Microstructure of Materials
Investigation of the Effect of Preweld Heat Treatment on the Liquation Cracking of GTD-111 Superalloy
2020 Volume 61 Issue 5 Pages 903-908
Details
2020 Volume 61 Issue 5 Pages 903-908
GTD-111 is a γ′ strengthened Ni-based superalloys, which is used in critical applications such as turbine blades. Liquation cracking during welding is recognized as being one of the most important problems of this type of components. The present paper outlines a new preweld heat treatment to control the liquation cracking occurred in HAZ regions of GTD-111 turbine blades. Different preweld heat treatments were carried out on the samples extracted from the root of turbine blades and then they welded by gas tungsten arc welding (GTAW) process The microstructural analysis of the joints was carried out using optical microscopy, scanning electron microscopy, and image analysis technique. The results showed that the preweld heat treatment regimes can obviously affect the intensity of liquation cracking. While the optimum solutionizing temperature and time were found to be 1180°C and 2 hours, the samples with no heat treatment showed the lowest resistance to liquation cracking. It was also concluded that the liquation cracking is mainly affected by the relative intensity of the base metal hardness, the average length and volume fraction of γ′ precipitates, presence of carbides, and the existence of low melting phases.
Fig. 5 The HAZ zone microstructure, under C (1125°C/2 h) condition.
The GTD-111 superalloy is among precipitation hardened nickel based superalloys. In 1980, this alloy was introduced as an alternative to the IN738LC in the first-stage blades of the gas turbines, due to its high temperature characteristics. This alloy is widely used in the gas turbine blades due to its excellent corrosion resistance and good creep resistance.1) These blades work in severe conditions including high temperatures, corrosion, oxidation and the presence of external objects suspended in the used gases. The blades are damaged due to the formation of thermal fatigue cracks, erosion occurred at their tip, and collision of external objects, result in reduced turbine output. Nowadays, for economic reasons, there is a high demand for the rebuilding and recovery of stationary blades, rotor blades and other parts of the gas turbines. Therefore, repair welding has been found to be a significant way for reuse the damaged turbine components. During the repair weld of these parts, hot cracking occurrence in the weld metal and heat affected zones, has always been one of the main problems.2–5)
For the first time, HAZ cracking due to grain boundary liquation in the Nickel-base superalloys was observed by Owczarski et al.6) They reported that the starting of liquation in adjacent with MC carbides and borides in Udimet 700 and Waspaloy alloys is due to the sudden decomposition of these particles during rapid heating rate of welding. Based on their results, the dissolution and constitutional liquation of MC carbide particles were identified as one of the main causes of intergranular liquation in the welding of many alloys. Further research showed that the liquation of other secondary particles such as γ′ precipitations, Cr–Mo rich boride particles, and Zr-rich sulfocarboides, as well as MC carbides, led to the formation of liquation film and cracking.7,8) Constitutional liquation of γ′ precipitations and their significant role in cracking of HAZ regions were observed and reported by Ojo et al.9) They observed that liquid films penetrate between the grain and precipitation boundaries towards grain boundaries, making them wet. Negative effect of γ′ on the weldability of superalloys due to the rapid rate of precipitation, heterogeneous distribution of strains during weld cooling, and the increasing of Al and Ti were found to be as the other sources of liquation cracking in nickel base superalloys may decrease solidus and liquids temperatures.
To eliminate or reduce the welding cracks of these alloys, some methods have been introduced, such as reducing the welding heat input, preheating, using low-strength solid solution filler metals, and development of preweld heat treatment.7,10) Preweld heat treatment has been identified as a major role in controlling the susceptibility to liquation cracking due to its effects on the phase microsegregation and also on the grain size. Aging, solutionizing, and homogenization (in some cases), have been found to be common pre-welding treatments used in superalloy welding. Many studies have so far done regarding the ability of different heat treatments on preventing the liquation cracking.11–13) Some of the researchers believe that age-hardened alloys have a higher cracking potential than solutionized alloys.11) However, there are still contradictions about the effect of preweld heat treatment. Owczakarski et al.6) showed that the solutionzing treatment of udimet700 alloy result in reducing the ductility at the welding temperature compared to the same alloy at the full aged condition, followed by increasing the sensitivity of cracking. Gordine14) found that the temperature of solutionizing treatment of 718 alloy, plays an important role in sensitivity to the HAZ liquation cracking. The research also revealed that using high solutionizing temperatures before welding, would lead to a high sensitivity to HAZ liquation cracking. Sidhuo et al.7) made a research on the IN738 superalloy under two preweld heat treatments, full aged and solutionized. The research revealed that the full aging heat treatment increases the resistance to liquation cracking. Extensive research has been carried out on the liquation cracking of a variety of precipitation hardened superalloys, but there is a lack of consistent studies on the cracking of GTD-111 alloy. In this study, an attempt has been made to gain a better understanding of the effect of different preweld heat treatments on the HAZ zone liquation cracking of GTD-111 alloy.
