Characterization of Properties of Laser Powder Bed Fusion Three-Dimensional-Printed Inconel 718 for Centrifugal Turbomachinery Applications

A form of the laser powder bed fusion (LPBF) three-dimensional (3D)-printing method was selected for manufacturing Inconel 718 impellers for blowers used for anode off-gas recycling in solid oxide fuel cell systems. These blowers, which were recently developed at Mohawk Innovative Technology, Inc. (MiTi^®) in Albany, NY, and their 3D-printed impellers are shown in Fig. 1 and presented in detail in Refs. [1] and [2]. Applications of this nature often require impellers, as well as other aerodynamic and flow components (e.g., volutes and housings), to be made from nickel-base super alloys that maintain high strength and corrosion resistance at homologous temperatures approaching 1, like Inconel 718, Hastelloy X, or Haynes 282 [3]. The selected LPBF method is capable of printing Inconel 718 and presents proprietary modifications [4] that enable it to provide the design flexibility, geometric precision, and part reliability necessary to achieve both higher-performance turbomachinery designs and reduced manufacturing cost with respect to that equivalent units fabricated by traditional methods. These performance and cost improvements are discussed in detail in Ref. [2].

To date, there is a serious lack of information on the metallurgical and mechanical properties of the alloys that are becoming available for additive manufacturing (AM), either due to the relative immaturity of the field or concerns about the protection of proprietary information. In the specific case of 3D-printed Inconel 718, the authors found a surprising sparseness of property data that are critical to the turbomachinery designer, such as the modulus of elasticity or 0.2% yield stress as functions of temperature, with few notable exceptions like [5], which shows results of testing on nine Inconel 718 samples 3D-printed by selective laser melting. The elevated stress, high temperature, and corrosive/reactive environments that often characterize the operation of turbomachinery rotating components can represent serious reliability and safety risks that can only be managed by knowledge of these mechanical properties, and not by an assumption that 3D-printed materials will behave like their bulk (wrought/forged or cast) counterparts.

This paper presents a compilation of material property studies conducted by the authors during the execution of multiple turbomachinery development projects, all incorporating LPBF 3D printing as a main manufacturing method and Inconel 718 as the selected nickel-base superalloy. It is expected that while limited in size, the experimental data sets presented in this paper provide sufficient information to glean the capability of LPBF Inconel 718, along with some guidance for the turbomachinery designer when selecting this material. The test data include (1) metallographic and chemical composition analyses of 3D-printed samples confirming the identity of the alloy; (2) basic mechanical properties derived from tensile strength tests, including 0.2% yield stress, ultimate stress, modulus of elasticity, and elongation to failure. The properties are based on a 108 sample population sorted by print orientation into three groups of samples, all tested as functions of temperature in the range from ambient to 871 °C; (3) creep rupture data and corresponding Larson–Miller parameter at temperatures between 649 °C and 871 °C, based on 21 samples; and (4) high cycle fatigue data. The particular choice of mechanical property tests was based on the authors' experience as turbomachinery designers and their need for specific sets of properties considered critical to carry out machinery designs with minimum risk.

This section presents the methodology followed to produce the 3D-printed Inconel 718 blanks that were subsequently fabricated into test specimens for mechanical property testing. This section also includes a brief description of LPBF method and the subsequent heat treatment processes, as well as results of chemical and grain structure analyses that were performed to determine how the 3D-printed material compares to the Inconel 718 specification and how its grain structure responds to the heat treatment.

Laser powder bed fusion is a method of additive manufacturing that allows complex geometries to be fabricated from high-performance materials. LPBF applies a laser beam to selectively melt metal powder layer-by-layer. Each layer is deposited on a build plate that retracts as the layer is completed; after the build plate retracts, a recoater evenly spreads a new layer of metal powder over the previous layer. LPBF is being implemented in a multitude of industries, including aerospace, automotive, and medical, and adoption of the technology grows every year [6].

