Abstract
A nickel alloy of a composition similar to that of the nickel based superalloy Inconel alloy 718 (IN718) was produced with the electron beam melting (EBM) process developed by Arcam AB. The microstructures of the as processed and heat treated material are similar to that of conventionally produced IN718, except that the EBM material showed some porosity and the δ phase did not dissolve during the solution heat treatment because the temperature of 1000°C apparently was too low. Mechanical testing of the layer structured material, parallel and perpendicular to the built layers, revealed sufficient strength in both directions. However, it showed only limited elongation when tested perpendicular to the built layers due to local agglomerations of pores. Otherwise, data for the hardness, Young's modulus, 0·2 yield tensile strength and ultimate tensile strength match those recommended for IN718.
Introduction
Rapid prototyping (RP) or layer manufacturing techniques for polymers have been on the market for about 20 years. Recently, a number of RP techniques for metals and ceramics have entered the market or are being developed.1 The RP techniques are suitable for rapid production of small series of parts possessing complex shapes. Besides electron beam melting (EBM),2 electron beam solid freeform fabrication,3 epitaxial laser metal forming,4 laser engineered net shaping (LENS),5–7 spray forming (SF),8 three-dimensional printing9 and shaped metal deposition using metal inert gas welding10 are some techniques that are being tested for the fabrication of nickel based superalloys. The component is built layer by layer as an electron or laser beam (except for SF) melts the material locally. However, only limited data of mechanical properties of nickel based superalloys processed by these techniques are available.6,7
Conventional Inconel alloy 718 (IN718) is used for a broad range of high temperature applications.11 In a previous study, the microstructure of a material similar to IN718, produced with EBM, was thoroughly investigated.12 These investigations were performed on a material without any subsequent heat treatment. In this study, the evolution of the microstructure and the related mechanical properties were evaluated by comparing them before and after heat treatment. Hardness measurements, compression and tensile tests were performed parallel and perpendicular to the built layers.
Experimental
A nickel based superalloy with chemical composition within the specification limits of IN718 (Ref. 13) was produced with EBM. A 10 mm thick plate of 90×90 mm in cross-section was heated with an electron beam in a bed of alloy powder in a vacuum chamber. A thin layer of the alloy powder of about 0·3 mm thickness was then added onto the plate and melted with the electron beam by scanning the electron beam in parallel lines of 0·3 mm distance. This procedure was repeated, building 0·1 mm thick layers until four blocks of 15×80 mm in cross-section and 80 mm in height were produced. In order to distribute the heat as evenly as possible, the electron beam was scanned over the surface in two fixed scanning directions: perpendicular to each other and in a predetermined pattern. A temperature of ∼1000°C was measured for the starting plate during the build-up using a thermocouple. After processing, the blocks were left in the insulating powder bed in the process chamber and cooled in stagnant helium gas in order to achieve a slow cooling rate. The blocks were cooled down to 100°C before taking them out of the chamber. This material is henceforth referred to as EBM718-4. The material was built in a laboratory EBM machine at Arcam AB, Sweden, and the machine settings were optimised for process stability only. The material was heat treated in a muffle type furnace, Nabertherm, according to the recommendation from Special Metals:13 solution annealing at 1000°C for 1 h followed by rapid cooling in water then precipitation hardening at 718°C for 8 h, furnace cooling to 621°C, aging at 621°C for 18 h, followed by air cooling.
The chemical compositions of the powder and of the as processed EBM718-4 are given in Table 1. The pre-alloyed powder was produced via the plasma rotating electrode process (PREP) by Advanced Specialty Metals. The oxygen content increased during the EBM process, which indicates low vacuum during processing. This was explained by the presence of residual oxygen in the EB vacuum chamber due to a leakage. The processing pressure was 1×10−3 mbar, which is 100 times higher than normal (1×10−5 mbar).
Chemical analyses of pre-alloyed powder and EBM718-4 and limiting chemical composition from Special Metals for Inconel alloy 718 (Ref. 13)
*Chemical data from powder supplier.
†Ni+Co.
‡Not analysed.
For microstructure investigations by light optical microscopy (LOM), scanning electron microscopy (SEM) and X-ray diffraction (XRD), the samples were embedded, ground and polished according to a procedure described in detail in Ref. 12.
For SEM investigations, a CamScan S4 with an energy dispersive spectrometer (EDS) Link ISIS was used. X-ray diffraction was performed on a Bruker AXS with Co Kα radiation equipped with a general area detector diffraction system with a high performance two-dimensional detector. Electron backscatter diffraction (EBSD) in combination with EDS was performed with a TSL/EDAX system in a field emission scanning electron microscope type JSM-6500 F (Jeol) and in a SEM type XB1540, (Zeiss).