The chemical composition of the base and filler metals used in this research, is given in Table 1. Plate samples with the dimensions of 5 × 3 × 0.3 cm were extracted from the root of GTD-111 turbine blade using wire-cut. Then, a longitudinal groove with 1 mm depth was made on each of the specimens. Different preweld heat treatments were carried out based on the Table 2. The welding treatment was then performed by gas tungsten arc welding (GTAW) process using IN625 filler metal, 60 A current, direct current (DC), and 10 V voltages. Transverse sections of the welds were prepared using a microcutter after which the samples were polished by standard metallographic techniques. The polished sections were electrolytically etched in a solution containing 12 ml H3PO4 + 40 ml HNO3 + 48 ml H2SO4 at 5 V for about 5–6 s. After wards, different zones and phases of the prepared samples were analyzed by a MIRA3 TESCAN scanning electron microscope (SEM) equipped with an energy dispersive spectrometer (EDS). The length of the cracks was measured in 5 different sections, and the mean value was taken as the Average Total Crack Length (ATCL). The volume fraction and the length of the precipitates were also calculated using MIP4 image analysis software. Finally, the hardness of the weld zone in different specimens, were measured at 5 points, to achieve the average hardness.
Figure 1 represents the SEM micrograph of the sample structure extracted from the blade root, which is selected as the reference. The microstructure consists of γ-matrix, γ′ precipitates (both with the FCC crystal lattice), carbides, and γ–γ′ eutectic. The γ′-precipitates is composed of Ni3(Al, Ti) with homogeneous and coherent distribution in the matrix. These final solidification products have also been reported by Ojo, et al.8)
SEM micrograph of sample microstructure extracted from the blade (Non heat treated).
SEM micrographs of the samples with different heat treatments, based on the Table 2, are shown in Fig. 2. Regarding the non-heat treated specimen, as shown in Fig. 2(a), γ′ precipitates have two distinct morphologies, the primary cubic γ′ and the secondary fine spherical γ′ distributed into the primary ones. In all of the received images from the blade root, secondary fine particles were observed, which could be considered as an indication of non-deterioration part of the blade. Figure 2(b) illustrates the microstructure of heat treated sample at the temperature of 1000°C for 24 hours. This microstructure could be almost the same as a microstructure of a used turbine blade. As shown in the image, the microstructure consists of the coarser primary cubic γ′ precipitates compared to the non-heat treated sample, which can be attributed to the growth phenomena. Moreover, as it can be seen, there is no any sign of secondary fine particles, which means that almost all of the secondary particles has been dissolved. Regarding the heat treatment C which consists of exposing at 1125°C for 2 hours (let say standard treatment, adapted from the reference Wangyao et al.1)), microstructure consists of primary precipitates as well as secondary fine particles that have precipitated during cooling from the dissolution temperature, as shown in Fig. 2(c). It is obvious that the mentioned heat treatment is not able to affect coarse γ′ precipitates. Figure 2(d) shows the image of dissolution heat treatment at 1180°C. As it is well illustrated in the image, this heat treatment has largely dissolved the primary precipitates in the matrix phase, which can be due to the higher level of the dissolution temperature. Moreover, the morphology of the mentioned phase has been transferred to the rosette shape. The shape of the precipitates depends on the interface energy and strain energy between the γ′-particles and matrix.14,15) There is a close relationship between cooling rate from dissolution temperature, size and shape of the γ′-particles, interface energy and strain energy. Shape evolution growth mechanism of the particles, is a result of competence between the interfacial and strain energies. As the particle size increases, the effect of elastic energy becomes more than interfacial energy, resulting in the particle shape evolution from sphere to the cubic and complex flower shape particles. Microstructure consists of primary dendrite shape precipitates as well as secondary fine precipitations formed during the cooling phase from the dissolution temperature. Figure 3 represents the effect of preweld heat treatment on the secondary phases. Grain boundaries of sample B contains coarse γ′ precipitate particles as well as semi continuous carbides (Fig. 3(a)). Regarding the images taken from all the samples, it can be observed that the precipitates in interdendrite regions are bigger than the precipitates formed inside the dendrite core, and a higher amount of them are dissolved in the structure due to the solution heat treatment. The difference in dissolution behavior, size and distribution of γ′ in the core, and interdendrite regions, can be attributed to microsegregation during alloy solidification. The separation and migration of the elements, especially Ti, into the interdendrite zones leads to a higher solvus temperature of γ′ precipitates and consequently, after cooling, the size of interdendrite particles is larger than that of inside the dendrites. This initial difference in the size of the precipitates also causes interdendrite precipitates remained in the matrix after solution heat treatment, and also having larger sizes compare to precipitations in the dendrite core (Fig. 3(b)). It has been reported that the solvus temperature in interdendrite is about 1170°C and in the dendritic core is about 1120°C.16)
SEM micrographs of the microstructure of different tested specimens in this study, based on the Table 2: (a) specimen A (non-heat treatment), (b) specimen B (1000°C/24 h), (c) specimen C (1125°C/2 h), (d) specimen D (1180°C/2 h).
