2017 Volume 58 Issue 1 Pages 52-58
Creep behaviors and microstructures for two Ni-based heat-resistant alloys with different carbon contents were investigated. The chemical compositions of the alloys were Ni-20Cr-15Co-6Mo-1Ti-2Al-2Nb-0.004 and 0.021C (mass%). The 0.004C and 0.021C alloys are referred to as the low- and high-C alloys, respectively. After solid-solution treatment at 1373 K for 1 h and isothermal annealing at 1023 K for 32 h, fine Ni3Al (γ') particles were formed in the grain interior of both alloys. The average diameter and number density of γ' particles were similar in both alloys. M23C6 carbides were formed on grain boundaries after the isothermal annealing. Coverage ratios with the carbides in the high-C alloy were higher than that in the low-C alloys. Creep tests were performed at 1123 K and 130 MPa. The rupture time for the high-C alloy was longer than that for the low-C alloy, though both minimum creep rates were similar. In the high-C alloy, the creep strain was stored uniformly in the grain interior and the formation of a precipitate-free zone during the creep deformation was suppressed. Therefore, intergranular carbides with a high coverage ratio decreased the creep rate in the acceleration region.
Advanced ultra-supercritical (A-USC) thermal power plants are being developed worldwide to increase energy conversion efficiency and reduce environmental load.1,2) High energy conversion efficiency can be achieved by using higher temperature and pressure steam conditions. For example, in Japan, the maximum steam temperature and pressure conditions of A-USC plants are 973 K and 35 MPa. Under these conditions, Ni-based heat-resistant alloys are promising candidate materials for the main steam pipes, and header, superheater, and re-heater tubes because of the superior creep properties of these materials.
In typical Ni-based heat-resistant alloys, the creep strength is increased by precipitation strengthening of the Ni3Al (γ') phase in the matrix (γ) phase.3–7) In a single-crystal Ni-based superalloy and a directionally solidified Ni-based superalloy with excellent creep strength, which are used in turbine blades, the volume fraction of γ' phase reaches 70%.8) In these alloys, the change in γ' phase morphology during creep deformation, called rafting, strongly affects the creep strength.9–11) Polycrystalline Ni-based heat-resistant alloys, such as Alloy 617,12) Alloy 740,13) and HR6W,14) are expected to be used for boiler tubes and pipes. In these alloys, γ' particles also increase the creep strength, although high volume fractions of γ' particles decrease the hot workability substantially. The volume fraction of the γ' particles must be less than about 25% at 1073 K in candidates for Ni-based heat-resistant alloys for boiler tubes and pipes.15) Accordingly, Ni-based heat-resistant alloys should be strengthened by the precipitation of γ' particles in the grain interior and by other structure control methods. Recently, the microstructure near grain boundaries has been the focus of much research. Takeyama et al. developed Fe-20Cr-30Ni-2Nb (mol%) austenitic heat-resistant steels. Grain boundaries in this steel were covered by an intergranular Laves phase and its creep properties were superior to conventional austenitic heat-resistant steels.16) We expected that Ni-based heat-resistant alloys could be strengthened by intergranular precipitates. In addition, intergranular carbides form in Ni-based heat-resistant alloys containing C. These carbides do not decrease the hot workability because they are dissolved at the hot-working temperature of around 1373 K.
In this study, we prepared two types of Ni-based alloys with low and high C contents to control the volume fraction of intergranular carbides, and we investigated the difference in creep deformation behaviors between the alloys. The effect of intergranular carbides on the creep strengthening mechanism was also discussed based on microstructural analyses near the grain boundaries.
