1. Introduction
Additive manufacturing (AM) has brought new capabilities for the development of Functionally Graded Material (FGM) in the past decade [
1,
2,
3,
4,
5,
6]. Unlike the conventional welding and cladding approach for multiple materials, AM makes it easier to deposit mixed powder and tailor location-specific material chemistry, microstructure, and properties of FGM structure. Thus, it offers a new degree of freedom in material and component design that could lead to innovations beyond state-of-the-art material/manufacturing technologies. Several AM modalities, i.e., directed energy deposition (DED), laser powder bed fusion (LPFB), electron beam melting (EBM), cold spray, etc., have demonstrated the feasibility of manufacturing FGM coupons or prototypes [
1,
2,
3,
4].
The motivation of the present work is to develop a graded IN718-René41 (718-R41) superalloy structure by laser blown-powder directed energy deposition (DED-LB/M) with balanced performance and cost, which potentially replaces a single cast/wrought alloy (with R41 equivalent temperature capability) or dissimilar metal weld structure. 718-R41 combination is chosen due to established DED build parameters for 718 and the availability of the commercial powder of the two alloys. DED-LB/M additive modality is chosen to manufacture FGM coupons owing to its flexibility in mixing powders from different hoppers to custom gradient compositions. The structural components operated in temperature gradient require higher temperature capability at one end (i.e., up to 900 °C) and lower temperature capability at the other end (i.e., up to 600 °C). Compared to using a higher temperature superalloy (900 °C capability) for the whole structure or joining dissimilar alloys, a compositionally graded superalloy structure offers several advantages, such as matching alloy capability to thermal gradient, reducing component cost by allowing the use of a less expensive alloy, eliminating welds or mechanical joints, and allowing gradual transition in physical properties. However, developing compositionally graded superalloys also faces several technical challenges, such as the compatibility of terminal materials and gradient path design, cracking susceptibility of medium to high γ′ superalloys, additive manufacturing process control to minimize defects, and heat treatment optimization for mechanical properties.
To the best of our knowledge, the published research on graded superalloys is quite limited [
7,
8,
9,
10,
11,
12,
13,
14] compared to the number of publications for other graded dissimilar alloys. Although the existing literature shows the feasibility of creating a unique graded superalloy or microstructure, the limitation of superalloy FGM is not well addressed, and microstructure–property relationship within the graded zone is not fully understood. Several papers have shown success in the microstructure grading of 718 by modifying AM process parameters such as heat input and scanning strategy, which drastically change grain structure (columnar vs. equiaxed grain), grain size, texture, and dendrite orientation [
7,
8]. Grain size grading, i.e., fine grain size at the bore and coarse grain size at the rim of a turbine disk in 718 was achieved by advanced heat treatment after forging to balance tensile, creep, and fatigue properties at the bore and rim of the disk [
9]. A gradient of nano to micron grain size as well as residual stress was generated at a surface depth of ~500 µm using an ultrasonic surface rolling process (a severe plastic deformation method), resulting in a significant increase in fretting fatigue life at 600 °C [
10]. Directional recrystallization heat treatment with varying draw rates produced a graded structure of fine equiaxed grains and columnar coarse grains in LPBF IN738LC to balance location-specific fatigue and creep properties [
11]. In terms of compositional grading, a complex gradient pattern between two superalloys was made by powder DED for turbine blisk application, and the selected heat treatment achieved tensile properties comparable to the terminal alloys in cast form despite a creep property debit [
12]. Although hardness profile across the gradient zone and bulk tensile curves were reported, the paper did not investigate microstructure response to heat treatment and local deformation within the gradient zone. IN625 coating with a gradient of γ′ precipitates in 10–80% was in situ synthesized by plasma-transferred arc deposition using a mixture of IN625 powder and Ni, Al elemental powders [
13]. This graded IN625 coating was reported to show good weldability, improved hardness, and oxidation resistance compared to the IN625 bulk. Compositional grading in a superalloy from low to high γ′ was built by powder DED as a high throughput method for superalloy design and property screening [
14]. However, this work did not intend to create a graded superalloy structure and perform a detailed evaluation of its gradient zone.