The effect of preweld heat treatment on the secondary phases, (a) Coarse γ′ and M23C6 boundary carbides (sample B, 1000°C/24 h), (b) Optical microscope image of the difference in precipitation dissolution in dendrite core and interdendrite regions (sample D, 1180°C/2 h).
The volume fraction and the average length of the precipitates for each sample were calculated and are presented in Table 3. As the table shows, the largest primary precipitate size belongs to samples B, C, A, and D respectively. During the heat treatment of specimen B, the γ′ precipitates join together to reduce their surface energy; and consequently precipitate size grows. During the heat treatment of sample C, due to prolonged exposure to higher temperature than dissolution, the secondary fine precipitates start to dissolve. At the same time, primary fine precipitates also continue to dissolve in the matrix and join to the coarse precipitates, result in reducing interfacial energy. During cooling from the heat treatment temperature, owing to the supersaturation of γ′ former elements in the matrix, secondary finer γ′ precipitates nucleate. Dissolution and joining of the primary precipitates to each other, results in decreasing the volume fraction and increasing the precipitate length compared to the sample A. Regarding sample D, prolonged exposure to elevated temperature, has generated higher dissolution rate and solubility of the matrix. Consequently, a higher amount of precipitates can be dissolved in the matrix, which results in reducing the volume fraction and precipitate size in this sample. Due to the high supersaturation of the matrix from the constituent elements of γ′ during cooling, the secondary fine precipitates start nucleation.
The results of hardness testing of the samples after various heat treatments are shown in Fig. 4. It shows that the specimens, which are subjected to partial solution heat treatment at the temperatures of 1125 and 1180°C, have highest hardness values. The least hardness belongs to the specimen B at 1000°C for 24 hours. Due to the dissolution of the most of γ′ Ni3(Al,Ti) precipitates, the amount of aluminum and titanium as the solute elements in matrix is increased. Therefore, the hardness of the specimen C and D is increased compared to the specimen B. It is also worth noting that higher hardness value of the specimen D compared to specimen C could be attributed to the contribution of Mo and W atoms as solid solution hardener agents of the matrix. As the Fig. 8(c) shows, these elements are present in carbide composition, consequently can be distributed in the matrix during heat treatment, while the temperature rises above the dissolution point of M23C6 and M6C carbides, i.e. 1065°C and 1150°C respectively.17) Lower hardness value of the specimen B compared to A, can be attributed to broader distribution of γ′ particles which result in higher amount of grain boundary area per volume between the particles and matrix, in specimen A.18)
Vickers hardness testing results for specimens after different heat treatments.
Moving from the unaffected base metal (BM) through the HAZ to the fusion zone (FZ) is accompanied by an increase in the rate of dissolution of precipitates, as shown in Fig. 5. This phenomenon can be attributed to exposure at higher temperature as well as being longer time in this temperature. It was also found that as the size of primary precipitates in different samples increases, the length of the dissolution zone reduces, which could be due to the fact that any increase in precipitate size, requires more time for dissolution in the matrix.
The HAZ zone microstructure, under C (1125°C/2 h) condition.
