Search for articles:
Jingxia Sun
Corresponding authors:
Received: 2018-11-16
Revised: 2019-01-19
Accepted: 2019-02-1
Online: 2019-11-05
Copyright: 2019 Editorial board of Journal of Materials Science & Technology Copyright reserved, Editorial board of Journal of Materials Science & Technology
More
Abstract
To examine the influences of minor modification of Al content on the microstructural stabilities and stress rupture properties, two alloys with minor difference in Al content were exposed isothermally at 1100 °C for 100 h, 500 h, and 1000 h, respectively. The microstructures were characterized before and after thermal exposure. It was found that when Al content was decreased by 0.4 wt %, the volume fraction γ′ decreased by 4 %, the size of γ′ increased by 40 nm, the matrix channel width increased by 5 nm, and the misfit degree of γ/γ′ phases increased by 0.006 % after heat treatment (HT). During thermal exposure, the alloy with low Al content had a better resistance to coarsening of γ′ phase and precipitation of μ phase. The γ′ particles of the alloy with high Al content tended to connect each other and coarsened along <100>directions; however, the γ′ particles of the alloy with low Al content remained cubic after 500 h. A serious coarsening behavior took place in the two alloys after aging for 1000 h. The structural stabilities were significantly improved. However, the change of 0.4 wt % Al content was found to have little effect on the high temperature stress-rupture properties.
Keywords：
Ni-base single crystal (SX) superalloys are widely used for advanced gas turbines due to their excellent mechanical properties at elevated temperature. Great efforts have been made to improve the high-temperature creep properties resulted in the development of several generations of SX superalloys [1,2]. The superior stress-rupture properties of recently developed SX superalloys are mainly becanse of their large contents of refractory elements such as Mo, Re and W [3,4]. However, microstructural instability at elevated temperatures, causing γ′ coarsening and the formation of topologically close packed (TCP) phases, can be a concern for the second and third generation alloys, since the mechanical properties can be adversely affected [5,6]. The occurrence of γ′ coarsening depends on the exposure temperature and duration, which can reduces the tensile and low-cycle fatigue properties of the SX superalloys [1]. Excessive amounts of Cr, Mo, W, and Re promote the precipitation of the TCP phases during service. TCP phases are very detrimental because of two main factors. At first, when they nucleate and grow, they deplete the γ-matrix phase from solid solution strengtheners. This leads to decreased creep resistance of the material [7]. Further, TCP phases are brittle and exhibit complex morphology [8,9]. Thus, cracks are easily nucleated. As a consequence the fatigue life-time and tensile ductility of the material will be decreased [10,11].
A lot of studies have carried out to improve the structural stability of superalloys by a judicious control of the alloying elements, but most of the works focused on the refractory elements. It is generally considered that Re additions has been shown to dramatically increase creep strength despite it provoked the TCP phases formation [11,12]. Some researchers have studied the platinum group elements such as ruthenium, which tends to inhibit the precipitation of TCP phases mostly by decreasing the driving force for TCP phase precipitation [4,13]. Zheng et al. [14] reported that Ta enhanced the stress-rupture life at intermediate temperature but weakened high temperature stress-rupture properties in polycrystalline Ni-base superalloys. The works of Wang et al. [15] showed that increasing W content accelerated the connection and deformation of γ′ phase and increased the area fraction of TCP phases during the thermal exposure. However, few studies have shown the effect of Al on the microstructural stability of SX superalloys.
The microstructure typically found in SX superalloys consists of the ordered L12 γ′ phase (Ni3Al) embedded in the disordered FCC γ matrix [16]. Obviously, Al plays a crucial role in the superalloys. Al is known to be γ′ former, which determines the volume fraction of two phases. It is well established that Al has the greatest influence on electron vacancy at the same mass fraction compared with other elements. Some studies proposed that the average electron vacancy of alloys relates to TCP phases [17]. When the average electron vacancy of the alloy is higher than the critical value, there is a tendency to form TCP phases. In contrast, when the average electron vacancy is lower than this, it is hard to form a TCP phase. The computer programs such as PHACOMP are used to predict alloy stability [18].
