Additive Manufacturing of Refractory Metals for Aerospace

High-temperature refractory metals are required for many high-temperature propulsion applications. Refractory metals are expensive and difficult to manufacture, with high buy-to-fly ratios and few vendors. Additive manufacturing (AM) is used to produce C103, Molybdenum (Mo), and Tungsten (W) reaction chamber and thrust stand-off, as well as Iridium ultra-fine lattice catalysts for integration into 1 N green propulsion thrusters. Refractory AM is in development and, like traditional AM alloys, requires substantial post-processing to include powder heat treatment, surface finish enhancement, inspection, and machining before being placed in service. The combination of limited feedstock sources, high-temperature processing, oxygen sensitivity, fracture-prone nature, and the need for elevated temperature mechanical testing limit the number of qualified facilities capable of post-processing AM refractory materials, which adds to cost and schedule constraints. However, adequately implemented refractory metal AM can overcome existing manufacturing limitations by significantly increasing design flexibility, new material options, reduced price, decreased lead time, and leveraging the ever-growing AM commercial supply base.

Nomenclature

AM	=	additive manufacturing
DED	=	directed energy deposition
EB-PBF	=	electron beam powder bed fusion
EDM	=	electro discharge machining
HfC	=	hafnium carbide
HIP	=	hot isostatic press
ICME	=	integrated computational materials engineering
L-PBF	=	laser powder bed fusion
LP-DED	=	laser powder-directed energy deposition
Mo	=	molybdenum
MSFC	=	Marshall Space Flight Center
Nb	=	niobium
NTP	=	nuclear thermal propulsion
PSD	=	particle size distribution
RCS	=	reaction control system
Re	=	rhenium
RHEA	=	refractory high entropy alloys
Sa	=	areal average surface roughness
SEM	=	scanning electron microscopy
t	=	layer thickness
Ta	=	tantalum
Tm	=	melt temperature
UTS	=	ultimate tensile strength
W	=	tungsten
WT	=	wall thickness
Z	=	build direction
ZrC	=	zirconium carbide
ρrel	=	relative density
µ-CT	=	x-ray micro-focus computer tomography
µm	=	micrometer
%RD	=	percent relative density
%wt	=	weight percent

Introduction

Refractory metals and alloys are used for service in extremely high-temperature environments. Examples of such applications in aerospace include reaction control system (RCS) thrusters, nuclear thermal propulsion (NTP) fuel-clad, hypersonic wing leading edges, hypergolic and green propulsion chambers and catalysts, power conversion systems, and electric propulsion. Refractory metals are desirable due to a high melt temperature (Tm) and the ability to retain strength and hardness at elevated operating temperatures. In power and propulsion, refractory metal parts tend to be primarily thin-walled geometries, such as converging-diverging nozzles, as shown in Figure 1.

Traditional refractory metal manufacture is expensive due to the high cost of the feedstock and the difficulty in achieving near-net shape forming and machining, heat treatment, joining, and inspection. Since refractory metals tend to be primarily thin-walled designs such as converging-diverging geometries or clad, part production routinely results in 95-98% of the stock machined away, constituting a 20:1 to 50:1 buy-to-fly ratio. Assuming a 20:1 buy-to-fly ratio, the cost of a part in feedstock alone is 5% and 95% in machining waste, not including machining time or machining waste disposal, which is proportional to machining mass. In addition, refractory metals are relatively difficult to machine, resulting in a limited number of machine shops with the requisite equipment and experience to meet exacting aerospace specifications. Ultimately, a machined refractory metal part can cost several thousand dollars, but the waste can be several tens of thousands of dollars.

