Influence of the inner roughness of powder channels on the powder propagation behavior in laser metal deposition

Laser metal deposition (LMD) has become an increasingly important process in recent decades.¹ In addition to the production of 3D-printed components, other important areas of application include the repair of high-quality components. In order to ensure economic efficiency, it is essential to consider both process efficiency and resource efficiency. The design of the powder feed is of great importance in both of these areas. In terms of resource efficiency, it is important that the powder utilization rate is as high as possible, as this minimizes the wastage of raw materials. At the same time, a high powder utilization rate also allows for a high buildup rate. A powder nozzle should also have the largest possible working distance and a large working area. The large working distance often protects the sensitive powder nozzle from heat generated by the process and from flying powder particles that can stick to it. The expansive working area enables the process to remain stable, even in the event of minor discrepancies in the distance between the workpiece and nozzle. It is crucial to maintain a narrow divergence angle of the powder stream to ensure a substantial working distance with a broad working range and precise focusing of the powder stream simultaneously. It is imperative that both of these aspects be met in order to guarantee the uninterrupted production of components or repairs.

Particle-wall collisions are known to have an impact on both divergence angle and particle velocity.² A real collision between two bodies leads to a reduction in velocity, as this impact is a mixture of elastic and plastic impact. The key figure to describe this relationship is the coefficient of restitution (COR). This results from the ratio of the velocities of the two bodies before and after the impact $v 1/ v 2$⁠. The studies by Odum et al. concentrate on the experimental determination of the normal COR. The velocities before and after an impact of 316L powder with a size distribution of 45–105 μm were investigated. Different materials with different roughnesses were selected as counterparts. The impact velocities were in the typical LMD range, between 0.65 and 13.55 m/s. The roughness of the substrates ranged from approximately Ra = 0.1 to Ra = 0.7 μm. The observed normal components of the COR values exhibited a range of 0.42–0.55. The normal COR was influenced by the material of the counter body, the speed before the impact, and the impact angle. However, the authors did not observe an influence of surface roughness on the normal COR.³ 

In addition to the influence on the particle velocity, particle-wall collisions also have an influence on the divergence angle of the emerging powder. The experimental and numerical investigations of powder nozzles by Polyanskiy et al. demonstrated that the divergence angle for copper injectors is between 10° and 11°, while it reaches a value between 30° and 35° for steel injectors. This phenomenon can be attributed to the differing scattering of rebound angles for different materials, which the particle trajectories undergo following their exit from the nozzle.⁴ 

Previous numerical simulations of powder flow have demonstrated that wall roughness plays a significant role in achieving a powder flow with minimal divergence. Additionally, it has been demonstrated that an increase in the value of $σγ$ results in a reduction in particle velocity.⁹ 

This paper examines the impact of varying inner powder channel roughnesses and the resulting standard deviation of the surface roughness angle $( σ γ)$ on the divergence of powder stream and particle velocities. To this end, a powder bed-printed HOVADUR CNCS cuboid with powder channels of varying inner surface roughnesses was employed. The powder stream from each channel was observed by a high-speed camera and subjected to comprehensive analysis.

The powder material was a nickel base alloy (EuTroloy 16625G.04) with a nominal particle size of −106 to +45 μm, sourced from Castolin Eutectic Ireland Ltd. The chemical composition of the powder material is presented in Table I. The scanning electron microscope (SEM) images (Fig. 2) of the powder demonstrate that the powder particles are primarily spherical, with satellite particles. A few particles exhibit an irregular shape.

A cuboid of HOVADUR CNCS was printed with powder bed fusion by the company Schmelzmetall. The chemical composition can be found in Table II. Subsequently, powder channels were placed into the cuboids by wire electrical discharge machining (by the Institute for Machine Tools and Factory Management, Technische Universität Berlin), resulting in the production of five channels with varying surface roughness. The channels were circular in shape, with a diameter of 1.54 ± 0.04 mm. The length of the powder channel is 19 mm (see Fig. 3) after placement of thread.

Following the completion of the experiments, the channels were cut in half by electrical discharge machining in order to measure their roughness. To measure the roughness, a Keyence VK 9700 confocal microscope was utilized with a 50× magnification and a numerical aperture (NA) of 0.55. Several images were combined to achieve a measurement length of at least 4 mm. Following the correction of the surface shape in order to flatten the curved data, the measurement data were filtered with a low-pass (λs) filter of 0.8 μm and a high-pass (λc) filter of 0.8 mm. The roughness profiles were then extracted using 10 parallel lines with a distance of 13.9 μm, and the values Ra and RSm were determined using these. The surface arithmetic average roughness Sa was also measured using the same values for low-pass and high-pass filters as for the line roughness. The diagram in Fig. 4 shows all measured values.

Figure 5 depicts a representative excerpt from the measurement topography of the five inner channel surfaces. The presence of defects within the surface can be observed in instances where a smoother finish is present. These defects are likely the result of imperfections in the HOVADUR CNCS samples, such as pores and bonding defects, which are inherent to the powder bed process.