The material studied in the present work is an experimental SX superalloy, which belongs to the third generation. The objective of this work is to improve the microstructural stability of the alloy without decreasing the stress-rupture properties. Therefore, two alloys with different Al contents were designed to investigate the microstructural stability and related stress-rupture properties. After thermal exposure at 1100 °C, the evolution of the γ/γ′ microstructure in the two alloys was compared, as well as the precipitation of TCP phases.
The master alloys of both experimental alloys were melted in a vacuum induction furnace. Al in different contents was separately added in the molten alloy, while the composition of other elements remained basically the same. The nominal compositions of both alloys, S1 and S2 alloy (S1 alloy denotes the alloy with high Al content, S2 alloy denoted as the alloy with low Al content), are listed in Table 1. The cylindrical bars with [001] orientation were cast in a directional solidification furnace using spiral selector. The two alloys were heat treated by following standard scheme: 1335 °C/16 h + 1340 °C/16 h (AC) + 1150 °C/4 h (AC) + 870 °C/24 h (AC). To study the microstructure of the two experimental alloys, specimens with length 10 mm were cut and thermally exposed at 1100 °C for 100, 500, 1000 h, respectively. The HT bars were machined into stress-rupture specimens to investigate the stress-rupture properties at 1120 °C and 137 MPa.
Table 1 Nominal chemical compositions of experimental alloys (wt%).
| Alloy | Cr | Co | Mo + W + Ta | Al | Re | Hf | Ni |
|---|---|---|---|---|---|---|---|
| S1 | 3 | 12 | 15 | 6.1 | 5 | 0.1 | balance |
| S2 | 3 | 12 | 15 | 5.7 | 5 | 0.1 | balance |
The specimens after thermal exposure were mechanically polished and etched in a solution of 5 g CuSO4 + 20 ml HCl + 80 ml H2O. The microstructure of these samples was examined using a JSM-7100 F field emission scanning electron microscope (SEM). Image-Pro Plus 6.0 software was used to calculate the average size and the volume fraction of γ′ phase as well as the width of matrix channel. The cubic γ′ particle is equivalent to a two-dimensional square. Thus the area percentage is the volume percentage of γ′ phase after heat treatment. Rigaku D/MAX 2500 X-ray diffraction (XRD) was used to measure lattice misfit of γ/γ′ phases at room temperature with the incident ray CuK, 50 kV voltage, current 300 mA, step width 0.02°. Thin foils for TEM observation were cut from HT and thermal exposed specimens, which were ground to 50 μm in thickness and finally prepared by twin-jet thinning electrochemically in a solution of 10 % perchloric acid in alcohol at -20 °C. A JEOL JEM-2100 F (200 kV) transmission electron microscope (TEM) was used to observe TCP phases and dislocation configuration after aging.
Typically, a lattice parameter misfit, δ, exists because the γ′ phase has a smaller lattice parameter than the γ matrix, which is temperature-dependent due to the different coefficients of thermal expansion [16,19]
δ=$\frac{2(α_{γ´}-α_γ)}{α_{γ´}+α_γ}$. (1)
The misfits of both experimental alloys were measured by XRD under HT condition at room temperature. The {004} diffraction peaks of γ and γ′ phase were treated by a three peak fitting model due to the lattice distortion of γ phase. Therefore, the horizontal and vertical matrix channels generated two separate γ matrix peaks γI and γII for the two alloys (Fig. 1) [20]. It should be noted that the Kα2 was removed. The average lattice parameter and misfit have been calculated by the following equations [21]:
$a_{γav}=\frac{a_{γI}I_{γI}+a_{γII}I_{γII}}{I_{γI}+I_{γII}}$, (2)
$δ_{av}=\frac{δ_{γI}I_{γI}+δ_{γII}I_{γII}}{I_{γI}+I_{γII}}$ (3)
where IγI and IγII are the integrated intensities of the γI and γII peaks, δγI and δγII are the lattice misfits between γ′ and γI and γII, respectively. According to Eqs. (2) and (3), the average lattice parameter and misfit are aγav S1 = 0.898 nm, aγav S2 = 0.897 nm, δS1 = -0.069 %, δS2 = -0.063 %, respectively.
