Feedstock powder processing research needs for additive manufacturing development

https://doi.org/10.1016/j.cossms.2018.01.002Get rights and content

Highlights

  • By completing these tasks researchers can promote rapid growth of AM.

  • Both rotating disk and close-coupled gas atomization can be improved to suit AM feedstocks.

  • The wide CC-GA industry base pushes it to the top for potential to accelerate AM.

Abstract

Additive manufacturing (AM) promises to redesign traditional manufacturing by enabling the ultimate in agility for rapid component design changes in commercial products and for fabricating complex integrated parts. By significantly increasing quality and yield of metallic alloy powders, the pace for design, development, and deployment of the most promising AM approaches can be greatly accelerated, resulting in rapid commercialization of these advanced manufacturing methods. By successful completion of a critical suite of processing research tasks that are intended to greatly enhance gas atomized powder quality and the precision and efficiency of powder production, researchers can help promote continued rapid growth of AM. Other powder-based or spray-based advanced manufacturing methods could also benefit from these research outcomes, promoting the next wave of sustainable manufacturing technologies for conventional and advanced materials.

Introduction

Additive manufacturing (AM) is an extremely active area of materials and manufacturing sciences research due to its promise to change the manufacturing game. AM can permit ultimate agility/customization for rapid component and system design changes in commercial products, as well as enabling component part consolidation and “impossible” composite materials or structures. For polymers and polymer-based composites with widely available low-cost feedstocks, the highly flexible AM platform is being quickly expanded and is now widely deployed with great success for both small and large-scale part builds. For example, big area additive manufacturing (BAAM) with polymers has been demonstrated for large structures, including cars and buildings. While the technical barriers for polymeric materials have been mostly overcome, additive manufacturing of metallic alloys remains challenging.

To be clear, this exploration of the status and prospects for improvement of metallic AM feedstock powders is targeted at two specific types of AM technologies, out of 7 types, that are being most widely developed for fabricating complex shapes of metals and alloys, namely powder bed fusion (PBF) and directed energy deposition (DED) [1]. A spherical powder shape is preferred for feedstock powders in all of these AM technologies to enhance flowability, layer spreading, and loose powder packing, particularly in powder bed types. DED processes are more tolerant of fragmented powder shapes, as long as the powder feeder employed can maintain a constant powder feed rate. Further, specific PBF processes include Selective Laser Sintering (SLS), Direct Laser Metal Sintering (DLMS), and Electron Beam Melting (EBM), all of which involve highly localized melting (typically) of powders and re-solidification of the “micro-weld” fusion zone. In fact, depending on the depth of penetration of the heat source and the scan pattern and speed, most volumetric regions of the AM build can experience multiple melting and re-solidification cycles during the AM process. Specific DED processes include Laser Engineered Net Shaping (LENS), Direct Metal Deposition (DM3D), Laser Deposition Technology (LDT), and Electron Beam Additive Manufacturing (EBAM). These “blown powder” methods utilize single or multiple powder feeders and either a laser (most commonly) or an e-beam as a highly localized heat source for melting a portion (typically 20–30%) of the injected powders to build up a free-form object [2].

A growing consensus within the AM community is that sufficient understanding of AM process fundamentals and process control is lacking to produce the desired microstructures and properties needed for robust metallic parts, particularly for operating in extreme environments or high stress and fatigue conditions [3], [4], [5], [6], [7]. There have been observations of degraded mechanical behavior in AM built test specimens that can be traced back to several types of microstructural defects that develop during the AM process. These processing/microstructure challenges that limit the properties of AM builds include reducing/eliminating residual stresses [3], [5], [7] and controlling as-built microstructure texture [8]. In some existing alloys designed for cast and wrought parts, AM processing results in cracking or other microstructure deficiencies due to inability to suppress unwanted inclusions/precipitates during processing [2], [4] and poor composition control from evaporation of some alloy components [5], [6], [9]. To overcome these challenges, it is desirable to develop a wider range of build parameters, e.g., solidification temperature gradient control and an increased pallet of alloy designs that are specific to AM processing. To accelerate verification of new alloy designs, the experimental alloys should be readily available in affordable small batches of high quality powder feedstocks for build trials.