In a previous study, we fabricated crack-free single-pass thin wall specimens in 718/R41 graded composition using DED-LB/M and mapped the residual stress profile [
15]. R41 is a medium γ′ alloy (~30% volume fraction of γ′) with up to 900 °C temperature capability and cost, whereas 718 is a γ′/γ″ strengthened alloy with up to 650 °C temperature capability and cost. Lately, defect-free 718-R41 graded thick blocks have been successfully built using the contour and hatch scan strategy. In the present work, we experimentally investigated tensile properties within the graded region to improve the understanding of the structure–property relationship and evaluate a baseline heat treatment condition. The composition-dependent stress–strain response of 718-R41 graded alloy was determined by in situ Digital Image Correlation (DIC) and tensile test. Diffraction elastic constants (DECs), grain and phase level yielding, and load partition as a function of composition were quantified using in situ neutron diffraction for the as-built and heat-treated specimens. The influence of heat treatment on microstructure and tensile behavior was discussed. This work presents novelty in revealing microstructure and mesoscale deformation of graded superalloys fabricated by DED-LB/M.
2. Materials and Methods
Plasma-atomized 718 alloy powder in the powder size range of 45–150 µm was acquired from AP&C (Colibrium, Saint-Eustache, QC, Canada). Gas-atomized R41 alloy powder in the powder size range of 44–125 µm was acquired from Linde (previously known as Praxair, Indianapolis, IN, USA). Alloy powder compositions measured by inductively coupled plasma mass spectrometry and LECO combustion testing are described in
Table 1.
718-R41 graded blocks were built by a RPMI 557 DED-LB/M AM system which melts and deposits powder using a continuous wave fiber laser capable of 3 kW focused to a beam diameter between 1.27 and 4 mm. Two hoppers containing 718 and R41 powders allowed the mixing of powder in the desired ratio for deposition. An argon gas environment with oxygen less than 10 ppm was applied during the building. Each block dimension had a build height of 133.3 mm, a length of 45.7 mm, and a width of 15.2 mm. The contour and hatch strategy was adopted to first build the outer contour (45.7 × 15.2 mm) and then build the interior material by hatching passes. The detailed build parameters are summarized in
Table 2. The DED build parameters were down-selected from machine learning-guided build parameter optimization, with the criteria of no lack of fusion, minimal porosity, no cracking, and good powder mixing for uniform composition. A wrought 718 build plate was used and 718 powder was first deposited to reduce thermal expansion-induced residual stress in the initial built material. The build process was performed at ambient temperature without heating the build plate or part. Monolithic 718 in 54 mm height was first built. The next 25.4 mm height (approximate mid-height of each block) contained 4 graded compositions in equal height with a decreased amount of 718 and increased amount of R41 by weight percent: 80% 718 20% R41 (80–20 mix), 60% 718 40% R41 (60–40 mix), 40% 718 60% R41 (40–60 mix), and 20% 718 80% R41 (20–80 mix). Monolithic R41 in 54 mm height was built on top of the graded region. The linear composition grades were predicted by Thermo-Calc to have minimal detrimental brittle phases (topologically close-packed phases, σ and µ) along the gradient path, and thus were chosen for build. These build parameters were verified to produce crack-free materials by X-ray CT scan and metallography on cut-up sections.
