J. Mater. Sci. Technol., 12, 28(3), 214 2. Modes of Grain Selection in Spiral Selector during Directional Solidification of Nickel-base Superalloys Xiangbin Meng 1,2), Qi Lu 1,2), Jinguo Li 1), Tao Jin 1), Xiaofeng Sun 1), Jun Zhang 3), Zhongqiang Chen 4), Yanhui Wang 4) and Zhuangqi Hu 1) 1) Institute of Metal Research, Chinese Academy of Sciences, Shenyang 1016, China 2) Graduate School of Chinese Academy of Sciences, Beijing 0049, China 3) State Key Laboratory of Solidification Processing, Northwestern Polytechnical University, Xi an 7072, China 4) Aviation Industry Corporation of China (AVIC) Shenyang Liming Aero-engine (Group) Corporation LTD., Shenyang 1043, China [Manuscript received March 24, 11, in revised form July 28, 11] The modes of grain selection in spiral selector were investigated by both a ProCAST simulation and experimental confirmation. The results show that the efficiency of grain selection in starter block is associated with the geometry shape. At the early stage of grain selection, the optimization of grain orientation is dominated by competitive grain growth, but the optimization of grain orientation in starter block is gradually dominated by geometry shape at the later stage of grain selection. Besides, the spiral part could also optimize the orientation of the selected single crystal when the initial angle is large enough, and the single crystal selection in spiral parts with different pitch lengths and initial angles is dominated by different modes. The simulation results agree well with experimental ones. KEY WORDS: Ni-base single crystal superalloys; Competitive grain growth; Geometry shape; Directional solidification; Spiral selector 1. Introduction In the past decades, Ni-base superalloys single crystal turbine blades have been used extensively in modern gas turbine for aircraft propulsion and power generation. Compared with conventionally casting blades, single crystal turbine blades can endure higher service temperature and possess superior mechanical properties due to the elimination of grain boundary. To obtain the single crystal structure, spiral selectors are usually employed during directional solidification. Since the spiral selector plays an important role in single crystal selection, many researches have been carried out to investigate the process [1 9] and the mode of grain selection [ 15] in the spiral selector during directional solidification. A spiral selector consists of two parts, a cylin- Corresponding author. Prof., Ph.D.; Tel.: +86 24 83978872; Fax: +86 24 83971857; E-mail address: jgli@imr.ac.cn (J.G. Li). drical starter block and a spiral part in the form of helicoids with 1 to 2 spiral turns, as shown in Fig. 1. The conventional viewpoint is that the starter block is used to optimize the grain orientation to acquire columnar grains close to <001> orientation by competitive grain growth [2,9], and the spiral part is used to efficiently and randomly select a single crystal by blocking dendrites growth, which depends on the geometry shape of spiral part [2,3,9]. However, the grain orientation cannot be optimized in spiral part as it is in starter block. In this paper, the process of grain selection in spiral selector was investigated by numerical simulation and experimental confirmation. For the models of starter block, the diameter is changed to study the effect of geometry shape on grain selection; and for the models of spiral part, the diameter and turn of spiral part are fixed, the pitch length and initial angle of spiral part are changed to study the effect of the grain selection route on single crystal selection.
