2020 官网升级中!现在您访问官网的浏览器设备分辨率宽度低于1280px请使用高分辨率宽度访问。
Source: Yangtze River Delta G60 Laser Alliance
Introduction: This article investigates the effect of powder oxidation on melt pool dynamics and defect formation during the LAM process.
Understanding the formation of defects in laser additive manufacturing (LAM) processes for raw, stored, and reused powders is crucial for producing high-quality additive manufacturing parts. This article investigates the effect of powder oxidation on the kinetics of melt pool and defect formation during the LAM process. During LAM, powder oxides are entrained into the molten pool, changing Marangoni convection from inward centrifugal flow to outward centripetal flow. Assuming that oxides promote pore nucleation, stability, and growth. We observed that splashing is more frequent under suspension conditions compared to layer by layer conditions. Droplet splashing can be induced by indirect laser driven gas expansion and laser induced metal vapor formation on the surface of the melt. Under layer by layer construction conditions, laser remelting reduces pore size distribution and quantity density by promoting gas release from keyholes or inducing liquid flow (partially or completely filling pre existing pores). We also observed that during the laser remelting process, holes located on the surface of the orbit may rupture, leading to the formation of droplet splashes and open holes, or the healing of holes through Marangoni flow. This study confirms that excessive oxygen in powder raw materials may lead to the formation of defects in LAM.
Figure Summary: Using a laser additive manufacturing process replicator with in situ and operational X-ray imaging (a) allows for the capture of (b) pores and (c) the formation of splashes during laser material interaction. In addition, we conducted post mortem X-ray computed tomography analysis (d), revealing two types of pores within the melt trajectory: (i) open pores and (ii) closed pores.
1 Introduction
Laser additive manufacturing (LAM) uses a focused laser beam to selectively fuse powder particles layer by layer to construct complex 3D objects. It has great prospects in aerospace, nuclear fusion and energy storage applications; However, the application of LAM technology in these fields is hindered by inconsistent component performance. Specifically, due to the accumulation of residual stress and the presence of defects such as porosity, spheroidization, and cracks, the mechanical, thermal, and electrical properties of additive manufacturing components are lower than those of forged components.
The nickel based high-temperature alloy (Inconel 718) turbine blades used in jet engines are produced through direct laser metal sintering (a form of metal additive manufacturing (MAM)).
MAM technology makes it possible to construct parts layer by layer from powder or wire raw materials (Figure 1). Laser or electron beam or plasma arc are typically used to selectively melt raw materials together (based on computer-generated design files), allowing for the construction of parts through continuous grating beams and supplementary materials. 3. Compared to traditional methods, this process has multiple advantages, including the ability to produce hollow and lightweight parts, parts with geometric shapes that cannot be produced traditionally, and the ability to perform on-site repairs. Compared to traditional metal processing, a special advantage of MAM is that it can produce very small batches of parts in a short period of time with less financial investment (compared to casting that requires expensive molds), making it an ideal choice for small batch or one-time parts and rapid prototyping. These advantages make MAM attractive in a wide range of industries, including biomedical engineering, transportation and national defense.
Although computer simulations can provide some physical understanding of additive manufacturing (AM) processes, they require experimental data for model validation and validation, especially regarding melt pool and defect dynamics. Some data can be collected using on-site monitoring equipment installed on the AM system. However, when single-layer or multi-layer tracks are formed, these devices cannot reveal the dynamic behavior inside the melt pool or melt track (such as the evolution of porosity and lack of fusion defects).
Strain map measured during in-situ strain in a scanning electron microscope.
Comparing the velocity values between three curves at the same angle, the velocity continuously decreases within a period of 0.344 – 0.731 ms. Due to the almost constant gravity and friction, the decrease in particle velocity may be caused by the decay of expanded metal vapor.
Powder splashing and droplet splashing are two other common defects found in LAM. They affect the synthetic porosity and surface finish of AM parts. They may also cause contamination of the powder bed, improper powder diffusion, and damage to the AM system.
Our goal here is to identify how different levels of powder oxidation affect the AM process, including their effects on melt pool dynamics and defect formation. To this end, we use in situ and operational synchrotron radiation X-ray imaging to monitor the LAM process in real time. We investigated the impact of powder oxidation by studying LAM using raw and oxidized (stored for about 1 year) invar 36 powder raw materials. Our results reveal how oxides reverse Marangoni flow, directly affecting the formation of different types of defects.
