Final Report Summary - WELDMECS (Welding Metallurgy and Cracking in Superalloys)
Predictive, metallurgy informed crack criteria was developed within the Weldmecs project. These crack criteria were developed in a format appropriate for implementation to finite element software. It will be useful for general thermo-mechanical simulations estimating the risk for various crack formations. The primary target in current topic description was to utilize the crack criteria for both welding and heat treatment simulations. Since cracking during welding or heat treatment is a many-faceted problem, care has been taken to strongly incorporate fundamental metallurgical understanding and testing into the models. The model was based on actual failure mechanisms, metallurgical changes and their relations to the material response and prevailing thermo-mechanical conditions. Up to date – simulations have focused on geometric tolerances and/or prediction of residual stresses. Only simpler, limited defect criteria have been used in this context. The current project was expected to advance the modeling of manufacturing processes in the area of predicting crack formations.
A review of relevant literature, especially with regard to crack criteria, has been made. A detailed investigation, to understand the microstructure of the selected materials, the development of microstructure during welding and its effect on cracking has been carried out.
All materials have been heat treated according to plan. Evaluation of microstructure and hardness of as-received and solution heat treated material, using optical microscopy, scanning electron microscopy and hardness measurements has been made. Test samples have been manufactured according to plan. Gleeble testing and Varestraint testing of samples have been made as well as differential scanning calorimetry measurements. Simulation routines of Varestraint and Gleeble testing have been made
Project Context and Objectives:
The main objectives of the project has been to use a number of testing methods to be able to judge the weldability of some nickel-based alloys of main interest to jet engines, with special attention to cracking of the alloys, and then combine this with modelling and simulation of the structures. Through this, the aim is to be able to more quantitatively describe the risk for high temperature cracking of objects to be used in aerospace engines.
Management meetings were held regularly. In total, ten meetings were held, where the situation regarding the project progress was reviewed, as well as the results obtained at the various sites.
In the technical work, first, a literature review was made where fundamental concepts of the super alloys, proposed to be used in the project, were described in detail. The alloys nominated for the investigation were 718, 718Plus, and Haynes 282. Then, the cracking criteria, especially for hot cracking of such alloys, were reviewed. Nine different criteria are described in some detail. This is then followed by descriptions of different kinds of fracture criteria, for example concepts like total crack length or maximum crack length. Finally, some words are spent on Strain age cracking criteria.
Apart from this literature survey, the main activities has been made on different testing techniques, modelling of Varestraint testing and validation of cracking criteria.
Three different analytical methods were used for testing the materials in different ways: Gleeble testing, Varestraint testing and Differential Scanning Calorimetry (DSC). Prior to performing these tests, the material was heat treated to two different conditions, which varied slightly between the alloys. 718 and 718Plus were solution heat treated at 954°C for 1 h and tested in this condition. To simulate coarse grains in the microstructure, these two alloys were also heat treated at 1050°C for 3 h, followed by the solution heat treatment at 954°C for 1 h. The third alloy, Haynes 282 was either solution heat treated at 1121°C for 30 minutes or treated to have a coarse grain size by heat treating at 1150°C for 2 h.
Gleeble and Varestraint testing were made on both fine and coarse condition of each of the alloys. DSC testing to disclose solid state phase transformations for long dwell times (>40 hours) were also made but it was soon found out that no real information was gained by this method. A few tests were made, to investigate the solidification process of the different alloys.
Gleeble testing was made at Swerea-Kimab in Stockholm. A large number of specimens on all three materials were tested, mainly at 700, 800 and 900°C, with some local variations of temperatures. The results showed that 718 had a minimum in ductility at 800°C, while the other two alloys had their minimum in ductility at higher temperatures, around 900°C. It was difficult to distinguish between these alloys when it comes to strain-age cracking. An effort was also made on materials with coarser grains. However, no significant effect was seen of grain size on ductility.
Varestraint testing was made on about 65 different specimens, with two different thicknesses. For the three main alloys a thickness of 3.2 mm was used. The bending radius was either 100, 200, 400 or 500 mm. Some differences could be noted between the different materials. Mostly, very little cracking was seen until the highest augmented strain. 718Plus actually showed no cracking until an augmented strain of 1.1%, while Haynes 282 showed some small cracking for all augmented strains. 718 was in between these two alloys. There also seemed to be more cracks in 718 and 718Plus in the coarse-grained condition, while Haynes 282 was not very sensitive to this.
A significant part of the work was also to characterize the microstructure of the three alloys, both on as-received material, after heat treatment and after testing, using advanced microscopy and microanalysis.
The as-received materials contained a small number of large (typically 5 µm) primary carbides, often located in groups of several carbides. The heat treatment of 718 and 718Plus resulted in the formation of precipitates in the range 0.1-1 µm, which were mainly δ-precipitates. In 718Plus a large number of very small precipitates (~10 nm), presumably γ´ or γ´´, seemed to have formed, although they were too small to be clearly resolved by SEM. The heat treatment of Haynes 282 resulted in carbide precipitation at grain boundaries and in precipitation of a dense dispersion of γ´ with size of 24±7 nm.
The microstructure of samples subjected to Gleeble and Varestraint testing was investigated using SEM but also atom probe tomography (APT) and transmission electron microscopy (TEM). Prior to Gleeble testing the samples underwent a HAZ simulation heat treatment. The samples studied here had been Gleeble tested at three different temperatures; above, at and below the ductility dip temperature. In all cases, the fractures followed grain boundaries. Small precipitates, presumably carbides, were frequently observed at the grain boundaries, in particular in the samples tested at the higher temperatures and more so in Haynes 282 and 718 than in 718Plus. In the samples tested at the highest temperature, 900°C, a dense distribution of small precipitates (20 nm) could be identified using SEM. This was confirmed by APT for 718 (which was the only 900°C-tested alloy that was analyzed with APT). Using APT a dense distribution of very small (5 nm) precipitates enriched in Nb, Ti and Al, could be observed in all three alloys tested at 800°C (718 and Haynes 282), or 850°C (718Plus). In the Varestraint samples cracks running perpendicular to the welding direction, usually starting in the weld and continuing into the HAZ, were observed. These cracks were associated with a eutectic consisting of Nb-rich Laves phase and the γ-matrix. The Laves phase was identified using electron diffraction in coarse-grained 718Plus.