X.B. Meng et al.: J. Mater. Sci. Technol., 12, 28(3), 214 2. 215 Table 1 Dimension parameters of starter blocks No. Diameter/mm Height/mm Position/mm Case 1 8 0 40 Case 2 12 0 40 Case 3 16 0 40 Case 4 0 40 Case 5 30 0 40 Table 2 Dimension parameters of spiral parts No. L p/mm θ/ deg. D/mm d/mm r Case 6 13 5 2 Case 7 15 30 13 5 2 Case 8 40 13 5 2 Case 9 30 50 13 5 2 Case 45 60 13 5 2 Case 11 70 70 13 5 2 Case 12 30 45 15 5 1 Fig. 1 Schematic diagram showing the parameters in the spiral selector Furthermore, the modes of grain selection in starter block and spiral part were discussed. These results are expected to deeply understand the process of grain selection in spiral selector, and instruct the design of spiral selector in practical production. 2. al 2.1 Model design The parameters of spiral selector are labeled in Fig. 1, including the height (H S ) and diameter (D S ) of starter block; the length of screw pitch (L P ), the turn of spiral (r), the initial angle of spiral (θ), the diameter of spiral (D), and the diameter of helicoid (d) of spiral part. However, the parameters of spiral parts (L P, θ, D and d) are not all independent. Only three of them are independent, and the fourth can be deduced from the equation: L p = 2(D d) tan θ (1) Geometry parameters of starter block and spiral part are listed in Table 1 and Table 2, respectively. These models include 12 cases which are divided into two groups. The starter block model was designed as cylinders with different diameters (φ8/12/16//30 mm 0 mm), as listed in Table 1. A φ16 mm 25 mm starter block was positioned at the bottom of spiral parts to simulate the entrance of columnar grains in spiral part [9], and the spiral parts were designed with the same diameter but different routes of grain selection by varying pitch lengths and initial angles, as listed in Table 2. 2.2 al procedure Directional solidification experiments were carried out in a Bridgman industrial vacuum furnace with two heating zones and a withdrawal chamber. A typical second generation superalloy CMSX-4 with the nominal composition of 2.9Re, 6.4Cr, 9.6Co, 0.6Mo, 6.4W, 6.6Ta, 5.7Al, 1.0Ti, 0.1Hf and balance Ni (wt%) was used for the present study. After induction melting in the crucible, the master alloy was poured at about 1500 C into a ceramic mold mounted on a water cooled copper chill plate in the center of the furnace chamber. After 600 s for thermal equilibrium, the mold was withdrawn at a rate of 5 mm/min. More details about the parameters during directional solidification have been reported previously [9]. After casting, the process of grain selection was investigated by electron backscatter diffraction (EBSD) and image-analysis instrument software. The grain structure and orientation evolution in the spiral selector were observed by EBSD in an S-3400N SEM, equipped with Channel+ software for automatic grain indexing and mapping. The EBSD was carried out over several zones multiplied by 50 times on every sample by area scan and linear scan with the steps of µm and 50 µm, respectively. The average grain size was analyzed by the use of image-analysis instrument SISC-IAS software to investigate the evolution of grain density at the corresponding positions. 3. Results 3.1 Thermal condition The ProCAST software based on a Cellular Automaton Finite Element (CAFE) model was used for simulation. Thermal condition and fluid flow during directional solidification was calculated by a macroscale ProCAST model with a finite element solution of energy, momentum equations, and then was coupled with the micro-scale model to predict grain structures and orientation evolutions [11 13]. The thermal condition plays an important role in
216 X.B. Meng et al.: J. Mater. Sci. Technol., 12, 28(3), 214 2. Orientation deviation / deg. 32 16 8 8 mm 12 mm 16 mm mm 30 mm 0 40 60 80 0 Height of starter block / mm Fig. 2 Predicted 3D thermal condition at 1265 s when the liquidus isotherm reaches spiral part of spiral selector competitive grain growth during directional solidification. ally measured temperature/time boundary condition and a second generation single crystal superalloy CMSX-4 with temperature dependent physical properties were used in this study. The temperature in heating zones was 1600 C and, and the temperature in vacuum withdrawal chamber was C. The pouring temperature of superalloys was 1500 C, the same as the mold heating temperature (1500 C). After heat preservation for about 600 s, the thermal condition reached equilibrium, and then the castings were withdrawn at a rate of 5 mm/min. More details about the parameters during directional solidification have been reported previously [9]. As an example, the calculated thermal condition at 1265 s when the liquidus isotherm reaches the spiral part is shown in Fig. 2. The calculated thermal condition clearly reveals the isotherm shape in spiral part, which is helpful for the analysis of grains growth. 