Macro- and microhardness measurements, compression and tensile tests were performed on samples of EBM718-4 without and with heat treatment cut parallel and perpendicular to the built layers. Brinell hardness (HB) was measured with a 2·5 mm diameter indenter with a load of 187·5 kPa (1839 N) in a macrohardness tester from Wolpert. Microhardness was measured with a Fischerscope H100 unit with a 136° Vickers indenter on polished samples with 0·5 N load. Points were measured automatically in a grid (five lines with 0·5 mm distance between the lines; 3–5 points per line, 0·75 mm step width). Samples of 5×5×8 mm were cut by electro-discharge machining (EDM) and tested in compression at a strain rate of 10−4 s−1 at temperatures ranging from room temperature (RT) to 1000°C. Tensile tests for evaluating the 0·2 yield stress at RT on the as produced material were performed on cylindrical samples 7 mm in diameter and 35 mm gauge length, according to the Swedish standard SS112111-7A35. The samples were tested in a universal tensile testing machine UTS 200. For the heat treated material, the tensile tests were performed on cylindrical samples with 4 mm in diameter and 20 mm gauge length, according to the German standard DIN 50125-B 4×20. Creep tests were performed on heat treated samples at 800°C in compression and tension. For tensile and compression tests, a universal testing machine from Schenck–Trebel was used. In compression, the load increased stepwise. All tests were performed in air.
Results and discussion
Microstructure
EBM718-4 consist of γ matrix with face centred cubic (fcc) (A1) crystal structure with round fcc MX precipitates, (Nb, Ti)(C, N), B1 type crystal structure and platelike orthorhombic δ phase precipitates, Ni3(Nb, Ti), D0a type of structure, as illustrated in Fig. 1.
Images (SEM-SE) of EBM718-4 after heat treatment showing section perpendicular to layers. Sample consists of γ matrix with round MX precipitates, (Nb, Ti)(C, N), and platelike δ phase precipitates, Ni3(Nb, Ti). Precipitates form rows, and porosity is observed in line with precipitates, perpendicular to layers
The γ matrix has up to millimetre sized large elongated grains with a very sharp (001) texture along the growth direction. Inside the building layers, the [001] directions approximately coincide with the scanning directions of the electron beam during build-up of the sample. This is the same orientation observed before in the EBM718-1 alloy sample.12 In Fig. 2, (111) and (200) pole figures from as produced and heat treated EBM718-4 measured by XRD are shown. The sample cross-section measured is parallel to the built layers with the normal in the building direction. The texture is getting even more pronounced after the heat treatment, as shown in Fig. 2b. The texture sharpness has also been confirmed with EBSD based orientation microscopy (see Fig. 3).
Representative (111) and (200) pole figures taken from EBM718-4 a, b as processed and c, d after heat treatment recorded with XRD
Inverse pole figure (IPF) of EBM718-4 sample recorded with EBSD: IPF colour legend of γ matrix shows that almost all grains are orientated in (001)||ND direction (indicated by the reddish colour of all grains). δ phase can be seen as platelike precipitates, and IPF colour legend gives orientation of the precipitates. All green precipitates have the same orientation. Grain boundaries that are marked black fulfil known orientation relationship between γ matrix and δ phase: {111} γ ||(010) δ ; 〈1¯10〉 γ ||[100] δ
The platelike precipitates, possessing a length of <10 μm and a thickness of some 100 nm, were confirmed with EBSD and EDS analysis to be δ-Ni3(Nb, Ti) phase with orthorhombic D0a structure (Fig. 3). The orientation relationship between the δ phase and the γ matrix is {111} γ ||(010) δ ; 〈110〉 γ ||[100] δ , which is in agreement with previous reports.14–16 Face centred cubic (Nb, Ti)(C, N) with B1 structure and round morphology (<1 μm in diameter) are also present in the microstructure and were analysed with EBSD and EDS, and the results are presented in Fig. 3. As expected, the fcc MX precipitates are orientated in the same crystal directions as the fcc γ matrix. Some TiN precipitates (<5 μm in diameter) were detected in the microstructure of the γ matrix.
Local agglomeration of porosity could be observed in all samples. The pores are arranged in strings perpendicular to the layers (Fig. 1b). Looking at a section parallel to the layers, the pore shape is round, and the pores are systematically arranged and seem to originate from a former dendrite microstructure. The precipitates also form rows, in line with the porosity, perpendicular to the layers (Fig. 1). The average distance between the rows is 5–10 μm. The porosity and precipitates most likely form in the spacing between dendrites or cells, which means as the last area to solidify.