NASA Marshall Space Flight Center (MSFC) has been developing additive manufacturing (AM) for propulsion systems to produce complex and optimized components from an array of materials with numerous advantages over traditional manufacturing. AM refractory investigations have found that L-PBF AM of refractory metals and alloys offer significant cost and schedule savings, as well as engineering advantages. A previous development effort of laser powder bed fusion (L-PBF) AM of C103 (Nb-10Hf-1Ti) found that the cost to generate the same part via AM as machined was significantly reduced even when taking into account powder feedstock (33% more expensive than wrought feedstock), print time, heat treatment, final machining, and waste disposal.

Unlike machining, AM allows for the desired near-net shape to be produced with minimal machining. AM waste is typically 5-10% of the printed part mass resulting from oversized powder, support structures, and sacrificial geometric features at interfaces to be machined away to meet surface finish requirements. Consequently, the AM buy-to-fly ratio is approximately 1.1:1. Most AM powder is reused for additional builds as long as PSD, morphology, and chemistry remain within feedstock specifications as with any other recycled AM powder feedstock. Therefore, like a machined part, an AM part will also cost several thousand dollars. However, unlike traditional manufacturing, the waste cost constitutes a few hundred dollars, not tens of thousands of dollars. The difference in feedstock costs alone to arrive at an equivalent part constitutes an order-of-magnitude cost reduction using AM. The cost savings, reproducibility, schedule control, and, in certain cases, properties of AM parts have significant aerospace industry implications for the broad implementation of an alloy that may have been previously avoided due to the aforementioned constraints. In addition, the number of commercially available AM vendors that are sufficiently qualified to print refractory metal parts increases annually. C103, tungsten (W), Molybdenum (Mo), Tantalum (Ta), and Rhenium (Re) are examples of refractory metals that are already utilized or in development using L-PBF, electron beam PBF (EB-PBF), electron beam wire DED (EW-DED), and laser powder DED (LP-DED) AM. However, there is wide variability in the maturity of these refractory metal options, with highly weldable C103 experiencing rapid development compared with fracture-prone W and Mo. Examples of small refractory L-PBF AM parts produced by NASA are shown in Figure 2.

One of the primary limitations of refractory AM is the limited number of refractory feedstock vendors that offer powder or wire that meet AM specifications, as well as the limited number of alloy formulations of engineering utility for AM. Refractory components have traditionally been produced via powder metallurgy, plasma deposition, and electro-deposition, with final geometry obtained with electro-discharge machining (EDM). Angular powder for each element was often mixed at the requisite weight percent (wt%), then consolidated through the hot isostatic press (HIP) or deposited, forged, and then heat treated to arrive at the desired homogenized microstructure.

Although mixed powder can be used for AM lab-scale feasibility studies, it is not well suited for industrial production applications where the desired elemental distribution, PSD, and morphology require consistency. Mixed powders can also segregate during sieving due to the differences in mass of the elements and powder particle size, requiring an additional mixing step before each reuse. Due to the high melting temperature of refractory metals, inert gas atomization (the most common form of AM powder production) cannot readily process most of these metals. Most refractory metal powders are produced through plasma spheronization of angular powder, plasma wire atomization, rotating electrode process, and plasma or induction atomization of the ingot. Additional feedstock methods are in development but are not commercially available at this time. Commercially available elemental AM grade powder options include W, Mo, Ta, and Nb, while available alloy powders include C103, Ta3W, T5W, and Ta13W, with FS85 (Nb-28Ta-10W-1Zr) in development.

Another major difficulty of refractory metal AM is the inherent fracture-prone nature of many of these materials during melt and solidification. A combination of the high thermal gradient that results in high residual stress across the part and low ductility results in micro-cracking. Low ductility is a well-known attribute of refractory metals, and many refractory alloys of interest were developed to increase ductility and enable traditional manufacturing. For example, Re is added to W and Mo (such as W-25wt%Re, Mo-44wt%Re, and Mo-47.5wt%Re) to provide sufficient ductility to allow for post-near-net shape machining. Re is advantageous due to a high Tm and increased ductility and enables the alloy to retain mechanical properties at elevated temperatures, as Re does not suffer from grain growth as drastically as the other refractory metals. Unfortunately, Re is extremely expensive, powder feedstock is mixed (non-alloyed), powder is angular, and the number of commercial suppliers is extremely limited. Pre-alloyed powder with spherical morphology and near fully dense particle density is desired for powder-based AM methods, which have become the AM industry standard. For these reasons, contemporary AM methods and materials require development to meet the need for high-temperature power and propulsion applications.