Fig. 1. Profiles of {004} diffraction peaks and peak fitting results of the: (a) S1 and (b) S2 alloy.
The average lattice misfit of S1 alloy is more negative than that of S2 alloy since the reduction of Al content leads to more Ta atoms to combine with Ni matrix to form Ni3Ta and the lattice parameter of γ′ phase increased due to a relatively large atomic radius of Ta. Moreover, because the volume fraction of γ phase increased with the reduction of Al content (Table 2), the partition of refractory elements to the γ phase has been reduced. According to Eq. (1), S1 alloy has a more negative lattice misfit.
Table 2 γ′ morphology, γ′ volume fraction, γ′ size and γ channel width in the dendrite core of experimental alloys after full heat treatment.
| γ′ morphology | γ′ volume fraction (%) | γ′ size (μm) | γchannel width (nm) | |
|---|---|---|---|---|
| S1 | Cuboidal | 70.0 | 0.30 | 60 |
| S2 | Cuboidal | 67.0 | 0.34 | 65 |
Table 2 summarizes the morphology, volume fraction, the average size of γ′ phase as well as the width of γ channel in dendrite core of the two alloys after HT. Apparently we can see the average particle size and the channel width of S1 alloy are smaller than those of S2 alloy, in contrast, the γ′ phase volume fraction of S1 alloy is larger than that of S2 alloy.
Apparently, the volume fraction of γ′ phase is directly determined by Al content. However, it is well accepted that the γ′ morphology is closely related to the sign and magnitude of the γ/γ′ lattice misfit [22]. As for SX superalloys, γ′ phase coherently precipitates from γ matrix, and thus the total elastic strain energy depends on the morphologies and elastic properties of γ matrix and γ′ precipitates. Assuming the total elastic strain energy ΔGS is independent of the shape of γ′ phase. Thus, the following equation is obtained [23]:
ΔGS∝μδ2∙V (4)
where μ is the shear modulus of matrix, δ is the lattice misfit and V is the volume of γ′ precipitate. It can be seen that the elastic strain energy induced by coherent strain is proportional to the volume of γ′ precipitate and the square of lattice misfit. Consequently, the larger absolute value of lattice misfit leads to a smaller γ′ precipitate size.
Fig. 2 represents the microstructures of the two alloys subjected to thermal exposure at 1100 °C from 100 to 1000 h. As can be seen from the diagram, the continuous variation of γ′ precipitates and γ channels changes with exposure time. Moreover, the coarsening behavior of γ′ phase aggravates with exposure time. After exposing for 100 h, both alloys still remain cuboidal shape except for the increase in size. At the same time the particle size and matrix width of S2 alloy are slightly less than those of S1 alloy. When exposed for 500 h the cuboidal γ′ precipitates have coarsened into rafts aligned along the <100> orientation in S1 alloy (Fig. 2(b)), whereas in S2 alloy there are still present many cuboidal γ′ particles (Fig. 2(d) and (e)). A serious coarsening behavior occurs in both alloys when exposed for 1000 h. The measured γ′ area fraction in dendrite core as a function of thermal exposure time is shown in Fig. 3(a). During thermal exposure, the γ′ area fraction does not vary significantly in the two alloys, whereas the average γ′ particle size and γ channel width, Fig. 3(b), reaches tens and hundreds of times more than the initial value of S1 alloy, respectively. That is to say, the resistance to γ′ coarsening of S1 alloy is worse than that of S2 alloy.
Fig. 2. Morphologies of γ′ phase in dendrite core of S1 alloy (a)-(c), S2 alloy (d)-(f) alloys after thermal exposure for 100 h (a, d), 500 h (b, e) and 1000 h (c, f) at 1100 °C.
Fig. 3. Evolution of γ′ phase in dendrite core during thermal exposure: (a) γ′ area fraction, (b) γ′ particle size, (c) γ channel width.