After the DED build, a baseline heat treatment (not optimized) was applied on one block as described in
Table 3. The block was positioned upright during the heat treatments. The stress relief temperature was selected to be above the γ′ solvus temperature of R41, above the δ solvus temperature of 718, and below the solidus temperatures of the terminal and graded alloys. It served as stress relief as well as homogenization to remove chemical segregation and Laves phase at the interdendritic region in the 718-rich compositions. There was no part distortion after the stress relief heat treatment. A hot isostatic pressing (HIP) cycle was applied to close the porosities in the as-built material. A super-solvus solution was intended to dissolve γ′ and γ″ precipitates in the entire block. The two-step aging condition was modified based on the time–temperature–transformation (TTT) diagrams of wrought 718 [
16] and heat treatment protocols of R41 [
17] to generate γ′ + γ″ precipitates in 718-rich compositions without going into the δ phase region.
Figure 1a,b show the as-built block and heat-treated block that had obvious surface oxidation in the 718 region. Screw-threaded cylindrical tensile specimens were machined along the build height (
Figure 1c,d). The graded region was in the center of the specimen gauge section.
Micro X-ray fluorescence (μXRF) was utilized to semi-quantify major composition profiles along the gauge section of the tensile specimens. Composition transitions were identified by the distinct signals of Fe, Co, Ti, and Nb elements. Lab tensile tests were performed to failure at 650 °C in lab air using an MTS servo-hydraulic loading frame and a clam-shell furnace. A test temperature of 650 °C was chosen as the operating temperature limit for 718. A constant crosshead displacement rate of 0.5 mm/min was applied. A high-temperature extensometer with a 25 mm initial length was applied to the graded region to record strain data. The stress–strain curves of the as-built and heat-treated specimens were collected to help determine the loading scheme for a neutron diffraction experiment. After tensile rupture, the specimen halves were measured by μXRF again to determine the failure location with respect to composition.
Separate lab tensile tests coupled with in situ DIC were also performed at the same temperature and with the same displacement rate control. The purpose was to provide the direct observation of non-uniform deformation in the graded material and quantify stress–strain responses within each composition zone. A furnace with an elongated viewport was picked to allow DIC access to the entire gauge length. High-temperature ceramic paint was applied along the gauge length except for a bare region used as a fiducial mark that identified composition locations. A dummy test using wrought 718 with a thermocouple placed near the center gauge was conducted to analyze thermal strain and verify the temperature uniformity of the entire gauge section at 650 °C. The actual DIC/tensile tests for the as-built and heat-treated specimens were interrupted at a few percent macroscopic engineering strain. The composition profile measured by μXRF was used to locate the six composition regions within the calibrated DIC X-coordinates from initial room temp positions. Strain as a function of composition was averaged from the uniaxial strain collected within each region, excluding the composition transition areas. Speckle paint started to spall and data quality suffered when applied stress was above 720 MPa for the as-built specimen and 860 MPa for the heat-treated specimen. Therefore, high-stress data for the specimens were chopped off.
Electron Backscatter Diffraction (EBSD) analysis was performed on the as-built and heat-treated materials to provide grain structure/morphology, grain orientations, retained plastic strain, and the fraction of recrystallization after heat treatment. A Hitachi SU-70 FEG-SEM (Hitachi High-Tech America, Inc., Schaumburg, IL, USA) with an Oxford Aztek Symmetry EBSD detector (Oxford Instruments, High Wycombe, UK) was used. An accelerating voltage of 20 kV and a step size of 2 μm were applied. The scanned area was a strip of 1 mm width along 37–38 mm build height where a 25 mm graded region was approximately centered along the height. Grain boundary maps, inverse pole figures, pole figures, and kernel average misorientation (KAM) maps were collected and analyzed. DED materials contained large grain in hundreds of μm to several mm and EBSD covered a relatively small sampling area. Therefore, the quantification of grain size was not conducted given insufficient grain statistics. The high-resolution backscattered electron images of representative microstructure were taken by Hitachi SU-70 FEG-SEM.