Refractory metal components can also suffer from changes in mechanical properties during operation at elevated temperatures as grain growth occurs with continued exposure to operating conditions. Materials that are strengthened by fine grain microstructure generally become ineffective at elevated temperatures as the grain growth occurs over cumulative operational service. In addition, there is potential for solid solution-strengthened alloys to suffer from elemental diffusion that can result in a non-uniform elemental distribution that impacts material chemistry (e.g., micro-segregation). To address the limitations of the existing refractory metal and alloy inventory, efforts are underway to develop new refractory metal formulations designed and optimized specifically for AM processes.

These AM-optimized refractory metal formulations are in development to address improved printability, sustained properties at elevated temperatures, cost, and availability to meet increased demand primarily from the aerospace, defense, medical, and energy communities.

Methodology

This article aims to conduct a technology gap review of the current state of refractory metal AM and to provide an example of developing a maturation path to meet needs. Efforts emphasize refractory metal formulation optimization for AM, evaluating potential feedstock production options, AM parameter development, heat treatment options, mechanical and microstructural characterization, inspection, and component-level tests. Three phases of the refractory metal development process were identified and are currently being utilized.

The first phase is to use AM with existing refractory metal powder feedstock such as W, Mo, Ta, and C103. High Tm materials combined with low ductility are inherently fracture-prone, and micro-cracking is often observed due to the high thermal gradient from melt to solidification, which results in thermally induced residual stress. The impact on heat treatment and surface enhancement/coatings is evaluated to determine if the micro-cracks can be overcome.

The second development phase utilizes additions of dispersoids (generally ceramic nano-powder). Dispersoids act as heterogeneous nucleation sites in the melt pool during solidification and induce the formation of refined equiaxed grains, decreasing residual stresses and associated cracking. The AM refined grain structure is retained at elevated temperature due to grain boundary pining of the metal matrix by nano-scale ceramic particulates, known as dispersoid strengthening. Dispersoids must have a chemical affinity for the metal matrix and be thermodynamically stable above metal solidification temperature, limiting options to carbide-based ceramics when applied to refractory metals. Oxide-based ceramic dispersoids are heavily utilized in lower-temperature Al-base and Ni-base alloys but will experience melting due to the high temperature of the melt pool in L-PBF and DED AM of refractory metals. For example, Tm for W is 3410 °C, Mo is 2610 °C, and C103 is 2349 °C. Carbide dispersoid candidates with the appropriately high Tm d include HfC at 3900 °C and ZrC at 3420 °C. Although shown to be advantageous in stabilizing Mo in EB-PBF, HfC is much more expensive than ZrC, and HfC has a high thermal neutron absorption cross-section, which is undesirable in nuclear applications. For this reason, ZrC is often considered the preferred dispersoid for refractory AM purposes. Dispersion-strengthened refractory alloys have improved printability and mechanical properties at elevated temperatures. One previous study of W-0.5wt%ZrC via L-PBF AM showed a significant decrease in micro-cracking of up to 88.7%, improved printability, and improved retention of mechanical properties after exposure to elevated temperatures. Clearly, dispersoid additions in AM can improve printability and mechanical properties but still suffer from constraints related to the parent metal characteristics.

