Generation of small batch high quality metal powder

Generation of small batch high quality metal powder Daniel Nils Ellendt 2 Lutz Mädler 2 Jörg Fischer- Peter Hofmann 3 Volker Schwenck Bühner 3 Uhlenwinkel schwenck@iwt.unibremen.de Ellendt@iwt.unibremen.de lm@iwt.unibremen.de jf@indutherm.de ph@indutherm.de uhl@iwt.unibremen.de : Universität Bremen 2: Stiftung Institut für Werkstofftechnik Badgasteiner Str. 3 Badgasteiner Str. 3 28359 Bremen 28359 Bremen Germany Germany 3: Indutherm Erwärmungsanlagen GmbH Brettener Str. 32 7545 Wössingen Germany For the development of a small batch metal atomization system, based on free fall atomization (ffa) and close-coupled atomization (cca), different powder quality features were defined to estimate the influence of the process parameters. Copper-tin alloys were used as feed stock material. Particle size distribution, appearance of satellite particles, particle circularity and the flowability were recorded as criteria of particle quality. Using data obtained in these experiments, main process parameters such as atomization pressure, mass melt flow, and height of the spray chamber as well as the atomizer system have been evaluated with respect to powder quality features. The atomization system was optimized to produce high quality powder with narrow particle distributions (d!",! d!",! =.6 for ffa) and high circularity with mass melt flows in the range of kg/h using melt volume between and ml. Experimental Setup The system used for the powder production is shown in Figure. The plant is built up of three main Vacuum vessel Stopper rod Nozzle system Spray chamber Filter Cyclone Fig. : Plant setup used for powder production Powder collection parts, the vacuum vessel including the inductive melting system, the spray chamber, and the nozzle system. The vacuum vessel contains a graphite crucible with a maximal melt volume of ml, and is closed by a stopper rod that is used to start the pouring when the melt has reached the desired conditions. In the conducted experiments a melt volume of 25 ml was used, which corresponds to 2 kg of material. The spray chamber is located underneath the melting chamber and is built up from cylindrical tube elements. The height of the cylindrical part of the spray chamber is about 4 m. A conical part is attached to the bottom of the plant to collect the powder. The atomization system is located at the interface between the two chambers. The system is modular so it is possible to use either a free fall- (atomization of a cylindrical melt stream below the melt nozzle) or a close coupled

atomization (lamella atomization close to the melt nozzle) system. The plant is equipped with sensors to measure different temperatures and pressures. The melt temperature and the crucible wall temperature are measured to characterize the state of the melt. The measured pressures are the overpressure acting on the melt surface, the overpressure within the spray chamber (for security reasons) and the atomization pressure. The alloy used for these experiments was a CuSn bronze with a liquidus temperature of 5 C. The standard superheat a for free fall and a close coupled experiment was K and 3 K, respectively. The atomization pressure used was between MPa and 2.5 MPa. Nitrogen was used for both, the free fall and the close coupled atomization. An overview of parameters is given in table. Tab. : Experimental Parameters Free Fall Atomizing Close Coupled Atomizing Gas Mass Flow 325-4 kg/h 245 kg/h Melt Mass Flow 8-3.5 kg/h 75 kg/h Number of nozzle Exits 2-46 - Gas nozzle Diameter. mm.7 mm Superheat K 3 K Melt Overpressure 3 mbar 3 mbar To characterize the produced powder, particles were measured with a Malvern Mastersizer 2 laser diffraction spectrometer (LDS) to determine the particle size distribution (i.e. mass distribution), and a Malvern G3 Morphology optical measurement system to measure the morphology of the produced powders i.e. circularity. Results In these experiments the main quality features of the powder that were assessed were the particle size distribution, particle morphology, and satellite particle generation of the process. The examined process parameters were the gas mass flow, the melt mass flow and the height of the spray chamber. Reproducibility To make sure that the process was only influenced by the desired parameter change, and