Euro PM2018: The influence of powder characteristics on processability in metal Additive Manufacturing

Framework development for the cleanliness assessment of metal powders for use in Additive Manufacturing

A contribution from Cameron Blackwell, Steven Hall, Jason Dawes and Nick Brierley (The Manufacturing Technology Centre (MTC), UK) described the development by the MTC of a framework for the cleanliness assessment of metal AM powders [1]. To date in AM, research has been largely focused on both improving the AM process itself (reducing production time and introducing in-process monitoring, for example) and on the post-processing of AM components and subsequent validation and testing of processed components. Until recently, less focus has been placed on the analysis of the material input, with little thought given to the assessment of cleanliness. At present, there are no established methods for the assessment of powder cleanliness and the task remains non-trivial.

As AM machines and auxiliary equipment move further towards closed loop systems throughout the component production chain, the ability to assess the cleanliness of metal powders becomes more pertinent. Chemical contamination of a powder batch can occur during all stages of its lifetime. There are common types of contamination that occur during specific stages and these are detailed in Table 1.

Whilst standards exist to analyse the cleanliness of components manufactured by PM processes, there are currently no standards to assess the cleanliness of metallic powders. Methods commonly used to assess powder cleanliness currently include optical microscopy, scanning electron microscopy and various chemical analyses.

Standard analysis of powder cleanliness can primarily focus on interstitial ingress into the powder chemistry. The oxygen, nitrogen, carbon, hydrogen and sulphur levels within a powder are commonly analysed as part of other testing for quality assurance. Testing is commonly performed using Inert Gas Fusion – Infrared and Thermal Conductivity Detection (IGF), alongside Induction Coupled Plasma – Optical Emission Spectroscopy or Mass Spectroscopy (ICP-OES or ICP-MS) for the analysis of other elements. However, as these techniques require prior selection of elements to be analysed, they cannot be used in isolation to determine cleanliness. Due to this, these analyses only provide definition of chemical variance from the nominal composition and, as such, do not provide information on foreign material contamination or information on the potential origin of the contamination.

Further to these chemical analyses, it is common to assess powder cleanliness via visual inspection or optical microscopy. This allows contamination to be categorised by its size, shape and colour. This analysis is often complemented by analysis using Energy Dispersive X-ray Spectroscopy (EDS) on an SEM. This provides analysis of size, shape and chemistry, key factors in determining the potential severity of the presence of a contaminating particle in a powder batch. A number of organisations have begun to use automated SEMs to not only identify contamination occurrences in metal powders, but also quantify the level (semi-quantitatively) of contaminating particles present in a powder batch. The use of an automated SEM allows a standardised cleanliness assessment method to be developed, which is operator-independent.

Sampling methodology is dictated by the total amount of the material to be represented and subsequently influences the sample size. Sample size is further defined by the analysis technique used to interrogate the material, as this presents a limit to the amount of material analysed due to issues of practicality and detection resolution. Due to the potentially catastrophic nature of a minute amount of contamination in a powder being used within PM applications, and the spatially localised nature of contamination, a limited sample size is not desirable when performing cleanliness assessments. However, considering the size range of potential contamination as detailed in Table 1, a high level of detection resolution is needed to capture all types of possible contamination. In order to circumvent these two conflicting concepts, metallic powder cleanliness assessment should use a variety of techniques in order to assess the full range of potential contamination occurrences, whilst still analysing a relevant and practical sample size.

Developed from knowledge gained from the study of many metallic powders for AM and considering the need for a larger sample size for analysis, the reported study has sought to understand the current capability of cleanliness assessment using large sample sizes. The MTC has proposed a workflow, detailed in Fig. 2, using optical microscopy (OM), scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM/EDS), dynamic image analysis (DIA) and X-ray Computed Tomography (XCT). Table 2 defines basic operating parameters for each of these analyses. This assessment should be performed alongside other powder characterisation analysis, such as size, shape, flow and chemistry analysis, using ICP-OES and IGF for instance, for overall chemical specification and performance validation.

