In laser melting processes of metallic parts, including welding and additive manufacturing, there are challenges in porosity formation and developing predictive multiphysics of the process. Surrounding a melt pool with an external magnetic field has promise in changing the Marangoni flow and reducing porosity formation. In-situ X-ray imaging enables the observation of melt pool behavior and porosity formation in real-time. This preliminary study shows that an external magnetic field can achieve both, with potential to scale up in industrial processes and to validate multiphysics models.
The objective of this work is to observe how a static external magnetic field can influence the dynamic melt pool behavior of a ferromagnetic material, particularly in porosity formation, porosity evolution, and melt pool size. This preliminary study uses high-speed X-ray imaging at the advanced photon source at Argonne National Laboratory to visualize the dynamic behavior in real-time. A deeper understanding of laser–matter interactions in the micro-scale or smaller is necessary in order to pave the way for new materials and better controlled advanced manufacturing processes. Growing needs in laser-based manufacturing processes include functional surfaces in laser texturing and cladding and improved structural materials in laser additive manufacturing. Laser-based processing is necessary for many refractory materials with high corrosion resistance, longer fatigue life, and grain refinement strengthening for applications in biomedical, aerospace, and energy industries.
Using static external magnetic fields to manipulate laser-based processing is nothing new. Permanent magnets were shown to induce magnetic flux great enough to change the shape of a laser-induced plasma plume used in microtexturing from a spherical plume into elliptical and to even line-shaped plasma plumes, resulting in machined areas with various geometries [1,2]. During solidification of molten metal in particular, electromagnetic fields have been historically used to surround large casting crucibles or Bridgman furnaces to induce a Lorentz force and stabilize stirring within the molten metal, leading to grain refinement and a reduction of defects like porosity and microcracks .
Several studies have demonstrated how magnetic fields with flux values in the mT scale have changed the melt pool size and could reduce porosity. An early welding study found that the influence of an electromagnetic field on a steel spot weld led to the removal of porosity and an increase in impact toughness . Zhou et al. developed a mathematical model that captured how electromagnetic fields influence porosity formation in pulsed laser welding, where an electromagnetic force can increase the backfilling speed of melt pool during solidification, avoiding the formation of a keyhole pore due to keyhole collapse . Bachmann et al. found that a magnetic field led to slower flow velocities in the melt pool during laser welding of aluminum, modifying the local temperature and decreasing the melt pool and keyhole sizes . Rong et al. was able to measure the change in keyhole depth during laser welding in real-time with an external magnetic field, observing the shortening of a keyhole with increasing magnetic field . The majority of these studies do not show much experimental evidence of the change of the melt pool or porosity formation in real-time with magnetic fields, hence the motivation for this study.
Methods include high-powered laser processing to emulate laser welding and a foundation for laser-based additive manufacturing, but without powder feedstock material for a layer-by-layer build. In order to capture the changes in the melt pool and porosity with high temporal and spatial resolution, a simplified laser melting process with permanent magnets was placed in a high-energy X-ray beamline for dynamic phase contrast imaging.
Materials and Laser Processing.
The material that underwent laser melting was an AISI 4140 ferrite-bainite steel sample in the form of shims with a thickness of 300 μm . The surfaces of the 4140 steel samples were polished to eliminate the influence of roughness on laser absorption into the material. In order to image the melt pool with the X-ray beam, the steel sample was placed so that the top surface of the steel was about 500 μm above the top surfaces of the permanent magnets, as seen in Fig. 1. The laser scanned along the top surface of the steel sample. The laser used was an IPG fiber optic laser with a Gaussian profile, a beam diameter of 80 μm at its focal point, a maximum power of 520 W and a wavelength of 1070 nm . The laser was used in continuous wave mode and used at 30% power for all laser melting experiments, or at a power of 156 W. The laser was integrated to a galvanometer scanner , with a scan speed of 100 mm/s used for all experiments.
In experiments without an external magnetic field, the sample was placed between two glassy carbon plates. In the magnetic field experiments, the shim was placed between NdFeB permanent magnets. Figure 2 shows a finite element method magnetics (FEMM) simulation  of the magnetic field distribution of throughout the steel sample when the two magnets surrounded the sample. The magnetic field direction was outward, with attraction toward the center of the steel sample, where the magnetic field strength was the greatest at 600 mT. For laser melting at various magnetic field strengths, the laser scanned at different areas along the top surface of the steel sample, with lower field strengths close to the edge or outside of the boundary of the magnets. The approximate magnetic surface field values used included 0 mT (no magnet), 300 mT, 450 mT, and 600 mT. The areas along the top of the steel sample surface where the laser scanned are denoted in Fig. 2, which correspond to the magnetic field strengths at those areas.
High-Speed X-Ray Imaging.