3.2 Grain selection in starter block For cases 1 to 5, the effect of the diameter of starter block diameters on the process of grain selection is studied to figure out the modes of grain selection in starter block during directional solidification. To achieve this goal, the same initial solidification conditions were applied for all the cases in Table 1. Fig. 3 shows the evolution of average orientation deviation with the increase of height in starter blocks with different diameters. Every case is repeated for times to consider the stochastic nature of the nucleation event. Differently shaped marks stand for starter blocks with different diameters. In Fig. 3, it can be seen that the average orientation deviation decreases rapidly with a distance of several millimeters away from the copper chill at first, and then decreases with a slower rate as the dis- Fig. 3 Effect of diameters of starter block on the grain selection in starter blocks with different diameters Required height for a given average misorientation / mm 1 8 deg. 9 deg. deg. 0 12 deg. 15 deg. 80 60 40 5 15 25 30 35 40 Diameter of starter block / mm Fig. 4 Effect of diameter of starter block on the required height for a given average misorientation tance increases. It implies that the efficiency of the grain selection in starter block decreases when the distance from the copper chill plate increases. The main reason is that grain orientations are arbitrary at first. During the following competitive grain growth, the misorientation of grains becomes gradually smaller due to the overgrowth of misoriented grains by the preferred grains. Therefore, the optimization of grain orientation becomes slower and the efficiency of grain selection gradually decreases along with the optimization of grain orientation. Besides, in Fig. 3, the curves of orientation deviation in the starter blocks with different diameters gradually separate from each other with the height increasing, which implies that the starter block with small diameter is more efficient to optimize grain orientation. To clearly show this tendency, the required heights for average orientation deviation of 8 /9 / /12 /15 are given as a function of the diameters of starter block, as shown in Fig. 4. The black solid marks are the required heights for the given orientation deviations, and every differently shaped mark stands for a given average orientation deviation. The red lines are the fitting lines to show
X.B. Meng et al.: J. Mater. Sci. Technol., 12, 28(3), 214 2. 217 the tendency of grain orientation optimization in the starter blocks with different diameters. In Fig. 4, it is shown that the rate of slope of fitting lines is positive and increases with the optimization of the average orientation deviation. The positive rate of slope implies that the starter block with small diameter needs a lower height to optimize the average grain orientation for a given deviation, which means that starter block with small diameter is more efficient to optimize grain orientation. Furthermore, the increased rate of fitting line implies that optimization of grain orientation by starter block with small diameter is more efficient for a small average orientation deviation (8 ) at the later stage of grain selection than that for a large average orientation deviation (15 ) at the early stage, which means that the advantage of optimization of grain orientation in the small diameter starter block is increasingly obvious when the average orientation deviation is gradually optimized at the later stage of grain selection. 3.3 Single crystal selection in spiral part For the cases 6 to 11, a φ16 mm 25 mm starter block was positioned at the bottom of spiral parts to simulate the entrance of columnar grains in spiral part, and the effect of the pitch length and initial angle of spiral part on the process of single crystal selection is studied to figure out the modes of single grain selection in spiral part during directional solidification. To achieve this goal, the same initial solidification conditions were applied for the cases 6 to 11 in Table 2. Fig. 5 shows the evolution of required height and orientation deviation with the increase of pitch lengths and initial angles of spiral parts. Every case is repeated for times to consider the stochastic nature of the nucleation event. Different shaped marks stand for spiral parts with different pitch lengths and initial angles, and the red curves show the average values of these calculations. As shown in Fig. 5(a), the required height for single crystal selection decreases when the pitch lengths and the initial angles decrease. The reason is lowering down the pitch length and initial angle could lengthen the route of grain selection at the same height, and offer more chances to select single crystal by blocking dendrites growth. Therefore, the efficiency of single crystal selection increases with the pitch lengths and initial angles decreasing. In Fig. 5(b), it is shown that the orientation of selected single crystal is not obviously optimized when the initial angle of spiral part is smaller than 50 ; however, when the initial angle is larger than 50, the trend of orientation optimization of selected single crystal is gradually obvious. 