The third phase relates to a new family of thermodynamically stable refractory alloys explicitly designed for the AM process. Cost and AM-optimized solid-solution alloy development takes the base element of interest and determines which alloying elements can be used to reduce or eliminate micro-cracking while providing sufficient strength for post-process machining. Although this is a relatively new field, feasibility studies using mixed powder for L-PBF have been conducted to produce W-5Nb and W-7Ni-3Fe solid solution alloys. Results show improved printability over elemental W while using cost-effective secondary elemental additions and show promise for additional development. Refractory high entropy alloys (RHEA) offer the potential for improved mechanical properties at elevated service temperatures and are under development by the AM community. The design of RHEAs is a computationally intensive process due to the very large number of possible alloy compositions. Therefore, an integrated computational materials engineering (ICME) approach that integrates melt-solidification mechanics simulations and machine learning is necessary. These simulations are intended to predict potential solid solution RHEA candidates optimized for a specific AM process while avoiding intermetallic phases, unfavorable ductile-to-brittle transition temperatures, micro-cracking, etc. The alloys must be weldable (printable) and not necessarily need properties to survive traditional manufacture, which means that significant reductions in Re content (single digit wt%) or eliminated entirely to reduce cost is highly desirable. This nascent approach is expected to dominate new alloy development across all AM materials and methods. An example of W L-PBF AM is now discussed to address the previously mentioned constraints that exist throughout the entire logistical supply chain, from powder availability to part inspection.

Powder Characterization and Optimization

An example development effort is discussed for L-PBF AM of W. W powder obtained from EOS North America for use in the MSFC EOS M100 L-PBF AM platform. The W powder was found to have angular powder morphology, near-full dense particles, with particle size distribution (PSD) of from 36 to 15 µm with a d90 of 25.9 µm and d50 of 16 µm measured using a Retsch Camsizer XT. The W powder was used to print a series of W specimens, which served well for the intended application. However, in efforts to improve flowability and increase the part density, the powder underwent spheronization using the MSFC Tekna Tek15 radio frequency plasma spherometer, as shown in Figure 3.

The advantages of the spheronization process are that it can process any high-temperature material and a limited number of process parameters variables such as process gas composition and powder flow rate. Figure 4 compares the as-received W powder and RF spheroidized powder. The powder morphology is improved considerably, measured using a Malvern Morphologi G3 optical analyzer at MSFC, and found to have a median circularity of 0.97, which indicates a highly spherical powder.

Subsequent L-PBF AM trials were also conducted using W and Mo powder obtained from Tekna with spherical morphology, near-full density, and a PSD from 45 to 15 µm with a d10 of 18 µm. As expected, spherical powder improved flowability and uniform powder distribution across the build plate.

L-PBF AM Parameter Development of W and Mo

W specimens were produced on the NASA MSFC EOS M100 L-PBF AM platform. Metallographic, heat treatment, mechanical property, surface finish, and gas leak test specimens were produced using either 304 stainless steel or CU110 commercially pure copper build plates. Mo parameter development was halted soon after initiation due to the COVID-19 pandemic protocols enacted at MSFC. Therefore, EOS North America was contracted to conduct Mo parameter development for the EOS M100 and was still in work at the time of publication. An example of W metallographic and mechanical test specimens produced by MSFC is shown in Figure 5. The angular W powder, spheroidized W powder, and commercially available spherical W powder were used for development prints. Improved powder flowability resulted in a more uniform powder distribution across the build plate when using spherical powder and improved printability; however, it does not necessarily equate to improved microstructure.

Heat Treatments

W specimens underwent a series of heat treatments at MSFC, including stress relief (SR), hot isostatic press (HIP), and recrystallization. SR was conducted from 1100-1200 °C in a 10-4 Torr vacuum for 1 hour. HIP was conducted at 1700-1800 °C from 172-193 MPa for 1-4 hours. Recrystallization occurred from 1250-1350 °C in a 10-4 Torr vacuum for 1 hour. It must be mentioned that heat treatment of refractory metals requires furnaces to hold a high degree of cleanliness, treated in vacuum or ultra-high purity argon, often wrapped in a foil with high oxygen affinity such as Ta. The resulting structure-property relations for AM W specimens are now discussed.