The chemical compositions of the two alloys prepared in this study were Ni-20Cr-15Co-6Mo-1Ti-2Al-2Nb-0.004 and 0.021C (mass%). The 0.004C and 0.021C alloys are referred to as the low-C and high-C alloys, respectively. The alloys were homogenized at 1373 K for 1 h and isothermally annealed at 1023 K for 32 h. The creep conditions were 1123 K and 130 MPa. Creep tests were interrupted at 2.8, 20, 186, 406, and 528 h in the low-C alloy and 2.8, 58.3, 353, 592, and 808 h in the high-C alloy. The microstructures in both alloys were observed by scanning electron microscopy (SEM; Ultra-55, Carl Zeiss), electron backscatter diffraction (EBSD), and scanning transmission electron microscopy (STEM; Titan G2 Cubed, FEI). The crept samples were cut along the loading direction. The SEM samples were wet polished with water-proof SiC sandpaper (#4000), buff polished using 1 μm diamond powder, and then mechanochemically polished with colloidal silica to remove the mechanically damaged surface layer. The crystal orientations were analyzed by SEM/EBSD. The thin foil samples for STEM were also cut parallel to the loading direction and mechanically polished to a thickness of about 0.1 mm. Subsequently, electropolishing was conducted with a mixture of perchloric acid and ethanol (HClO4/C2H5OH = 1:10, vol%) at 243 K and a voltage of 20 V.
From low-magnification SEM observations, the average grain sizes of the γ phase in both alloys were measured as 150 μm in diameter. Figures 1(a) and (b) show SEM images of the isothermally annealed low- and high-C alloys, respectively. Numerous spherical γ' particles were distributed homogeneously in the γ matrix of both alloys. The mean diameters of γ' particles are about 49 nm in the low-C alloy and 50 nm in the high-C alloy. The volume fractions of γ' particles are 15.2% and 14.6% in the low- and high-C alloys, respectively. The average surface distance, $\bar{\lambda}$, between γ' particles is estimated by17)
| \[ \bar{\lambda} = \alpha \bar{l}_s - \bar{d} \] | (1) |
| \[ \overline{l_s} = 1/\sqrt{N_s} \] | (2) |
SEM images around a grain boundary in isothermally annealed (a) low- and (b) high-C alloys, showing spherical γ' particles and intergranular M23C6 carbides.
As shown in Fig. 1, large precipitates with the darkest contrast are observed at the grain boundaries. These precipitates were identified as M23C6 carbides by selected area electron diffraction pattern. The coverage ratio of the intergranular M23C6 carbides to the total length of grain boundaries in the unit area in the high-C alloy is higher than that in the low-C alloy as discussed later. Moreover, in the high-C alloy, γ' particles are characteristically formed around the M23C6 carbides.
3.2 Creep deformation behaviorsThe strain, $\varepsilon$, as a function of time obtained in the creep tests is shown in Fig. 2(a). The semi-logarithmic and double-logarithmic plot of the creep rate, $\dot{\varepsilon}$, and time are shown in Figs. 2(b) and (c). The arrows in Fig. 2(b) indicate the time that the creep was interrupted. The rupture times are 574 h in the low-C alloy and 909 h in the high-C alloy. The creep strength in the high-C alloy is 1.6 times higher than that in the low-C alloy. The rupture strains of the low- and high-C alloys are 24% and 46%, respectively. Thus, the high-C alloy possesses higher creep strength and higher ductility. The creep behaviors of both alloys in the transient creep region coincide well (Fig. 2(c)). Subsequently, the steady-state creep region is hardly observed, and the acceleration creep region appears promptly. Generally, the increase in rupture time and minimum creep rate is constant, which is known as the Monkman-Grant relationship.20) In this relationship, the increasing rupture time could be explained by the decreasing minimum creep rate. However, the higher strength of the high-C alloy is caused by the decrease in the creep rate in the accelerated creep region, rather than by the reduction of the minimum creep rate.
(a) Creep curves of the low- and high-C alloys crept at 1123 K and 130 MPa. (b) Semi-logarithmic and (c) double-logarithmic plots of creep rate ($\dot{\varepsilon}$), and time (t). Arrows in (b) indicates the interrupted time to obtain the samples for microstructural characterizations.