Dislocations deposit on the γ/γ′ interfaces when γ′ coarsening takes place as seen in Fig. 4. Combined with Fig. 2, the precipitates of S1 alloy remain discrete and coherent when thermal exposed for 100 h. It was observed that the sizes of the precipitated phase and the matrix channel increases compared with the HT condition (Fig. 4(a) and (b)). At this stage, the driving force for coarsening is the decrease in interfacial energy. This process follows the cubic kinetics called Ostwald ripening [24], i.e., larger particles grow at the expense of smaller ones to dissolve. When thermal exposed for 500 h, the dislocations deposit on the γ/γ′ interfaces with regular distribution (Fig. 4(c) and (d)) under assistance of thermal activation and time.
Fig. 4. Dislocation configuration in samples exposed with different time at 1100 °C of S1 alloy: (a) 0 h, (b) 100 h, (c) 500 h, (d) 1000 h.
The creep resistance of superalloys is controlled by γ′ coarsening which is related to the lattice misfit of γ/γ′ phases. During thermal exposure, γ′ phase exhibits isotropic growth driven by the reduction of chemical interface energy on the early stage, and then the γ′ phase only becomes greater in size but still maintains cuboidal shape. With the increase of the exposure time, the size of γ′ phase increases to such an extent that the lattice misfit cannot be accommodated by elastic strain [19]. So the preexisted dislocations during heat treatment can overcome the Orowan resistance in γ matrix channels driven by the misfit stress, then the misfit dislocations deposit on the γ/γ′ interfaces (Fig. 4(c)) where the matrix channels with the highest misfit stress due to inhomogeneous distribution of component [4,25,26]. Therefore, the misfit stress is largely (but not completely) relieved, and then chemical potential of γ forming elements in these matrix decreases. Furthermore the γ forming elements such as Co, Cr, Mo, Re in the matrix without misfit dislocation will diffuse towards the regions where those misfit dislocations are deposited. On the contrary, the γ′ forming elements such as Al, Ta will diffuse along inverse direction in γ′ phase, and thus the γ′ phase extends and connects each other along the direction parallel to the interface with misfit dislocation deposition [27]. That means the decrease of misfit strain energy is the driving force of γ′ coarsening during thermal exposure, And the driving force of the is enhanced with the increase of lattice misfit.
The precipitation of TCP phases in dendritic core was observed by SEM under backscattering mode, as presented in Fig. 5. The TCP phases are imaged as white particles, while the γ/γ′ microstructure corresponds to black color. As can be seen in Fig. 5(a), in the early stage of precipitation TCP phases exhibit “needles” morphology inclined to γ/γ′ phase boundaries with 45° angle. With increasing the exposing time, the number and size of TCP phases increase and three different morphologies are presented, which are needle-like, rod-like and blocky. According to Ref. [28], it can be determined that the TCP phases in the alloys are all μ phase, which is commonly observed in Re-containing Ni-based single crystal superalloys during thermal exposure. The μ phase is mainly composed of Ni, Cr, Co, W and Re, most of which segregate into the matrix. The decrease of Al content increases the volume fraction of γ phase, which leads to the decrease of the concentration of μ phase forming elements in the matrix. Therefore, the precipitation of TCP is reduced to a large extent.
Fig. 5. Morphologies of TCP phases in dendrite core under backscattering mode of S1 alloy (a)-(c), S2 alloy (d)-(f) alloys after thermal exposure for 100 h (a) and (d), 500 h (b) and (e) and 1000 h (c, f) at 1100 °C.