While some defects that occur during a build are build parameter or alloy design related and can be minimized/healed by post-processing, e.g., hot isostatic pressing (HIP) and/or annealing, many defects related to porosity have their origin in the “quality” attributes of initial powder feedstock and cannot be healed by these methods. Limits on fatigue strength and fracture toughness due to voids in the build are probably the most important type of microstructural defect that must be avoided for wide acceptance of critical parts made by AM [3], [5], [9], [10]. Thus, it is typically total void volume, void size distribution, and void shape that are characterized in detailed studies of AM build samples in an attempt to recognize an optimum “minimum void” condition [9]. Powder quality related defects include internal porosity of large size (pore dia.  >10–90% of powder dia.) from trapped atomization gas [11] that is most prevalent in coarser size powders (dia. >70 µm), which are usually used for EBM/PBF and for LENS/DED, to some extent. It also should be noted that pores of very small size (pore dia. ≪5% of powder dia.) from trapped interdendritic solidification (“micro”) porosity are related to alloy “mushy” zone (liquid + solid) range. These also are more apparent in coarser powders due to slower solidification rate [12], but do not typically present a problem in build microstructures. Another type of problematic larger porosity in AM builds can result from powder that has attached “satellites” or projections [13] that prevent smooth flowability and impede uniform powder packing during spreading of successive powder layers. Surface impurities (e.g., adsorbed water vapor) also can promote powder agglomeration-induced spreadability deficiencies and porosity of larger size [14] in an AM build. These large pores may contain trapped hydrogen from decomposition of physi-sorbed water molecules or of chemi-sorbed hydroxides [15] during AM processing. Although this may be as-produced powder quality issue [15], it is typically due to inadequate atmosphere control during powder storage or handling [14].

It is important to describe the probable origin of two as-produced powder quality deficiencies (leading to larger, problematic build porosity) to add background about the challenges that are presented in this paper to the powder making community. To trace back the source of trapped atomization gas porosity defects, it is necessary to examine the droplet formation mechanisms that are active, particularly during gas atomization (GA). As described previously [11], [13], [16], [17] in any GA process there are many types of liquid breakup mechanisms occurring at any one time that can be ranked according to the energetics of the atomization gas interactions with the molten metal. Melt break-up into droplets also occurs in a dynamic sequence with droplet cooling and solidification. Break-up also can occur during a dramatic melt viscosity increase [11], [16], [18], especially in early solidification stages of mushy alloy fragments or droplets. When one of the most energetic mechanisms, “bag” break-up (see Fig. 1), is stimulated at high gas velocities, a melt fragment (or large droplet) becomes shaped into a bag-like sheet that spreads in a direction normal to the gas flow. The bag sheds small droplets from its periphery and may shatter into fine droplets. Alternatively, if the viscosity rises sufficiently, the sheet collapses on itself to form a large drop (hollow sphere) with a trapped pocket of atomization gas inside. Thus, it can be reasoned that to suppress the generation of hollow spheres, one should reduce the energetics of the breakup process to avoid operation of bag break-up, but this is difficult to achieve without precise control of the atomization process.

Identification of the source of attached “satellites” or projections on GA powders that prevent smooth flowability and packing appears to be complex, with at least two separate causes that have been proposed. One widely cited source [13] attributes satellite attachment to the (unavoidable) “rendezvous” of fine powders with coarser powders during their flight downstream in the spray of an atomizer. It was proposed that the fines would cool and solidify before larger droplets in the droplet size distribution of the atomized spray and would accelerate faster in this high velocity gas flow, eventually impinging/welding onto the larger (fully or partially) molten droplets. The other explanation considers spray chamber designs typical for industrial production of metal powders and offers alternative design options that can reduce satelliting. This alternative satellite attachment mechanism [19] relates to entrainment of “clouds” of fine solidified powders into the exterior of the atomization spray “cone,” where they weld themselves onto larger droplets. Anyone who has watched a gas atomizer in operation through a side viewport of a (typically large diameter) spray chamber is aware that a vertical upward flow of fine powders will always be visible along the chamber walls that continues to feed fines into this recirculating cloud. One way that was developed experimentally to suppress much of the fine powder cloud is by use of a smaller (30 cm dia.) diameter spray chamber and this was demonstrated to have a significant improvement in spherical smoothness of the resulting powder [19]. As with the suppression of internal porosity, there are considerable challenges to this method of satellite suppression, including the avoidance of “splatted” particulate from premature collision with the spray chamber wall, especially during atomization that produces a broader spray cone [20].