Figure 3 shows the evolution of γ' particles during creep testing in the gauge portion. In all samples, spherical γ' particles are distributed uniformly, as in the isothermally annealed alloys (Fig. 1), although the size of the γ' particles increases during creep testing. The $\bar{d}$ and $\bar{\lambda}$ values of the gauge and grip portions in the crept samples (Fig. 4) are evaluated quantitatively from the SEM images. $\bar{d}$ is approximately 50 nm after isothermal annealing, and then $\bar{d}$ increases gradually during creep deformation. $\bar{\lambda}$ shows the same trend. Consequently, the threshold stress by γ' particles becomes lower during creep deformation. It causes the creep deterioration. Generally, the growth rate of precipitates changes with annealing under applied stress.21,22) In this study, the growth of γ' particles is same in the gauge and grip portions. Thus, the applied stress does not affect the changes in the size of the γ' particles. Moreover, the coarsening behaviors of $\bar{d}$ and $\bar{\lambda}$ and the precipitation strengthening are similar in the low- and high-C alloys. Figure 5 shows a logarithmic plot of the creep strain rate (log $\dot{\varepsilon}$) versus the average surface distance, $\bar{\lambda}$, of interrupted alloys in the acceleration creep regions. Log $\dot{\varepsilon}$ and $\bar{\lambda}$ are fitted with a linear relationship. The lines for the low- and high-C alloy do not coincide, although the coarsening behaviors of $\bar{\lambda}$ in both alloys are the same (Fig. 4(b)). $\dot{\varepsilon}$ is smaller for the high-C alloy than for the low-C alloy, indicating that different strengthening mechanisms may occur in the high-C alloy.
Microstructural evolution of γ' particles during creep deformation in low- (top) and high-C alloys (bottom).
Changes in (a) average diameter ($\bar{d}$) and (b) surface distance ($\bar{\lambda}$) of γ' particles as a function of creep time.
Logarithm creep strain rate ($\dot{\varepsilon}$) versus the average surface distance ($\bar{\lambda}$) of interrupted alloys in the acceleration creep regions.
Figure 6 shows the microstructure near a grain boundary in the crept low-C alloy. In the crept sample interrupted at ε = 1.0% (Fig. 6(a)), there is a precipitate-free zone (PFZ) along the grain boundary (white arrow). However, there is no PFZ around the intergranular M23C6 carbides (black arrow). Similar PFZ formation has been reported in other crept alloys.23,24) Hornbogen25) proposed that the PFZ was generated by the grain boundary migrating and the γ' particles redissolving. The main driving force of the grain boundary migration is the energy stored near the grain boundary during creep. Therefore, the width of the PFZ increases gradually during the creep test along the grain boundary, except for around the intergranular M23C6 carbides (Fig. 6(b)). Notably, there is still no PFZ around the intergranular M23C6 carbides (black arrow). However, no PFZ is observed in the crept high-C alloy interrupted at ε = 0.7% (Fig. 7 (a)). In the crept sample interrupted at ε = 14.3% (Fig. 7 (b)), several PFZs are formed in places where the grain boundary is not covered with M23C6 carbides. Based on these observations, the intergranular M23C6 carbides prevent the formation of PFZ due to the pinning of grain boundary migration. We discuss the relationship between the intergranular carbides and PFZ quantitatively by the following methods. Coverage ratio f of the intergranular M23C6 carbides and formation ratio F of PFZ in a unit area are defined as
| \[ \boldsymbol{f} = \boldsymbol{f_{c}}/\boldsymbol{f_{GB}} \] | (3) |
| \[ \boldsymbol{F} = \boldsymbol{f_{PFZ}}/\boldsymbol{f_{GB}} \] | (4) |
SEM images near the grain boundary in low-C alloy creep interrupted at (a) 186 and (b) 528 h. The white arrows indicate the PFZ and the black arrows indicate the intergranular M23C6 carbides.
SEM images near the grain boundary in the high-C alloy creep interrupted at (a) 353 and (b) 808 h. The white arrows indicate the PFZ and the black arrows indicate the intergranular M23C6 carbides.
(a) Coverage ratio (f) of carbides and (b) the formation ratio of PFZ (F) to the total grain boundaries in each of the creep interrupted alloys.
