasonic wave propagation and heat conduction. This approach, known as image-based modeling, is expected to become a standard practice within NDE4.0. This simplified simulation contributes to the improvement of the reliability of NDT and facilitates engineering education. And our concept lays a foundation for cost-effective digital twin–based inspection systems.

As the line width of integrated circuits continues to decrease, the resulting increase in current density leads to a serious issue of electromigration—the diffusion of metal atoms driven by electron flow, which ultimately causes wire failure. Monitoring void formation and degradation in nanowires at the early stages of electromigration is essential for ensuring device reliability. Although early diagnosis using 1/f noise has been proposed as a simple method, it cannot detect local defects in long wires. In-situ and localized observation techniques using transmission electron microscopy or X-ray can visualize voids, but they require special sample preparation.

In this study, we developed an in-situ and localized monitoring technique for electromigration in a single nanowire using picosecond ultrasonics. A femtosecond laser was employed to generate and detect sub-THz acoustic waves within a focused laser spot. Single copper and aluminum nanowires, with widths ranging from 50 to 1000 nm, were fabricated and evaluated. We observed pulse-echo signals from 150 nm-thick aluminum wires and acoustic resonance signals from copper wires. When direct current was applied, the amplitude and frequency of the acoustic signals changed significantly, even at an early stage when the electrical resistance remained nearly constant. These results demonstrate that picosecond ultrasonics enables real-time, non-destructive assessment of local defects that electrical measurements cannot capture.

This approach provides attendees with a practical and adaptable framework for understanding and applying high-frequency ultrasonic sensing to nanoscale reliability evaluation in advanced electronic devices.

We developed a fast computational method to find the optimal non-destructive-testing (NDT) inspection conditions—while considering the uncertainty of detection—that maximize probability of detection (POD). An exhaustive search for maximum POD typically requires a large number of NDT-signal simulations. They thereby incur a large computational load that poses a challenge for practical implementation of POD-based fast computational methods, particularly in the case of complex inspection systems involving a lot of parameters. In this study, we propose a method using support vector machine (SVM) to reduce the number of required simulations. To validate the proposed method, we evaluated POD of a simple ultrasonic-testing model by a conventional exhaustive search and the proposed method. The results demonstrate that the proposed method reduces the number of required simulations by 98% compared to a conventional exhaustive search while maintaining accuracy.

Austenitic stainless steel is widely used in nuclear power plants (NPPs), and various cases of stress corrosion cracking (SCC) near welds have been reported. Ultrasonic testing is a critical technique for detecting and sizing SCCs. However, the anisotropic and heterogeneous properties of these welds make ultrasonic testing challenging because ultrasonic waves are significantly distorted when propagating through the weld metal. A deeper understanding of wave characteristics through wave propagation simulation is therefore essential for improving inspection accuracy. To this end, numerical models must account for both crystal orientation and grain boundaries. In this study, we developed a three-dimensional Cellular Automaton–Finite Difference (CAFD) model to predict the solidification structures of multi-layer and multi-pass austenitic stainless steel welds (ASSWs) based on welding conditions. The model was validated by comparing wave propagation simulated using the predicted solidification structure with experimental measurements for a seven-layer, seven-pass ASSW. We then predicted the solidification structure of an actual 15-layer, 37-pass ASSW and incorporated the forecasted structure into an FEM code to simulate ultrasonic wave propagation. Finally, we compared the numerical results with experimental findings obtained from angle-beam testing.

KEY WORDS: Austenitic stainless steel weld, Cellular automaton method, Wave propagation, Solidification structure, Finite element method