Complementing the experimental investigations reported above, two simulation activities have also been carried out. The first one was a thermo-metallurgical-mechanical finite element model of the Gleeble and Varestraint welding tests and simulations of the tests have been performed. The simulations predict the evolution of strains, stresses and temperatures in different regions of the material. They also predict the risk of warm and hot cracking with the aid of two models, one for warm cracking (simulating Gleeble testing) and one for hot cracking (simulating Varestraint testing).
The implemented hot cracking criteria is based on the following assumptions:
• The material cannot carry any tensile stress above the coherent temperature.
• Cracks cannot form at temperatures above the coherent temperature.
• The material cannot carry any real load above the solidus temperature.
• Cracks form above the solidus temperature.
• Cracks only form when the temperature is decreasing.
For warm cracking the following criteria was used
• The material will crack when a typical amount of plastic work is reached.
• There is a temperature region where the material is extra sensitive to plastic deformation due to metallurgical phenomena.
Note that even if the specimen breaks in the test, it never breaks in the simulation. The formation of the crack is not simulated numerically, but an indicator of the correct point in time that the specimen is about to break is achieved.
The criteria’s were implemented into the commercial finite element code MSC Marc.
The hot crack criteria are based on the theory for solidification of metallic materials alloys. It is assumed that there is a range of temperature, above the solidus temperature, where the material is extremely sensitive to plastic deformation. The model takes this as an indicator and converts the information info a parameter that indicates the risk of a hot crack to form. Input data for the model can be evaluated from Varestraint tests performed within other parts of the project.
The criteria for warm cracking are based on the amount of plastic work. The basic assumption is that a metallic material only can withstand a certain amount of plastic work before different kinds of defects starts to develop in the material. It is also assumed that the critical amount of plastic work depends on the temperature of the material. The model calculates the plastic work done at different temperatures and converts the information into a parameter that indicates the risk for warm cracking to occur. The input parameters for this model can be evaluated from Gleeble tests performed within other parts of the project.
Simulations using MSC Marc with the incorporated crack criteria’s show that the warm and hot crack indicators can be used to get a reasonably good indication for where and when a crack is likely to form.
Simulation of Gleeble testing
The result from the simulations of the Gleeble tests can be summarized in Figure 124 in the full report in the Appendix. This figure show the relationship between the calculated area reduction at fracture, and test temperature for all six materials tested (three materials, two heat treatments each). The area at fracture is calculated from the deformed geometry at the time of fracture for all simulations presented under point 3 in the full report. The time of fracture is the time when the warm crack indicator reaches the value 1.0. One dot represents one test and 33 simulated Gleeble tests were performed. The figure should be compared with the test result found in other project reports. The similarity is impressive. The calculations catch the shift from ductile to brittle fracture over the tested temperature range for all materials. Also the difference among the materials is described correctly.
Simulation of Varestraint testing
The result from the Varestraint simulations is shown in figure 125, 126 and 127. Figure 125 shows the hot crack indicator as a function of anvil radii used in Varestraint tests. A small radius of the anvil leads to large deformations (strains) in the specimen and increase the risk of hot cracks to occur. In Figure 126 the calculated hot crack indicator is plotted against augmented strain in the test. This is almost how the result is presented in the Varestraint test report, and this figure is therefore easy to compare directly with measurements from the test. Observe that the hot crack indicator is never mentioned in that report. Instead the total detected crack length is used.
The augmented strain is a theoretical strain on the top surface of the test plate.
The real total strain may differ somewhat from the augmented strain due to unexpected effects of the Varestraint test. In reality the centre of the test plate may not be in contact with the anvil. Instead two contact regions slightly offset to each side of the centre of the plate may appear. This will influence the strain distribution on the top surface of the plate. But from the simulations we know the actual total strain as well as its elastic and plastic components. It is therefore possible to plot the hot crack indicator as a function of plastic strain. This is done in figure 127, but the information is basically the same as in figure 126. The calculations catch the sensitivity to hot cracks for all materials and the difference among the materials is also described correctly.
Note that the numeric values for the hot crack indicator shown in figure 125 to 127 are taken from the maximum value found on the top surface of the plate in the tests. The figures can be found in section 4 of this report, but the maximum values is not read directly from the figures involved, but are found automatically by computer aid during the evaluation of the simulations. This is also true for the effective plastic strain of each plate. The value is taken on the top surface of the plate, half way between the weld and the edge of the plate. The total number of test plates (and simulations) is 33 in this case.
As validation of the techniques developed in the project, samples of 718Plus, with a thickness of 4 mm was tested.
Project Results:
This report will be split into the different work package structure, to make it more easily to follow.
Work package 1 Project management
The project management group consisted of the following members:
Lars-Erik Svensson, University West (chairman)
Joel Andersson, University West
Lars Nyborg, Chalmers University of Technology
Mattias Thuvander, Chalmers University of Technology
Jonas Edberg, Luleå Technical University
The project started 130301 and ended 150831. The management group had ten meetings (130305 (Kick-off meeting), 130603, 130821, 131111, 140217, 140422, 140610, 150206, 150409 and 150612).
All management group meetings contained a review of the present position of the project, what tasks that had been fulfilled, the outcome of these tasks and how this reflected on the other stages of the process. The remaining work, according to the project planning, was always mentioned as a last point of the meeting. All meetings were held in a very constructive atmosphere. All minutes and most of the presentation material from meetings are made in Swedish and therefore not suitable for presentations here.
Work package 2 Literature review and survey of weldability criteria for super alloys
This work package was mainly carried out by University West. The literature survey conducted can be found wholly in the Appendix. Parts of it is presented here.
Literature Review and Selection of Crack Criteria
2.1 Introduction
Superalloys are heat-resistant alloys based on nickel, nickel-iron, or cobalt that exhibit a combination of mechanical strength and resistance to surface degradation [1]. Superalloys are primarily used in gas turbines, coal conversion plants and chemical process industries and for other specialized applications requiring heat and/or corrosion resistance [1]. The modern high-performance aircraft engine could not operate without the major advances made in superalloy development over the past 60 years [1]. Nickel alloys being commercially used today and the ones at development stage range from single-phase alloys to precipitation-hardening superalloys, oxide dispersion strengthening alloys and composites [1]. Actually, elements such as cobalt, chromium, iron, molybdenum and tantalum are used as solid solution strengtheners since they exhibit similar atomic radii, electronic structure and crystal structure relative to Ni which allows them to remain in solid solution [2].
