Euro PM2019: Effects of humidity and storage conditions on Additive Manufacturing powder quality

Influence of humidity in Ti-6Al-4V powder during storage – Part 1

A paper from K Dietrich, T Arunprasad and P Foret, of Linde AG, Germany; O Messe and B Szost, of Oerlikon AM GmbH, Germany; A Schobert, of Airbus Central Research and Technology, Germany; and G Witt, of the University of Duisburg-Essen, Germany, investigated the potential degradation of powder flowability in powder-bed AM through exposure to humidity, specifically in relation to the alloy Ti-6Al-4V [1].

Reactive metal powders such as this alloy could be highly influenced by humidity through the resultant formation of oxides and consequent degradation of flowability. If this proposition holds true, it could lead to inhomogeneity in the final product. The authors’ study comprised two parts; in the first part (reported in this paper), the evolution of powder attributes as a function of storage duration was investigated, while, in a second part (reported in a separate paper in a later technical session at the congress), the effects of the moisture level on the build process and the resultant mechanical properties were considered.

The powder used in the study was Ti64 grade 23, supplied in containers and sealed under argon to limit the possibility of any degradation during transportation. The supplied powder characteristics are given in Table 1. The powder was recombined and subsequently blended to remove any possibility of batch to batch variation. Two-litre, wide-mouth, low-density polyethylene (LD-PE) bottles were used to store 3 kg of powder. The powder was sampled to fill thirty-six of these plastic containers. For every twelve, the remaining powder was blended to ensure homogeneous sampling through the entire process.

The containers were randomly assigned and stored at three different moisture contents (MC) (18%MC, 58%MC and 75%MC). The containers were stored as opened to the box’s atmosphere. The boxes were hermetically sealed and the humidity content and temperature were monitored using a humidity and temperature transmitter probe. The temperature remained stable at 22 ± 5°C throughout the experiment, whereas the humidity in each container was kept stable by using different salts such as potassium acetate (18%MC), sodium bromide (58%MC) and sodium chloride (75%MC).

One container from each of the humidity levels was taken from its box every three weeks to carry out an AM build job in a Trumpf Tru-Print 1000 equipped with Linde’s ADDvance^® O₂ precision to monitor oxygen and humidity content in the build chamber. The powder was sampled from the top (directly exposed to the humidity), after the powder was blended and after the build on the reclaimed powder from the baseplate. Subsequently, these conditions were referred to as ‘top’, ‘conditioned’ and ‘reclaimed’. Powders to be used for the build process were then stored in sealed aluminium containers to prevent any subsequent degradation.

Particle size distribution was measured for each of the powder storage conditions, to ensure that powder in each of the containers was identical and that no difference was measured between the top and conditioned states. The fact that both the top and conditioned states displayed the same powder size distribution, shown in Fig. 2 with D10, D50 and D90, illustrated that no granular convection occurred when the containers were placed in the storage box or during storage (Fig. 2a–b). This also demonstrated that, for the results of each of the tests performed, no normalisation was required for the moisture content. A noticeable shift in the PSD of the reclaimed powder was, however, observed. The measured moisture contents in the powder in both the top or conditioned states did not show any trend (Fig. 3a–b). However, visual inspection revealed loosely bound agglomerates on the surface developing as the storage duration increased (Fig. 3c). These were most visible in the powder stored under 75%MC and were not observed at all for the powder stored at 18%MC. The formation of these agglomerates was also accompanied by slight but distinguishable darkening of the powder (surface of containers only) stored at both 58 and 75%MC.

Oxygen and hydrogen contents were measured for all of the powders to evaluate the presence of oxide, water adsorbed at the particle surfaces or hydrogen seeping into the particle because of oxide formation. Both oxygen and hydrogen contents did not show any significant variation over storage time, suggesting that the powder sampled from the top did not markedly oxidise (below the detection limit) during storage, despite the formation of the loosely bound agglomerates and powder darkening for both 58 and 75%MC (Fig. 4).