To maximise the value of these cleanliness analyses, a comparison must be made between different analyses and samples using a global classification system. Using such a system allows a comparative level assessment to be made of each contamination type, providing a framework for future cleanliness and quality assurance assessments. The system that the MTC has proposed, as detailed in Table 3, classifies contamination based upon its chemistry and its morphology. It is important to make both levels of distinction as both give insights into the origin of the contamination. An example of this would be finding stainless steel foreign metallic contamination. If this were spherical, this raises the possibility of alloy cross-contamination occurring sometime within the powder lifecycle; however, if it were an irregular shard-like particle, then this may be due to degradation of the atomiser or AM machine components.
The reported study did not look to provide contamination severity levels, but rather to put forward a proposed classification system to allow the assessment of comparative levels of various contamination types. Significant further work is needed to compute severity levels, which should take into consideration contamination type and size, and which would vary in accordance with alloy and component end-use sector.

SEM/EDS is the key analysis technique for the system, although this is closely linked to optical microscopy, due to the sampling methodology used in the workflow. OM analysis allows the identification of contamination based on size, shape and colour. In order to capture all contamination within a large sample, the authors have suggested that the powder be placed in a wide tray and, using a vibratory and rotating sifting action, contamination will segregate and concentrate in one area given sufficient sifting time. The area surrounding the area of concentrated contamination is where OM analysis should be conducted, the concentrated area sitting centrally within it, and all contamination occurrences imaged and mapped. Once OM analysis has been conducted, the powder in the concentrated area should then be loaded onto an SEM stub for manual or automated analysis. Contamination should first be identified by either a contrast difference to the bulk material in the electron-backscatter image, or by a variation in morphology. Each contamination instance should then be imaged (to a practical extent for contamination classes VII and VIII (Table 3)) and chemical analysis should be performed via EDS.

Dynamic image analysis (DIA) is currently used to perform size and shape analysis of particulate material. Given its evaluation of shape, it was chosen as a detection technique for the assessment of contamination based on morphological differences from the bulk. DIA allows the assessment of a large powder sample (Table 2) in a fast and automated way. In order to do this in an automated fashion, there is a level of development needed. DIA machines record particulate shape characteristics using various measurands, e.g., sphericity, aspect ratio and convexity amongst others. A user can specify shape characteristic threshold limits, which can be used to automatically calculate the volume percentage of material between these limits. If a user develops threshold limits using size and shape characteristics unique to different contamination types, an automated detection and calculation of comparative contaminant level may be developed. It is important to note that standard two-camera based systems do not provide a user with an explicit number of contamination instances within a sample due to the counting algorithms used to correlate measurements between the two cameras. However, these data can still serve to provide comparative level assessments between samples.

To develop shape characteristic threshold limits, the MTC has undertaken work as detailed in Fig. 3, using three samples with different morphological properties to simulate highly elongated, highly irregular and irregular contamination occurrences, and was able to develop threshold limits to define highly elongated and highly irregular particles. Using the thresholds developed, the MTC was able to automatically semi-quantitatively assess the level of contamination classes I-VI. DIA is now used in conjunction with OM and SEM-EDS to corroborate findings to a larger scale sample size.

The use of X-ray Computed Tomography has been explored for cleanliness assessment of powder feedstocks as a complementary technique allowing larger sample characterisation compared with SEM. The MTC study was therefore to look at the use of polychromatic XCT with a large sample for the analysis of a known possible contamination within powder. A known example of possible contamination was that of tungsten particles found within Ti-6Al-4V. These contamination instances are a similar size to those of the bulk material and so an analogous tungsten alloy powder of size range < 75 μm was mixed into 40 g of Ti-6Al-4V powder in a split sample arrangement. Such samples were scanned in a set-up that produced a 22.5 µm voxel size. Contamination occurrences were identified by eye and, for each, a Region Of Interest (ROI) volume was selected manually. The volume was then extracted for size analysis. This manual analysis could be automated; however threshold levels need to be developed and so, for the reported study, manual analysis was used. The extracted volumes were then compared to the associated size fraction of the tungsten powder to assess applicability of the use of large scale samples in this system. The measured size of contaminant, in fact, correlated well with the sieved size fraction of powder and therefore this set-up shows promise for the identification of tungsten contamination in Ti-6Al-4V in a large sample sizes with regard to contaminant size analysis.

The authors’ overall conclusion was that the proposed contamination assessment framework should be developed in order to perform cleanliness assessments of metal powder. A holistic approach, using various measurement technologies, must be applied to perform a full assessment of metal powder cleanliness, due to the variety of contamination sources, measurement sample sizes and detection limits.

