This year the author's company celebrates its 30th birthday and with it a lengthy period in which countless process faults have been found. The ongoing development in laser sources, processes and applications continuously offers new challenges for the photodiode-based process monitoring to find even the smallest deviations. Especially in applications with many weld seams, such as batteries or bipolar plates, not only detection but also classification is required to quickly identify the root causes of faults.

In this study, a conventional broadband approach is compared to a multi-spectral monitoring. Welding experiments are carried out with highly electrically conductive materials using two sensor systems. The experiments are then evaluated for both systems and the results are compared to identify the individual benefits of the sensor approach. As a result, the multi-spectral system is particularly beneficial when it comes to classifying and differentiating between faults which enables individual rework strategies for production.

We present a novel Optical Coherence Tomography (OCT) system for industrial metrology with higher imaging power (>50 mW) and greater imaging depth field of view (>30 mm) than any existing industrial OCT system. This system includes a low noise fiber amplifier to generate an imaging light source with very high power (>60 mW) suitable for OCT imaging with sensitivity > 100 dB. Using this system and a 50 kW Yb:fiber laser, we have produced in situ direct measurements of laser weld keyhole geometry exceeding 40 mm in depth. This represents the deepest in situ direct keyhole measurements performed to date without the use of X-rays. The capabilities of this system extend the application space for in situ weld monitoring of laser processes with OCT.

In laser deep penetration welding a constant weld depth is an essential quality measure that is usually obtained by on-line process monitoring techniques such as optical coherence tomography (OCT). However, due to the lack of referential information about the keyhole shape, the extraction of meaningful information about the keyhole from the OCT signal is limited to the keyhole depth which is acquired by means of statistical analyses. In this research on-line X-ray-recordings alongside OCT measurements were conducted during the welding of pure aluminium (EN AW-1050A) in the Rosenthal regime to enable a more profound interpretation of the OCT signal. The comparison between both measurements enabled a correlation between geometric keyhole features and characteristics of the OCT frequency analysis, resulting in an improved time-resolution for the extraction of the keyhole depths. Furthermore, metallographic longitudinal cross-sections were used to investigate how close the keyhole depth is linked to the actual welding depth.

Laser beam welding offers unique capabilities for producing components in several industries, such as electric vehicle battery manufacturing. However, the high solidification rates and high coefficient of thermal expansion when welding 6082-T6 aluminium alloy result in strains that lead to hot cracking. This highlights the need for cost-effective process monitoring approaches. Optical systems such as high-speed cameras and sensors, including infrared and acoustic emission, are capable of monitoring strain-induced hot cracking. However, most of these methods are limited to surface region and are unable to investigate the sub-surface solidification dynamics where hot cracks originate. Therefore, high-speed X-ray diffraction is used for measuring sub-surface strain and stresses, but the process is time-consuming and expensive. Physics-informed models have emerged for reconstructing sub-surface dynamics using surface data from sensor measurements, however, these methods are still inaccurate. Hence, the present review explores machine-learning approaches for correlating real-time surface measurements with complex XRD-derived sub-surface dynamics.

Welding of aluminium alloys from the 6xxx series is challenging due to the formation of hot-cracks, however, their favorable mechanical properties require laser micro welding processes for housings in battery technology or microelectronics. In that regard, The determining factors for hot crack formation are residual strains and stresses that occur during solidification. In order to examine these in more detail, in-situ and ex-situ investigations using the high-energy X-ray diffraction beamline ID31 at the European Synchrotron Radiation Facility were conducted. The measurement data provided the possibility to calculate temporally and spatially resolved stresses in welding direction and in z-direction. Various prevention strategies, such as laser pulse modulation and application of an additional heating laser in various alignments were used to determine their effect on the residual stresses and hot crack formation. In addition, microsections allowed further insights on phenomena related to the melt zone and the heat affected zone.

Pyrometric temperature measurements are essential for quality monitoring of laser processes, but still face some major challenges. Temperatures usually need to be measured with high accuracy and high spatial resolution due to the typically small process scales of laser processes. However, the temperature accuracy is affected by the angle-dependent emissivity, which can lead to incorrect temperature measurements if the local inclination angle changes, and the spatial resolution is limited by the measurement spot size, leading to reduced spatial resolutions for larger spots.

This contribution addresses these issues and presents a method to correct the angle-dependent emissivity by simultaneously measuring surface temperature and geometry, and a method to increase the spatial resolution by compensating for the effect of the transfer function of the system, caused by the size of the measurement spot, using a reconstruction algorithm. The results show good agreement with reference measurements, demonstrating the effectiveness of the proposed methods.