Particle shape was monitored as part of the experiment. In this case, powder circularity has been plotted as a function of the storage time. The data were discretised so that only C10, C50 and C90 are shown. Both Aspect Ratio (not shown) and Circularity remained constant, or within their initial range, suggesting that the powder does not change or degrade during storage and is marginally affected by the build process.

Pycnometric density was measured for the conditioned powder to assess any change in the density that may be linked to a change in the powder. Table 2 shows the pycnometric densities measured for the conditioned powder. The results do not show any trend over storage time or for specific humidity levels. The results obtained are also similar to those for the virgin powder.

The combined data from PSD, Particle Morphology, Oxygen and Hydrogen suggest that the powder degradation, if occurring, is below the detection limit of these methods.

The results obtained are at odds with the visual changes observed from the top surface of the containers and suggest that any change in moisture, oxygen or hydrogen may be very local beyond the sampling and instruments’ resolution or detection limits. Also, these storage conditions do not affect powder attributes, which could affect, through varying powder rheology, the build process and bulk properties. The colour change observed is presumably attributed to small changes from the powder surface located at the top of the container. To test this hypothesis, additional sampling was carried out at the very top of the container for SEM investigation for all the powder which were stored for more than 150 days. These powders were sampled by gently pressing an aluminium stub covered with a carbon tape on the powder to obtain only particles located at the surface.

SEM micrographs showed very few differences between 18 and 75%MC. The overall powder morphologies for these powders were nearly identical. The lack of a relationship between the formation of agglomerates and any measured powder attributes during this study suggested that the Ti-6Al-4V powder was not affected by the storage condition.

It is possible that the observed colour change may be attributed to a TiO₂ oxide layer increase combined with impurities in the oxide. A few parts per million of certain metals (Cr, V, Cu, Fe, Nb, Al) can distort the crystal lattice, generating defects. The reported study demonstrated that, for L-PBF and titanium alloys, specifically Ti-6Al-4V, storage condition is not a process variable. Additional work would need to be carried out to evaluate the origin of the darkening observed at the powder surface.

The influence on additively manufactured part quality using AlSi10Mg powder aged in different humidity levels

Attention was next switched to the possible influence of humidity levels in aluminium alloys, specifically AlSi10Mg, in a presentation from Matthew Schultz-Sciberras of SLM Solutions Group AG, Germany. Aluminium materials, including powders, form a native oxide film on the surface once exposed to oxygen. This oxide layer is typically only a few nanometres thick and acts to passivate the particle against additional oxidation. While this holds true for aluminium-based powders in relatively dry air at ambient temperatures, it is not necessarily the case for powders in humid air at ambient or elevated temperatures. When the metal powder is exposed to water vapour, the water molecules adsorb onto the metal powder surface and become either chemi- or physisorbed. In the former case, water molecules attach to the oxide surface and chemically react to form metal hydroxides. [2]

The surface of the oxide layer is always hydroxylated to a degree under typical powder storage conditions. It is, however, not known what exposure limits to humidity exist for such metal powders before the surface layer changes enough to influence powder and built part quality. In addition, while reaction kinetics may be very slow at ambient temperature, they could become significant at elevated temperatures. This has implications not only for build jobs with some water vapour in the build chamber, but also for the various powder drying approaches using ovens in air.

The reported study aimed to show the changes in powder and part quality using AlSi10Mg powders exposed to humidity in their lifetime and then dried before use.

A single batch of gas atomised AlSi10Mg powder was used. The PSD of the batch of powder was measured by laser diffraction. For each experimental condition, 10 kg of AlSi10Mg was used. Powders were aged in their original containers under different conditions in a climate chamber. The conditions used to age the powders are presented in Fig. 5. One method to mitigate powder exposure to environmental humidity is to add a desiccant bag to the powder container; for this reason, the desiccant bag was removed prior to ageing experiments. The containers were placed open in the climate chamber for 168 hours (1 week) at the specified condition. After this period, typically the containers were removed from the climate chamber, a fresh desiccant bag was added and the containers were sealed tightly by hand. However, powders aged under the conditions 20°C, 70% RH, 20°C, 90% RH and 30°C, 90% RH were dried in a fan forced oven at 60°C for approximately 24 hours.