The demonstrated technologies that currently show promise for cleanliness assessment are optical microscopy, SEM-EDS, DIA and XCT. Cleanliness severity levels for each contamination type, alloy and component end-use sector must be developed within the wider scientific and industrial community.

Development of a reliable method for contamination detection in raw metal powders for AM

A second paper continued on the theme of contamination detection in raw metal powders for Powder Bed Fusion (PBF) AM. This paper came from an Italian academic consortium comprising Eleonora Santecchia, Paolo Mengucci and Gianni Barucca (Universita Politecnica delle Marche), Andrea Gatto, Elena Bassoli and Lucia Denti (Universita di Modena e Reggio Emilia) and Federica Bondoli (Universita di Parma) [2].

Currently, major technical limitations for metal AM relate to the lack of specific qualification standards for AM parts and feedstock materials. Raw powders for Additive Manufacturing are subjected to potential contamination through the full supply chain, from production to the storage and usage (AM) steps. In the reported study, the challenge of cross-contamination detection in feedstock powder materials was addressed. Various scenarios of contaminants and contamination sources during the production and sintering processes were taken into account and batches of two powders, having the typical compositions of Ti-6Al-4V (EOS Ti64) alloy and maraging steel (MS1) and containing a controlled cross-contamination, were prepared. The contamination was detected using SEM and EDS techniques and a statistical treatment of the collected data allowed quantification of the cross-contamination.

Controlled contamination was introduced to the powder samples following assumptions concerning damage to the PBF equipment (i.e. breaking of the recoater blade), cross-contamination of the powder taking place in the Additive Manufacturing equipment (i.e. the same PBF machine used for different powders) and cross-contamination during powder production or transportation (i.e. sieving equipment, tools or gloves used for different powders). The inspected scenarios are summarised in Table 4.

Scanning electron microscopy observations were performed on a field emission SEM equipped with microanalysis for the energy dispersive spectroscopy (EDS) inspections. Powders were accurately spread and attached on stubs for SEM; three stubs for each cross-contamination condition were characterised.
The chemical compositions of the pure and contaminated powders were checked by collecting three EDS spectra on areas at the same low magnification (200 x), using 20 keV accelerating voltage. Deconvolution of the elemental peaks was used in order to resolve peak overlaps and quantitative analysis was performed by the EDS software. In order to ease the detection of the contamination particles, the backscattered electrons (BSE) signal was used together with EDS elemental maps. The low atomic weight of Ti- and Al-based oxides, in the MS1_Oxi samples, resulted in a high BSE contrast and, therefore, contaminant particles in these samples were spotted using SEM-BSE micrographs only.

Quantification of contamination was performed by collecting fifty micrographs for each stub and the elemental map for each position, looking for the major contamination element, i.e., Ti in MS1_Ti64, Fe in Ti64_MS1 and Zr in Ti64_ZrO₂.

The scanning electron microscope working parameters were kept constant for all of the investigations and can be summarised as: (i) 60 μm aperture (beam spot), (ii) 500 x magnitude, (iii) 8.3 mm working distance and (iv) 15 keV accelerating voltage. The latter parameter was chosen in order to achieve an optimised balance between the EDS signal and the BSE contrast/brightness for SEM imaging. The contaminant particles spotted on each micrograph/map were counted and a statistical procedure was then applied, in order to define a calculated contamination. Firstly, the frequency of contaminant particles μ is calculated as in the equation:

• μ = Counted Contaminant Particles / Inspected Area

The total area of the stub is known to be 122.6 mm². The total contaminant particles (TCP) number on the overall stub area is therefore given by:

• TCP = (μ ∙ 122.6)

Therefore, the calculated contamination (CC) is obtained as the ratio between the contaminant particles and the total number of virgin powders particles on the stub:

• CC = TCP / TOT

The nominal compositions, as well as the EDS quantification results of the data acquired from MS1 and Ti64-based powders samples, are reported in Tables 5 and 6, respectively.

With reference to the MS1_Oxi samples, the average concentrations reported in Table 5 show no remarkable variations that could be linked to the TiO₂ and Al₂O₃ contaminations. On the other hand, the average Ti concentration in the MS1_Ti64 samples is higher compared with the nominal and MS1_Oxi values, suggesting higher concentrations of titanium.