The standard deviations of the surface roughness angle $σγ$ for both cases described in Formula (1) were calculated (Fig. 6).

The velocity of the carrier gas was quantified using a Prandtl tube. The tube was positioned 10 mm beneath the channel exit, oriented directly into the gas stream. Each measurement was repeated five times.

High-speed videos were captured using an i-Speed 7 high-speed camera from iX Cameras. All videos were captured at a frame rate of 45 kHz to observe the particles ejected from the powder channel and had a minimum resolution of 21 μm/px. The videos were illuminated using a Cavilux illumination laser from Cavitar. Particle tracking was conducted with TrackMate¹⁰ within imagej software.

The high contrast achieved in the videos through the use of the illumination laser enabled automated particle detection after background reduction. Particles were tracked and information on each detection was saved. From these, individual trajectories could be calculated, including the mean velocity within each particle track and their flight angle. Rebounding particles (having a negative velocity) were filtered from the results. The mean particle velocity ±2σ and their corresponding particle flight angles were divided into bins (according to the Freedman–Diaconis rule). For each of these bins, the 2σ of the particle flight angles was calculated.

To calculate the divergence angle of the powder, a background model of the video was first calculated from all 10 000 frames. The background model was then subtracted from each individual frame, and the resulting binary images were summed up to create an intensity map of the particles’ locations over 10 000 frames (see Fig. 7). The intensity along a line, orthogonal to the powder channel, situated close to the channel exit and at a distance of approximately 15 mm to the channel exit, was extracted. A Gaussian fit was calculated for this intensity, with 2σ marking the boundary of the powder stream. The divergence angle was then calculated between these boundaries.

Particle velocities as a function of increasing powder mass flow rate for different inner channel roughness values are presented in Fig. 8. It can be observed that with increasing powder mass flow rate, the particle velocity remains constant. The lowest velocities are observed for the highest Ra value channel. Conversely, the highest particle velocities are recorded for the second lowest Ra value powder channel.

Figure 9 depicts the particle velocities for varying carrier gas flow rates across all inner channel roughnesses. It can be observed that an increase in carrier gas flow rate is accompanied by a seemingly linear increase in particle velocity for all channels. The powder channel with the highest Ra value exhibits the lowest particle velocities. The highest particle velocities are observed across all carrier gas flow rates for the second lowest Ra value powder channel.

The divergence angle of the powder stream for different powder mass flow rates can be observed in Fig. 10. For the two roughest powder channels, the divergence angle appears to decrease with increasing powder mass flow rate. In contrast, the other three surface finishes exhibit a seemingly constant value with the increase in powder mass flow rate. The lowest values of divergence angle can be observed for the smoothest surface finish of Ra = 0.27 μm. The divergence angle values also increase in ascending order according to the Ra value.

With regard to the influence of inner channel roughness, similar observations could be made for an increasing carrier gas flow, as seen in Fig. 11. A slight increase in divergence angle can be observed for all inner channel roughnesses. Once again, the smoothest surface finish leads to the smallest divergence angles. Conversely, the highest surface roughness leads to the highest divergence angles.

The standard deviation of the surface roughness angle $σγ$ was selected for d_p > RSm, as a greater proportion of the particles is larger than the respective RSm values. Figure 12 illustrates the particle velocity for varying carrier gas flow rates as a function of the angle $σγ$⁠. For each of the carrier gas flow rates, the particle velocities exhibit a slight decreasing trend with increasing angle $σγ$⁠. As the angle $σγ$ increased from 0.42° to 2.17° (an increase of 423%), the mean particle velocity for a 2 l/min carrier gas flow rate decreased from 4.49 to 3.5 m/s (a decrease of 22%), while for a 5 l/min carrier gas flow rate, the mean particle velocity decreased from 9.85 to 7.57 m/s (a decrease of 23%).

Figure 13 depicts the divergence angle as a function of the surface roughness angle $σγ$⁠. An almost linear increase in the divergence angle can be observed as the angle $σγ$ increases. The divergence angle does not exhibit significant differences for varying carrier gas flow rates. As the angle increases from 0.42° to 2.17° (an increase of 423%), the divergence angle increases by at least 140% for all carrier gas flow rates.

The velocity of the carrier gas demonstrates a nearly linear increase with an increase in the carrier gas flow rate (see Fig. 14). Additionally, the carrier gas velocity exhibits a slight decrease with an increase in inner wall roughness.

Figure 15 depicts the ratio between particle velocity and carrier gas velocity. For higher values of angle $σγ$⁠, a slight decrease in the ratio can be seen across all carrier gas flow rates.

Figure 16 presents a scatter plot of the tracked particle velocities with their associated particle flight angles, derived from 222.2 ms of the high-speed videos. Each dot represents a tracked particle. The particle flight angles are distributed symmetrically around 0°, with the lowest particle velocities observed at approximately 2 m/s, where the particle flight angle also exhibited the greatest spread. As the particle velocity increases, the particle flight angle shifts toward 0°. The highest particle velocities are at approximately 7 m/s.