Microstructural Characterization

As-built and heat-treated W specimens had Relative density (ρrel) determined using Archimedes’ method or helium pycnometer. Density ranged from 93-95 %RD, and Mo was found to be 97.9 %RD. These values are substantially lower than the standard L-PBF AM density cutoff threshold of 99.5 %RD but are not unexpected considering the inherent micro-cracks that form during the AM process. The microstructure was evaluated using optical microscopy, as shown in Figure 6, for both W and Mo.

Red boxes indicate areas of substantial surface-connected micro-cracks.

Mechanical Testing

Meso-scale tensile specimens in the as-built and heat-treated conditions underwent tensile testing at ambient temperature with load applied parallel to the build direction (Z). The median ultimate tensile strength (UTS) is shown in table 1:

Table 1: Tensile data of L-PBF AM W at 20 °C

Condition	UTSm (MPa)
As-Built	157.65 ± 17.9
Stress-Relieved	164.34 ± 29.7
Recrystallized	177.30 ± 4
HIP	In work

For comparison, the UTS of wrought W at 20 °C is 349.26 MPa, which means the L-PBF AM W is 50.7 % the strength of wrought. This result is unsurprising when we consider ρrel ranged from 93-95 %RD and the micro-cracks inherent in the microstructure act as stress concentration sites to an already fracture-prone material undergoing tensile loading at well below the ductile to brittle transition temperature. Future tasks include tensile testing as a function of elevated temperature since when W AM components are in service, they will not operate in a temperature range that would make them brittle.

Surface Finish

As-built areal surface roughness (Sa) as a function of geometric angle was measured on several specimens using a Keyence VR-3200 wide-area 3D scanner with a magnification of 80x in super-fine mode. The surface finish was typical for L-PBF AM, considering the layer thickness, laser focus diameter, and powder PSD. Table 2 shows the resulting AM W surface finish values measured as a function of angle from Z.

Table 2: Surface Finish of As-Built L-PBF AM W.

Condition	Sa (µm)
	0° (vertical)		15°		30°		45°		90° (horizontal)
	Blade-Facing	Back	Up-Facing	Down-Facing	Up-Facing	Down-Facing	Up-Facing	Down-Facing	Up-Facing
W As-built	5.5	6.3	5.4	7.3	6.1	8.9	9	15.7	–

Additional surface finish improvement is readily achievable using commercially available surface modification methods such as shot peen, chemical etch, slurry hone, electro-polish, etc. Future tasks will focus on thoroughly evaluating the applicability of these surface finish modification methods to W, Mo, C103, and other refractory metal options as they become implemented. Such surface enhancement methods will be critical to optimize AM surfaces to improve the fatigue life of refractory AM components and to impact protective coating adherence, commonly used to protect oxygen-sensitive refractory metals at elevated temperatures. Coating examples include molybdenum-silicide or R512E (Si-20wt%Fe-20wt%Cr), which are commercially available; however, the deposition process will likely require optimization based on the AM part surface condition. In addition to adherence, the performance of the coating to protect the refractory metal substrate will also require characterization.

Leak Testing and Coatings

Due to the known issue of micro-cracks that are volumetrically distributed throughout the W AM material, it is crucial to characterize the impact on the mechanical properties and gas permeability since many of the applications used in propulsion will require some degree of gas retention. Leak test specimens were printed from L-PBF AM W with 0.5 mm, 0.76 mm, and 1.0 mm wall thickness (WT), as shown in Figure 7. Four specimens for each wall thickness were produced to undergo heat treatment schedules previously discussed before testing. Specimens were tested with an as-built surface finish, and no additional post-processing was applied beyond powder removal and heat treatment.