In addition to their use in aircraft, marine, industrial gas turbines, nickel-based alloys are also used in space vehicles, rocket engines, experimental aircraft, nuclear reactors, submarines, steam power plants, petrochemical equipment and other high-temperature applications [3]. Nickel-based alloys are also known as some of the most difficult-to-machine/weld [4, 5]. The principal characteristic of nickel as an alloy base element is the high phase stability of the face-centered cubic (fcc) nickel matrix. The surface stability of nickel is readily improved by allowing with chromium and/or aluminium [1].
2.2 Microstructure of nickel-based alloys
This chapter provides a general background of nickel-based superalloys, including the different phases which can be found in these alloys. The full description is found in the full report.
2.3 Base materials
Alloy 718 is the predominant nickel-iron base superalloy, representing almost half of the total tonnage of superalloy used throughout the world [29]. The 53% nickel - 19% iron matrix is strengthened mainly by 5.3% niobium that forms γ΄΄ (~18 to 20%) giving alloy 718 a higher yield strength than other superalloys strengthened by an equivalent amount of γ΄ [29]. However, γ΄΄ being metastable can transform to δ after long periods at temperatures at and above ~650 0C resulting in some loss of strength [30-35]. In other words, above ~5%, niobium promotes Laves and δ that in alloy 718 are potentially deleterious to both toughness and strength. The composition of Laves phase can typically tie up 19% of the niobium and 10% of the molybdenum contained in alloy 718 preventing these alloying elements to contribute to strength [29]. In addition to this, Laves phase (Ni2Nb) particles form in cast alloy IN 718 by eutectic-type reaction during the last stage of solidification [36]. Alloy 718 of nickel-based superalloys is an age-hardenable, high-strength and iron-containing alloy suitable for service temperatures from -250 to 7050C [37]. It also provides moderate high temperature strength as well as good resistance to strain-age cracking in welding [3]. The fatigue strength of 718 is high and the alloy exhibits excellent stress-rapture properties up to 7050C, besides oxidation resistance up to 982 0C [38]. These unique properties have resulted in use in a wide range of applications such as gas turbine components, cryogenic storage tanks, jet engines, pump bodies, thrust reversers, nuclear fuel element spacers, etc. [37]. 718 is reported to be resistant to strain-age cracking as a result of the sluggish precipitation kinetics of its principal strengthening precipitate γ΄΄ (Ni3Nb) [39]. δ phase (Ni3Nb) is a major equilibrium secondary phase in niobium-bearing nickel- and iron-based superalloys, including IN 625, IN 706, Incoloy 903 and the newly developed Allvac 718+ and is a most stable secondary phase in Alloy 718 [36]. δ phase particles form in wrought 718 material by a solid-state precipitation reaction during heat treatment in the temperature range 860-995 0C [40] and as such, it has been generally assumed that they dissolve by solid-state reaction above their equilibrium solvus temperature during thermal exposure in weld HAZ [36].
Allvac 718Plus is a newly developed nickel base superalloy. Alloy 718Plus is a derivative of alloy 718. The alloy composition reflects a continuation of work performed at ATI Allvac in the early 1990’s directed at improving the creep and rapture properties [41-47]. ATI Allvac introduced the superalloy Allvac® 718PlusTM in 2004 to combine the high temperature resistance of Waspaloy and the workability of Alloy 718 [48]. The differences of alloy 718Plus chemistry compared to alloy 718 are the increase in sum of Al + Ti, the ratio of Al/Ti, the addition of W and especially Co to replace Fe in order to eliminate the unstable γ΄΄ phase (Ni3Nb, D022 tetragonal) [49]. This newly developed alloy 718Plus appears to have increased temperature capability compared to alloy 718 by 55 0C and excellent thermal stability [49]. The reduction of Fe and the addition of Co along with the increase in Al content found to promote the formation of the ordered L12 γ΄ phase (Ni3(Al, Ti)) as opposed to the DO22 bct γ΄΄ phase (Ni3Nb) [50-52]. Alloy 718Plus has a much larger content of γ΄ and γ΄΄ than alloy 718 and a smaller amount of δ phase [53]. Grain-boundary δ phase has a significant impact on the mechanical properties of both alloy 718 and Allvac 718Plus, although the precise effect of the phase can vary considerably and appears to depend strongly on the location and morphology of the phase [55-63]. Solvus temperatures for γ΄ and γ΄΄ are also higher in alloy 718Plus. All of these points likely contribute to improved high temperature properties [53]. Also, tungsten is added in order to increase the high-temperature deformation resistance of Allvac 718Plus [56].
Allvac 718Plus microstructure contains two types of intermetallic phases. Firstly, the γ΄ phase Ni3(Al, Ti) with its L12-ordered face-centered cubic crystal structure [64] acting as nanometer-sized obstacles for the migration of dislocations and therefore primarily accounting for the mechanical properties of the alloy [54]. Secondly, the δ phase Ni3Nb of orthorhombic crystal structure, which precipitates in plate-shaped morphology predominantly at grain boundaries [65]. The δ phase is of technological importance as it is used for controlling the grain growth during processing, e.g. hot forging [66]. It is generally accepted that a small amount of δ precipitation is essential for grain-boundary pinning and that excessive levels of coarse δ are detrimental to properties due to solute depletion and easy crack initiation and growth. The precipitation of the phase can lead to serration of grain boundaries in 718 and 718Plus and reports suggest that such a microstructure is likely to improve intergranular fatigue crack resistance [55, 57, 67, 68].