Figure 17 also represents a scatter plot for the particle velocities and their associated particle flight angles. In this measurement, compared to Fig. 16, the angle $σγ$ decreased and the carrier gas flow rate increased. A shift of the particle velocities to higher velocities is observed. The lowest particle velocities start at approximately 6 m/s and spread up to 14 m/s. Once again, a symmetrical distribution of particle flight angle around 0° is evident. The distribution of angles is less distinct in this instance.

In order to facilitate comparison, only the two sigma values for the particle flight angle are plotted in Fig. 18. It is evident that for both angles $σγ$ fast particles have a smaller flight angle. Increasing the carrier gas flow rate leads to a shift toward higher particle velocities, but does not change the distribution itself.

The particle velocity is influenced by more than one parameter. For the carrier gas flow rate, an almost linear correlation to the particle velocity can be seen (Fig. 9). In the case of a static powder channel inner diameter, an increase in carrier gas flow rate leads to a higher carrier gas velocity (Fig. 14), thus increasing the acceleration of powder particles. In addition to the carrier gas velocity, the angle $σγ$ also influences the velocity of the powder particles. A decrease in Ra (and thus the standard deviation of the surface roughness angle $σγ$⁠) leads to a small increase in particle velocity (Fig. 12). A reduction in the standard deviation of the angle at which particles rebound from the inner powder channel wall will lead to a more direct particle trajectory through the powder channel. Consequently, particles traversing smoother powder channels are likely to experience a smaller number of particle-wall collisions, resulting in a diminished loss of kinetic energy and an augmented period for the carrier gas to accelerate the particles in between collisions. This is supported by the decrease in the ratio between particle velocity and carrier gas velocity with increasing angle $σγ$ (Fig. 15).

The particle velocity is not influenced by the powder mass flow rate (Fig. 8). For the powder mass flow rates utilized in this study, particles contribute only a small amount to the total mass of the gas-particle stream. This limits the particle-particle interaction, thus minimizing the effect of powder mass flow rate on the particle velocity. A reduction in the angle $σγ$ from 2.16° to 0.42° results in a reduction in carrier gas consumption while maintaining the same particle velocity distribution.

The divergence angle of the powder stream is primarily influenced by the inner surface roughnesses of the powder channels. The divergence angle increases almost linearly with the increase in $σγ$ and is the same across different carrier gas flow rates (Fig. 13). This leads to the conclusion that the carrier gas flow rate does not have an influence on the divergence angle. The divergence angle is the result of the last collision of a particle inside the powder channel. As previously stated, a smaller angle $σγ$ increases the probability of a particle having a more direct trajectory.

Figures 16 and 17 demonstrate that particles exhibiting a high velocity relative to the other particles in the experiment are more likely to display a straight trajectory (flight angle ∼ 0°) with respect to the powder channel. It is postulated that the last collision for these particles occurred further upstream, thereby allowing the carrier gas to accelerate the particles to a greater extent. Conversely, it is postulated that particles exhibiting a high flight angle can only achieve this if a collision with the powder channel occurs just before exiting the powder channel. Figure 18 illustrates that a smaller angle $σγ$ leads to an overall smaller deviation in the particle flight angle. This phenomenon can be attributed to the assumption that particles within a powder channel with a low angle $σγ$ exhibit straighter trajectories, resulting in a greater proportion of particles exiting the powder channel with a flight angle approaching 0°.

The impact of the inner surface roughness of powder channels on the divergence angle of the powder stream and the velocity and flight angle of individual particles was investigated.

• The parameter of the standard deviation of the surface roughness angle $σγ$ provides a theoretical insight into the possible particle trajectories within a powder channel.

• The divergence angle of the powder stream is primarily determined by the angle $σγ$⁠. A reduction in the angle $σγ$ from 2.16° to 0.42° (81%) resulted in a mean reduction in divergence angle of 61%. For a specific desired powder spot size, this implies an increase in working distance, thereby enhancing the robustness of the powder nozzle against process emissions and back reflection.

• The particle velocity is influenced by the angle $σγ$⁠. With smoother inner powder channels if a specific particle velocity is required, a lower carrier gas flow rate can be used to achieve this. This results in a potential reduction in process costs.

The authors gratefully acknowledge the collaboration with the project partners Nutech GmbH, Schmelzmetall Deutschland GmbH, and Technische Universität Berlin regarding the support of knowledge, material, and equipment over the course of the research.

ZIM-Project No. KK5274710EW2 was funded by the Federal Ministry for Economic Affairs and Climate Action (BMWK) via the German Federation of Industrial Research Associations (AiF) in accordance with the policy to support the Central Innovations of Medium-Sized Enterprises (ZIM) based on a decision by the German Bundestag.

LUNOVU 835 LMD-Machine is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Project No. 434424600 (Highly flexible material synthesis and microstructure adjustment through combined laser deposition welding and short-term heat treatment for high-throughput materials development).

The authors have no conflicts to disclose.

Annika Bohlen: Data curation (equal); Investigation (equal); Methodology (equal); Writing – original draft (lead). Thomas Seefeld: Funding acquisition (equal); Project administration (equal); Writing – review & editing (lead).