The vanadium concentrations in Table 6 are below the nominal values in all the samples. Despite the same amount (wt.%) of contamination in the Ti64-based samples, no iron was detected or quantified (Table 6) while a clear signal for zirconium, corresponding to 0.3 wt.%, was obtained, as shown in Fig. 4. Here, two representative spectra, collected on areas of the Ti64-based contaminated samples, are reported in the form of de-convoluted peaks with the background already subtracted. While the peak corresponding to the Lα characteristic energy of zirconium is excited by the electron beam (Fig. 4(b)), no peaks related to iron are observed in the Ti64_MS1 spectrum (Fig. 4 (a)). A feasible explanation for this result can be achieved by accounting for the density of the contamination particles under consideration, i.e., 8.0-8.1 g/c m³ for MS1 and 5.8 g/cm³ for ZrO₂.

By collecting the elemental map of the major contamination element, namely that with the highest percentage, it is possible to accurately highlight the cross-contamination powder particles. During the first scan, the mapped element is shown on many points of the micrograph, but, when a very high concentration is detected in a certain area, the software uses this information to adjust the detected chemical element amount on the overall frame area.

For the MS1_Oxi samples, the high BSE-signal contrast, given by the very low atomic weight of the contaminant particles, was sufficient to distinguish them from the virgin powders, without the aid of elemental maps. However, the composition of each contaminant particle still needed to be verified using EDS point analysis. The EDS spectrum confirmed that the major elements in the contaminant particles were aluminium, titanium and oxygen. Other unindexed peaks corresponded to Fe, Ni and Mg.

Table 7 shows the average values of calculated contaminations obtained for all the inspected samples. The lowest level of average calculated cross-contamination is given by the samples having an unknown cross-contamination level, or rather MS1_Oxi. Results on specular samples, namely MS1_Ti64 and Ti64_MS1, are of particular interest. The average calculated contamination value of the MS1_Ti64 samples was almost two times higher than that of the Ti64_MS1 samples. Given that the density values of the two virgin powders correspond to 8.0-8.1 g/ cm³ for MS and 4.41 g/cm³ for Ti64, in order to obtain the same amount of 0.5 wt.% of cross-contamination, a lower number (approximately a half) of MS1 particles is required. This explanation also justifies the high level of calculated contamination obtained for the Ti64_ZrO₂ samples, as zirconia has a density equal to 5.81 g/cm³. This larger number of zirconia particles leads to a higher number of nuclei of characteristic X-ray signal generation. This is particularly remarkable for the EDS spectra collected from large areas, or rather those used for the chemical composition quantification (Fig. 4).

A systematic approach for understanding powder influence in powder bed-based AM

The next paper turned the attention to the influence of powder physical characteristics, as opposed to chemical contamination, on processability in powder bed-based Additive Manufacturing. This paper came from Silvia Vock, Solomon Jacobs, Burghardt Kloeden, Thomas Weissgarber and Bernd Kieback (Fraunhofer IFAM, Dresden, Germany) and Michael Haertel (AM Metals GmbH, Germany) [3].

The reported study introduced an approach for the systematic assessment of powder influence along the process chain. As a first step, a database was evaluated in order to identify suitable characterisation techniques and parameters for the reliable and sensitive detection of powder quality changes. In future applications, the continuously growing database would serve as a source for the predictive modelling of process and part properties, based on measured powder characteristics. It was anticipated that this would pave the way for efficient quality control and accelerate the development of process windows for new powder materials.

The authors proposed that powder can be characterised on different levels. On the one hand, the individual particles can be characterised by their morphology, size distribution, composition (main elements and impurities), moisture content on the particle surface and their individual particle density, for instance. On the other hand, physical properties of the powder describe the collective behaviour of the particle assembly, such as the packing of the particles (apparent density, tap density) and the mechanical behaviour of the particle assembly when it is forced to move (flowability). All of these characteristics contribute to a specific process behaviour represented, for instance, by the raking quality and the optimum processing parameters (e.g., the necessary volume energy) and finally will translate into the part properties.

In order to be able to track the powder influence along the complete process chain, a comprehensive database must contain all of the significant characteristics of each chain link. The necessary inputs for such a database are schematically shown in Fig. 5.