Haynes International introduced the Haynes 282 in fall of 2005. It is a new, wrought, and gamma-prime precipitation strengthened nickel-base superalloy developed for high temperature structural applications, particularly for aero and land-based gas turbine engines [69]. Haynes 282 alloy is solid solution strengthened by cobalt (Co), chromium (Cr) and molybdenum (Mo). It was designed for improved high-temperature creep resistance, and its creep resistance surpasses that for Waspaloy and approaches that for R-41 alloy [70]. Haynes 282 alloy possesses a unique combination of creep strength, thermal stability, weldability and fabrication possibility. It is being considered for the transition sections and other hot-gas-path components in land-based gas turbines and for critical aircraft gas turbine applications, such as sheet fabrications, seamless and flash butt-welded rings, exhaust and nozzle components and cases found in compressor, combustor and turbine sections [71].
Major compositional design of Haynes alloy 282 as compared to other alloys in its class are the lower concentrations of Al and Ti, which is a major limiting factor that controls the volume fraction and precipitation kinetics of the main strengthening phase, γ΄ [72]. Cobalt is added to stabilize γ΄ by controlling the γ΄ solvus temperature apart from being a solid solution strengthener [72]. The addition of chromium provides resistance to both oxidation and hot corrosion while molybdenum is primarily added to enhance creep strength, weight fraction and solvus of γ΄ [69]. The concentration of iron, silicon and manganese are strictly limited to a maximum and also there is a noticeable preclusion of Nb in the alloy [72]. The commercially recommended pre-weld solution heat treatment (SHT) procedure for the alloy ranges from 1121 0C to 1149 0C followed by rapid cooling [72]. Major phases reported in the alloy prior to and after SHT include: primary gamma phase (γ) and MC type carbide and carbonitride [69]. The size of MC type carbides ranges from 2 μm to about 15 μm and they are intra- and intergranularly dispersed within the γ matrix [72].
Inconel 718 C Ni Cr Mo W Co Fe Nb Ti Al
min - balance 17.00 2.80 - - 18.00 5.00 0.65 0.20
max 0.08 balance 21.00 3.30 - 1.00 19.00 5.50 1.15 0.80
Table 3: Chemical composition of Inconel 718 in wt. % [100]
Allvac® 718PlusTM C Ni Cr Mo W Co Fe Nb Ti Al
min - balance 17.00 2.50 0.50 8.00 8.00 4.75 0.50 1.20
max 0.06 balance 21.00 3.10 1.50 10.00 10.00 5.80 1.00 1.80
Table 4: Chemical composition of Allvac® 718PlusTM in wt. % [100]
Haynes® 282 AlloyTM C Ni Cr Mo Si Co Fe Mn Ti Al B
max 0.06 balance 20.00 8.50 0.15 10.00 1.50 0.30 2.10 1.50 0.005
Table 5: Chemical composition of Haynes® 282 AlloyTM in wt. % [70]
2.4 Cracking theory
The head-lines of the different cracking mechanisms are given here. The full text is given in the Appendix.
2.4.1 Solidification cracking
2.4.2 Liquation cracking
2.4.3 Centerline grain boundary cracking
2.5 Cracking criteria
2.5.1 Prokhorov criterion
2.5.2 Yamanaka criterion
2.5.3 RDG criterion
2.5.4 Ploshikhin criterion
2.5.5 Martikainen criterion
2.5.6 Dye criterion
2.5.7 Hier-Majumber criterion
2.5.8 Chadwick criterion
2.5.9 Rosenthal criterion
2.6 Essential work of fracture criterion
2.6.1 Gao criterion
2.6.2 Total crack length and maximum crack length criteria
2.6.3 Benhadad criterion
2.6.4 Clyne and Davies criterion
2.6.5 Katgerman criterion
2.6.6 Novikov criterion
2.6.7 Feurer criterion
2.6.8 Farup and Mo criterion
2.6.9 Magnin criterion
2.6.10 Criterion of Braccini
2.6.11 The Niyama criterion
2.7 Strain age cracking
2.7.1 Pierce-Miller patch test
2.7.2 Criterion of minimum elongation
2.7.3 Fracture toughness criterion
3 Testing and microstructural investigation
3.1 Testing (Gleeble and Varestraint)
Gleeble testing was performed at Swerea-Kimab in Stockholm and at Manitoba University in Winnipeg, Canada, since none of the participating labs has got a Gleeble testing equipment.
Gleeble testing was made on the three different base materials, with two different conditions of each material. Prior to testing the materials were subjected to a heat treatment, simulating the Heat Affected Zone at the appropriate temperatures.
A large number of specimens were tested, mainly at 700, 800 and 900°C, with some local variations of temperatures. The results showed that 718 had a minimum in ductility at 800°C, while the other two alloys had their minimum in ductility at higher temperatures, around 900°C. It was difficult to distinguish between these alloys when it comes to strain-age cracking. A small effort was also made on materials with coarser grains. However, no significant effect was seen of grain size on ductility.
The full results of the testing are given in Appendix, with micrographs of the specimens and plots the curves of reduction of area versus temperature, as indicator of the ductility of the specimens.
Varestraint testing
Varestraint testing was made on about 65 different specimens, with a thickness of 3.2 mm. The bending radius was either 150, 200, 400 or 500 mm. Some differences could be noted between the different materials. Mostly, very little cracking was seen until the highest augmented strain. 718Plus actually showed no cracking until an augmented strain of 1.1%, while Haynes 282 showed some small cracking for all augmented strains. 718 was in between these two alloys. There also seemed to be more cracks in 718 and 718Plus in the coarse-grained condition, while Haynes 282 was not very sensitive to this.
The full results of the testing are given in Appendix, since a graphical presentation is necessary.
3.2 Material Characterization – SEM
This report presents results from an investigation using scanning electron microscopy (SEM) of the base materials of nickel base superalloys Alloy 718, ATI 718Plus and Haynes 282, and complements other reports in the Weldmecs project. The microstructure was studied using SEM, and chemical analyses were performed using energy dispersive spectroscopy (EDS). The microstructure was studied on the as-received materials (Alloy 718, ATI 718Plus and Haynes 282), and after solution heat treatments giving a fine microstructure. The final heat treatment was made at 954°C for 1 h for Alloy 718 and ATI 718Plus, and at 1121°C for 30 minutes for Haynes 282. The as-received materials contain a small number of large (typically 5 µm) primary carbides, often located in groups of several carbides. The heat treatment of Alloy 718 and ATI 718Plus resulted in the formation of precipitates in the range 0.1-1 µm, which are mainly δ-precipitates. In ATI 718Plus a large number of very small precipitates (~10 nm), presumably γ´ or γ´´, seemed to have formed, although they are too small to be clearly resolved by SEM. The heat treatment of Haynes 282 resulted in carbide precipitation at grain boundaries and in precipitation of a dense dispersion of γ´ with size of 24±7 nm
3.2.1 Introduction
The base materials have been reported in chapter 2 of this report, where also background theory and references to literature are included. In this chapter, additional scanning electron microscopy (SEM), often at higher magnification, and energy dispersive spectroscopy (EDS) analyses are included.
