The highly dynamic, complex, and short-lived physical phenomena that occur during the 3D-printing process have eluded detection and understanding since the advent of the technology more than three decades ago. Now, however, research conducted at the U.S. Department of Energy’s (DOE) Advanced Photon Source (APS) at Argonne National Laboratory in Illinois is revealing for the first time how microscopic defects arise during the production of 3D-printed metals. The research harnesses the power of the APS, the nation’s leading source of hard (short-wavelength) X-rays, to image the 3D-printing process, also called additive manufacturing. Check out the video:
“We look to expand our capabilities and do research more broadly to understand the additive manufacturing process,” says Aaron Greco, the lead principal investigator for Argonne’s Additive Manufacturing Consortium. “Those capabilities include computer simulation, machine learning, and materials characterization to determine where defects form during additive manufacturing and how to minimize or eliminate those defects.”
Argonne’s first major contribution to the field, in collaboration with researchers at Carnegie Mellon University and the Missouri University of Science and Technology, was an X-ray analysis published last June in Scientific Reports. The study, led by Tao Sun, a physicist in Argonne’s X-ray Science Division, applied high-speed X-ray imaging and diffraction to observe a commonly used 3D-printing process.
“The experimental and data analysis approaches developed here will provide a compass pointing to the fundamental understanding of the physics in the AM process, and will further accelerate the coming of the Additive Manufacturing age,” wrote Sun and his co-authors.
The team carried out its research on the titanium alloy Ti-6Al-4V at the APS 32-ID beamline, which offers high-speed X-ray imaging and diffraction among its capabilities. X-ray images were taken at the rate of 50,000 images per second. Imaging was coupled with the diffraction technique, which involves the scattering of high-intensity X-rays that enabled the team to visualize the dynamical interactions taking place during the transient laser-metal interaction.
The titanium alloy that the team experimented with is a commonly used 3D-printing material. Other often-used materials are aluminum alloys, stainless steel, and nickel superalloys. APS X-rays can penetrate the millimeter-sized titanium and aluminum alloys, but not yet stainless steel or nickel superalloys for single X-ray pulse experiments.
That could change, however, under a proposed $750 million upgrade to the facility, which would make the beam much brighter and more focused. With that upgrade, Sun said, the APS would have a better chance of going through stainless-steel samples (less than two millimeters) in megahertz imaging and diffraction experiments.
Argonne selected snapshots from larger datasets (i.e., each complete set includes a total of 340 images covering a 7.5-ms timespan), to show the details of laser-sample interaction within a ~110 ms time window (see figure below). The laser beam positions are marked in red in the first image of each image series. On the left of the laser beam, a “printed” line can be observed, while the un-melted powder bed is on the right side. Underneath the laser beam, a vapor depression zone inside the metal base can be detected. In the case shown in (a), a moderate laser power is applied. A small depression zone is formed and some particles (including molten metal and raw powders) are spatter ejected from the powder bed by the metal vapor. This is a heating mode often referred as conduction mode.
Argonne researchers have provided a first-of-its-kind in-situ observation and measurement of the metal additive manufacturing process. Shown here are dynamic x-ray image series of two laser powder-bed fusion processes of Ti-6Al-4V with different laser powers (a: 210 W; b: 360 W). In both cases, the laser beam with a size of ~100 μm is scanned from the left to the right with a speed of 0.4 m/s to “print” one line. The frame rates of two data series are both 45.3 kHz, and the exposure time for each image is 0.1 ns. Note the straight lines in the middle of each image are the top edges of the glassy carbon plates, which sandwich a Ti-6Al-4V base plate and a layer of powders on top. The thickness of the metal base and powder bed along the X-ray beam direction is 0.5 mm.
In the case shown in (b), the higher laser power generates a much deeper depression zone, and at the same time, more particles are ejected. Also, a track of pores can be observed along the laser path near the depth of the depression zone. The formation of one such pore is clearly revealed in this image series. This heating mode is called “keyhole” mode, in which an excessive thermal power is deposited on the sample. The porosity formed in this heating mode is often referred as “keyhole” porosity, which is one of the major defects in the additively manufactured metal parts.
Experiments with X-Rays and 3D Printing
The initial APS experiments were a simplified version of the actual commercial 3D-printing process, conducted with support from Argonne, DOE's Office of Science, DOE’s National Nuclear Security Agency, DOE's Office of Energy Efficiency and Renewable Energy, the Grumman Corp., and the University of Missouri Research Board. In future work, Sun would like to incorporate a small, commercial 3D printer into the beamline, or otherwise develop the experimental conditions to more accurately reproduce the industrial 3D-printing process.