Evaluation of the cracking phenomenon in HAZ zone of all the samples reveals that the cracks start mainly close to the liquation boundary in the HAZ region, while extends to the base metal and weld metal in some cases. Most of the observed cracks found to be grain boundary cracks with a zigzag appearance, which is one of the characteristics of liquation cracking.7) Figure 6 shows the liquation cracking in samples B and C. The re-solidified liquid on one side of the crack which is another characteristic of the liquation cracking, is visible in the images 6(a) and 6(b). While the partially solidified grains are surrounded by liquid, strong contraction stresses cause liquation cracking to be occurred.2)
The SEM micrograph of liquation cracking in HAZ regions (a) sample C (1125°C), (b) sample B (1000°C).
The constitutional liquation of γ′ precipitates widely observed in the present study (Fig. 7). Figures 7(a) and 7(b) illustrate a magnified view of the crack edge and the obtained EDS analysis from point A, respectively. Results showed that the solidified dendrites toward the crack edge contain higher total amount of Al and Ti than the mass of the alloy (13.8 Vs 8.7 mass%), which confirms the constitutional liquation of γ′ particles. Around the primary γ′, fine precipitates are also seen. After solutionizing during the heating stage of the welding, fine γ′ particles are precipitated from the supersaturated areas by elements former γ′ particles (Al,Ti), during the cooling stage.
The constitutional liquation cracking of γ′ precipitations in the welding: (a) constitutional liquation cracking of γ′ precipitations in sample B (1000°C), (b) EDS analysis from point A.
As shown in Fig. 8(a), cracking mostly occurred in places where several secondary phases are contiguous. So that by melting, formation of a continuous liquid film is promoted. It has been reported that the front region of γ-γ′ eutectic structure, is one of the most susceptible areas to cracking. The reason can be attributed to the fact that low melting phases such as Cr–Mo carbides, sulfides, and borides are repelled and accumulated in those areas, which makes them vulnerable to cracking.8,19) Meanwhile, the constitutional liquation of carbides on one hand, and their intersection areas with matrix on other hand, makes their interface susceptible for cracking. For example, Fig. 8(b) shows the cracking in the adjacent with carbides. Figure 8(c) represents an EDS analysis from carbides showing richness of carbides by Ta, W, and Ti. Figure 9 shows the average total crack length (Av. TCL) per section for various welded specimens. As it can be seen, the maximum crack length belongs to the non-heat treated specimen and the minimum crack length belongs to the specimen under solutionized condition at 1180°C for 2 hours. Variables affecting the liquation cracking phenomenon can be summarized as the hardness of the matrix, the average length and volume fraction of γ′ primary precipitates, presence of secondary phases such as carbide and the existence of low melting phases such as borides. Simultaneous impact and the relative intensity of the mentioned parameters could largely influence the liquation cracking. Investigation of the hardness, inter-granular liquid, and total crack length of the samples revealed that while almost all of the HAZ region of sample B is covered by liquid film, but the Av.TCL is shorter than that of specimen A. The reason can be attributed to the fact that lower level of hardness of the sample B compared to A, resulted in a better damping capacity of sample B.19,20)
Cracking in HAZ regions (a) The SEM micrograph of cracking in front of γ-γ′ eutectic (sample A), (b) SEM micrograph showing cracking of the MC carbide particles, (c) EDS analysis from point A.
The average of total crack length of HAZ regions, after various heat treatments.
Decreasing trend of the crack length from the samples B, C and D respectively, implies that the size and volume fraction of γ′ particles, and the absence of low melting, are dominant. By examining the HAZ region of sample D as shown in Fig. 10, it is found that while the base metal has the highest hardness, no cracking has occurred. This could be due to the absence of any evidence of grain boundary liquid film, smallest size of γ′ particles, and the minimum volume of γ′ particles. Larger precipitates require more time for complete dissolution in the matrix during the rapid heating cycle of the welding process. They may change to liquid, based on γ + γ′ → L reaction, within the passing sequence from the eutectic temperature.21,22) In this sample, although the hardness is much higher than the other samples, but due to the dissolution phenomena and also reduction of the size of secondary phases, including γ′ precipitates, the amount of inter-granular liquid is reduced. These results are in agreement with the findings of previous studies.8,9,19)
Optical microscopy image of liquid film in the grain boundaries of sample D.
This research demonstrates that a crack-free welding of a very susceptible nickel base alloy can be performed successfully by a comprehensive understanding of all the factors involved. The following conclusions can be drawn:
The authors acknowledge Mr. Atabak Alizadeh, CEO of Turbine Atlas Company, for his technical assistance regarding the experimental work.