3.2.2 Experimental Procedure
Materials and Heat Treatments
Three different precipitation-hardened Ni-based superalloys containing about 20 wt.% Cr have been investigated. The specified chemical compositions are listed in tables 1-3. The compositions of Alloy 718 and ATI 718Plus are similar in many respects. The main differences are that Co to a large extent has replaced Fe and that the Al content has been increased in ATI 718Plus. Haynes 282 is more different, and is characterized by high Mo, Co and Ti content, and the absence of Nb, which is a key element in the two other alloys.
The materials were studied in the as-received condition and after a final heat treatment intended to give a fine grain structure. The final heat treatment was made at 954°C for 1 h for Alloy 718 and ATI 718Plus, and at 1121°C for 30 minutes for Haynes 282.
Table 1. Chemical composition of Inconel 718 (Alloy 718) in wt.%.
Inconel 718 C Ni Cr Mo W Co Fe Nb Ti Al
min - balance 17.00 2.80 - - 18.00 5.00 0.65 0.20
max 0.08 balance 21.00 3.30 - 1.00 19.00 5.50 1.15 0.80
Table 2. Chemical composition of Allvac® 718PlusTM (ATI 718Plus) in wt.%.
Allvac® 718PlusTM C Ni Cr Mo W Co Fe Nb Ti Al
min - balance 17.00 2.50 0.50 8.00 8.00 4.75 0.50 1.20
max 0.06 balance 21.00 3.10 1.50 10.00 10.00 5.80 1.00 1.80
Table 3. Chemical composition of Haynes® 282 AlloyTM in wt.%.
Haynes® 282 AlloyTM C Ni Cr Mo Si Co Fe Mn Ti Al B
max 0.06 balance 20.00 8.50 0.15 10.00 1.50 0.30 2.10 1.50 0.005
Metallography preparation
Standard automatic metallographic preparation techniques were used for microstructural analysis. The preparation was carried out at University West. The metallographic specimens were etched electrolytically at 3.5-4 V in oxalic acid for 10 to 15 seconds. Haynes 282 samples were also electrolytically etched in a mixture of 12 mL H3PO4 + 40 mL HNO3 + 48 mL H2SO4 at 6 V for 5-6 seconds to reveal the γ΄ phase in the microstructure. In some cases also unetched samples were studied.
SEM and EDS
Metallography characterization was carried out using the SEM instrument LEO Ultra 55. Chemical analyses were performed using EDS equipment from Oxford Instruments, with the INCA software. Both SEM and EDS were performed at an accelerating voltage of 20 kV. Imaging was performed using secondary electrons, either using the standard SE2 detector or the InLens detector. It should be noted that the volume contributing to the EDS signal is of the order of 1 μm3, which means that when small particles are analyzed, the signal comes both from the particle and from the surrounding matrix. The capability to measure carbon is not very high using EDS, so in most cases C is not included in the results presented. Both point analyses and mapping were performed.
3.2.3 Results and discussion
Alloy 718: as-received
The as-received material contains some primary carbides that are a few µm large. It was shown in chapter 2 that the carbides are both NbC and TiC, with some intermixing of other elements. The grain structure is equiaxed, with a grain size of ASTM 10 (about 10 µm).
Alloy 718: 954°C, 1 h
The heat treatment has resulted in a large number of precipitates, both at the grain boundaries and inside the grains. Some precipitates are globular and some are plate-like, and there is a large spread in the particle size. The platelet precipitates in the interior of grains are often aligned, indicating that they have a crystallographic relationship with the fcc matrix. EDS point analysis results (typical examples) from matrix, a large carbide and a small plate-like precipitate have been made. As expected the large carbide is NbC, also containing some Ti (Ti/Nb ratio of 0.13). The small precipitate is too small to be analyzed accurately with EDS in SEM, but as the concentration of Nb and Ni are increased compared to the matrix, and Cr and Fe are depleted, it suggests that the precipitate is either δ-Ni3Nb or γ''-Ni3Nb (as Ni is not enriched in carbides). From literature it appears to be more likely that the precipitates are δ. An EDS elemental maps confirm that the large particles are Nb(Ti)C. The small precipitates are too small to be visible in this type of maps.
ATI 718Plus: as-received
The as-received ATI 718Plus contains some primary carbides, similar to Alloy 718. The primary carbides are of NbC and TiC types. Contrary to as-received Alloy 718, the as-received ATI also contain small, mainly rounded, precipitates in the range from 0.1-0.6 µm. The number density and volume fraction of these precipitates are rather low. EDS analyses of two small precipitates was made. The larger of the two precipitates is clearly enriched in Nb, but also in Al, Ti and Ni (compared to the matrix concentrations), which suggests that the precipitate is of Ni3Nb, i.e. either δ or γ''. The round shape could indicate that it is γ'', at least the morphology is different to that of δ in Alloy 718. Also the smaller precipitate is enriched in Nb, but even more so in Al. Ti is slightly enriched, whereas Ni, Cr and Fe have a lower concentration than in the matrix. Although the decrease in Ni is small, it is assumed that the Ni content should increase in the expected Ni3Nb intermetallic phases. Still, the high Al content could indicate that the smaller precipitate is γ'-Ni3(Al, Ti, Nb).
ATI 718Plus: 954°C, 1 h
The heat treatment has resulted in an extensive formation of blocky, faceted precipitates, some of which are acicular). Using XRD these precipitates were determined to be δ-Ni3Nb. There are also round precipitates present, which have a similar appearance as the round precipitates in the as-received material. After heat treatment, they have become smaller (now smaller than 0.1 µm), and at the same time the number density has increased. If it is the same phase, one possibility would be that the precipitates dissolved, and re-precipitated on cooling. If they were stable at 954°C, they would have coarsened. Another possibility is that the original precipitates dissolved, and at the same time a new phase formed. However, as the morphology is the same as in the as-received material, they are potentially the same phase, presumably γ''. At higher magnification (100 kx), it looks like there might be a dense dispersion of very small precipitates (≈10 nm). If they are really precipitates, they are probably γ'-Ni3(Ti, Al).