Also, powder reuse is not a straightforward issue. The powder quality after use can depend on factors such as build time, the heat evolved during the build and the part surface area in direct contact with the powder. Further, an established method for powder recovery includes sieving of powder from the build platform and remixing it with virgin powder, so that there is a sufficient amount of powder for the next build jobs. This procedure is not repeated after each build job, but only when the powder volume drops below a certain limit. It is advisable to allow the operator to classify the powders with regard to how heavily they were exposed during the process (heat, time, contact surface) and how much they are diluted with virgin powder. Ten so-called recycling levels (RL) are defined as classes, ranging from virgin powder (RL=0) to very heavily exposed and more than ten times reused powder (RL=10).

Humidity is an environmental condition, which can affect several parameters describing powder behaviour. Humidity can be measured in the ambient air, but also in the powder itself (e.g., by thermogravimetric methods).

Archiving all of the defined parameters in a single source repository enables the application of meaningful statistical evaluation tools, data mining and visualisation, and facilitates data sharing and comparison with project partners.

Fraunhofer IFAM characterised powders in the database, using both standard and dynamic methods. In relation to standard characterisation, particle size measurements were carried out using a laser diffraction device and apparent and tap densities were determined according to DIN ISO 3923/2 and DIN ISO 3953, respectively. The Hausner ratio HR is the ratio between the tap density TD and the apparent density AD: HR=TD/AD. An HR value close to unity means that the powder cannot be further consolidated and hints towards good flowability. The larger the value, the lower the powder’s flowability rating will be. The Hall and Gustavsson flowabilities were measured according to DIN ISO 4490 and DIN EN ISO 13517, respectively.

Dynamic flowability measurements were conducted using an FT4 Powder Rheometer from Freeman Technology. This device measures the rheological properties of powders in an analogous manner to fluid rheometry. Blades are rotated and moved downwards through the powder at a blade velocity specified in the test programme. During measurement, both the vertical force and the torque are detected and the total energy is derived. Based on this, the following parameters can be defined:

• Basic Flowability Energy BFE (mJ): the energy required to move the blades anti-clockwise downwards through the powder, high stress mode
• Specific Energy SE (mJ/g): the energy required to move the blades clockwise upwards through the powder, low stress mode
• Stability Index SI: measure of the energy change after repeated test procedures (down and up movement of the blade)
• Flow Rate Index FRI: measure of the energy change rate, when the blade velocity is varied
• Aerated Energy AE (mJ): flowability energy at a defined air velocity
• Aeration Ratio AR: factor by which the BFE is reduced by aeration at a defined air velocity
• Normalised Aeration Sensitivity NAS (s/mm): measure of sensitivity of the powder to the introduction of air
• Pressure Drop PD (mbar): pressure drop across the powder bed at a defined applied normal stress at a defined air velocity
• Compressibility CPS (%): percentage by which the bulk density has increased with an applied normal stress

To date, the database consists largely of powders for Laser Beam PBF (LB-PBF). The inclusion of the coarser EBM powders has now begun and will continue. The majority of powder types consist of Al-based alloys. Cu-, Ti-, and Fe-based alloys are also included. The powders are available in virgin form and in varying recycling levels.

With the current status of the database, it is possible to address the following queries:

• Which powder characterisation technique and which parameters are most suitable for identifying changes in powder quality?
• How are the measured parameters correlated with one another?
• Which powder characteristics are affected by recycling?

Fig. 6 shows a correlation matrix of the complete database based on the Bravais-Pearson correlation coefficient. The intensity of the colour represents the strength of the correlation, whereas the colouring itself denotes the type, i.e., the sign of the correlation. In the upper (left) part, the rheometer characteristics are shown. In the middle part, the characteristics, as determined by standard analytics, and, in the lower (right) part, the particle properties (D10,D50, D90, span) can be found. The last column represents the powder condition (recycling level).

Focusing on the correlation strength, the plot readily reveals the parameters which are most sensitive to powder quality changes, i.e., variations in D10, D50, D90 and span. In the case of the standard characteristics, strong correlation (correlation coefficient > 0.8) occurs only for the Hausner ratio. Bulk density and tap density show only weak correlation (correlation coefficient < 0.5). Hall flow and Gustavsson flow could only be measured for well-flowing powders; therefore, the available data do not allow for a calculation of the correlation coefficient. In the case of the powder rheometer, the FRI, AE, PD and CPS can be identified as parameters strongly correlated with the powder quality. The basic FT4 Rheometer parameters BFE and SE show only weak (BFE) or medium (SE) correlation with D10, D50, D90 and span. Interestingly, the sign of the correlation is opposite for these two parameters. A physical explanation for this could be that more cohesive and thus loosely packed powders are more easily moved during the downward movement of the rheometer blades. In the case of the SE measurement, the blades move upward and the energy needed for powder displacement is mainly influenced by the cohesive forces between the powder particles rather than their packing properties.