not by any undesired outside effects, four experiments using free fall atomization with the same set of parameters were conducted. The gas mass flow was about 26 kg/h and the melt mass flow at 8 kg/h. The melt superheat was at K and the melt overpressure was about 3 mbar. These experiments were compared using their particle size distribution as measured by LDS. cumulative mass distribution Q 3 (x),9,8,7,6,5,4,3,2, d5,3 = 4,8 µm; d84,3/d5,3 =,63 d5,3 = 3 µm; d84,3/d5,3 =,59 d5,3 = 5,9 µm; d84,3/d5,3 =,6 d5,3 = 3,5 µm; d84,3/d5,3 =,63 particle diameter [µm] Fig. 2: Reproducibility of particle size distribution using free fall atomization In the figure above the cumulative mass distribution Q 3 (x) is plotted versus the particle diameter. The results are shown in terms of the size distribution. The mass median diameter d 5,3 varies between 3 µm and 6 µm. Since the LDS measures the volume/mass distribution of a powder, the mass median 2

diameter d 5,3 was chosen as the particle mean size. The ratio d!",! d!",! is between.59 and.63. These results show that a process with a given set of parameters is stable and reproducible. Particle Size Distribution Metal powder is used for different application such as selective laser melting (SLM) and metal injection molding (MIM) or in the area of Research and Development to examine the properties of new alloy. For different applications the produced powders must have certain properties. One important property is the particle size distribution. To characterize the size distribution of a powder the mass median diameter d 5,3 and the ratio d!",! d!",! was used. To adjust the size of the powder the gas mass flow and the melt mass flow was varied. First the influence of the gas mass flow was varied between 4 kg/h and 325 kg/h, the corresponding gas pressures were between MPa and 2.5 MPa. All experiments were conducted at a constant melt flow rate of about 8 kg/h, corresponding to a melt nozzle diameter of 2 mm. The results are shown in Fig. 3. Here, the mass median diameter d 5,3 is plotted versus the atomization pressure and the corresponding geometric standard deviation. mass median diameter d 5,3 [µm] 8 6 4 2 8 6 4 2 5 2 25 3 35 gas mass flow [kg/h] Fig. 3: Mass median diameter versus gas mass flow for the free fall atomization 2,9,8,7,6,5,4,3,2, geometric standard deviation d 84,3 /d 5,3 mass median diameter d 5,3 [µm] 8 6 4 2 8 6 4 2 2 4 6 8 melt mass flow [kg/h] Fig. 4: Mass median diameter versus melt mass flow for free fall atomization 2,9,8,7,6,5,4,3,2, geometric standard deviation d 84,3 /d 5,3 The results in Fig. 3 show a decrease of the d 5,3 with an increase of the gas mass flow. The measured mass median is d 5,3 between 98 µm and 63 µm. The corresponding d!",! d!",! varies between.6 and.7. To estimate the influence of the mass melt flow, experiments with different melt flows between 3.25 kg/h and 8 kg/h were conducted. The melt flow was controlled by using the melt nozzle diameter and by adjusting the overpressure. Nozzle diameters were in the range of.5 mm up to 2 mm. The overpressure was set to 3 mbar, and the gas mass flow was about 26 kg/h. Results in figure 4 show the mass median diameter d 5,3 versus the melt flow, and the corresponding geometric standard deviation. The trend of the d 5,3 shows no correlation between the melt mass flow and the median diameter. To compare the results of the different experiments, the gas to metal ratio (GMR) has been calculated, based on the experimental data. It is defined as: GMR = m G m L () With ṁ G as the gas mass flow and ṁ L as the melt mass flow. The GMR is a controlling parameter to adjust the mean particle size. A higher GMR represents more gas that usually leads to smaller particles (see eqn.: 2), colder spray conditions, and a faster cooling of the particles []. Figure 5 shows the d 5,3 versus the GMR for the conducted ffa and cca experiments. 3