2. Results and Discussion
The particle size distribution of the oxidized powder is 5 – 70 μ m. Mode is 10 Å μ M (Figure 1). The SEM image in the illustration shows the powder surface before B2 was covered by oxides (Figure 1a), however, it shows a similar morphology and shape of the original powder before B1. XRD pattern (Figure 1b) and expected face centered cubic γ- (Fe, Ni) are consistent.
Figure 1. Powder characteristics of Invar 36: (a) Particle size distribution. Illustration: Oxygen EDS image overlaid on SEM secondary electron image. (b) XRD pattern shows presence γ Phase.
The effect of powder surface chemistry on melt pool dynamics is not yet clear, so we used XPS to examine the powder surfaces of the original (reference) and oxidized invar 36 powders. Figure 2 shows high-resolution scans of Ni 2p, Fe 2p, O 1s, and C 1s in two powder samples, showing the presence of Fe, Ni, FeO, Fe2O3, NiO, Ni (OH) 2, and amorphous carbon pollutants.
Figure 2 (a – d) XPS spectra of the original and (e – h) oxidized invar 36 powders.
From the high-resolution XPS scans of Ni (Fig. 2a and e) and Fe (Fig. 2b and f), the shapes and peak area percentages of metals, metal oxides and metal hydroxides are very similar. This indicates that metal oxides/hydroxides are easily formed during powder processing, including during powder packaging and transfer processes. During LAM, metal hydroxides are likely to be thermally decomposed into metal oxides and then released into the molten pool. The presence of iron oxide and nickel oxide in the molten pool will cause the temperature coefficient of its surface tension to change from negative to positive, leading to the reversal of Marangoni convection and the generation of centripetal convection.
2.1. LAM of raw and oxidized Invar 36 powder
The initial, intermediate, and final stages of the evolution of the original powder melt trajectory are shown in Figure 3a. The high-power density laser beam melts the invar 36 powder particles to form a molten pool, and then evaporates the top surface of the molten pool to form a metal vapor jet. We assume that the metal vapor jet indirectly heats the argon gas in the laser material interaction zone; Both of these effects will promote powder entrainment into the melt pool, splashing, and trajectory growth. When the movement speed of the laser beam exceeds the growth speed of the molten pool, it will generate a separate molten pool before the trajectory of the molten pool.
Figure 3 shows a time series radiograph of the melt characteristics observed during the LAM of the first layer Invar 36 melt trajectory.
3.2. Splash evolution mechanism
From single-layer melt trajectory experiments using raw and oxidized powders, we observed powder injection and droplet splashing throughout the entire LAM. Our results indicate that the laser melt orbit interaction generates a laser induced steam jet and a recoil pressure perpendicular to the surface of the melt orbit, which sprays the powder while generating the erosion zone (Figure 3 and Figure 4). We speculate that the erosion zone is inverted bell shaped and contains high concentrations of metal vapor (Figure 4a). High temperature metal vapor indirectly heats the surrounding argon gas, generating convective or inward argon airflow in the erosion zone, promoting steam driven powder entrainment and extending the melt trajectory.
Figure 4 shows a schematic diagram of the influence of the position of the laser beam in the melt trajectory on the splatter evolution: (a) the formation of powder splatter when the laser beam is located on the melt trajectory; (b) When the laser beam is located in front of the melt trajectory, the formation of droplet splashing occurs.
During suspension construction, the melting trajectory extends horizontally and deeper into the powder bed, as the powder particles near the melting trajectory are removed by a combination of metal vapor and hot argon gas. The laser beam melts the powder deeper into the powder bed and before the melting trajectory (Figure 4a), thereby reducing the growth rate during the extension of the melting trajectory. The laser beam continues to move and ultimately illuminates the powder in front of the melt trajectory, forming new molten beads (Figure 4b). Sometimes, the laser beam moves before the first molten bead, forming another molten bead while growing the first molten bead, because the laser beam contour is wide enough to interact with the powder between the molten bead and the two molten beads.
Figure 5a shows a positive correlation between splash size and velocity, despite significant scattering.
Figure 5 LAM splash analysis of raw and oxidized powders, divided into three categories: first, only powder splash; 2、 Powder splashing/agglomeration+droplet splashing; 3、 Only droplets splash. (a) Splash size and speed, as well as (b) the splash form of each type.
Figure 5b illustrates the different splashing patterns during LAM. For the original powder, the splashes are roughly spherical in all size categories. For oxidized powders, Class I and Class II splashes have irregular shapes and are formed by agglomerated powders. It seems to hinder the coarsening of spherical droplets, indicating that oxides are chemically and/or physically different. Class III splashes are mainly composed of droplet splashes covered by agglomerated powder on the surface. The evidence clearly indicates that powder oxidation strongly affects powder agglomeration, pore formation, and pore stability.
