Efficient production of a high-performance dispersion strengthened, multi-principal element alloy

Additive manufacturing currently facilitates new avenues for materials discovery that have not been fully explored. In this study we reveal how additive manufacturing can be leveraged to produce dispersion strengthened (DS), multi-principal element alloys (MPEA) without the use of traditional mechanical alloying or chemical reactions. This new processing technique employed resonant acoustic mixing to coat an equiatomic NiCoCr powder with nano-scale yttrium oxides. Then, through laser powder bed fusion (L-PBF), the coated powder was successfully consolidated into 99.9% dense parts. Microstructural analysis confirmed the successful incorporation and dispersion of nano-scale oxides throughout the build volume. Furthermore, high temperature mechanical testing of the DS alloys showed significant improvements in strength and ductility over the baseline NiCoCr. As a result, this recently discovered processing route opens a new alloy design and production path that is synergistic between additive manufacturing and dispersion strengthening, possibly enabling a new generation of high-performance alloys.

www.nature.com/scientificreports www.nature.com/scientificreports/ have successfully used AM to produce DS components 18,19 and results on the mechanical properties of these alloys have been overall promising. However, a process such as mechanical alloying is still required to incorporate the dispersoids into the metal matrix prior to building with AM 20 . Furthermore, AM processing with mechanically alloyed powder is problematic because the highly deformed powders have poor flow and, thus, reduced feedstock delivery properties. Studies have also shown that powder shape and size distribution contribute significantly to the quality of AM builds 21 . Other methods of incorporating nanoparticles into additive powder lots through chemical reactions or depositions have been explored; though the added complexity and expense of these techniques may limit their commercial viability 15,17,[22][23][24][25][26] .
For this study, we produced AM NiCoCr mechanical test specimens with dispersed nanoscale yttrium oxides (Y 2 O 3 ) that are suitable for use in high-temperature applications through a new, economically viable method. Mechanical and microstructural analysis confirm the success this new fabrication pathway had in creating dispersion strengthened metallic alloys.