3.4 al confirmation The process of grain selection in starter block Required height for single crystal selection / mm Orientation deviation / deg. 0 8 (a) deg. 30 deg. 40 deg. 50 deg. 60 deg. 70 deg. 30 40 50 60 70 80 90 Initial angle / deg. 24 (b) deg. 30 deg. 40 deg. 50 deg. 16 60 deg. 70 deg. 12 30 40 50 60 70 80 90 Initial angle / deg. Fig. 5 Effect of initial angle and pitch lengths of spiral part on (a) required height; (b) orientation deviation for single crystal selection Fig. 6 Grain structure of spiral selector in starter block of φ16 mm 40 mm is generally shown in Fig. 6. Many small equiaxed grains nucleate at the bottom of starter block. They compete with each other, and columnar grains close to <001> are acquired at the top. Fig. 7 shows the evolution of grain density and orientation with increased distance from copper chill plate. The curve is the simulated results, and the
218 X.B. Meng et al.: J. Mater. Sci. Technol., 12, 28(3), 214 2. 8 (a) 00 (a) Grain density 7 Grain number 0 6 0 30 40 50 Height of starter block / mm 1 0 5 15 25 30 35 Height of spiral part / mm Orientation deviation from [001] / deg. 35 30 25 15 5 (b) 0 0 30 40 50 Height of starter block / mm Fig. 7 Evolution of grain density (a) and orientation deviation (b) in starter block solid triangle marks are the experimental results. It can be seen that grain density (Fig. 7(a)) and average orientation deviation (Fig. 7(b)) decrease stably with the increase of distance from copper chill plate, but the decrease rate slows down with the distance increasing. These results prove the optimization of grain orientation in starter block and imply that this process is lasting and stable. Besides, the simulation results are in accordance with the experimental ones. A φ16 mm 25 mm starter block was positioned at the bottom of spiral part to simulate the entrance of columnar grains in spiral part, and both simulation and experiments were carried out to show the process of single crystal selection in the spiral part with 30 mm pitch length, 45 initial angle and 15 mm in diameter (Case 12). To remove the influence of initial solidification conditions on the grain selection, the initial solidification conditions in case 12 was the same as other cases in Table 1 and Table 2. In Fig. 8, the evolution of grain number and average orientation deviation are shown as a function of the height of spiral part. The simulation and experimental results are represented by black lines and solid triangles, respectively. The amount of grains in spiral part gradually decreases from 60 to 1 with the height increasing, as shown in Fig. 8(a). As shown in Fig. 8(b), the average orientation deviation is about at the bottom of the spiral part, which is a little higher than the simulation results, and then the orientation deviation Orientation deviation from [001] / deg. 15 5 (b) 0 0 5 15 25 30 35 Height of spiral part / mm Fig. 8 Graphs of grain number (a) and the average orientation deviation (b) measured by experiments and simulated by a CAFE model in spiral part fluctuates in a ± 3 range around 7 all the time. The simulation result could be proved by experiment in Fig. 8(b). Both the simulation and experimental results show that the single crystal selection of this spiral selector is efficient and a single crystal has been successfully selected. The simulation results also agree well with the experiments. 4. Discussion In the conventional views, starter block is used to optimize the grain orientation to acquire grains close to <001> orientation by competitive grain growth [2,9], and spiral part is used to efficiently and randomly select a single crystal by the geometry shape but cannot optimize the grain orientation [2,3,9]. However, the results above show that the efficiency of grain selection in starter block is associated with the geometry shape, and the orientation of selected single crystal could be optimized in spiral part when the initial angle of spiral part is large enough. 4.1 Model of grain selection in starter block In the above work, the starter block with small diameter is more efficient to optimize grain orientation, and this advantage is increasingly obvious at the later stage of grain orientation optimization in