Powder flowability was measured using an SLM^® Flowmeter. This device follows a similar principle to the Hall Flow test, but has a modified orifice in the funnel. Using this device, the flow time and packing density could be determined. For each ageing condition, if the powder passed the flow test, then it was used in a single build job on an SLM^® 125 machine. The SLM^® 125 was equipped with a 700 W laser and the layer thickness used in the build was 60 μm.

Samples of the virgin AlSi10Mg powder were analysed to determine the PSD. The following D values were obtained: D10 = 24.84 μm; D50 = 43.68 μm; D90 = 72.66 μm; mean particle size = 46.65 μm. When conducting powder flow tests, the laboratory environment was 19°C with a relative humidity of 52.0%. The residual humidity in the powder container prior to the test was 4.0% at the same temperature. Flow tests indicated that 82.0 g of powder had good flowability of 47.0 s through the SLM^® Flowmeter. This amount of powder corresponds to an apparent density of 1.45g.cm³.

Following ageing experiments, all powders showed visible signs of exposure to the humid environment. For all conditions examined, except 30°C, 90% RH and 50°C, 90% RH, the static flowability and packing density data were between 40 to 50 sec and 1.42 to 1.43 g.cm³, respectively. Powder agglomerates formed during exposure to the condition 30°C, 90% RH were large and mainly confined to the top volume of powder, while the top 4 cm of powder from the condition 50°C, 90% RH had sintered together into a solid mass that was difficult to break apart. For these reasons, powders from these conditions were not used in build jobs.

Fig. 6 displays the results for oxygen analysis in the virgin and aged powders, after drying, as well as the oxygen in the spatter from the respective build job. The oxygen level trended upward from an initial value in the unaged virgin powder of 530 ppm to 1125 ppm for the powder aged at 50°C, 90% RH. The oxygen analysis of the spatter powder from each respective build job showed the rapid accumulation of oxygen in the build chamber for such particles. There was no observable increase in the oxygen accumulation due to the drying procedure in an oven at 60°C for about 24 hours.

The hydrogen analyses for aged virgin powders are displayed in Fig. 6. No significant increase was observed in the hydrogen content of the powders aged under the conditions 10°C, 50% RH, 20°C, 30% RH, 10°C, 70% RH and 20°C, 50% RH, when compared with the value for unaged virgin powder. All the following conditions showed a gradual increase in the hydrogen content of the sample from 20.3 ppm in the unaged virgin sample to 48.9 ppm in the sample aged at 30°C, 90% RH. Samples dried in an oven showed a greater increase in the hydrogen content compared with samples dried at room temperature with a desiccant. The sample aged at 50°C, 90% RH showed a rapid increase in the hydrogen content to 691.9 ppm.

The hydrogen analysis of the powder spatter after one build showed an almost constant average baseline of hydrogen content at 19.4 ppm across all samples. As the solubility of hydrogen in aluminium increases rapidly at the melting point of the metal, the lack of a hydrogen increase beyond levels seen in the unaged virgin powder indicated that the temperature of the ejected powder spatter did not exceed the alloy’s melting point.

Analysis of particle surface morphology by SEM revealed differences in surface features between unaged virgin powder and spatter after one build job from the unaged powder. The latter showed the formation of nodules approximately 50–100 nm in size, which are not present in the original virgin powder. These surface nodules are probably the result of oxide layer growth from exposure to heat near the laser scanning track during a build job. The hot powder spatter is ejected from the powder bed and reacts with the low residual oxygen and other reactive gasses in the inert gas stream. This observation fits with the increase in oxygen content of the unaged virgin powder versus its spatter after one build job (Fig. 6). The corresponding hydrogen analysis (Fig. 6) for this sample maintained the same level between the virgin and spatter particles, indicating that the oxide growth in the spatter is due to increases in the Al2O3 amount, probably through the reaction 4Al(s) + 3O2(g) = 2Al2O3(s). In contrast, the increase in oxygen content of the samples aged under different conditions was not observable here through changes in the surface morphology of the particles.