mass median d 5,3 [µm] 8 6 4 2 8 6 4 2 Shifting of the trend of the d 5,3 by varying the melt mass flow FFA reproducibility FFA gas mass flow FFA melt mass flow FFA chamber height CCA 2 4 6 8 GMR Fig. 5: Mass median versus gas/metal ratio for comparison of the different atomization runs The values in black represent the ffa experiments, and the red symbols correspond to the cca runs. The ffa runs show that with increasing GMR the particle size decreases. The change of the GMR is only due to the change of the gas melt flow. This behavior is similar to the one shown in figure 3 and satisfies the empirical correlation by Lubanska [2] for the mean diameter given by: d 5,3 = kd N )( + GMR ) ϑ L ϑ G We (2) as mentioned in e.g. [3] and [4]. Where k is a constant determined experimentally and θ! and θ! are the kinematic viscosity of the melt and the gas, and d N as the melt nozzle diameter. We is defined as follows: We = v rel 2 ρ L d N γ L (3) Where ρ! is the density, γ! is the surface tension of the melt respectively and v!"# is the kinematic viscosity of the gas. At a GMR of around 3 the particle size does not decrease with increasing GMR by varying the GMR only due to a decreasing mass melt flow. The gas mass flow was at a constant 26 kg/h. This could either indicate a state of saturation for a certain combination of melt mass flow and gas mass flow or a shift of the GMR curve to lower or higher diameters as illustrated in fig. 5. Referring to eqn. and 2 it should be possible to adjust the particle size via the melt mass flow and the gas mass flow. The cca results for similar GMR s as the ones for the ffa showing much smaller particle sizes. This is due to the difference in the atomization process. Cca offers a higher kinetic energy of the gas for the atomization. More detailed investigations in the different atomization processes are to find in [5] and [6]. The actual cca values also already indicate a similar trend compared to the ffa Particle Circularity As an influencing parameter for particle morphology, the height of the spray chamber, the resulting cooling distance, was varied. The experiments were conducted with four different heights from 4 m (full chamber height) down to.6 m in steps of.5 m. The height was varied with a metal sheet plate as aperture to artificially shorten the chamber. All experiments were conducted with a gas melt flow of 262 kg/h and a melt mass flow of 8 kg/h. As a measure to evaluate the morphology, the circularity was used. The circularity is defined as: Ci = 4πA P! (4) Where A is the area and P is the perimeter of the particle. In Figure 6 the circularity of a particle is plotted versus the particle size for the different spray chamber heights. 4

circularity [-],95,9,85,8,75,7 FFA; h =.6 m; d5,3 = µm; d84,3/d5,3 =.66,65 FFA; h = 2. m; d5,3 = 8. µm; d84,3/d5,3 =.64,6 FFA; h = 2.6 m,; d5,3 =.2 µm; d84,3/d5,3 =.67,55 FFA; h = 3. m; d5,3 = 7 µm; d84,3/d5,3 =.64 FFA; h = 4 m; d5,3 = 5.9 µm; d84,3/d5,3 =.6,5 8 2 4 6 particle diameter d P [µm] Fig. 6: Circularity versus particle diameter for ffa 3 circularity [-],95,9,85,8,75,7,65,6,55 CCA; h = 2, m; d5,3 = 46.6 µm; d84,3/d5,3 =.62,5 2 3 4 5 6 7 8 particle diameter d P [µm] Fig. 7: Circularity versus particle diameter for cca powder mass flow [kg/h] 25 2 5 5 Fig.8: Flow rate versus spray chamber height,5 2 2,5 3 3,5 4 chamber height [m] The results in a particle size range between 8 µm and 6 µm are displayed because this is the typical size range achieved by the free fall system with the afore mentioned set of parameters. This also contains about 5% of the produced powder according to fig. 2. The trend shows a slow decrease of circularity for increasing particle sizes. Particles with a diameter around 8 µm show circularity between.95-.97 while particles with a diameter of 6 µm have a circularity around.9. A circularity of would be a perfect sphere. In fig. 7 the result of a cca run is shown with a particle size range from 2 µm up to 8 µm (about 7% of the produced powder). Here the former trend is visible again. With an increase of particle size, circularity decreases. For a particle with a diameter of 2 µm the circularity is about.95 while the circularity of a particle with a diameter of 8 µm drops down to.83. Similar observations have been made by Singh [7], who observed the effect of particle size distribution and flight distance on particle shape, and Freyberg [8]. So there is a higher probability for bigger to be still partly liquid or in some kind of a mixed state where they can be deformed when colliding with the bottom of the chamber. Since the cca powder is smaller and therefore faster cooling, it would be assumable that the circularity is similar to the ffa powder. But the comparison of the powders shows that the circularity is far below the ffa. This indicates