In addition to improved powder quality, current certified feedstock powders have excessive cost and limited availability due, in part, to the need for specific, narrow powder size distributions that are optimal for each AM process. Generally speaking, for laser beam melted/PBF processes, powders with a size range of +15 µm/−45 µm dia. are usually specified, for EBM/PBF processes, powders of +45 µm/−106 µm are specified, and for LENS/DED methods, powders of +45 µm/−75 µm dia. are used [3]. Actually, the size range for LENS/DED may be adjusted to finer powders and non-spherical powders may be used, as noted above [3]. Thus, for experimental alloy powders produced by common atomization methods (see Fig. 2), oversize and undersize powder size fractions of each batch [21], e.g., about 80–90% for free-fall gas atomization (GA), can limit the yield of AM powder and result in increased prices to cover costs from excessive inventory. As Fig. 2 shows, the selection of powder size distributions from “classic” spherical, aerospace-quality powder atomization methods includes three types of centrifugal atomization methods, with one type of (high rpm) spinning rod method (REP) that is fed from a (high cost) precisely machined solidified ingot and plasma melted at the tip [21]. Note that the “rotating electrode process” is now termed PREP (see Table 1) to recognize current use of a controllable plasma-melting source [22]. PREP now also has higher rpm capability and can produce a narrow particle size in the RSR range of Fig. 2. Two rotating disk methods of centrifugal atomization, RSC and RSR, appear in Fig. 2 with different rpm capability and are fed from a (versatile, low cost) molten alloy stream source. The current commercial version of the “rapid solidification rate” process (Ervin Industries of Ann Arbor, Michigan) has a much higher rpm capability that now can access the finer powder size range (relevant to AM) of the experimental “rapidly spinning cup” method [16]. There also are two types of spray nozzle methods represented in Fig. 2, one called vacuum (soluble gas) atomization (VA) that is not prominent now and one for (free-fall) gas atomization (GA) that is widely practiced with a typical broad size distribution, both starting with (low cost) molten alloy that is either forced upward through a gas atomization nozzle or poured down through a free-fall nozzle, respectively. The currently available gas atomization options also include electrode induction melting gas atomization (EIGA) that is practiced by numerous powder makers [22], especially for Ti. EIGA starts with a (low cost) roughly cylindrical alloy ingot that is slowly rotated within an induction coil to drip/drool melt through a free-fall gas atomization nozzle, again producing a broad size distribution (see Table 1) like GA in Fig. 2 [22]. The current AM powder feedstock marketplace also includes plasma methods for spherical powder production that start from a (high cost) wire feed or a fragmented particulate feed and produces a fairly narrow size distribution with a maximum size of about 200 µm, again relevant for AM, but without long range potential for reduced cost [22]. Although not included in either Fig. 2 or in the review (see Table 1) of modern powder making methods for AM [22], close-coupled [17] gas atomization (CC-GA) can stand with the current configuration of the RSR method as having great potential for achieving average particle size control [23] that can be extended to AM feedstock needs and for developing particle size distributions that rival the RSR method.