The density rods produced on the SLM^® 125 and analysed in the as-built state revealed some minor porosity. Parts produced using powder that was not aged resulted in a porosity of 0.34% in the specimens. This level dropped on average for specimens built using powder that was aged in the respective condition, up until the condition 20°C, 90% RH where the residual porosity peaked in the sample at 0.48%.

Fig. 7 shows the mechanical properties of parts built using powder from each condition. The yield strength and tensile strength values for all conditions tested were similar to reference values for AlSi10Mg, using the same machine and parameter set. The ductility of the tensile rods did vary between samples from about 6% to 8% elongation, but average values were either similar to or better than the reference value. It was considered likely that the gradual increase in oxygen content of the virgin powders through ageing did not exceed a critical limit. While these static mechanical properties did not show significant deviations from the reference values, this study did not investigate dynamic mechanical properties of the built parts. The oxidation kinetics under the tested conditions are not well understood and it is considered likely that the exposure time of one week for each ageing condition may be too short to observe significant oxidation.

The author concluded that further research should be conducted in order to determine the oxidation kinetics under different ageing schemes.

Study of acceptance criteria and good practices to follow during powder handling to limit hydrogen trapping in aluminium L-PBF

Finally, a paper from Olivier Rigo, David Reuter, Nathan Routiaux, Regine Van Den Berge, Pauline Tritiaux, Hanane Mekkaoui and Celia Parmentier, of Sirris, Belgium, continued the focus on the effects of humidity in the Laser Powder Bed Fusion (L-PBF) of aluminium alloys, considering the definition of acceptance criteria and good practice for powder handling to limit hydrogen trapping. The L-PBF of parts made from aluminium alloys, such as AlSi7Mg0.6, is extremely problematic due to porosity problems related to the tendency of such alloys to absorb hydrogen when powder with traces of moisture is used. The main source of hydrogen is the reduction of moisture by the oxidation reaction of aluminium [3].

Up to now, standardisation organisations and large companies active in Additive Manufacturing have established acceptance criteria for aluminium powders based on the usual method of measuring free moisture level. However, since moisture in the presence of aluminium powder gradually binds to form hydroxides, these methods are only valid to show the free moisture content of the powder at the time of analysis. This means that the result should depend on the initial exposure of the powder batch to a humid atmosphere, the condition of the powder at that time and the storage duration and conditions of the batch following exposure, but also to the storage time of the samples analysed. The goal of this reported study was therefore to establish acceptance criteria to obtain better process practice for aluminium powder handling.

A single aluminium-silicon alloy (AlSi7Mg06) powder was investigated in the study. A stainless steel type 316L powder was also studied, in order to compare the reactions of the two materials with moisture. The two powder samples were chosen to be as close as possible in terms of size range and morphology (Table 3).

All moisture measurements were carried out following the standard test method for Loss-On-Drying (LOD) by thermogravimetry by using an automated moisture balance. This method measures the humidity level of the powder by heating a specimen of known mass at a constant temperature, while its mass is continuously measured as a function of time. At the end of a pre-determined time interval, the specimen is recorded as a percentage of the original mass, this value being identified as the LOD value.

In the reported experiments, the 316L powder and AlSi7Mg06 powder were exposed to two different climatic atmosphere conditions for a range of durations. The first condition was fixed at a high humidity level of 95%, to verify the sensitivity of the LOD method and estimate the time needed to expose a powder to force a drift.

The moisture content of the AlSi7Mg06 powder reached 0.12%rH after one week of exposure (Fig. 8a). Particles deviated from the initial spherical shape and the bayerite surface layers started to form bridges between particles. The scatter of the data led to the conclusion that probably the bounding reaction of water with the aluminium alloy in hydroxide would have affected the measurement. To highlight this link between the measurement difficulties and specific characteristics of aluminium powder, the same approach was adopted with the 316L powder, which is not subject to hydroxide formation.