In order to investigate the impact of splashing on the quality of parts, the laser single scan trajectories on AlSi10Mg samples were characterized using an optical microscope after in-situ X-ray imaging experiments. The loose powder on the sample is blown away by compressed air before taking optical images. The top and side views of the track are shown in the above figures (a) and (b), respectively. Some particles were observed to be sintered on the track. A typical type of residual particles is solidified liquid splashes, as shown by the red circle. The size of liquid splashes can be much larger than the size of the original powder. When some cold particles splash (the same size as raw powder) into the laser beam area, the particles can melt into small droplets. Large liquid splashes can be formed by the collision of small droplets. Solidified large splashes can cause defects in completed parts, as (1) splashes may carry high levels of oxygen content, thereby reducing the wetting of the substrate. (2) During the laser scanning process, large splashes may not completely melt, becoming potential sites for pore formation.
3.3. Effect of molten pool dynamics on pore fracture
In the original powder study, under the processing conditions used in this study, pore fracture was not significant. However, Leung et al. demonstrated that pore rupture occurs through pore aggregation and migration during the solidification process.
In the study of oxidized powders, we revealed different pore breaking mechanisms in the LAM process of the second layer melt trajectory, as shown in Figure 6. The laser beam forms a small hole, penetrates the second powder layer and remeltes the top surface of the first layer’s melt trajectory (Figure 6a). Laser remelting promotes the transmission of pores in the melt pool, allowing pores to escape into the atmosphere through locked pores. Similar observations are shown in the first layer of melt trajectory and Figure 3b. In order to carry out gas entrainment, these pores must be located approximately 1-mm from the surface of the powder bed (based on our settings).
Figure 6: Time series radiograph showing the second layer of invar 36 melt trajectory of oxidized powder (P=150).
Figure 6b reveals a new mechanism of pore formation during LAM. At 7 Å ms, the laser remeltes the surface of the first layer of molten orbit and forms a liquid bridge (represented by a purple dashed line). Between 10 and 34.8 ms, the laser beam doubles the size of the liquid bridge while accelerating its internal melt flow, thereby promoting pore aggregation, growth, and transmission. Through 34.8 Å ms, the Marangoni driven airflow entrains pores at both ends of the liquid bridge, significantly weakening its structural integrity. The laser beam will increase the temperature of the material around the pores, heat the pores (see the red dashed arrow), and proportionally expand the volume of the pores. Once the gas pressure exceeds the surface tension of the liquid bridge, the liquid bridge will rupture (35 Å ms) and act as a liquid metal flow (36 Å ms), forming droplet splashes. Therefore, the closed pores rupture and open, leaving dents or pits on the surface of the melt track.
Figure 7 shows the evolution of the third layer melt trajectory in oxidized powder LAM. Similar to the second layer melting trajectory, the laser beam melts the powder above the open hole in the front of the melting trajectory, forming a liquid bridge that temporarily closes the hole. With the development of LAM, the laser beam causes indirect laser driven gas expansion in the hole, which overcomes the strength of the liquid bridge and leads to hole rupture, followed by the formation of open holes and droplet splashing. This reproducible observation indicates that pore rupture is the key formation mechanism for droplet splashing and pore opening in oxidized powder LAM, which may be applicable to the original powder LAM.
Figure 7 Time series radiograph shows the LAM of the third layer Invar 36 melt trajectory (B2.3) at 150 W.
Figure 7b reveals a new pore healing mechanism during LAM. The laser beam penetrates the trajectory of the second and third layers of the melt, opening pre existing holes (361 milliseconds). The gas expands towards the inner diameter of the pre existing pores, pushing the liquid metal upwards (362 milliseconds). Subsequently, the liquid metal driven by high surface tension rotates back to the top track (indicated by the red dashed arrow), healing the pores. The inward flow of liquid metal may be driven by Marangoni convection, combined with the weight of the liquid metal, causing the molten pool to rotate upwards and fall back, thereby repairing the pores (361-363 milliseconds).
3.4. Time resolved quantification of melt pool geometry and porosity
Using X-ray photographs, we quantified the changes in the geometric shape of the melt trajectory and its internal porosity throughout the LAM, as shown in Figure 8. Figure 8a shows that the trajectory length of the oxidized powder (L-B2.1) is about 20% longer than that of the original powder (L-B1.1). This is due to 1) the reduced surface tension leading to further expansion of the molten pool, and 2) an increase in splashes in front of the laser beam, extending the trajectory towards the bottom of the powder bed.