The microstructure and constituting phases of EBM718-4 are similar to those described for EBM718-1,12 namely a γ matrix with oriented γ′ Ni3(Al, Ti, Nb) precipitates of 2–5 nm, and γ″ Ni3Nb precipitates of 5–10 nm. In addition, 0·2–2 μm sized precipitates of MX (Nb, Ti)(C, N) aligned in rows along the growth direction were detected. The distance between the rows in EBM718-1 was 10–20 μm, i.e. larger than in the present material. In addition to the phases listed above, a hexagonal Laves phase (Ni, Fe, Cr)2(Nb, Mo, Ti), δ phase Ni3(Nb, Ti) or the σ phase CrFe could be present in the material, but none of these phases has been detected in the previous investigation.12 All these topological closely packed phases are undesirable, as they may have a detrimental effect on the mechanical properties.15 In case of the δ phase, the degradation has been attributed to a combined result of the loss of niobium from the matrix plus the γ″ coarsening that accompanies the δ phase formation.15
The occurrence of the δ phase in the current investigated material may be attributed to slight differences in EBM process parameters and/or to differences in the compositions of the pre-alloyed powder between EBM718-1 and EBM718-4. Also for IN718, it has been reported that the solvus temperature for the δ phase, i.e. the temperature above which the δ phase dissolves, may vary.15,16 For example, it has been reported that the solvus temperature increases when the Nb content is raised.17 Possibly, the temperature of 1000°C used in the present study for solution annealing was too low, which could explain why the δ phase is present in EBM718-4. Actually, XRD analysis of the highest peaks of the δ and of the MX phases showed that the volume fraction of both phases has increased after the heat treatment (see X-ray spectrum in Fig. 4).
X-ray diffraction pattern from EBM718-4 a as processed and b after heat treatment. Integrated signal from all measured angles
Hardness
The Brinell hardness of EBM718-4 increased from ∼340 HB in the as processed condition to ∼430 HB after the heat treatment (Table 2). The microhardness increased as well from ∼500 to ∼640 HV after the heat treatment. These differences show the effectiveness of the precipitation hardening treatment at 718°C for 8 h. No significant difference in macro- and microhardness was observed for measurements parallel and perpendicular to the layers. In addition, no difference in hardness was observed between the bottom (starting layers) and the top (initial layers) of the EBM processed sample. The Young's moduli derived from the microhardness measurements are 180±6 GPa for the as processed material and 198±8 GPa after the heat treatment.
Brinell hardness HB (d = 2·5 mm, load = 187·5 kPa) and microhardness HV (load = 0·5 N) and Young's modulus derived from microhardness measurement of EBM718-4
*Average of 10 measurements.
†Average of 25 measurements.
The hardness of 450 HV (recalculated from Brinell hardness) for EBM718-4 after the heat treatment is higher than the minimum reference value of 355 HV given by Special Metals for IN718.13 However, the Young's modulus of IN718 of ∼200 GPa13 is similar to that found for EBM718-4.
Yield strength
The 0·2 yield tensile strength (YTS) has been evaluated in compression and tension, and the results are illustrated in Fig. 5. The difference between compression tests parallel and perpendicular to the layers is small for both the as produced and the heat treated material. After the heat treatment, the compressive yield strength increases considerably below 800°C due to precipitation hardening. At 800°C, the YTS of all samples is about 550 MPa but at 1000°C less than 50 MPa.
Yield strength as function of test temperature of EBM718-4 tested in compression (C) and in tension (T)
When tested at 400°C, some of the specimens showed serrated yielding. Serrated yielding has previously been reported for IN718 at this temperature.18–20 As a rate controlling mechanism for dynamic strain aging at this temperature, the migration of interstitial solutes, most likely of carbon, has been proposed.
Results from tensile tests at RT of EBM718-4 in the as processed condition and after heat treatment show that the YTS and the ultimate tensile strength (UTS) increase after the heat treatment (see Fig. 5). The elongation and the area reduction were much lower perpendicular to the layers than parallel, as described in Table 3. From studying polished sections next to the fractures (parallel to the pulling direction) and the fracture surfaces (Figs. 6 and 7), the conclusion is drawn that local groups of porosity led to the low elongation in those specimens that were tested perpendicular to the built layers (Fig. 7). These local groups of porosity are arranged in parallel strings, perpendicular to the built layers (Fig. 1b), which gives a weak area when pulling perpendicular to the layers. After tensile testing, large voids can be seen, which were not present before testing (Fig. 7a).