Neutron diffraction coupled with in situ loading is a unique experimental approach that provides the mesoscale elastic and plastic behavior of matrix and secondary phases in the constrained state, whereas other conventional methods are not able to probe [
18,
19]. Additionally, it offers deep penetration in Ni-based alloy, a large beam size beneficial for capturing statistics of large grain materials such as DED-built alloys, and the ability to detect superlattice reflections of the γ′ phase. Mesoscale deformation behavior at grain- and phase levels was studied by in situ neutron diffraction at BL-7, the VULCAN beamline at Spallation Neutron Source, Oak Ridge National Laboratory [
18,
19]. The experimental setup is displayed in
Figure 2. The in situ tensile loading and unloading at 650 °C was performed with interrupted engineering strains of 1% and 4.8% for the as-built and heat-treated specimens, respectively. The specimen was held horizontally on a customized MTS loading frame. The specimen axial direction was positioned at a 45° angle with respect to the incident neutron beam. Neutron diffraction data along and transverse to the loading directions were collected by two area detectors (Banks 1 and 2) positioned at ±90° with respect to the incident neutron beam. Neutron experimental parameters included the high-resolution mode, a chopper speed of 30 Hz, a data collection of 5 min at each location, and a defined beam size of 5 mm × 5 mm horizontal and 5.5 mm height (irradiated volume of ~137.5 mm
3) to prevent diffraction signals from thermocouple wires. A custom induction heating coil was placed several mm above and below the gauge section to heat the specimen. The macroscopic strain was measured by a high-temperature extensometer with a 12 mm gauge length that was positioned at the center of the gauge section (two ceramic rods in contact with the specimen gauge top in
Figure 2a). Based on the measured μXRF profiles along gauge length, thermocouple welding locations and neutron beam probing locations were determined for the six studied compositions. Before loading the specimen, six pairs of K-type thermocouple wires were spot-welded to each composition region. One thermocouple at the center was assigned as a control thermocouple (set point of 650 °C), and the rest five thermocouples were assigned as readout thermocouples. The thermocouples shown in
Figure 2a from left to right were TC1 at 718, TC2 at 80–20 mix, control TC at 60–40 mix, TC3 at 40–60 mix, TC4 at 20–80 mix, and TC5 at R41. The grips of the loading frame were water-cooled. The specimen grip section had a lower temperature, which created a temperature distribution along the gauge length. For the as-built specimen test, TC1 and TC5 had stable temperatures of 580–590 °C, while TC2, TC3, and TC4 had stable temperatures of 630–640 °C. For the heat-treated specimen test, a stable temperature of 610–640 °C at various thermocouple locations was obtained. Stepwise loading was applied during the in situ neutron measurements, which included 4–5 loading levels within the elastic regime, several loading levels within the elastoplastic and plastic regimes, and 5 levels during unloading. At each loading step, the load frame was translated on the VULCAN sample stage along the loading direction to measure, respectively, the nominal graded region determined from the μXRF measurement.
The diffraction peaks of (111), (200), (220), and (311) grains were obtained. These fundamental reflections had overlapped γ and γ′ peaks. Single peak fitting was conducted to extract peak position, intensity, and width as the representation of grain orientations that contain the γ and γ′ phases. Superlattice peaks from the γ′ phase including (210), (211), and (300)/(221) were only observed for R41 and R41-rich compositions. γ′ superlattice peaks in 718-rich compositions were not observed. Given the limit of lattice spacing with neutron instrument setup, the γ′ (100) peak was outside the measurement range. In theory, (300) and (221) reflections of γ′ share the same lattice spacing and both contribute to the diffraction intensity of the same peak. However, they behave differently in elastic and plastic responses. Orientation parameter
Ahkl (
) of (300) and (221) are 0 and 0.296, respectively. Therefore, it was considered not appropriate to use the γ′ (300)/(221) peak to determine the γ′ peak position at (200) and further deconvolute the γ/γ′ (200) peak. Instead, the γ/γ′ (200) peak was deconvoluted into γ and γ′ peaks by using the Lebail fit in GSAS [
20] with the consideration of the instrument profile. The presence of some low-intensity γ″ peaks in 718-rich compositions did not result in a reliable fitting, and thus was not reported.