Options for Improving Performance of Additively Manufactured Nickel-Base Superalloys for Gas Turbine Applications

Additive manufacturing (AM) has been increasingly used by the gas turbine (GT) industry and has provided an opportunity for innovation of design concepts in various components. More recent developments have involved the implementation of AM build components into the hot-section area of gas turbines. The Advancement of hot-section components will provide a method to increase overall efficiencies of gas turbines by increasing the operating temperature and/or reducing the consumption of cooling air. In turn, this will drive reduced carbon emissions and allow gas turbine technology to lead the way for the energy transformation.

The performance of hot-section stationary vanes can be improved through new design concepts that implement improved internal cooling features to reduce the consumption of cooling air [1–4]. Recently, this technology has also been implemented into commercial gas turbines with significant benefits. Vistra Corp. installed AM row 1 vanes (R1) in two of their GTs at the Ontelaunee Energy Center, Pennsylvania, USA. This combined-cycle gas turbine is a 600 MW, 2 × 1 design. The new Power System Manufacturing (PSM) GTOP7 [5] upgrade package (including the R1 AM vanes) was installed on the two Siemens SGT6-5000F units, and this resulted in a 10-MW increase from the GTs [6]. These vanes have been operating >25,000 h and are close to reaching the full 32,000 h planned run. Additionally, the increased firing temperature allowed an increase in exhaust temperature and subsequently added ∼4 MWs of power output from the steam turbine. In total, the combined-cycle gas turbine plant was able to increase power output by ∼14 MW's, which represents an overall increase in efficiency of ∼2.5%. Additional benefits involve the continuation of operating at the traditional 32,000-h life cycle for the components due to the decrease in overall part temperature, which was a decrease of about ∼85 °C compared to standard cast components. Figure 1 shows the design of the AM produced R1 gas turbine guide vanes, estimated part temperature, and a schematic of how Vistra has obtained increased power output at the Ontelaunee Energy Center.

To further improve GT technology, additive manufacturing will play a key role, and more hot-section components will utilize AM methods. However, key challenges remain due to insufficient mechanical test data to support the higher temperature operation of AM built components. Additional research is needed to better understand the performance of AM nickel-base superalloys and their resulting properties for the GT industry to further capitalize on AM technology.

This paper discusses on-going research studies and key findings in the processing–structures–properties–performance relationships in ABD®900-AM material, which has been identified as a potential candidate material to further expand AM opportunities and push the boundaries in GT technology.

Recent studies have focused on developing relationships between alloy composition, processability, and high temperature mechanical behavior for ABD®900-AM material [7,8]. This alloy has been shown to have excellent printability at a wide range of processing parameters and exhibits properties comparable to nickel-base superalloys in this class (∼35% γ′ vol. fraction) [7,8]. Laser-based powder bed fusion (LBPBF) and electron beam powder bed fusion (EBPBF) were two methods utilized for manufacturing, each with advantages and disadvantages. LBPBF material was processed with a Renishaw 400AM machine using a 200 W laser power, 30 μm layer thickness, 90 μm hatch distance, 70 μm point distance, and 1000 mm/s laser speed. EBPBF material was processed with an Arcam Q10+ machine using a 60 kV accelerating voltage, 16 mA reference current, ∼800 mm/s scanning speed, 70 μm hatch spacing, and 50 μm layer thickness.

Various builds were made for each process, and multiple postbuild heat treatments were also utilized to understand the effects on the resulting microstructure and high temperature mechanical properties. LBPBF material was assessed in both a near-solvus and super-solvus solution anneal (solvus temperature γ′ ∼1060 °C). EBPBF material was only evaluated in the super-solvus solution anneal. Each heat treatment also involved multistep aging to produce a bimodal distribution of γ′. Table 1 shows a comparison of the near-solvus and super-solvus conditions.

Hot isostatic pressing (HIP) methods were also utilized on the EBPBF material to determine if part density and performance could be improved. Specimens were given a HIP at 1120 °C (2050 °F) for 4 h at 100 MPa (14.5 ksi). Comparisons of overall build density were determined using imagej software.

Microstructural evaluations were carried out in each test condition using optical and scanning electron microscopy (SEM) tools for both transverse and longitudinal orientations relative to the build direction. The SEM electron backscattered diffraction (EBSD) detector was used to better distinguish differences in grain structure due to build method, orientation, and heat treatment. Density was measured optically using automated large area image analysis of unetched mounts.

High temperature mechanical testing (creep) was conducted in accordance with ASTM E139 [9], and some of the results are discussed in the paper. Additional details on creep modeling and creep mechanisms are discussed in previous work published by the authors [10].