Haynes 282: as-received
In the as-received Haynes 282 some primary carbides are present, much in the same way as in Alloy 718 and ATI 718Plus. However, as Haynes 282 does not contain Nb, they are MC, mainly containing Ti and Mo. In addition to the large primary carbides (several µm) there are also some scattered precipitates with sizes of 0.2-0.5 µm. They are faceted with cubic, hexagonal or blocky morphology. The precipitates are enriched in Ti and Mo, with more Ti than Mo. The composition is similar to the primary carbides (taking the matrix contribution to the signal into account), so they could be secondary MC or possibly M23C6 or M6C, although Cr would be expected to be enriched in the two latter cases. At higher magnification (100 kx), signs of very small precipitates (10-20 nm) can be observed. If correct, the small precipitates are likely γ'-Ni3(Ti, Al).
Haynes 282: 1121°C, 30 minutes
There are still primary carbides, as in the as-received material. Also the sparsely distributed Ti- and Mo-rich carbides remain. However, two main changes have occurred. Firstly, a large amount of grain boundary precipitates has formed. Secondly, a dense distribution of very small precipitates has formed throughout the matrix. Some of the grain boundary precipitates are several µm in size, and they might have been present already in the as-received material ((Ti, Mo)C), but the majority are smaller than 1 µm, and were not present before the heat treatment. Most of the grain boundary precipitates are enriched in Ti and Mo, and might be MC. Some of the particles, however, contain very little Ti, but much Mo. These precipitates are likely M6C, as the Cr content would be expected to be higher if they were M23C6. The very small precipitates are most likely γ'-Ni3(Ti, Al). The size of these precipitates was determined to be 24 ± 7 nm. Using EDS it was noted that slightly higher concentrations of Ti, Al and Mo were measured when the beam was positioned on a precipitate than when it was positioned between precipitates.
Conclusions
The as-received materials contain a small number of relatively large primary carbides, Nb(Ti)C in the case of Alloy 718 and ATI 718Plus, and Ti(Mo)C in Haynes 282. The solution heat treatment created a large volume fraction of δ-Ni3Nb. In ATI 718Plus, the heat treatment also resulted in precipitation of what is believed to be γ''-Ni3Nb (0.1 µm) and Ni3(Ti, Al, Nb)-γ'' (10 nm). In Haynes 282 the solution heat treatment resulted in the formation of grain boundary carbides (Ti- and/or Mo-rich) and a fine dispersion of γ'-Ni3(Ti, Al), with a size of 24 ± 7 nm.
3.3 Material Investigation, Atom-probe and TEM
The microstructure of samples subjected to Gleeble and Varestraint testing has been investigated using scanning electron microscopy (SEM), atom probe tomography (APT) and transmission electron microscopy (TEM). The three nickel-base superalloys Alloy 718, ATI 718Plus and Haynes 282 were studied. Prior to Gleeble testing the samples underwent a HAZ simulation heat treatment. The samples studied here had been Gleeble tested at three different temperatures; above, at and below the ductility dip temperature. In all cases, the fractures followed grain boundaries. Small precipitates, presumably carbides, were frequently observed at the grain boundaries, in particular in the samples tested at the higher temperatures and more so in Haynes 282 and Alloy 718 than in ATI 718Plus. In the samples tested at the highest temperature, 900°C, a dense distribution of small precipitates (20 nm) could be identified using SEM. This was confirmed by APT for Alloy 718 (which was the only 900°C-tested alloy that was analyzed with APT). Using APT a dense distribution of very small (5 nm) precipitates enriched in Nb, Ti and Al, could be observed in all three alloys tested at 800°C, Alloy 718 and Haynes 282, or 850°C, ATI 718Plus. In the Varestraint samples cracks running perpendicular to the welding direction, usually starting in the weld and continuing into the HAZ, were observed. These cracks were associated with a eutecticum consisting of Nb-rich Laves phase and the γ-matrix. The Laves phase was identified using electron diffraction in coarse-grained ATI 718Plus.
3.3.1 Introduction
In this report some complimentary scanning electron microscopy (SEM) of samples from Gleeble testing and Varestraint testing is presented, as well as an Atom Probe Tomography (APT) study of five samples from Gleeble tested samples and a transmission electron microscopy (TEM) study of the crack region in a sample from Varestraint testing.
3.3.2 Experimental Procedure
Materials and Heat Treatments
Three different precipitation hardened Ni-based superalloys have been investigated, Alloy 718, ATI 718Plus and Haynes 282, with compositions according to tables 1-3. The alloys were solution heat treated in two different ways, one to give a fine-grained structure and one to give a coarse-grained structure. The heat treatment of Alloy 718 and ATI 718Plus were the same; 954°C for 1 h to give small grains and 1050°C for 3 h followed by 954°C for 1 h to give small grains. Haynes 282 was instead heat treated at 1120°C for 30 minutes to give fine grains and 1150°C for 2 h to give coarse grains.
Table 1: Chemical composition of Inconel 718 (Alloy 718) in wt. %.
Inconel 718 C Ni Cr Mo W Co Fe Nb Ti Al
min - balance 17.00 2.80 - - 18.00 5.00 0.65 0.20
max 0.08 balance 21.00 3.30 - 1.00 19.00 5.50 1.15 0.80
Table 2: Chemical composition of Allvac® 718PlusTM (ATI 718Plus) in wt. %.
Allvac® 718PlusTM C Ni Cr Mo W Co Fe Nb Ti Al
min - balance 17.00 2.50 0.50 8.00 8.00 4.75 0.50 1.20
max 0.06 balance 21.00 3.10 1.50 10.00 10.00 5.80 1.00 1.80
Table 3: Chemical composition of Haynes® 282 AlloyTM in wt. %.