The laser melting experiments were conducted at the 32-ID-B high-speed X-ray imaging beamline at the advanced photon source at Argonne National Laboratory. The X-ray beam penetrated through the 4140 steel sample, with the X-ray signal converted to visible light with a scintillator, which was collected by a high-speed camera. Initiating the laser melting experiments triggered the X-ray shutters and image collection. A simplified schematic of the experimental setup in the beamline is seen in Fig. 1, with more details in past literature describing the X-ray imaging technique . Several studies have used this setup for fundamental research in laser–matter interactions for applications in additive manufacturing, including keyhole formation , spattering , porosity elimination , melt pool flow comparison with multiphysics models , melt pool variation in laser powder bed fusion additive manufacturing , and directed energy deposition additive manufacturing [17,18].
The X-ray images were collected at a shutter speed, or exposure time, of 10 μs and at a frame rate of 50 kHz, or a temporal resolution of 20 μs between each frame. The spatial resolution of the images were 640 by 600 pixels, with a pixel size of about 1.9 μm, resulting in a field of view of about 1.2 mm wide and 1.1 mm tall. The open-source image analysis software, imagej  and simple matlab scripts were used to analyze the melt pool and porosity geometries of the images. A representative image outlining the melt pool and pointing to the laser-induced keyhole cavity, induced porosity at the liquid–solid boundary of the melt pool, and the moving laser toward the right is shown in Fig. 3.
Results and Discussion
Melt Pool Geometry.
The changes in melt pool depth were measured and averaged for experiments, with an example of the melt pool boundary of an X-ray image in Fig. 3. The maximum melt pool depth was measured at each time-step, with the depth defined by the distance from the surface of the steel sample to the bottom of the melt pool boundary. The depth was measured for each surface field magnetic strength, as seen in Fig. 4.
Based on Fig. 4, the melt pool depth fluctuated the most without an external magnetic field and at 300 mT field strength, as the Marangoni convection within the melt pool led to an unstable keyhole. With little to no magnetic field strength, fluctuations could be defined as the root-mean-square error (RMSE) of the depths compared to the logarithmic melt pool depth growth at a RMSE of 24. With increasing external magnetic field, fluctuations decreased to a RMSE of 22 at 450 mT and at 10 at 600 mT. The melt pool depth decreased from an average of about 420 μm at 0 mT surface field to an average of about 360 μm at 600 mT surface field. Without a magnetic field, or at low field strengths, the melt pool flow stirring direction was mostly in the direction of the laser. With greater magnetic field strengths, field lines moved inward due to the attraction between the two magnets, leading to more flow movement along the direction of the magnetic field and surpassing the forces of Marangoni convection and gravity within the melt pool.
During laser melting and solidification in keyhole mode, porosity formed at the bottom of the keyhole cavity due to rapid collapse of the cavity and an unstable keyhole . However, porosity in a moving melt pool with several forces at play can go through several stages, starting with the formation, then stirring within the melt pool, and finally solidification or dissipation. At the varying surface field magnetic strengths, these porosity life cycles also differed.
As seen in Fig. 3, porosity formed at the bottom of the keyhole and moved against the laser scanning direction due to the drag force and Marangoni force in the melt pool. Figure 3 shows the representative porosity structure of a solidified laser melt without an external magnetic field. The porosity pinned at the solid–liquid boundary resembled most gas-entrapped, keyhole spherical pores. In addition, the pore at the center of the melt pool stirred within the melt pool with velocities ranging from 550 to 700 mm/s before it was also pinned due to solidification.
With an increase in magnetic surface field strength, porosity behavior followed the melt pool flow in the liquid region. During porosity formation, porosity coalesced at periodic locations, signifying the cooling rate due to the scanning speed. The coalesced pores as seen in Fig. 5 were angled in the direction of the magnetic field, signifying upward, epitaxial molten flow.
In addition, pores that were not pinned at the liquid–solid boundary and were in the center of the melt pool did not stir with high velocities. Instead, they stirred with velocities of only a few mm/s, and mostly were pinned within the melt pool before solidification. This indicates that an external magnetic field stabilizes and drastically slows down the convection within the melt pool.
Finally, at the maximum magnetic surface field strength of 600 mT, very little porosity formed, as seen in Fig. 6. This is because the melt pool and keyhole cavity were stable with less fluctuations, in addition to having less depth due to the magnetic field pushing the flow upward. Because of this, no keyhole porosity formed.
Overall, high-speed X-ray imaging elucidated how melt pool fluctuations decreased over time and how porosity reduced with an external magnetic field. Further, better controlled experiments with various magnetic field orientations are required. Comparisons with multiphysics models could also reveal how magnetic fields influence the melt pool. Understanding and controlling a metal molten pool can lead to improvements in laser processing.
The authors would like to acknowledge Benjamin Aronson, Alex Deriy, and Kamel Fezzaa who helped with the experiments. This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory (ANL) under Contract No. DE-AC02-06CH11357 in addition to support through Laboratory Directed Research and Development (LDRD) funding from the DOE Office of Science under the same contract.
Laboratory Directed Research and Development (Grant No. DE-AC02-06CH1135; Funder ID: 10.13039/100007000).