The stress-rupture properties under 1120 °C/137 MPa of S1 and S2 alloys as HT state are shown in Fig. 6. It shows that the stress-rupture life increases with the decrease of Al content, and the elongation does not change much. The direction of the rafted microstructure in dendrite core is perpendicular to longitudinal section, Fig. 7, which is called N type rafting resulted from γ′ phase directional coarsening. One way to explain this process is to consider the negative lattice misfit and the stress direction during service temperature [29,30]. The effect of this inevitable microstructure evolution on creep properties is not yet well understood, but it is certain that the rafting structure will gradually develop towards topological inversion when alloy is thermally exposed for a prolonged time (if the volume fraction of γ′ in both alloys are > 50%). Topological inversion, i.e., the γ′ phase surrounds the γ phase and becomes topologically the matrix, has been considered in the frame of microstructure stability [5]. Comparing Fig. 6(a) and (b), it is striking that the tendency of topological inversion of S1 alloy is more serious, although the duration of S1 alloy is shorter than that of S2 alloy. Recently, it has been reported that topological inversion lowers yield stress and tensile ductility at room temperature [31].
Fig. 6. Stress-rupture properties of two alloys as heat treated state under 1120 °C/137 MPa.
Fig. 7. Rafted microstructures of two experimental alloys stress ruptured under 1120 °C/137 MPa, light phase is γ, dark phase is γ′: (a) S1 and (b) S2 alloy.
Several important factors can influence the strengthening mechanism, to different extents, in the SX superalloy. These factors are: lattice misfit, volume fraction of γ′ phase, width of γ-channel and the refractory elements content in the superalloy. As precipitation strengthening alloys, γ′ precipitates are the primary strengthening phase. Harada and Murakami [32] reported that the creep property of Ni-based single crystal superalloys was improved by changing the lattice misfit towards more negative, because the net stress in the γ-channel of alloys with negative lattice misfit was lower than that of alloys with positive lattice misfit for loading along the <001> direction. Pollock and Argon [33] argued that the Orowan resistance to the bowing of dislocations through narrow γ matrix channels to be inversely proportional to the width of γ-channel. Therefore, the decrease of Al content significantly improved the structural stabilities, but there are some changes may have an adverse effect on the creep resistance, such as the decrease of the lattice misfit and volume fraction of γ′ phase and the increase of the width of γ-channel. Under the combined action of the above factors, the stress-rupture life slightly improved after decreasing Al content. Additionally, it is frequently suggested that the precipitation of μ phase would reduce the stress-rupture strength of the superalloys by scavenging -Re, W, Cr and Co from the γ matrix, thereby lowering the solution strengthening effects [7]. However, Pessah [2] confirms that a significant softening effect will occur only if the volume fraction of μ phase exceeds (5-vol.%). According to Fig. 5, a large number of precipitated phases appear after 500 h in the current study. Therefore, it is indicated that the failure of both alloys may be attributed to the coalescence of the strengthening γ′ phase rather than to the precipitation of μ phase. However, although lowering Al content has little effect on the rupture properties, the improvement of the structural stabilities is very important for other properties of superalloys. Liu et al. [31] reported that topological inversion by tensile creep in the SX superalloys N5, CMSX-2, CMSX-6, CMSX-11B and PWA1483 caused a marked drop in the yield stress at room temperature. Recently, it has been reported by Nazmy et al. [1] that the rafted structure exhibits low-cycle fatigue strength at all temperatures. Similar conclusion was reported in the reference [28].
The aim of the present study was to find the mechanism of Al content on the microstructural stability in the experimental alloy belonging to the third generation. Based on the investigations, the following conclusions can be drawn:
(1) γ′ morphology varied slightly by changing 0.4 wt % Al. Lowering Al leads to larger size of γ′ particle and smaller volume fraction.
(2) Lattice misfit is more negative with high Al content. Large γ/γ′ lattice misfit promotes severe γ′ coarsening.
(3) Minor modification of Al can bring significant influence on microstructural stability, resulting in retardation of γ′ phase coarsening and smaller volume fraction of TCP phases. The precipitation of μ phase was basically eliminated due to Al reduction.
(4) The high temperature stress-rupture properties changed slightly when Al content was lowered. The tendency to topological inversion is enhanced with the increase of Al content after high-temperature stress-rupture.
This work is partly sponsored by The National Key Research and Development Program of China (No. 2018YFB 1106600).
WeChat
/
| 〈 |
|
〉 |