Chemical composition of the material was measured using inductively coupled plasma optical emission spectrometry for Al, B, Co, Cr, Mo, Nb, Si, Ta, Ti, and W. Combustion methods were used for C and S. Inert gas fusion was used for O and N.

The measured chemical composition for both the LBPBF and EBPBF produced material is shown in Table 2. The nominal composition is also shown for comparison. The measured boron content in the EBPBF material was much lower (not to specification) than in the LBPBF material. All other controlled elements were within specification for both materials. Measured oxygen was lower in the EBPBF material, which may have some implications on HIP processibility, and the prevention of transformation induced porosity. The impacts of chemical composition on performance are detailed in Sec. 4.

The measured powder size ranged from ∼15 to 53 μm for the LBPBF material and ∼15 to 90 μm for the EBPBF material. LBPBF material exhibited high build density (>99.9%) in the as-built condition, but the EBPBF material had some porosity in the as-built condition (∼98.4% density). However, the HIP process improved density considerably in the EBPBF material (∼99.6%). Figure 2 shows a comparison of measured part density in the EBPBF material in both the no-HIP and HIP conditions. It is also important to note that the material was given a full solution anneal and multistep aging both before and after the intermediate HIP cycle. Also, the HIP temperature (at 1120 °C) was below the full super-solvus solution anneal (at 1240 °C).

A total of 12 different conditions were assessed with variations in the build process, heat treatment, and orientation relative to build direction. Seven different conditions were used for high temperature mechanical testing. Table 3 shows details for the different conditions, general findings, and measurements for grain size, MX carbide size, primary γ′ size, and secondary γ′ size. In MX, the “M” represents metallic elements (Ti, Ta, Nb, or W), and the “X” represents interstitial elements (C or N). “AB” indicates as-built, “HT1” indicates near-solvus heat treatment, “HT2” indicates super-solvus heat treatment, “T” indicates transverse to the build direction, and “L” indicates longitudinal to build direction. The “GSOA” was defined as the measured grain size relative to the stress direction. This was determined due to orientation effects, and the fact that creep data were observed at boundaries perpendicular to the stress axis (see Sec. 3.3).

Lower magnification SEM images are shown in Fig. 3. The LBPBF material exhibited a very fine grain structure with circular/hexagonal substructure in the transverse direction and epitaxial growth in the longitudinal direction (parallel to build). The near-solvus heat treatment showed a similar grain structure, but the substructure was removed due to the supersolvus solution anneal. The super-solvus heat treatment showed recrystallization and some grain growth. EBPBF material showed a much larger grain size in the as-built condition and clear epitaxial growth parallel to the build. The applied super-solvus heat treatment did not alter the overall grain structure or size for the EBPBF builds.

Higher magnification SEM images are shown in Fig. 4. The LBPBF material clearly shows substructure (likely due to high densities of dislocations), and MX carbides are very fine (<10 μm). There was also no observed γ′ in the as-built condition up to 50,000× magnification. The primary and secondary γ′ sizes were the same for the near-solvus and super-solvus, which were 100 to 240 nm and 20 to 40 nm, respectively. MX carbides were shown to have some growth at the higher temperature heat treatment and were ∼20 to 80 nm for near-solvus and 90 to 250 nm for the super-solvus condition. EBPBF processed material showed both primary and secondary γ′ precipitates that were 180 to 280 nm and 40 to 80 nm, respectively. The coarser primary γ′ was also in individual “islands,” and MX (∼200 to 500 nm) was also shown to precipitate in these regions. The super-solvus heat treatment process (solution anneal + aging) dissolved all γ′ and reprecipitated to provide a bi-modal distribution of γ′ that was more evenly dispersed throughout the matrix. Primary γ′ was 100 to 240 nm, and secondary γ′ was 20 to 40 nm, which was equivalent to the LBPBF.

Electron backscattered diffraction data were collected for each condition, and inverse pole figures are shown in Fig. 5. This data not only helped show clear differences in grain shape/size but also in structure, such as misorientation differences and overall texture. EBSD clearly confirmed that the subgrain structure was eliminated with just a near-solvus heat treatment and completely recrystallized for the super-solvus condition in the LBPBF produced material. Results also showed no differences in both grain shape, size, or orientation for the as-built and super-solvus conditions in EBPBF produced material. Texture, defined as the distribution of crystallographic orientations relative to each other, was observed in some conditions. Modest texture was observed in the as-built in near-solvus conditions for LBPBF material, but no texture dependence was shown in the super-solvus condition. However, the EBPBF produced material showed very strong texture and major changes in preferred crystallographic orientation relative to build orientation (transverse versus longitudinal).