The correlation between measured parameters and the recycling level RL shows only weak to medium correlation (0.2–0.65) in all cases. This points either towards negligible effects for the given RLs in the range of 0-5 or to the fact that other parameters have to be included in order to reliably judge the recycling effect on powders, e.g. humidity in the powder.

As the volume of data in the database increases, further tasks will be addressed in future:

• Modelling the relationship between particle properties and powder behaviour (regression analysis) to allow for the prediction of particle size distributions from given rheometer data
• Pattern recognition: Are powder morphology variations reproduced by a combination of parameters? How are batch-to-batch variations condensed in the database?
• Including process and part characteristics to allow for a prediction of part properties based on process and powder properties

Static and dynamic characterisations of IN625 powder for Powder-Bed Fusion application

The final paper, from Kewei Li, Amy Nommeots-Nomm, Jose Muniz-Lerma and Mathieu Brochu (McGill University, Canada), also addressed static and dynamic characterisations, in this case for IN625 powder for Powder-Bed Fusion AM [4].

All powder-bed fusion processes rely upon the distribution of thin layers of powder feedstock with a thickness typically ranging between 20 and 60 μm. An ideal powder would easily spread across the powder bed, resulting in a uniformly distributed layer with a high particle packing density. Research has shown that the particle size distribution and inhomogeneities within the powder layer can influence the melt pool dynamics during processing and can result in defects in the parts being built.

To obtain thin powder layers with high packing density, there is a tendency to use as fine powders as possible. The utilisation of fine powders is beneficial to the printing resolution and the reduction of incomplete melting of previous layers. This can minimise thermal distortion and pore defects and thus result in products with low surface roughness and high dimensional accuracy. However, fine powders tend to agglomerate due to the occurrence of inter-particle forces such as Van der Waals attractive force, gravitational force, cohesive force and liquid bridges. The resulting effect of such inter-particle forces is the reduction of powder flowability due to the formation of large aggregates.

It is known that powder flowability decreases with reducing particle size. This is not only due to the agglomeration behaviour, but it is also postulated that both moisture and surface chemistry affect both static and dynamic behaviour.

A knowledge gap exists in the relationship between moisture and finer superalloy powders. Thus, quantitative studies and comprehensive theoretical insights into the particle surface chemistry, surface energy and surface roughness effects are still lacking.

To gain an insight into these issues, a newly developed semi-automatic experimental set-up was employed to understand the dynamic behaviour of IN625 superalloy powder. This tool mimics the powder flow during a powder raking step by subjecting the powder to dynamic flow via a rotating drum. In the current paper, special attention was paid to the investigation of the effects of moisture and surface chemistry on the flowability of IN625 nickel-based superalloy powder.

Argon atomised IN625 powder left in an AM system’s powder storage container for 90 days and named ‘as-received powder’ was used as the starting material in the current study. Its particle size distribution was measured by laser diffraction. The spherical particles had a size distribution between 22 μm (D10) and 45 μm (D90) with mean volume diameter around 30 μm (D50). Although most of the particles were spherical in geometry, a small content of non-spherical particles could be visually identified. The majority of these were caused by fine spherical particles attaching to the large spherical particles during production to form satellites.

Two batches of the as-received powder were dried at 200°C for 1 h in a vacuum (termed vacuum dried powder). One batch of the dried powder was then spread in a large pan with thickness of ~0.5 mm and left in air with humidity of 23% for 6 days (termed 23% RH powder). The other batch of dried powder was kept in a glass container settled in a 0% humidity incubator (termed 0% RH).