The particular LPBF method used in preparation of the material test specimens for this work is characterized by a number of control, diagnostics, and process modifications by Velo3D, with headquarters in Campbell, CA [4], which has recently risen as one of the leading manufacturers of LPBF systems [6]. Such modifications aim to increase the quality of the 3D-printed parts and are implemented in an integrated 3D-printer and diagnostics software system, shown in Fig. 2(a). Amongst these system characteristics are: (1) high-resolution lasing control and smoke plume management that prevents unwanted laser attenuation; (2) automated optical layer-by-layer validation process that guarantees a material matrix free of voids, flaws, or inclusions; (3) a one-touch self-calibrating system that ensures part consistency between different printers; and (4) use of a noncontact recoater that does not contact the part at the build plane [7]. In particular, the noncontact recoater is fundamental to the operation of this system, as it minimizes distortion of parts with protrusions and decreases the likelihood of a disruption to the build. In a nutshell, the recoater performs a number of operations simultaneously, as shown schematically in Fig. 2(b). First, it applies a layer of powder over the build plate, as the recoater blade shears the layer down. Finally, at the trailing end of the recoater, a high-precision vacuum system removes powder to leave a perfect powder layer above the previously lased surface. In this manner, the next lasing layer is prepared without disturbing the built part, minimizing distortion and allowing manufacturing of small features, thin edges, and floating parts.

This work aggregates results from multiple programs conducted over the last two years in which 3D-printed Inconel 718 was used as a primary manufacturing material of various critical turbomachinery components. As a result, the types of tests and the methodologies presented here vary according to the needs of the different programs, the level of funding dedicated to testing, and the authors' own evolving understanding of the problem. The clearest evidence of this variation lies in the difference in geometries of the 3D-printed Inconel 718 blanks from which the different testing specimens were prepared.

Figure 3 shows three batches of 3D-printed parts, still attached to the build plates delivered by the 3D printer. The build plate shown in Fig. 3(a) features sixteen simple cylindrical blanks for preparation of tensile property test specimens printed in two orientations (eight vertical and eight horizontal), and three blocks for chemical composition and metallographic analysis. Along with a group of test impeller prints, Fig. 3(b) shows a build plate with three rectangular blocks printed to allow later preparation of tensile test dog-bone specimens oriented along three orthogonal axes (X, Y, and Z). Figure 3(c) shows a build plate with 42 blanks oriented along three orthogonal axes and all principal diagonals (X, Y, Z, XY, YZ, ZX, and XYZ) for creep rupture and high cycle fatigue testing. The rationale for the different geometries, print orientations, and detailed specimen dimensions will be discussed below.

It should be mentioned that while ultimately the 3D-printing process is expected to yield final geometry parts (i.e., ready-to-spin impellers), in all of these trials, the process was only used to deliver test specimen blanks. This was done to allow a true one-to-one comparison to standard material testing results available in the literature, where base material (forgings or castings) blanks are machined to extremely tight tolerances and with controlled surface finishes that might not yet be available directly from additive manufacturing.

As with forged and cast materials, before machining to final dimensions, the blanks must be subjected to a multistep heat treatment process to guarantee the material becomes dimensionally stable and achieves its maximum mechanical properties. While the process discussed next is generally applicable to all 3D-printed metals, the temperature, pressure, and time parameters provided here are specific to Inconel 718. The first step in the process occurs before cutting the parts off the build plate. This is known as stress relief and generally consists of heating the build plates in a vacuum oven at a high temperature; for Inconel 718, this is around 1065 °C. The duration depends on the size of the build plate, to allow thermal stabilization (for these plates it was of the order of 1.5 h). The build plates are then allowed to oven cool. The parts are subsequently cut from the plate. For parts like these, a preferred method is wire electrodischarge machining (EDM). Following wire EDM, the individual 3D-printed parts follow a series of thermal treatments. First, the parts undergo a hot isostatic pressing (HIP) process per standard ASTM F3055-14a, which consists of heating the parts in a high-pressure argon atmosphere with the purpose of eliminating any residual microporosity and achieving maximum material density. The parameters for the HIP process are shown in the first column of Table 1. It should be noted that while the process was strictly followed, no verification was made of what level of residual porosity or densification was achieved. After HIP, the parts are heat treated in a two-step process, as used with forgings and castings. This process consists of a solution treatment in a vacuum followed by an aging process, both of which follow standard AMS 5663. The temperature, pressure, and gas conditions for the heat treatment processes are shown in Table 1.