not only the cooling distance is of importance but other parameters as well, for example the probability of impact between small and large particles. Comparing the different chamber heights for the ffa experiments shows no significant difference in morphology. Therefore it seems that a major fraction of the particles must already be solidified after a flight distance of.6 m this also corresponds with the observations made in [7]. Additionally G3 measurements (see fig.: 6 and 7) the powder mass flow rate of the produced powders was determined to double check the height independency. In Figure 8 the results for the mass flow measurements are plotted versus the chamber heights. The powder mass flow rates are in all cases around 2 kg/h, which indicates that the particle properties are similar in shape and satellite droplet occurrence. Additionally, the figures 9 and show digital pictures taken by the G3 of single particles for ffa and cca processes. The particles in fig. 9 indicate a high sphericity of the particles, but they also show a certain amount of satellite particles on the surface of the primary particles. According to the flow rate data from figure 8, the occurrence of the satellite particles should also be similar for all ffa powders since there is no difference in the flow rates. The particles in fig. show a high amount of satellite particles and deformation. Since these are particles from a cca run, the pictures correspond well with the measurements shown in fig. 7. 5

d! 4μm Fig. 9: Particle images for free fall atomization d! 2μm d! 8μm Fig. : Particle images for close coupled atomization d! 5μm Conclusion With the current configuration of the powder plant, it is possible to produce particles within in a narrow size distribution with a controllable mean particle size. As shown above, the particle size can be influenced by the gas flow rate until a certain combination between gas mass flow and melt mass flow is reached. The influence of the melt mass flow has to be further examined, since no change of particle diameter could be observed with decreasing mass melt flow. Although it is assumable that the particle size can be controlled with the melt mass flow since it is influencing the GMR. Using the melt mass flow to adjust particle size at lower gas mass flow levels could lead to a desired low gas consumption. The sphericity of the powder is influenced only by the particle size and showed no correlation to the spray chamber height in the experiments. For the cca system, the circularity is influenced by particle size as well and showed a lower circularity than the produced ffa powders. More investigations are planned to further investigate the influence of the spray chamber height on the circularity, and the GMR s influence on the particle size distribution. The system will also be equipped with a gas recirculation system to avoid occurrence of satellite droplets, since it is an undesired effect and has to be dealt with to increase the powder quality especially for the cca process. References. Achelis, L. and V. Uhlenwinkel, Characterisation of metal powders generated by a pressuregas-atomiser. Materials Science and Engineering: A, 28. 477( 2): p. 5-2. 2. Lubanska, H., Correlation of spray ring data for gas atomization of liquid metals. J. Met., 97. 22: p. 45. 3. Bauckhage, K., Das Zerstäuben metallischer Schmelzen. Chemie Ingenieur Technik, 992. 64(4): p. 322-332. 4. Zhou, Y., et al., Size distribution of spray atomised aluminium alloy powders produced during linear atomisation. Materials Science and Technology, 999. 5(2): p. 226-234. 5. Markus, S., U. Fritsching, and K. Bauckhage, Jet break up of liquid metal in twin fluid atomisation. Materials Science and Engineering: A, 22. 326(): p. 22-33. 6. Ünal, R., The influence of the pressure formation at the tip of the melt delivery tube on tin powder size and gas/melt ratio in gas atomization method. Journal of Materials Processing Technology, 26. 8( 3): p. 29-295. 7. Singh, D. and S. Dangwal, Effects of process parameters on surface morphology of metal powders produced by free fall gas atomization. Journal of Materials Science, 26. 4(2): p. 3853-386. 8. Freyberg, A., et al., Droplet solidification and gas-droplet thermal coupling in the atomization of a Cu-6Sn alloy. Metallurgical and Materials Transactions B, 23. 34(2): p. 243-253. 6