Haynes® 282 AlloyTM C Ni Cr Mo Si Co Fe Mn Ti Al B
max 0.06 balance 20.00 8.50 0.15 10.00 1.50 0.30 2.10 1.50 0.005
Metallographic Preparation
Standard automatic metallographic preparation techiques were used for making samples for microstructural analysis. Metallographic specimens were etched electrolytically at 3.5-4 V in oxalic acid for 10 to 15 seconds. Haynes 282 samples were also electrolytically etched in a mixture of 12 mL H3PO4 + 40 mL HNO3 + 48 mL H2SO4 at 6 V for 5-6 seconds to reveal the γ΄ phase in the microstructure.
Sample Preparation using FIB/SEM
Lift-out samples for APT and TEM were prepared using focused ion beam/SEM (FIB/SEM), with the instrument FEI Versa 3D. The APT samples were taken from the fine-grained Gleeble samples, about 5 mm from the crack, in the centre of the sample.
Metallographic Characterization
Metallographic characterization was carried out using the SEM instrument LEO Ultra 55, which is a high resolution instrument with a field emission gun (FEG). Chemical analyses were performed using EDS equipment from Oxford Instruments, with the INCA software. Both SEM and EDS were performed at an accelerating voltage of 20 kV. Imaging was performed using secondary electrons, either using the standard SE2 detector or the InLens detector. It should be noted that the volume contributing to the EDS signal is of the order of 1 μm3, which means that when small particles are analyzed, the signal comes both from the particle and from the surrounding matrix. The capability to measure carbon is not very high using EDS, so in most cases C is not included in the results presented. Both point analyses and mapping were performed.
The APT analyses were performed using the instrument LEAP 3000X HR in laser mode. The pulse energy was 0.3 nJ, the pulse frequency was 200 kHz and the analysis temperature was 30 K. The data was evaluated using the software IVAS 3.6 from Cameca. For compositional analysis, peak decomposition was applied to take some minor peak overlaps into account.
The TEM investigation was performed using an FEI Tecnai T2 equipped with an EDAX detector for EDS, and an FEI Titan 80-300 equipped with a monochromator and a spherical aberration corrector, and EDS equipment from Oxford Instruments.
Gleeble Testing
Prior to testing the material was subjected to a heat treatment simulating the heat affected zone (HAZ) in a typical welding operation.
Varestraint Testing
The samples investigated here were of the coarse-grained type and had been tested with a bend radius of 150 mm.
3.3.3 Results and Discussion
Scanning Electron Microscopy of Gleeble Samples
Ductility dip cracking is most pronounced at temperatures around 800°C. In this investigation the purpose was to study samples tested at the temperature where ductility is minimum, at one temperature above and at one temperature below the most critical temperature. The choice of temperatures is shown in table 4.
Table 4: Gleeble samples investigated by SEM/EDS.
Alloy Grain size Temperature (°C)
Alloy 718 Fine 900, 800, 700
Alloy 718 Coarse 900
ATI 718Plus Fine 900, 750, 700
ATI 718Plus Coarse 900
Haynes 282 Fine 900, 800, 700
Haynes 282 Coarse 900, 800, 700
Alloy 718: Fine-grained
Observations were made at both 900, 800 and 700°C. At 900°C some large primary carbides, probably Nb(Ti)C, were observed. Some grain boundaries appear to have opened up, and some precipitation could be observed at the grain boundaries. The nature of these precipitates is unknown, but a guess is that they are carbides. In some regions a dense distribution of very small matrix precipitates could be observed. These are probably Ni3(Nb,Ti), as determined by APT, see below. At 800°C no particular observation was made. At 700°C much fewer grain boundaries had opened up, and there were no secondary cracks. The fracture had a sharp appearance.
ATI 718Plus: Fine-grained
For the sample tested at 900°C several secondary cracks could be seen, which is also the reason why there is a big step in the fracture surface. Some grain boundary features are similar to those in fine-grained Alloy 718 tested at 800°C. Some examination was also made at 750°C. This temperature was studies because at the time the choice was made it was considered to be the temperature with the lowest ductility. Later it was realized that the ductility is lower at 800°C. At 700°C the macroscopic fracture surface is rather flat. Some grain boundaries have opened up, but there are no signs of grain boundary precipitation.
ATI 718Plus: Coarse-grained
Only one coarse-grained ATI 718Plus sample, tested at the highest temperature, was investigated. The secondary cracks in this specimen were very wide.
Haynes 282: Fine-grained
For samples tested at 900°C small precipitates are visible at grain boundaries at high magnification. In the grain interior there are also numerous small precipitates (probably Ni3(Ti,Al)).
For sample tested at 800°C, also here precipitates at the grain boundaries can be resolved, although they are smaller than in the material tested at 900°C. Small precipitates can also be discerned in the matrix, as confirmed by APT (below).
For sample tested at 700°C there are almost no grain boundary precipitation.
Haynes 282: Coarse-grained
For samples tested at 800°C, small particles are present at the grain boundaries
Atom Probe Tomography of Gleeble Samples
Gleeble samples of Alloy 718, fine-grained material, were analyzed using APT. The samples had been tested at 700, 800 and 900°C. It was found that a large number of precipitates were present in the 800°C sample. In the 900°C sample fewer and larger precipitates were observed, whereas in the 700°C sample more or less no precipitates were found. However, there is a tendency to form precipitates also in the 700°C sample, as seen in the (bulk normalized) radial distribution functions constructed with respect to Nb. The high values at short distances for Nb-Nb, Ti-Nb and, to a smaller extent, Ni-Nb indicate a positive interaction as a result of initial clustering. In the sample tested at 800°C, there are two types of precipitates. One type is enriched in Nb and one in Al, and both contain Ni and Ti. This probably means that Ni3(Ti, Nb) and Ni3(Ti, Al) are forming.
In order to compare the three alloys, also Gleeble samples of ATI 718Plus, tested at 850°C, and Haynes 282, tested at 800°C, were analyzed using APT. It is clear that all three materials contain a dense distribution of small precipitates. The precipitates in ATI 718Plus are larger, but that is partly caused by the higher temperature.
Alloy 718 has a lower volume fraction of precipitates, which is expected as the total concentration (in at.%) of the precipitating elements Al, Ti and Nb is lower than in the other two alloys.