The plastic strain accumulated during the creep deformation is analyzed by using SEM-EBSD. Figure 9 shows kernel average misorientation (KAM) maps in the creep-interrupted low-C samples. The KAM value is an average misorientation between a given point and all of its neighbors.27) The KAM value rarely changes between near the grain boundary and the grain interior in the transient creep region (Fig. 9 (a)). In contrast, because the KAM values are increased locally near the grain boundary in the acceleration creep region (Fig. 9(b)), the plastic strain is inevitably stored near the grain boundary. Figure 9(c) shows an enlarged micrograph of the framed area in Fig. 9(b). The plastic strain is localized in the region 5 μm from a grain boundary; thus, we defined this region as the area near grain boundaries. The KAM values near the grain boundary, $\bar{\theta}_{gb}$, and in the grain interior, $\bar{\theta}_{gi}$, were estimated from KAM maps. Figure 10 shows the KAM values as a function of creep time. The changes in $\bar{\theta}_{gb}$ and $\bar{\theta}_{gi}$ are different for the low- and high-C alloys, although $\bar{\theta}_{gb}$ is larger than $\bar{\theta}_{gi}$ in all the samples. $\bar{\theta}_{gb}$ increases earlier in the low-C alloy than in the high low-C alloy (Figs. 10(a) and (b)). Figure 10(c) shows the relationship between creep time and $\bar{\theta}_{gi}$/$\bar{\theta}_{gb}$. In the isothermally annealed low- and high-C alloys, both $\bar{\theta}_{gi}$/$\bar{\theta}_{gb}$ ratios are about 1.1. This value is generally accepted as being without deformation, because the Kikuchi pattern is unclear near the grain boundary. In the low-C alloy, the ratio gradually increases up to 186 h, and then abruptly increases to 1.76 at 528 h. The ratio in the high-C alloy increases little up to 592 h and then increases to 1.41 at 808 h. Therefore, grains in the high-C alloy are deformed homogeneously until the later acceleration creep region.
KAM maps of low-C alloys creep-interrupted at (a) 20, (b) 406 h, and (c) enlarged micrograph of square area with dotted lines.
Plots of $\bar{\theta}_{gi}$ and $\bar{\theta}_{gb}$ as a function of creep time in (a) low- and (b) high-C alloys. (c) $\bar{\theta}_{gi}$/$\bar{\theta}_{gb}$ as a function of creep time in low- and high-C alloys.
The formation of the PFZ and the accumulation of plastic strain near grain boundaries occur simultaneously (Figs. 8 and 10). In other words, the PFZ decreases the creep strength. Therefore, the relationship between PFZ and the localization of strain should be clarified.
Figure 11 is a STEM bright field image taken near the PFZ in the high-C alloy crept for 808 h. Contrary to expectations, few dislocations are observed in the PFZ, although there are many dislocations and sub-grain boundaries in the γ matrix adjacent to the PFZ. Therefore, the increase of $\bar{\theta}_{gb}$ (Fig. 10(b)) is caused by the dislocation structure developed in the γ matrix near the PFZ, since a dislocation density is quite low as mentioned above. Then, we discuss the mechanism of the development of the dislocation structure near the PFZ by the PFZ formation. There are no differences in the chemical compositions measured by using STEM-EDS between the PFZ and the adjacent γ matrix. In addition, the hardness of PFZ is lower than that of the γ matrix because there are no γ' particles in the PFZ. Tomota et al. investigated stress redistribution between soft ferrite and hard cementite phases in plastic-deformed pearlite steels by neutron diffraction and found that the stress is accumulated in the cementite phase.28) Hirata et al. also reported that the acceleration of the creep rate in cold-rolled high-Mn austenitic stainless steel is due to stress redistribution between soft recrystallized and hard deformed grains.29) In the Bailey-Orowan model, the effective stress is proportional to the internal stress in the particle-distributed hardening alloys.30,31) In the region adjacent to the PFZ, the internal stress is increased by stress redistribution. The dislocation substructure is developed to reconcile the external stress and internal stress. For example, Sherby et al.32) and Sawada et al.33) revealed that the sub-grain width increases to the stationary values determined by the creep stress. Additionally, the development of the dislocation substructure assists the expansion of the PFZ region. Thus, the development of the dislocation substructure and the formation of PFZ occur in an autocatalytic reaction and accelerate the creep rate. The high-density intergranular precipitates in the high-C alloy retard the formation of the dislocation substructure and PFZ due to the pinning of the grain boundary, suppressing the acceleration of the creep rate (Fig. 2). Consequently, intergranular M23C6 precipitates are an effective strengthening factor equal to interior γ' precipitates.
STEM images of the dislocation structure near the PFZ at a grain boundary (dashed line) in a high-C alloy creep interrupted at 808 h, showing stress-redistribution between the PFZ and matrix regions.
Creep tests at 1123 K and 130 MPa were performed in Ni-based heat-treated alloys with low and high C contents, and the effect of intergranular precipitates on creep properties was examined. Our results are summarized as follows.