The machining of the blanks to their final dimensions will be discussed below in the context of the specific tests for which they were used. First, however, samples of material taken from these fully heat treated blanks were used for metallurgical and chemical analyses with the purpose of determining how the 3D-printed material grain structure, level of recrystallization, and chemical composition compare to those of its bulk counterparts.

After undergoing the stress-relief, plate separation, and heat treatment processes described above, samples were cut from each of the three blocks shown in Fig. 3(a), and subjected to a chemical composition analysis according to standard ASTM E1086-14, using an optical emission spectrometer. The results of the analysis for each of the blocks are shown in Table 2. The first column shows the individual chemical elements detected by the spectrometer. Columns 2–4 show the chemical element weight proportion percentage detected for each of the three blocks. The last column shows the proportions specified in UNS N07718 for what constitutes Ni–Cr alloy precipitation hardenable Inconel 718. From the table, all of the sample chemical element proportions lie below the maximum thresholds or within the ranges specified by UNS, which confirms that the 3D printed material in question is in fact Inconel 718.

Grain structure characterization of the 3D-printed Inconel 718 was performed via electron backscatter diffraction (EBSD). To this end, a material sample was cut from each of three fully heat treated blanks, which were selected from build plate shown in Fig. 3(c). The selected blanks were printed along diagonal orientations denoted by XY, ZX, and XYZ.

For each of the three samples, two orthogonal areas of interest were identified for characterization utilizing EBSD and a 5 μm step size. These two areas are denoted as the parallel and perpendicular regions, which are relative to the build plane of the material. For each sample, a total surface area of 40 mm² was characterized to assemble grain average statistics, which encapsulated approximately 7500 grains. A section of one EBSD inverse pole figure map is shown in Fig. 4 for the parallel region of sample XYZ. The figure shows that most of the material achieved complete recrystallization after the HIP and heat treatment processing outlined above. The regions that did not recrystallize presented with a large amount of intragranular misorientation as shown in Fig. 5. All samples and orientations displayed similar behavior. Grain boundaries were defined between crystals that had misorientation of 15 deg or greater, or conformed to the definition of coherent twin boundaries. The regions of misorientation may play a role in the mechanical properties of the material.

Each build orientation had a distinct average mean angular deviation (MAD), but it is difficult to say whether the variation is the result of the print orientation or of whether it may stem from the differing scan strategies [8]. As calculated from EBSD, the average grain size and MAD are shown in Table 3.

The $γ″$ strengthening phase was characterized via transmission electron microscopy (S/TEM). A representative high-angle annular dark-field (HAADF) image is shown in Fig. 6(a). First, a 3 mm disk was extracted. The through thickness of the disk was thinned to 37 μm using silicon carbide polishing media ranging from 180 to 600 grit. The disk was ion milled on a Fischione 1051 TEM ion mill to approximately 100 nm in thickness, enabling electron transparency. Spatial identification of the $γ″$ phase was performed by utilizing selected area diffraction patterns and X-ray energy dispersive spectroscopy (XEDS) maps, as shown for niobium in Fig. 6(b). These techniques were performed on a 200-KeV FEI Talos F200X (S/TEM) equipped with a Super-X (Quad) EDS system. The size of the $γ″$ phase was measured utilizing the niobium elemental map and imagej [9]. The average length and width of the $γ″$ precipitates were found to be 60.2 nm ± 10.9 nm and 16.5 nm ± 3.1 nm, respectively. The niobium elemental map was utilized to perform size measurements for two reasons (1) the contrast difference or “haloing” created around the precipitates (indicative of $γ′$ [10,11]) in the HAADF images created uncertainty in spatial position, and (2) niobium is an element known to preferentially segregate to the $γ″$ phase [3,10,11].