Figure 8 Quantification of melt characteristics during LAM: (a) Length and depth of the first layer (B2.1), second layer (B2.2), and third layer (B2.3) melt trajectories, and (b) Changes in porosity over time in each melt trajectory.
3.5. Post mortem3D analysis
The pixel resolution of the synchrotron radiation X-ray imaging device is 6.6 μ m. This means that we cannot quantify diameters less than approximately 20 μ m. In addition, radiographic analysis did not consider the depth of pores along the X-ray beam path. Therefore, we conducted high-resolution XCT scans to examine samples made from raw and oxidized powders, visualize and quantify morphology and pore size distribution in 3D.
Figure 9a shows that the melt trajectory generated by the original powder shows a porosity of 0.08%. According to the resolution of XCT data, the melt trajectory shows no open holes, but includes some areas with an equivalent diameter of 10 Å μ m。 Figure 9b shows that the total porosity of the melt trajectory generated by the oxidized powder is 15.1%, with two-thirds (8.6%) being open pores and one-third (6.5%) being closed pores.
Figure 9 3D volume rendering of melt trajectories made from (a) raw powder (B1) and (b) oxidized powder (B2). (c) Their corresponding pore size distribution.
Adjusting the layer thickness is an effective method to reduce powder splashing. As shown in the figure below, the steam jet has an inverted cone, which is more limited to the vicinity of the molten pool but expands when away from it. Therefore, thinner powder beds have a better chance of being completely melted by the laser and leaving less loose powder in the steam injection path for spraying.
The optical image shows the effect of powder layer thickness on the microstructure of AlSi10Mg samples manufactured with additives.
By comparing AM components constructed with different layer thicknesses, evidence was found to confirm the impact of layer thickness. Two AlSi10Mg samples were manufactured in commercial AM machines with layer thicknesses of 50 respectively μ M and 30 μ m。 The cross-section of the two samples is cut perpendicular to the direction of the building. The optical images of the two cross-sections are shown in the figure above (a, b). The pores caused by splashing can occur in thicker layers (50 μ m) Found in the sample, as shown in Figure (a). However, in the presence of a relatively thin layer (30 μ m) Few defects caused by splashing were found in the samples. The density of the two samples also supports the above observations. fifty μ The density of the m-layer sample is 2.5648 g/cm3, below 30 μ Density of M-layer samples
4. Conclusion
This study explores the influence of powder oxidation on the kinetics of melt pools and reveals new evolutionary mechanisms of splashing, porosity, and exfoliation zones in the LAM process of raw and oxidized powders.
Three types of powders were characterized by SEM-EDS, IGF-IR, and XPS, including the original powder of B1, the oxidized powder of B2, and the reference original powder. Oxidized powders exhibit an increase in oxide layer thickness due to oxygen generated during powder processing and/or long-term storage under non ideal conditions.
Our results confirm that bath wetting and steam driven powder entrainment are key orbital growth mechanisms for LAM. The oxygen content from the oxidized powder is sufficient to change the temperature coefficient of the surface tension of the molten invar 36 from negative to positive, thereby transforming the Marangoni convection from outward centrifugal flow to inward centripetal flow. Oxides can serve as nucleation sites for pore formation and subsequently stabilize these pores.
We have discovered two new phenomena related to pore rupture in the LAM process: (1) promoting pore healing through liquid supply, and (2) inducing pore opening through the formation of droplet splashes. This indicates that droplet splashing can be achieved through indirect laser driven gas expansion within the melt trajectory and laser induced steam jet formation on the melt surface.
The quantitative results and proposed mechanisms indicate that defects in additive manufacturing can be minimized by using low oxygen content metal powders. The new formation mechanisms of opening and droplet splashing can enhance existing process simulation models to predict these defects. The quantification of melt trajectory geometry over time can be used to calibrate simulation models to accurately predict fluid flow behavior during LAM. Finally, the quantification of porosity over time can be used to validate and enhance existing process simulations for defect prediction under layer by layer construction conditions.
source：The effect of powder oxidation on defect formation in laser additive manufacturing, Acta Materialia, doi.org/10.1016/j.actamat.2018.12.027
reference：Policy needed for additive manufacturing, Nat. Mater., 15 (2016), pp. 815-818, https://doi.org/10.1038/nmat4658
Alloy design for aircraft engines, Nat. Mater., 15 (2016), pp. 809-815, https://doi.org/10.1038/nmat4709