A high temperature creep testing program is being carried out for each condition at both high stress and lower stress conditions over a range of temperatures and stresses. Some early results for the LBPBF-HT1 are discussed in detail in Ref. [10]. Figure 6 shows the strain (%) versus time to rupture (hours) plot for high stress creep tests conducted at a temperature of 800 °C (1472 °F) and stress values of 400 MPa (58 ksi) and 325 MPa (47.1 ksi). Figure 7 shows the strain (%) versus time to rupture (h) plot for low stress creep tests conducted at a temperature of 900 °C (1652 °F) and a stress value of 100 MPa (14.5 ksi). These test results were selected because they are generally representative of the larger test program and data were available for all the different conditions.

Each creep tested sample was sectioned parallel to gauge axis and polished for metallographic imaging. SEM imaging was used at a variety of length scales to document the extent of damage in the gauge section of the post-test specimens. SEM images for the high stress tests are shown in Fig. 8 and for the low stress tests in Fig. 9. For the high stress test condition, creep damage was observed at the grain boundaries perpendicular to the stress axis, and creep cavities formed as spherical voids. Similarly, creep damage in the low stress test condition was also identified at grain boundaries perpendicular to the stress axis. However, there were substantial differences in the morphology and surrounding microstructural features. At grain boundaries, there were regions denuded of γ′ precipitates, also known as precipitate free zones (PFZs). Creep damage appears to be associated with these PFZ areas. There were also larger MX carbides near the PFZ regions that were of the order of ∼1000 nm in size, suggesting growth during the creep test. The creep voids were blocky in shape compared to the spherical cavities in the higher stress tests. Each material condition showed similar damage characteristics at the same test conditions, but variability in overall rupture life.

Material characterization and high temperature creep testing results in ABD900-AM have provided new insights into the effects of build process, heat treatment, and orientation on additively manufactured nickel-base superalloys. This alloy has been shown to be effectively printed with standard parameters for both LBPBF and EBPBF methods. A key focus of this research is to understand the effects of microstructural features on high temperature performance and to also see if performance was improved compared to other alloys in this class (∼35% γ′ vol. fraction). Results strongly suggest that there are benefits of this alloy compared to prior work in a similar alloy. Figures 10 and 11 show a comparison of rupture life for ABD900-AM in the various material conditions to IN939 casting material and a similar AM produced IN939 derivative alloy for the high and low stress conditions [4].

The creep testing results for ABD900-AM clearly show an improvement in properties (rupture life and ductility) at all test conditions compared to testing in the prior AM alloy of the same class. At high stress and lower temperatures, the material showed comparable properties to the IN939 castings. However, at lower stresses and higher temperatures, the AM produced materials show a debit in creep performance compared to the castings but better ductility and small improvements in rupture strength compared to other similar AM alloys. The improved AM performance of ABD900-AM is likely due to the computationally optimized chemical composition to balance processing and properties. To provide improved printability, Ti and C content is lowered compared to similar alloys to reduce the susceptibility of strain age cracking while Al and Nb are balanced to still provide sufficient quantities to obtain ∼35% γ′ vol. fraction after aging. The alloy composition also provides solid solution strengthening from judicious additions of refractory elements (Nb, Mo, Ta, W). The only major difference in chemistry due to processing was oxygen content. Since EBPBF is conducted under a vacuum condition, and the overall build is conducted at a higher temperature, the oxygen content was lowered by over 50% from that in the powder and the LBPBF material. However, it did not reduce to the single PPM level typical of today's GT castings. It is unclear if the oxygen content has a major influence on the creep performance but historical research on castings suggests reducing oxygen can be beneficial to rupture life [11]. Also, it should be noted that the reduced boron content in the EBPBF material may have a slight impact on rupture behavior, as boron is a grain boundary strengthener.

A comparison of the creep rupture properties was also made to common nickel-base superalloys, such as IN625, IN718, IN939, and IN738 [12–15]. This is shown as a Larson–Miller plot in Fig. 12. The performance of ABD900 shows a clear benefit compared to IN625 or IN718 and would be a useful alternative for common GT components fabricated from these materials. Previous studies have also shown that tensile properties are improved for yield strength, tensile strength, and ductility compared to IN939 [16]. IN939 and IN738 are higher creep strength alloys commonly used for hot-section components, such as guide vanes or blades. Results from this study suggest ABD900 could also be used as an alternative to these alloys if performance is improved elsewhere, such as improved cooling features (as shown by example in Sec. 1).

Heat treatment was shown to have a major influence on the resulting microstructure. The near-solvus condition led to the elimination of the substructure and suggested reduced dislocation densities, but grain size was comparable to the as-built condition. However, the super-solvus heat treatment provided enough driving force for full recrystallization and some grain growth. This led to an increase in performance of ∼2 to 3× in rupture time in the LBPBF material. This suggests that traditional solution anneal heat treatments for nickel-base superalloys may not be sufficient, and higher temperatures may need to be used. Previous literature research has also shown higher solution anneal temperatures leading to recrystallized structures in AM nickel-base alloys [17–21].