Scanning Electron Microscopy of Varestraint Samples
Samples with coarse grains of Alloy 718, ATI 718Plus and Haynes 282 were investigated after Varestraint testing using a mandrel of radius 150 mm. The samples are of the plan view type, which means that the weld and the HAZ are studied from above. The samples were only ground to remove the weld cap. In each sample a few cracks are present in the weld/HAZ region running approximately perpendicular to the welding direction, of length around 1 mm.
Alloy 718: Coarse-grained
A low magnification SEM image of a crack was observed. EDS analysis of matrix, eutectic precipitation and grain boundary precipitates were recorded It appears as if the composition of the eutectic and the small grain boundary precipitates are very similar.
ATI 718Plus: Coarse-grained
A streak of eutectic precipitation running from the weld into the HAZ was observed. Here a crack cannot be identified, but probably there was originally a crack. Also, some short cracks, not associated with precipitation, can be seen in the HAZ. A long crack in the weld, which is associated with eutectic precipitation was also found.
Haynes 282: Coarse-grained
This sample contained very few cracks.
Transmission Electron Microscopy of ATI 718Plus Varestraint Sample
The Varestraint sample of coarse-grained ATI 718Plus was investigated using TEM. Using FIB/SEM two samples from the HAZ were prepared; one at a crack with eutectic precipitation and one at a nearby grain boundary without precipitates, or cracking.
Some eutectic precipitates was seen by TEM. Using TEM/EDX, it was concluded that the precipitates were Nb-rich. Selected area electron diffraction was used to identify the crystal structure of the precipitates. This diffraction pattern matches the Laves phase, with zone axis [3-3-1]. A high-resolution TEM image of this precipitate was made, with the corresponding fast Fourier transformation (FFT), which matches the Laves phase.
At the grain boundaries not associated with cracks, very small precipitates were observed. It was not possible to identify the type, but it is likely that they are carbides.
Conclusions
The SEM investigation of Gleeble-tested samples has shown that the cracks follow grain boundaries in all cases. At the higher testing temperatures grain boundary precipitation is usually observed. Using APT it was concluded that the samples having poor ductility also contained a large number of small precipitates, Ni3(Nb,Ti) in Alloy 718 and ATI 718Plus, and Ni3(Ti,Al) in Haynes 282. The importance of these precipitates for the ductility dip is unclear. The cracks studied in Varestraint samples where associated with massive Nb-rich eutectic precipitation in Alloy 718 and ATI 718Plus, but any corresponding precipitates were not observed in Haynes 282 (which does not contain Nb). Using TEM and EDS it was concluded that the eutectic precipitation consists of Nb-rich Laves phase and the matrix. Grain boundaries that had no cracks, contained very small precipitates, assumed to be carbides.
4 Simulation of Gleeble and Varestraint testing
Gleeble testing
The result from the simulations of the Gleeble tests can be summarized in Figure 124 in the full report in the Appendix. This figure show the relationship between the calculated area reduction at fracture, and test temperature for all six materials tested (three materials, two heat treatments each). The area at fracture is calculated from the deformed geometry at the time of fracture for all simulations presented under point 3 in the full report. The time of fracture is the time when the warm crack indicator reaches the value 1.0. One dot represents one test and the total number of simulated Gleeble tests is 33. The figure should be compared with the test result found in other project reports. The similarity is impressive. The calculations catch the shift from ductile to brittle fracture over the tested temperature range for all materials. Also the difference among the materials is described correctly.
Varestraint testing
The result from the Varestraint simulations is shown in figure 125, 126 and 127. Figure 125 shows the hot crack indicator as a function of anvil radii used in Varestraint tests. A small radius of the anvil leads to large deformations (strains) in the specimen and increase the risk of hot cracks to occur. In Figure 126 the calculated hot crack indicator is plotted against augmented strain in the test. This is almost how the result is presented in the Varestraint test report, and this figure is therefore easy to compare directly with measurements from the test. Just observe that the hot crack indicator is never mentioned in that report. Instead the total detected crack length is used.
The augmented strain is a theoretical strain on the top surface of the test plate.
The real total strain may differ somewhat to the augmented strain due to unexpected effects of the Varestraint test. In reality the center of the test plate may not be in contact with the anvil. Instead two contact regions slightly offset to each side of the center of the plate may appear. This will influence the strain distribution on the top surface of the plate. But from the simulations we know the actual total strain as well as it’s elastic and plastic components. It is therefore possible to plot the hot crack indicator as a function of plastic strain. This is done in Figure 127, but the information is basically the same as in figure 126. The calculations catch the sensitivity to hot cracks for all materials and the difference among the materials is also described correctly.
Please note that the numeric values for the hot crack indicator shown in figure 125 to 127 is taken from the maximum value found on the top surface of the plate in the tests. The figures can be found in section 4 of this report, but the maximum values is not read directly from the figures involved, but are found automatically by computer aid during the evaluation of the simulations. This is also true for the effective plastic strain of each plate. The value is taken on the top surface of the plate, half way between the weld and the edge of the plate. The total number of test plates (and simulations) is 33 in this case.
Potential Impact:
This project is related to development of jet engines by using thin sheets and welding to join these structures. From the nature of the base material it is known that welding can create problems and it is necessary to increase the knowledge about how these problems can be overcome. To solve this problem, a more collective approach is necessary, to bring in knowledge from both production, metallurgical and simulation side. The aim of this project has been to create a team working in this context, with the aim to solve the fundamental problems of joining very technically demanding alloys.
The potential impact of the project is in principal very high, since it may in the long run help us to make lighter, stronger jet engines that can operate at higher temperatures and thus be more environmental friendly. However, it must be realised that this is a very challenging subject, which needs to be treated in a number of consecutive projects. The present project, Weldmecs, is a part of these consecutive projects. Thus, it is difficult to point to this single project, claiming that it will bring immediate solutions to a very wide range of subjects. The merits of this project lies more in the hardware that it is creating, being more or less directly useful for companies creating the jet engines of tomorrow.
The project has not yet created any dissemination activities, since a lot of the important results arrived very late in the project. However, it is planned to write at least three scientific papers on subjects examined in the project and also use the information for presentations at scientific conferences.
It is also planned to continue the work, by setting up a new project consortium.
List of Websites:
No public website made