Results
Production of dispersion strengthened MPEAs via additive manufacturing. To produce AM material, equiatomic NiCoCr medium entropy alloy (MEA) powder which exhibited a diameter size range between 10-45μm and Y 2 O 3 particles rated between 100-200 nm were acquired. A Resodyn LabRAM II resonant mixer was employed to coat the NiCoCr powder with one weight percent of nanoscale Y 2 O 3 27-29 . This acoustic mixing process is illustrated in Fig. 1 along with images of the resulting powder.
In Fig. 1(a), a diagram of the mixing mechanism is shown. Here, acoustic mixing employs a longitudinal pressure wave composed of a short amplitude and high frequency by attaining a resonance between a vibrating spring system and the stored mass of the powder and container 29 . This process quickly homogenized the powder ( Fig. 1(b)), eventually coating the larger NiCoCr powder with a thin film of Y 2 O 3 after an hour of mixing in a polyurethane container (Fig. 1(c)). A cross section of a coated powder particle imaged using a scanning electron microscope (SEM) and Everhart-Thornley detector is shown in Fig. 1(d,e). The high-resolution SEM image of the Y 2 O 3 coating reveals the film to be roughly 250 nm thick. Energy dispersive spectroscopy (EDS) analysis confirmed that the particle is fully coated with the nano-scale Y 2 O 3 as shown in Fig. 1(f). It is still not well understood how this Y 2 O 3 coating is obtained and on-going work is being performed to better understand the underlying physics of this process. Qualitative Hall Meter flow tests indicated that the coated powder had similar flowability www.nature.com/scientificreports www.nature.com/scientificreports/ as compared to the virgin (V) uncoated NiCoCr powder. Optical and SEM microscopy revealed the morphologies of the NiCoCr powder to be unchanged after the mixing process in Fig. 2. Figure 2 also further confirms that the acoustic mixing step successfully coated the NiCoCr powder. In addition, the powder maintains its spherical morphology as compared to the highly deformed platelet-like powder produced through MA making it more suitable for AM. Baseline NiCoCr (V-MEA) and oxide dispersion strengthened NiCoCr (DS-MEA) mechanical and metallurgical test bars were successfully produced via L-PBF after a thorough exploration of the build parameter trade space. A recent study by Carnegie Mellon successfully captured the formation of a meltpool using L-PBF and the subsequent production of keyhole porosity. Their study successfully illustrated that the meltpool created by a laser during the L-PBF process is turbulent, mixing the liquid multiple times over before solidification can take place 30 . In fact, multiple studies have recently revealed that L-PBF melt pool turbulence can sufficiently disperse satellite ceramic particles into a metal matrix 18,23,24 . Figure 3(a) illustrates how it is believed this turbulent melt pool successfully dispersed the coated oxides into the metallic matrix of the build volume.
SEM micrographs of the dispersed oxides ( Fig. 3(b)) confirm that the oxides are dispersed throughout the microstructure without any noticeable pattern that may have resulted from the laser scan path. EDS analysis in Fig. 3(c) confirms that the darker particles are Y 2 O 3 and not some other contamination. Numerous other micrographs taken randomly throughout multiple builds reveal similar results, confirming the successful incorporation of the Y 2 O 3 into the AM parts ( Supplementary Fig. 1). The chemical maps in Fig. 3(c) also confirm that the NiCoCr remained a solid solution FCC phase during the L-PBF process. SEM analysis of the oxides using images represented in the Supplementary Fig. 1 found a 0.6% volume fraction of Y 2 O 3 present in the as-built samples. This number should represent a lower bound of oxide content as some oxides may not be counted due to resolution limitations of the SEM. Separate build parameters were developed to produce the 99.9% dense parts for both the V-MEA and DS-MEA alloys. It should be noted that the energy density needed to produce fully dense builds using the coated powder was much higher than that needed for the uncoated feedstock. In fact, trying to use the DS-MEA build parameters on uncoated feedstock resulted in print failures due to the excessive energy input. Examples of the micrographs used to determine the porosity for both builds are displayed in Supplementary  Fig. 2.