Grain size showed some dependence on creep rupture life in the LBPBF material where increased grain size led to longer life, but other factors may have influenced behavior (such as grain structure and orientation). Additionally, the EBPBF material exhibited much larger grains, but performance was reduced in the transverse direction compared to LBPBF material with much finer grains. While many have suggested grain size as the primary reason for reduced life of AM components [22], this study has clearly shown there are many other factors to consider.

Carbide structure (MX) may also have some effect on the resulting creep behavior. In the higher temperature tests, it was observed that PFZs were occurring near grain boundaries, and MX increased in overall size in these regions. Literature has also suggested that Ti-rich M(C,N) precipitates may grow at the expense of γ′ leading to PFZs [23].

Grain texture and orientation were shown to have a major impact on creep performance. The EBPBF produced material showed a very strong texture with epitaxial growth in the 〈100〉 build direction. The difference between crystallographic orientation is likely to play a role in the deformation response. It is well known that the modulus values are different in the 〈100〉, 〈110〉, and 〈111〉 directions for FCC nickel-base structures [24] as well as the stress rupture properties [25]. The change in orientation in addition to a larger number of grain boundaries perpendicular to the stress axis is postulated as the significant microstructural changes leading to reduced life in the transverse direction in EBPBF material.

Porosity was shown to have a minor/modest effect in the creep performance on the EBPBF material. At the higher stress condition, the rupture times were nearly identical, but there was a factor of ∼2× at the lower stress test condition in the HIP versus non-HIP conditions. Much of the porosity was randomly dispersed in the matrix and only a limited number of voids were near grain boundaries (dominnat creep initiation location). There were no issues with transformation-induced porosity. The lower oxygen content (30 ppm) in the EBPBF and the HIP temperature being below the supersolvus temperature may have both contributed to successful HIP processing improving overall part density while avoiding transformation induced porosity.

A key benefit to the LBPBF method is that complex parts can be printed with very high feature resolution. Compared to previous research on a 939Derivative AM alloy, ABD900 may accommodate at least a ∼15 °C increase temperature capability for a given stress in a creep-failure-dominated design. Alternatively, a stress increase of ∼25% could be accommodated at a given temperature. However, other considerations such as creep deformation response (such as time to 1% creep strain) and fatigue behavior also need to be considered.

In addition to increased temperature capability in the build direction compared to other similar materials, a potential benefit to the EBPBF process is that it may provide a method for repair of GT blades and, in particular, those with directionally solidified structures. The EBPBF method can build at high part temperatures of 900 °C and produce directionally solidified structures like grain structures while also minimizing the possibility of strain-age or solidification cracking. Future work is being explored to determine the feasibility of repairing higher ∼70% γ′ vol. fraction superalloys with ABD900-AM using the EBPBF method.

This detailed study has shown that there are many factors that can influence the high temperature creep behavior of AM nickel-base superalloys. ABD900-AM is clearly a candidate for future hot-section gas turbine components based on the comparable high temperature mechanical performance to similar nickel-base superalloys that are traditionally cast. Several different processes were identified to improve overall performance. A super-solvus heat treatment improved creep rupture strength by ∼3× in the LBPBF material. HIP cycling was effective in EBPBF produced material and increased overall rupture time for low stress conditions by ∼2×. EBPBF produced material in the longitudinal direction produced columnar grains similar to directional solidified structures and exhibited the highest overall creep strength, matching closely to that of IN939 castings. The two different build methods may be useful for various applications, such as building complex vanes with advanced cooling schemes using the LBPBF process and tip repairs using the EBPBF method.

The authors would like to acknowledge Vistra Corp. for their participation in rainbow testing that led to a full installation of additively manufactured R1 guide vanes in their F-class turbine fleet. The collaboration between EPRI, PSM, and Vistra has helped set a roadmap for future innovation of hot-section gas turbine technology.

This research was sponsored by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy (EERE), Advanced Materials and Manufacturing Technologies Office under Contract No. DE-AC05-00OR22725 with UT-Battelle LLC and performed in partiality at the Oak Ridge National Laboratory's Manufacturing Demonstration Facility, an Office of Energy Efficiency and Renewable Energy user facility.

This paper has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a nonexclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this paper or allow others to do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan².

AM =

additive manufacturing

EBPBF =

electron beam powder bed fusion

EBSD =

electron backscattered diffraction

GS =

grain size

GT =

gas turbine

LBPBF =

laser-based powder bed fusion

MCR =

minimum creep rate

MW =

megawatt

PFZ =

precipitate free zone

RA =

reduction of area

SEM =

scanning electron microscope

γ′ =

gamma prime phase