This section presents the results of comprehensive tensile strength testing performed on 3D-printed Inconel 718 during the course of two separate turbomachinery development programs. The results are presented together, as they are complementary to each other, but it is evident that one of the studies was of an exploratory nature, with only 12 specimens tested at two temperatures, whereas the other was more comprehensive, with 108 specimens tested at a broader range of temperatures.

The tests were performed in a universal testing machine as that shown in Fig. 7. In the machine, material specimens of prescribed geometry (shown on the left side of Fig. 6) are placed under an increasing tension force that causes elongation; the specimens elongate until rupture. This is done at ambient conditions or as a function of temperature. For ambient conditions, the procedure is regulated by ASTM standard E8, and for testing as a function of temperature, it is regulated by ASTM standard E21-03a. The raw data obtained from these tests are in the form of stress versus strain curves, which are used to derive fundamental mechanical properties like the 0.2% yield stress, modulus of elasticity (Young's modulus, ultimate strength, and elongation to rupture. The aggregated data are then cast into curves of these properties as a function of temperature.

For ordinary materials, the problem is limited to determining the variation of mechanical properties as a function of temperature. However, 3D-printed materials may introduce a new complication, since they are produced in a layered manner, with the question naturally arising as to whether their properties are isotropic, i.e., whether they are uniform/invariant with respect to build orientation. This question was addressed by both of the testing efforts reported herein.

The specimens required for tensile strength testing consist of material coupons with geometries and dimensions that are regulated by the standards mentioned above (E8 and E21-03a). These coupons are colloquially known as “dog bones,” and generally consist of bars of cylindrical or rectangular cross section, with a narrow midportion known as the gage length, where any deformation or breakage that the part might undergo should occur. The regulating standards provide ranges of dimensions that are acceptable for both types of dog bone specimens. The exact dimensions of the specimens selected in these trials comply with the standards and are shown in Fig. 8. For the small scale exploratory study, the authors selected a threaded cylindrical geometry for fabrication of the dog bones. To this end, sixteen cylindrical blanks were 3D-printed, half in a vertical and half in a horizontal orientation, as shown in Fig. 3(a), and fully heat treated as described in Sec. 2. Six blanks from each print orientation were then machined in a lathe to the dimensions shown in Fig. 8(b). Six of the finished specimens are shown in Fig. 9, before testing.

Given the relatively high cost of lathe-turning each of the cylindrical specimens, the rectangular cross section geometry was selected for the larger scale follow-on study. For this, three blocks of material were 3D-printed along three orthogonal directions, as shown in Fig. 3(b). Figure 10 shows different stages of the dog bone sample fabrication process. The relative print orientation of the blocks and the directions along which the individual specimens would be cut are indicated in Fig. 10(a) with the colored X, Y, and Z-marked arrows. The blocks, removed from the build plate and fully heat treated, are shown in Fig. 10(b). An end face of each of the blocks was embossed during the 3D printing process with X, Y, or Z markings to identify the print orientation. Thirty-six samples were cut from each of the blocks via wire EDM to produce the individual samples shown in Fig. 10(c). The three finished sample groups are shown in Fig. 10(d). The samples were further marked with paint according to their orientation to avoid confusion during handling at the test facility.

For the exploratory study, the 12 threaded cylindrical specimens were 3D printed as two groups of six vertical and six horizontal samples. Three of each set were tested at ambient conditions and the rest were tested at 593 °C.

For the follow on the study, 36 rectangular cross section dog bones (seen in Fig. 10(d)) were printed along three orthogonal axes, X, Y, and Z, for a total of 108 specimens. Within the cost constraints of the project, this number was considered minimally adequate to build curves of tensile properties versus temperature with sufficient resolution along the temperature axis to capture important curvature or inflection points in the material behavior, as well as with enough samples to calculate means, standard deviation, and confidence intervals with some level of statistical significance for each of the three build orientations. For each orientation, the population of specimens was allocated for testing at the temperature conditions as shown in Table 4.