Although new alloy designs for AM feedstock powder have been mentioned, the primary focus of this paper will be on research to improve powder production efficiency (yield) and powder quality for AM feedstocks, which, in turn, can enable much broader testing and application of alloy design work for AM. As mentioned above, there is a potential for development of high powder yields for both the current RSR method and the CC-GA method in the size ranges of the targeted (widely practiced) AM methods of this review, namely PBF processes, which include SLS, DLMS, and EBM, and DED processes that include LENS, DM3D, LDT, and EBAM, since RSR and CC-GA both are high throughput, mass production methods for spherical powders. Moreover, selection of these atomization methods for further development also is justified by their low production cost with versatile alloy melting methods, accepting any form of cast alloy, pure metal, and recycled over-size or recycled AM-processed powder. Thus, PREP (especially) [2], EIGA, and the plasma methods should not be preferred for a greatly expanded AM industry [22], since all require a high degree of charge preparation for starting material, although they all produce smooth spherical powders with low internal porosity content and (essentially) no satellite projections. Further, when viewed with the criteria of an existing broad base of industrial practitioners to permit greatest impact of fundamental processing developments, the CC-GA method [13], [16], [19] rises to the top for potential to accelerate AM as the advanced manufacturing method of choice for a wide spectrum of applications. Therefore, this review will describe several process research challenges (some mentioned above) that must be overcome for CC-GA to meet the needs for efficient production of experimental powder batches, providing the critical experimental control needed to develop both optimal alloy designs and build parameters. Eventually, broad adoption of the process research results can lead to full-scale commercialization of high quality gas atomized feedstock powders at reduced cost for robust and cost competitive AM parts [24].

Section snippets

Current status of feedstock powder

In general, the highly localized melting and re-solidification that occurs during the AM processing of metallic alloy powders in the most popular PBF and DED processes is sensitive to defect incorporation (primarily porosity) from non-uniform powder flowability. These problems are due to powder agglomeration or off-spherical shapes that lead to non-uniform packing or feeding, along with internal porosity (trapped in the powders) that can persist even after multiple re-melt cycles [2], [3]. Such

Approach for improvement

One challenge for industrial powder makers desiring to serve the AM market is that there are different optimal powder size distributions for each AM process. This is especially troublesome if the requested powders are of experimental alloys for which no other market exists and oversize/undersize powders might sit in inventory for years. Also, considerable labor is involved in extensive screening of the full powder yield just to capture only a small portion (10–20%) of salable AM powders from a

Recent results

The results of recent work in our laboratory [19] that has an approach modeled after the above description has provided initial confirmation of the effectiveness of some of the selected experimental directions. It can be stated that suppression of retained porosity in an AM build is a major challenge for many techniques and that post-build hot iso-static pressing (HIP) has become a common practice for the closure of some types of porosity.

Unfortunately, HIP is not effective for one typical type

Discussion

As described in the Introduction, current AM technologies for metals lack adequate fundamental process knowledge and control that leads to deficiencies in repeatability for fabrication of products with consistent desired properties and structural features. Moreover, the AM community needs to conduct fundamental investigations of standardized material testing techniques for different forms of feedstock to better understand the impact on final AM part quality of: (1) feedstock powder size range

Conclusions

Much fundamental process research needs to be done on gas atomization to convert this high throughput method for production of metal powders into a precision process for efficient generation of high quality powders that target the needs for AM feedstock powders. This truly “smart” powder manufacturing will require fundamental process research advances to improve the state of the art of close-coupled gas atomization nozzles and atomization spray chamber designs. The process research goals should

Acknowledgements

The authors appreciate the detailed experimental preparations and performance of the gas atomization trial that was reported by Ross Anderson, Dave Byrd, Trevor Riedemann and Jordan Tiarks. We also are grateful for the microstructural analysis results of Tim Prost and extended discussions with Bill Peter, Mike Kirka, and Todd Palmer. Work performed with support from the USDOE-EERE-Advanced Manufacturing Office through Ames Laboratory contract no. DE-AC02-07CH11358.

References (24)

  • M.M. Kirka et al.

    Strategy for texture management in metals additive manufacturing

    JOM

    (2017)
  • R. Cunningham et al.

    Synchrotron-based X-ray microtomography characterization of the effect of processing variables on porosity formation in laser power-bed additive manufacturing of Ti-6Al-4V

    JOM

    (2017)
  • Cited by (172)

    View all citing articles on Scopus
    View full text