Leak specimens were attached to a flexible plastic tube, and the joint between the specimen and the tube was sealed with epoxy. The other end of the tube was attached to a regulator and gaseous nitrogen supply. The specimen was immersed in water, and the nitrogen pressure was gradually increased incrementally from 68.5 kPa (10 psig), 137.9 kPa (20 psig), 206.8 kPa (30 psig), and 275.8 kPa (40 psig) to observe gas leakage through the walls. All AM W specimens in as-built and heat-treated conditions experienced leakage at a range of relatively low internal gas pressures. The 0.5 mm WT specimens experienced a slight leak at 10 psig, as observed by the formation of a single bubble that slowly grew over time and experienced bubbles free-flowing through the walls at 20 psig. The 0.76 mm WT specimens experienced a slight leak at 20 psig and a free-flowing leakage at 30 psig. Finally, the 1.0 mm WT specimens experienced a slight leak at 30 psig and free-flowing leakage at 40 psig. An example of the leak testing qualitatively showing gaseous nitrogen leak through the leak specimen wall as a function of pressure is shown in Figure 8. Specimens in the HIP condition showed the same leak result, indicating that the process was ineffective at closing micro-cracks, which was not unexpected considering the max HIP furnace temperature of 1700-1800 °C is low compared to the desired HIP temperature of 70% of W Tm or approximately 2400 °C. One potential option to prevent gas leakage is to increase wall thickness, but this has obvious drawbacks, such as increased weight, and does not address the underlying issue of micro-cracking. Obviously, gas-permeable materials are not desirable for most propulsion applications, except perhaps transpiration cooling. For this reason, ongoing efforts are underway to determine the influence of coatings on mitigating the gas permeability of AM W and Mo parts. These coatings are simply a near-term option until more advanced AM methodologies (dispersoids and new alloys) become widely available, which will have been optimized to minimize, if not eliminate, unnecessary micro-cracking or excessive porosity.

Potential coating methods include electro-forming, vacuum plasma spray, and LP-DED to add a thin layer (0.1-0.5 mm) coating of the same material as the AM substrate or a compatible material. Coatings used in refractory metals are typically applied to protect oxygen-sensitive materials during elevated service. However, this may also help address gas permeability for low-pressure applications where AM parts with micro-cracks may still be considered functional. Coating adherence will also be characterized as a function of surface finish from as-built to polished using various surface enhancement techniques previously discussed.

Non-Destructive Evaluation Limitations

Non-destructive evaluation methods such as x-ray computer tomography (CT) and x-ray micro-focus computer tomography (µ-CT) are heavily relied upon to inspect complex AM components. The high spatial resolution provided by µ-CT is able to discern fine geometric features, particularly for parts printed from relatively low effective atomic number elements that have low radio-opacity, meaning that a sufficient x-ray flux can penetrate into the part and to the detector to yield a sufficiently high signal-to-noise ratio that allows for images to be resolved with adequate resolution. Unfortunately, x-ray CT and µ-CT are unsuitable for the inspection of refractory metal parts since refractory metals have high atomic numbers, which have high radio-opacity resulting in high scatter, low penetration depth, and poor signal-to-noise ratio, which result in unresolvable images let alone defect detection. Neutron CT is a potential option in theory but traditionally requires access to national user facilities, which is impractical in a production environment. However, commercial neutron CT vendors are now a potential option, although part activation and cool-down may have significant implications on part schedule and handling.

Destructive methods such as automated sectioning and imaging produce a CT-like model but are not suited for production. Ultrasonic inspection is generally difficult to apply to AM due to the inherently rough surface finish of AM parts and would require surface enhancement to allow for appropriate application. Alternative non-destructive evaluation methods are required to use AM refractory metals and alloys pragmatically.

Conclusion

Refractory metal AM has been demonstrated as viable with W, Mo, Ta, and C103, as well as several refractory alloys in development. Recommendations for future work include surface finish enhancement evaluations, coating process and material evaluations, elevated temperature mechanical testing, component hot fire testing, and continued development of dispersoid-strengthened refractory metals and RHEAs to offer a more extraordinary array of formulations to meet the increasing demand for complex components operating at extreme temperature environments.