The smallest sample group consists of three samples and corresponds to ambient temperature testing. As the testing temperature is increased, the number of samples in each group also increases, since information availability is most critical for the designers at elevated temperatures where the material might begin to weaken. The maximum temperatures tested (760 °C and 871 °C) represent a practical limit of Inconel 718 for turbomachinery applications, as material strength is significantly diminished. Higher temperatures were not considered relevant since at those conditions, higher strength superalloys and/or forced cooling strategies would be necessary.

The dog bone specimens described above were individually installed in the universal test machine (Fig. 7) allowed to equilibrate to a predetermined test condition temperature, and subjected to a gradually increasing tensile force, under which they stretched, deformed, and eventually ruptured. The force and elongation data were converted to stress and strain, respectively, and presented in the form of stress versus strain curves. Figure 11 shows one such curve, corresponding to the test of a dog bone 3D-printed along the Y direction and tested at 260 °C. The data shown in this figure are typical of all of the data generated for each of the 120 specimens tested during both testing campaigns. The Fig. 11 curve has been labeled to indicate how key properties of the material like 0.2% yield stress, modulus of elasticity, ultimate stress, and elongation to rupture relate to the stress–strain data, and a schematic has been provided of the elongation, similar to what is presented in mechanics of materials textbooks [12].

Data for 0.2% yield stress (or 0.2% offset yield strength) are shown in Fig. 12 as a function of temperature and for each of the tested orientations. The data connected with dashed lines correspond to the aforementioned larger of the two studies. The individual traces represent each of the tested orientations (red is X, green is Y, and blue is Z). The symbols on those traces represent the means of each of the population samples shown in Table 4, and the error bars represent 95% confidence intervals assuming that the data are normally distributed and can be quantified using Student's t-distribution. The individual symbols that are not connected by dashed lines represent the results of the exploratory 12-sample test. As expected, the yield stress decreases as a function of temperature, with a rate of decrease that becomes steeper around 649 °C. At 750 °C, the 0.2% yield stress is approximately 599.8±4 MPa (87±5 ksi). This is a limit that a turbomachinery designer must keep in mind when using Inconel 718 in uncooled applications. Observation of the curves reveals that at lower temperatures, the z-direction specimens are significantly weaker than the corresponding x and y specimens, however, at higher temperatures, the data become more cohesive and show less dispersion.

Figure 13 presents the tangent modulus of elasticity (equivalent to Young's modulus in the linear elastic regime), which is the instantaneous rate of change of stress as a function of strain. It is obtained from the slope of elastic deformation portion of the stress–strain diagram of each of the individual test specimens (see Fig. 11). As with the 0.2% yield stress, the figure includes plots of the mean modulus of elasticity as a function of the test temperatures and for each of the print orientations. The error bars represent 95% confidence intervals for the true population mean. The dashed lines connect the average values at different temperatures of the comprehensive 108-sample study, and the individual points are the results of the 12 samples tested in the exploratory study. The large error bar for the Z-orientation at 260 °C appears to be due to one of the samples in the group having an anomalously small calculated module (possibly due to a flaw in the print), which would have a disproportionate impact due to the small population size. The ultimate stress and elongation results are shown in Figs. 14 and 15, and are presented in the same manner as the 0.2% yield stress and the modulus of elasticity charts.

The collective data of Figs. 12–15 appear to indicate that at lower temperatures, the material printed along the Z-axis is noticeably weaker than the horizontal X- and Y-orientations. The differences seem to become insignificant at higher temperatures (>600 °C), where the material appears to behave in a more homogenous manner. This is intuitively consistent with the layer-by-layer nature of the building process, however, the number of tested samples is small, and these observations should be considered conjectures until larger data sets become available.