Specimen from as heat treated EBM718-4 after tensile testing parallel to built layers
Specimen from as heat treated EBM718-4 after tensile testing perpendicular to built layers: a image (LOM) taken from polished section showing area next to fracture; b–e SEM images (SE contrast) of fracture surface. Location of magnified areas shown in d and e are indicated in b. In area d, fracture occurred in local group of voids, leaving particular surface visible in d. Area e, which is on different level than area d, has dimpled surface, which indicates ductile fracture where the material broke outside of arrangement of pores
Results of tensile tests perpendicular and parallel to layers in as processed condition and after heat treatment
Minimum tensile strength values from Special Metals for IN718 at RT after the standard heat treatment used in this study are UTS of 1241 MPa, YTS (0·2 offset) of 1034 MPa, elongation to failure ϵf of 6–12 and reduction of area of 8–15.13 Those requirements are met for the heat treated EBM 718-4, except that it shows lower elongation and area reduction in the direction perpendicular to the layers due to local agglomerate of pores.
At high temperatures, data provided for IN718 (hot rolled bar material, annealed at 980°C and aged) by Special Metals for the YTS are 550 MPa at 800°C and about 120 MPa at 1000°C.13 Data of the compression tests (Fig. 5) show that, at 800°C, the electron beam deposited material EBM 718-4 has the same strength but exhibits a considerable lower strength of less than 50 MPa at 1000°C. Perhaps, this is caused by softening due to porosity.
Creep
Figure 8 shows the results of the creep tests, which were carried out on heat treated material in compression at 800°C by stepwise increase of the load. Creep tests were performed parallel to the layers (test duration, 368 h) and perpendicular to the layers (413 h). In both cases similar values were obtained, and for both samples, the stress exponent n, as derived from Norton's power law for creep, changed from about n = 4 at lower stresses to about n = 10 at higher stresses. A change in stress exponent in dependence of the applied stress, i.e. an increase of n at higher stresses, has been frequently observed in precipitation strengthened superalloys. The increase has been attributed to an additional back stress at high stresses, caused by precipitation coarsening.21 In addition to the compression samples, a number of samples for tension testing have been prepared by EDM. All of them broke, either at the threads of the sample or at the measuring length, within the first 70 h of the creep test, i.e. before secondary creep was attained. Again, it is believed that porosity causes the premature failure.
Logarithm creep rate ˙ϵtrue versus logarithm creep stress σtrue. Compressive creep tests were carried out at 800°C with stepwise increased load. Material was tested parallel and perpendicular to built layers. For both materials, a stress exponent of n = 4 is observed at lower stresses, which changes to n = 10 at higher stresses
Comparison of EBM718-4 with IN718 produced by different RP techniques
Because RP techniques become more readily available, a number of investigations have been undertaken where IN718 has been processed by SF17 or laser rapid forming (LRF; also known as LENS).6,7 In case of the SF material, specifically, the precipitation behaviour of the δ phase has been studied in greater detail. It was found that, compared to wrought IN718, the SF material contained a larger volume fraction of δ phase precipitates, which dissolved at about 1030°C. This is in line with the microstructural observations on the present EBM718-4, which also contained a considerable volume fraction of δ phase that was not dissolved by heat treating the material at 1000°C.
The as heat treated EBM718-4 shows a YTS, which is of the same order of magnitude as the same heat treated material manufactured by LRF/LENS.6,7 However, the elongation and area reduction are lower for the EBM718-4 material. Like EBM718-4, the LENS material shows very poor elongation when tested perpendicular to the built layers.6 After hot isostatic pressing, the LENS material showed higher yield strength, which is explained by an improved cohesion of the successive layers of the deposit.
Conclusions
It has been shown that nickel based superalloys like IN718 can be successfully produced by EBM processing. Despite some porosity stemming from the EBM process, where the process parameters have to be optimised, and the presence of δ phase, EBM718-4 shows a promising mechanical behaviour in that most data match those of the respective recommendations.13 The mechanical properties will be improved by optimising the EBM process for IN718 in order to minimise the porosity. Furthermore, a higher solution annealing temperature may eliminate the δ phase precipitates.
Footnotes
Acknowledgements
The authors would like to acknowledge the support of Gerhard Bialkowski for performing mechanical testing and Dr Stefan Zaefferer and his group for texture measurements. A special thanks also to Arcam AB for processing IN718 by EBM for this investigation.