Microstructural characterization of V-MEA and DS-MEA AM builds.
Following the successful L-PBF builds, the two alloys were characterized both before and after a high stress hot isostatic press (HIP) cycle at 1185 °C which was performed to relieve residual stress. The grain structures derived from electron back scatter diffraction (EBSD) maps of the V-MEA and DS-MEA builds with and without post processing are presented in Fig. 4 with the build direction denoted by the Z axis. www.nature.com/scientificreports www.nature.com/scientificreports/ Large variations in grain structure and average grain diameter between both the as-built and post-HIP conditions for the V-MEA and DS-MEA specimen were observed. Average grain diameters (d) calculated from the EBSD maps are presented for each image. Interestingly, the laser path is evident in the as-built XY plane EBSD maps for both samples -producing a grid like grain structure of large grains surrounded by finer grains. For the case of the V-MEA builds, the HIP cycle promoted grain recrystallization and growth resulting in a much more equiaxed structure though some large elongated grains were still captured in the ZY plane EBSD maps. In contrast, the Y 2 O 3 in the DS-MEA specimen clearly suppressed grain growth and recrystallization by pinning the grain boundaries both during the L-PBF build process and subsequent HIP cycle. Indeed, minimal recrystallization was observed in the post-HIP DS-MEA specimen, retaining the grain texture and finer average grain size, as compared to the V-MEA builds. Twin formation was also suppressed in the DS-MEA samples. Despite the minimal grain boundary movement during the HIP cycle for the DS-MEA sample, XRD analysis revealed that the residual stress (σ r ) was successfully reduced from>300 MPa to near zero. SEM-EDS and X-ray diffraction confirmed that the NiCoCr MEA matrix remained a disordered solid solution in every stage of processing explored ( Supplementary Fig. 3).
Mechanical testing of AM samples. All four sample conditions that were analyzed in Fig. 4 were subsequently tensile tested at room and 1093 °C according to the ASTM standards E8 and E21. Figure 5 presents the stress-strain curves from the room temperature tensile tests.
Notably, Fig. 5(c) reveals the post-HIP yield strength of the DS-MEA to be 50% higher than the V-MEA sample (496.4 MPa vs 331.6 MPa) further confirming the successful production of a dispersion-strengthened alloy. The DS-MEA specimen also retained the pronounced elongation and strain hardening properties inherent in the NiCoCr MPEA 31 . In addition, the strength and ductility exhibited by the HIPed V-MEA specimen is comparable to conventionally produced NiCoCr alloys while the finer grain structure in the as-built V-MEA sample provides higher strength and less ductility 8,32 . Figure 6 reveals the 1093 °C tensile test results. Figure 6 reveals that the DS-MEA alloys possessed notably improved high temperature properties over the baseline NiCoCr specimen. In the as-built condition, the inclusion of the Y 2 O 3 nano particles provided a > 35% increase in ultimate strength and >2.5x improvement in ductility compared to the V-MEA sample. Consistent improvements were found for the HIP specimen as well. Fracture surfaces from the elevated temperature tests were examined to better understand the large strength and ductile differences between the V-MEA and DS-MEA specimen. Fractography analysis reveals tensile failure, with some evidence of oxidation damage. The DS samples displayed significantly more ductility as seen in Supplementary Fig. 4. This increase in ductility may be due to the Yttria dispersions in the DS specimen maintaining a finer grain structure compared to the baseline NiCoCr samples. The baseline samples also appeared to be more susceptible to localized oxidation damage. Future work www.nature.com/scientificreports www.nature.com/scientificreports/ aims to further explore these failure differences and better understand the effects the nano-scale dispersions have on different high temperature properties, such as creep.

Discussion
Comparison to present day state of the art high temperature materials. Figures 5 and 6 revealed that the DS-MEA specimen exhibited superior mechanical properties over the baseline V-MEA samples. This was most evident at the higher temperature were the oxides provided significant improvements in strength and ductility. To better understand how these properties compare to the present day state-of-the-art wrought and DS superalloys, the 1093 °C ultimate strength vs. density of these alloys are plotted together in Fig. 7 [33][34][35][36][37][38] .
The plot in Fig. 7 compares the high temperature properties of both DS superalloys (blue), which are currently quite difficult and resource intensive to produce, and conventional wrought superalloys (red). The DS-superalloys clearly present a strength improvement compared to similarly dense superalloys. Most notably is the finding that the DS-MEA specimen presents similar properties as other DS-superalloys. Indeed, compared to the wrought superalloys, only Hastelloy S presented a higher ultimate strength than the DS-MEA specimen. This finding confirms that this new DS manufacturing process can be successfully leveraged to produce alloys with comparable high temperature properties to current DS-Superalloys. These findings are even more impressive considering that many aspects of the DS-MEA production have not yet been optimized, including oxide amount, grain structure, or heat treatment.
In conclusion, we present for the first time the ability to produce DS alloys through AM processes without requiring resource intensive steps, such as mechanical alloying, to incorporate dispersoids. The resulting DS material exhibited a 50% increase in room temperature yield strength over the baseline alloy with both samples in the HIP' ed condition. More notably, the addition of nanoscale dispersed Y 2 O 3 in the NiCoCr improved its high temperature ultimate strength by >35% and increased its ductility almost three-fold over the baseline NiCoCr. This economical fabrication technique represents a new, unexplored approach to producing high temperature and high strength materials that until now have been either difficult or impossible to fabricate.