Figure 16 shows how the 0.2% yield stress of the 3D-printed Inconel 718 compares to its hot-rolled and cast counterparts [13]. As can be seen, the yield strength of the 3D-printed material along any of the tested orientations lies neatly between both “bulk” versions of the material. The relative strength of the hot rolled material may be the result of the mechanical work hardening that characterizes the forging process, which may introduce a level of extra residual energy to the material matrix that is simply not available in the laser fusion process alone. Moreover, as seen from the EBSD analysis above, there are regions of incomplete recrystallization that may be responsible for weak material regions in the 3D-printed material. The relative weakness of the casting may be due to similar arguments combined with the amorphous material aggregation that may hinder the recrystallization process, though these are merely conjectures at this point and further study is warranted.

This section is a summary of results of a testing effort to expand the property dataset of 3D-printed Inconel 718 to include creep rupture (CR) and high cycle fatigue (HCF) data. Both of these sets of tests aimed to determine the creep and fatigue behavior of the material as a function of temperature and specimen print orientation, and were performed on populations with 21 samples each. This number per set primarily obeyed budget constraints due to the high costs associated with both fatigue and creep testing. As before, these experiments were performed following the guidance of appropriate ASTM standards; the CR tests were conducted according to ASTM E139-11 and the HCF tests followed ASTM E466-21.

The potential for anisotropy of the 3D-printed material was a primary concern, thus the material blanks were printed in different orientations. Due to the small number of samples (21 per set of tests), and in an effort to extract as much information as possible from the reduced dataset, the samples were printed along three orthogonal axes and all principal diagonals (X, Y, Z, XY, YZ, ZX, and XYZ). The build plate on which the total 42 blanks for these tests were 3D printed is shown in Fig. 3(c), and the specimen blanks are shown in Fig. 17 before complete separation from each other and finish machining. As before, these specimens adhered to the heat treatment processes described in Sec. 2.

The groups of three blanks were all cut from each other, and all 42 blanks were then machined to the specimen dimensions in accordance with the aforementioned ASTM standards. For this, a low-force low-heat grinding method was utilized to avoid the introduction of residual stresses in the test material. The HCF specimens were further subjected to a low stress powder glass blasting polishing process with the purpose of reducing the possibility of the presence of surface scratches, which could introduce stresses and affect the test results.

According to ASTM E139-11, creep testing subjects the test specimen to a constant tensile force at constant temperature, and determines the amount of deformation as a function of time (creep test) and the measurement of the time for fracture to occur (rupture test). Table 5 shows the test matrix to distribute the twenty one 3D-printed Inconel 718 samples, as identified by their print orientations, over stress, and temperature conditions that are characteristic of the operation of uncooled high-temperature anode off-gas recycle blowers, turbochargers, microturbine wheels exposed to combustor gases, and similar turbomachinery. This study concentrated primarily on determining the time to onset of deformation and the time to rupture of the samples.

The results of the creep time to failure tests for each of the 21 individual specimens are shown in Fig. 18. The test temperatures are recorded on the horizontal axis and the time to failure is reported in hours on the vertical axis. The dashed lines connect data obtained at a given stress level. The print orientation of each of the samples is shown as a label next to each of the data points. As would be expected, the time to failure is seen to decrease as the temperature and stress levels increase. Neither visual inspection nor statistical reductions allowed to make any statement about the influence of print orientation on the measure times to failure, and at this time it is not possible to say whether this is due to the small size of the population tested or to an insensitivity to orientation.

In the equation, t represents the time to failure in hours and T_abs is the test absolute temperature. The LMP was calculated from CR data for each of the tested stress level by averaging and by least squares regression, and the results are shown in Fig. 19, along with the curve reported in the special metals datasheet [13]. It should be noted that the data reported in that reference are calculated using Eq. (1) with the absolute temperature in degrees Rankine. The curves in Fig. 19 were calculated using Kelvin, including the curve from the reference, which was adjusted accordingly. The strong effect of temperature on the LMP can be seen as a drop in stress to fracture as the LMP increases. The 3D-printed Inconel 718 appears to present lower fatigue stress to rupture than hot rolled bar.