The dynamic flowability of the powder presented in this work was determined using the GranuDrum™ from the company Granutools. The testing device is composed of a horizontal aluminium cylinder of diameter D=84 mm and length L=20 mm with coated glass side walls. The drum is half-filled with 50 ml of powder. The cylinder rotates around its axis at a dictated angular velocity ranging from 2 to 20 RPM. To perform the measurements, a CCD camera was used to capture images of the surface of the powder at different times, as shown in Fig. 7. For each angular velocity, 50 images of the rotating drum are recorded at 0.5 second intervals. The positions of the air/powder interface are measured by inbuilt image processing software and analysed. The average air/powder interface position and the fluctuations around this average position are computed. With the fluctuations of the interface, the standard deviation σf can be calculated and this is denominated as cohesive index. The avalanche angle (αP) is the angle of a linear regression of the free powder surface as shown in Fig. 7. The cohesive index and avalanche angle are directly related to the flowability of the powder inside the drum. The flowability measurements were repeated three times and the average values are presented.

Fig. 8 shows the dynamic angle of repose (a) and cohesive index (b), obtained using the GranuDrum™ method, as a function of rotation speed. The results indicate that the dynamic angle of repose of the IN625 powders did not change significantly between powder storage conditions, whereas the cohesive index of the as-received powder was slightly higher than the other three samples, suggesting that drying the powder at 200°C for 1 h can improve the powder flowability by eliminating moisture-related cohesion formed between the particles themselves. Interestingly, the cohesive index remained unchanged even after being exposed in humid air for 6 days or in a sealed glass container for 2 months, suggesting that humidity-related cohesion effects in a 23% RH environment take longer than 6 days to develop.

To further understand the improvements seen in flowability with drying, XPS was conducted on the dried and as-received IN625 powder to determine if the surface chemistry was altered by the drying process; the results are summarised in Table 8. These results show the presence of nickel, chromium, niobium, iron, molybdenum, silicon and oxygen at the particle surface, suggesting that changes within the surface oxide layer occur with drying. This implies that drying the powders not only enhances flowability by the elimination of liquid bridges, but that it may also alter the oxide formation upon the particle surface. Oxygen binding energy analysis via XPS (Fig. 9) showed a shift in the relative ratios of hydroxide to metal oxide species with drying. However, further analysis to understand the localised metallic bonding is needed. Changes in surface oxide state could have potential downstream effects on laser absorption during the build process. Further analysis would need to be conducted to gain a full understanding of these relationships.

The overall conclusion was that the results of the investigation of the flowability of IN625 powder, using the newly developed rotating drum instrument, indicate that the flowability of the as-received IN625 powder can be improved by drying at 200°C for 1 h. Preliminary XPS analysis suggests that the surface elemental chemical composition changes with drying. Further work to understand the particle surface oxidation behaviour and topological features may be key to understanding the flowability of IN625 powders in this context.

References

[1] Framework development for the cleanliness assessment of metal powders for use in Additive Manufacturing, Cameron Blackwell, et al. As presented at the Euro PM2018 Congress, Bilbao, Spain, October 14-18, 2018, and published in the proceedings by the European Powder Metallurgy Association (EPMA).

[2] Development of a reliable method for contamination detection in raw metal powders for Additive Manufacturing, Eleonora Santecchia et al. As presented at the Euro PM2018 Congress, Bilbao, Spain, October 14-18, 2018, and published in the proceedings by the European Powder Metallurgy Association (EPMA).

[3] A systematic approach for understanding powder influence in powder bed based Additive Manufacturing, Silvia Vock et al. As presented at the Euro PM2018 Congress, Bilbao, Spain, October 14-18, 2018, and published in the proceedings by the European Powder Metallurgy Association (EPMA).

[4] Static and dynamic characterisations of IN625 powder for Powder-Bed Fusion application, K W Li et al. As presented at the Euro PM2018 Congress, Bilbao, Spain, October 14-18, 2018, and published in the proceedings by the European Powder Metallurgy Association (EPMA).

Authors and contacts

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

Cameron Blackwell
National Centre for Additive Manufacturing at the Manufacturing Technology Centre
[email protected]

Eleonora Santecchia
Universita Politecnica delle Marche
[email protected]

Silvia Vock
Fraunhofer Institute for Manufacturing Technology and Advanced Materials – IFAM
[email protected]

K W Li
Taiyuan University of Technology & McGill University
[email protected]

Euro PM2018 Proceedings

The full proceedings of the Euro PM2018 Congress are now available to purchase from the Europen 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

Euro PM2019

The Euro PM2019 Congress and Exhibition will be held in Maastricht, the Netherlands, from October 13-16, 2019.

www.europm2019.com