The Scientific Reports study dissected the details of the laser powder-bed fusion method of 3D printing. In this method, a laser beam heats metallic powders, which melt and agglomerate to form the desired part one layer at a time. But the introduction of powders to laser processing is relatively new. 3D printing has its roots in laser welding, which involves no powders.
The powder exhibits extremely dynamic laser-driven motion—powder spatter injection is the umbrella term—during the laser powder-bed fusion process. When the laser hits the powder bed, some metal powders vaporize immediately. The metal vapor carries molten materials and many raw powder particles away from the bed with high velocity (up to tens of meters per second). The particles solidify into irregular sizes and shapes and fall back onto the powder bed.
“When they fall back on the powder bed, they become the source of defects,” says Sun.
The direct measurement of the powder ejection velocity using high-speed X-ray imaging can help understand the underpinning physics to optimize the laser melting condition and develop efficient ways to minimize the defects caused by powder ejection.
“As we have demonstrated with our work at 32-ID, visualizing these high-speed processes is immensely revealing and only APS/Argonne can help with that,” says Anthony Rollett, a co-author of the Scientific Reports study and a professor of materials science and engineering at Carnegie Mellon Univ. in Pittsburgh. “There are also many aspects of materials microstructure that diffraction probes can help with.”
The Advanced Photon Source at Argonne National Laboratory is a national synchrotron-radiation light source research facility funded by the United States Department of Energy Office of Science. The facility "saw first light" on March 26, 1995. (Source: Wikipedia)
One objective of the APS studies is to understand the mechanisms responsible for porosity formation. For example, why are there so many different porosities in the additive-manufactured metal parts?
Another objective is to measure various structural parameters with in situ X-ray techniques that previously could not be measured by other means. How does the melt pool develop, for example, and how does the dynamic fluctuation of the vapor depression zone affect the “keyhole” pore formation?
“Nobody could see that before,” says Sun. With that data in hand, Sun and his collaborators now can validate the many numerical models that computational materials scientists at Argonne and elsewhere have developed. Until now, researchers have “guessed” at what causes defect formation in 3D printing.
Sun says, “Now we have direct observations. We will be able to confirm or disprove a lot of the models people have proposed. And more importantly, we can help them build new models by involving more physics than before.”
One example is powder flow and injection, which is so complex that modelers currently omit it from their simulations. That’s due to the high computational costs of simulating the dynamics of millions of particles in the powder bed.
“In their models, the powders are pretty much stationary,” explains Sun. “The dynamic motion of the powder is so dynamic you actually can’t ignore this.”
Since that first APS study, Argonne has invested more than a million dollars into its Additive Manufacturing Initiative, which is part of the laboratory’s larger Manufacturing Science Initiative. Although still in its early stages, Argonne’s AM Initiative already has attracted interest from more than 10 companies in the aerospace, defense, and automotive industries.
Eyeing the Future through Collaboration
One attractive aspect of Argonne’s AM Initiative is the opportunity to collaborate with universities to train the next-generation workforce in additive manufacturing, says Lianyi Chen, a co-author of the Scientific Reports article.
Chen, an assistant professor of mechanical and aerospace engineering at the Missouri University of Science and Technology in Rolla, Mo., further noted the multiple strengths that Argonne collaborations with academia and industry bring to additive manufacturing. Argonne has world-leading capabilities in hard X-rays and computation. Universities offer educational resources and fundamental additive-manufacturing research expertise. And industry knows from experience the problems and barriers that block more widespread adoption of additive-manufacturing technologies from commercial production.
“The synergistic collaboration of Argonne and its academic and industry partners will have great potential to solve the challenges in AM technology,” says Chen. “For example, understanding the dynamics of AM processes by in-situ hard X-ray imaging and diffraction will guide the development of better processing technologies and better feed stock materials. The combination of in-situ characterization and computation can lead to a high fidelity model or machine-learning algorithm to predict optimized processing parameters to achieve optimized part quality.”
Selective laser melting of a powder bed currently dominates 3D-printing technology, but changes are afoot, says Rollett. “The increasingly broad scope of application of 3D metals printing means that other technologies such as binder jet and robotic wire feed will become more significant. Robotic wire feed is important for larger parts. Binder jet is a lower-cost technology that is equally suited to ceramics as it is to metals.”
Whatever future directions are taken by 3D printing, Sun, Rollett, Chen, and their associates will work toward the day when the technology is limited only by an engineer’s imagination.
To learn about tapping into Argonne’s facilities and expertise in this area, contact [email protected] The APS is a DOE Office of Science User Facility operated for the DOE Office of Science by Argonne.