Per standard ASTM E466, high-cycle fatigue testing involves the application of a controlled periodic axial force to obtain the strength of the material in the fatigue regime where the strains are predominately elastic. The specimens were divided into groups for testing at three temperatures: 649 °C, 769 °C, and 871 °C, which as with the fatigue tests are typical during operation of uncooled microturbine wheels. For all the tests, a 40 Hz force-controlled loading was used under which the samples were tested until their rupture or until they reached 10 million cycles. The results of these tests are presented in Fig. 20. In the figure, each of the symbols represents a tested specimen and is labeled with its corresponding print orientation. The horizontal axis represents the number of loading cycles until specimen failure, denoted as Nf, and the vertical axis represents the stress magnitude imposed by the load force. As would be expected, Nf is inversely related to stress and temperature. Data points labeled with a single star (*) represent samples that reached the runout condition of 10 × 10⁶ cycles. To maximize the amount of data from the reduced number of specimens, specimens that reached runout were then retested at a higher stress level until they reached 10 × 10⁶ cycles again or until they ruptured. Specimens labeled with a double star (**) represent the specimens that ruptured after previously reaching runout condition. It is a matter of conjecture whether these retested specimens were work-hardened or weakened by the testing at lower stress levels. As with the creep data, the number of samples appears to have been insufficient to discern an influence of the print orientation on the material fatigue response to periodic loading.

The data in Fig. 19 are compared to fatigue tests results reported in Ref. [13] for fatigue strength of hot-rolled bar, showing a general consistency in the trends of the material response. Direct comparison of results is only possible between the data for LPFB 3D-printed Inconel at 649 °C and hot rolled bar at 659 °C, showing that the rolled bar is measurably stronger than the 3D-printed material, likely due to the same argument made in Secs. 3 and 4, conjecturing the influence of work hardening that occurs during the forging process that is not available during the 3D-printing process.

The chemical analysis confirms that the 3D-printed material is consistent with the UNS N07718 specification of Inconel 718. The EBDS grain structure analysis and the presence of the γ″ phase revealed by the XEDS appear to indicate that the fully heat-treated material is consistent with the expected structure of conventional presentations of the material. While the material appears to be stronger than that resulting from casting, the data for tensile strength, particularly 0.2% yield stress, and creep rupture as quantified by the LMP, and HCF, indicate that the material is sensibly weaker than corresponding forgings. A key observation of this study based on the EBDS results is that while the material appears to approach full recrystallization following heat treatment, there is a detectable fraction of the material that does not fully recrystallize. This may be the reason for the LBPF 3D-printed material presenting with mechanical properties measurably lower than those of forgings, however, it is not possible to exclude other mechanisms as contributing factors to this mechanical property deficit. Due to the small sample sizes, it is difficult to conclude whether print orientation has a significant impact on material strength at temperatures higher than 660 °C, where uncooled fuel cell anode recycle blowers, turbochargers, and microturbines generally operate.

Generally, the authors conclude that with awareness of the potential reduction of mechanical properties with respect to forgings, the LPBF 3D-printed Inconel 718 is adequate for high-stress centrifugal turbomachinery component applications.

The authors acknowledge the support of the Velo3D team that supplied the 3D-printed material used for sample preparation.

• U.S. Department of Energy (DOE) (Award Nos. EE0008374, FE0027895, and SC0020793; Funder ID: 10.13039/100000015).

• Air Force Research Laboratory (AFRL) (Contract No. FA8649‐21‐P‐0932; Funder ID: 10.13039/100006602).

This report was prepared as an account of work sponsored by agencies of the United States Government. Neither the U. S. Govt. nor any agency thereof, nor their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not constitute or imply its endorsement, recommendation, or favoring by the U.S. Govt. or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the U.S. Govt. or any agency thereof.