Fig. 8b shows that, in the case of 316L powder, there was a clear evolution of the powder moisture content during the first ten minutes of exposure. This led to the conclusion that, for a typical L-PBF powder, ten minutes of exposure to ambient air would be sufficient to reach the maximum moisture level.

To evaluate the effect of hydroxide formation on the moisture measurement, firstly AlSi7Mg0.6 powder was exposed to atmospheres with high humidity level (95%rH) and controlled room condition (44%rH). All samples were exposed for one day. Fig. 9a and Fig. 9b show the evolution of free humidity rate during different storage times in airtight containers. For both conditions, the storage duration had a direct impact on the LOD moisture content measurement. At the time that the samples were stored, the reaction of the passivated powder grain surface to bayerite was not yet complete. As a result, this reaction was completed in the container, continuously modifying the measurable free moisture content. Therefore, to have consistent measurements for all further analyses, a storage time was applied for each sample in the sealed container before proceeding with analyses. This step was referred to as a stabilisation step.
The next issue of interest was to evaluate the impact of the hydroxide layer on quality after melting. Four different AlSi7Mg0.6 powder batches were exposed to different controlled atmospheres (Table 4) with different moisture levels from extra dry to extremely wet. The exposure period was for twenty-four hours and this was followed by a stabilisation time of one week.

The evolution of Hausner ratio and apparent density was studied. Climate conditions, shown in Table 4, were chosen to represent process conditions linked to real powder handling in the L-PBF process chain (e.g. 44%/22°C/24 h), but also to consider extreme conditions (e.g. 95%/22°C/24 h). The sensitivity of the physically measurable variables to surface powder changes impacted by hydroxide formation was studied. It seems that the drift due to hydroxide layer formation is detectable by controlling the combined evolution of bulk density and Hausner ratio (Fig. 10a).

Fig. 10b shows the impact of the conditions experienced by the different powder batches on the mechanical properties after using the resulting powder in the L-PBF process. The same build configuration in the L-PBF machine was maintained, with horizontal tensile samples followed by a post-machining step. No impact of the first three conditions applied to the powder on the mechanical properties was noted. However, a clear increase in yield strength and an indication of a decrease in elongation at fracture were noted for the last condition, which corresponded to the longer exposure time in moisture. This result was not anticipated. However, the authors intend to continue the investigations; in particular, the analyses of the specimen fractographs, in order to understand the mechanism of this strengthening.

Author

Dr David Whittaker
Tel: +44 1902 338498
[email protected]

References

[1] Influence of Humidity in Ti-6Al-4V Powder during Storage – Part 1, K Dietrich et al. As presented at the Euro PM2019 Congress, Maastricht, the Netherlands, October 13-16, 2019, and published in the proceedings by the European Powder Metallurgy Association (EPMA).

[2] The Influence on Additively Manufactured Part Quality Using AlSi10Mg Powder Aged in Different Humidity Levels, Matthew Schulz-Sciberras. As presented at the Euro PM2019 Congress, Maastricht, the Netherlands, October 13-16, 2019, and published in the proceedings by the European Powder Metallurgy Association (EPMA).

[3] Study of Acceptance Criteria and Good Practices to Follow During Powder Handling to Limit Hydrogen Trapping in Aluminium Alloy During Laser Beam Melting Process, Olivier Rigo et al. As presented at the Euro PM2019 Congress, Maastricht, the Netherlands, October 13-16, 2019, and published in the proceedings by the European Powder Metallurgy Association (EPMA).

Proceedings

The full proceedings of the Euro PM2019 Congress are available to purchase from the European Powder Metallurgy Association. Topics covered include:

• Additive Manufacturing
• PM Structural Parts
• Hard Materials & Diamond Tools
• Hot Isostatic Pressing
• New Materials & Applications
• Powder Injection Moulding

www.epma.com