Materials.
NiCoCr powder was purchased from Praxair Inc. The powder's composition in weight percent was 34.66% Cobalt, 30.29% Chromium, 34.90 Nickel, combined with trace amounts (<0.02%) of Silicon, Iron, Nitrogen, and Oxygen. The powder was sieved using +270 and −325 mesh to acquire an average diameter of 14.1um as determined using a Horiba PSA300 Static Image Analysis System Particle Size. The dispersoid used in the AM process was nanoscale Y 2 O 3 powder acquired from American Elements with a diameter range between 100-200 nm. The powder was certified 99.999% pure Yttrium Oxide. www.nature.com/scientificreports www.nature.com/scientificreports/  www.nature.com/scientificreports www.nature.com/scientificreports/ Dispersion strengthened AM process. Batches comprising of 500 g of NiCoCr and 5 g of Y 2 O 3 powder were poured together in a polyeruthane container and sealed using electric tape around the lid. Each container was mixed using a Resodyn LabRAM II Acoustic Mixer in 10-minute increments for a total of 60 minutes, letting the powder cool between each session. Post-mixed powder was then sieved using a 230 mesh screen to remove any large oxide or metallic powder particles. Both unmixed (V-MEA) and mixed (DS-MEA) NiCoCr powders were used to additively build microstructural and mechanical test components using an EOS M100 selective laser melting machine. Two-inch tall vertical test specimens were built upon 304 stainless steel build plates. All samples were then removed from the build plates using electric discharge machining (EDM).

Microstructural characterization.
For SEM analysis, samples were polished using SiC grit paper followed by 0.5 diamond suspension. Afterwards a final polish using 0.05 colloidal silica for 24 hours was employed on samples used for EBSD analysis. Oxide volume fraction analysis was performed on a Zeiss Auriga-FIB using an Everhart-Thornley secondary electron detector with low accelerating voltage (3 kV). By utilizing low accelerating voltage sub-surface oxide particles were avoided, thereby ensuring a more accurate volume fraction measurement. EBSD orientation mapping was performed using an EDAX Hikari EBSD detector with an 800 nm spot size. Post-processing of the maps was done using the TSL OIM Data Collection 7 software. Average grain diameters extracted from the maps did not include twin boundaries or grain sizes less than 3 um to remove effects from scan noise. High resolution imaging of the Y 2 O 3 coating on the NiCoCr powder was performed using a Tescan MAIA3 in the ultra-high resolution (UHR) configuration at 15 kV. Chemical maps were performed using an Oxford Ultim Max Silicon Drift Detector and Aztec Software. Residual stresses were measured at the surface using a Bruker D8 Discover (area detector) X-ray diffractometer aligned in accordance with the approach and error bounds specified in ASTM E 915-10 but applied to the side-inclination rather than iso-inclination method. Data was gathered using Mn Ka radiation and the (311) crystallographic plane on a specimen target area of 1.2 mm 2 . Each residual stress dataset consisted of 24 area detector frames taken at 4 sample tilt (psi) angles (0°, 15°, 30°, and 45°) and 6 sample rotation (phi) angles (0°, 45°, 90°, 180°, 225°, and 270°). X-ray penetration depth decreased with increasing psi angle, going from 29 μm to 20 μm, representing depths that correspond to a 99 percent contribution to the diffracted beam. These X-ray results were analyzed using the Bruker LEPTOS v.7 software. Peak width was measured using the TOPAS program.
Specimen processing after AM. To better understand the effect post-processing may have on both the microstructure and mechanical properties of the L-PBF MEA samples different a HIP cycle was performed. Samples of both the V-MEA and DS-MEA underwent a HIP cycle at 1185 °C while wrapped in Ta foil to mitigate oxidation. The HIP cycle also had the benefit of removing residual stress. This provides a better comparison between the DS and V-MEA samples as residual stress has been shown to affect mechanical properties 39 .
Mechanical testing. Room temperature and elevated tensile tests were performed by Metcut Research Inc.
for both the V-MEA and DS-MEA test specimen after different post processing pathways. All tensile tests were performed using ASTM E8 standard.

Data availability
The data that support the findings of this study are available from the corresponding author on reasonable request.  www.nature.com/scientificreports www.nature.com/scientificreports/