X.B. Meng et al.: J. Mater. Sci. Technol., 12, 28(3), 214 2. 219 Fig. 9 Model of grain selection in starter block (a) at early stage of grain selection; (b) at later stage of grain selection starter block. Obviously, the efficiency of grain selection in starter block is associated with the geometry shape. As follows, a model was constructed to study the process of grain selection in starter block. The process of grain selection in starter block is controlled by two factors: competitive grain growth (big red circle in Fig. 9) and blocking grain growth by geometry shape (small green circle in Fig. 9). The mode of grain growth between different oriented grains is dominated by competitive grain growth, and blocking grain growth by mold wall is dominated by geometry shape. At the early stage of grain selection in starter block (Fig. 9(a)), grains are of small size, large amount and random orientation; thus, competitive grain growth is rapid and dominates at this stage. However, at the later stage of grain selection in starter block (Fig. 9(b)), the grain size is large, grains are of small amount and oriented close to <001>. Therefore, it is increasingly difficult for grains to overgrow each other by competitive grain growth. At this stage, competitive grain growth is increasingly weak and blocking dendrites growth by mold wall gradually dominates to optimize grain orientation. Theoretically, without the consideration of the development of secondary and tertiary dendrites by the primary dendrites, starter block with dimension of φ=r H (in mm) (Fig. ) could block the growth of dendrites with misorientation larger than θ=arctan(2r/h) which is a critical angle, as marked by dashed line in Fig.. For starter block with small diameter (Fig. (a)), the ratio of diameter to height is smaller than large diameter starter block (Fig. (b)) and the critical angle (θ R ) is smaller. It is implied that at the same height the starter block with small diameter is more efficient to block dendrites growth and optimize grain orientation by geometry shape. Therefore, this advantage of geometry shape is stronger in starter block with small diameter and this kind of starter block is more efficient to optimize grain orientation by geometry shape (Fig. 3). At the same time, since the competitive grain growth is gradually weak at the later stage of grain selection in starter block, the advantage of grain orientation optimization by geometry shape in starter block Fig. Schematic diagram showing blocking grain growth by mold wall in starter block with (a) small diameter; (b) large diameter with small diameter is increasingly obvious (Fig. 4). In conclusion, for the start block, the grain selection is dominated by competitive grain growth at the early stage; however, with the optimization of grain orientation and the increase in grain size, competitive grain growth is increasingly weak and the geometry shape gradually dominates to optimize grain orientation at the later stage. 4.2 Mode of grain selection in spiral part In this work, the orientation of selected single crystal could be optimized when the initial angle of spiral part is larger than 50, as shown in Fig. 5(b). To figure out the reason for this orientation optimization and modes of grain selection in spiral parts with different initial angles, the thermal conditions were calculated at the time when the liquidus approached and reached the spiral parts with different initial angles. In the spiral part with initial angle, at 00s after withdrawal for 400 s, the liquidus isothermal approaches the spiral part and becomes concave (Fig. 11(a1)). About 300 s later, the liquidus inclines (Fig. 11(b1)). This result implies that the liquidus isothermal gradually transforms from a concave plane to an inclined plane when it enters the spiral part with initial angle. The transformation of thermal condition could bring in the alteration of the preferred orientation and the variation of grain growth state [9]. These instable states will result in the fluctuation of orientation in this spiral part; thus, grain orientation cannot be optimized in this spiral part with small initial angle (<50 ). However, with the increase in initial angle, the liquidus is relatively stable in these spiral parts (Figs. 11(a2), (b2), (a3), (b3)); therefore, the state of grain growth is increasingly stable, and the competitive grain growth gradually dominates in these spiral parts. Consequently, when the initial angle is large enough (>50 ), the orientation of selected single crystal in spiral part could be optimized by competitive grain growth.
2 X.B. Meng et al.: J. Mater. Sci. Technol., 12, 28(3), 214 2. (1) The grain selection is dominated by competitive grain growth at the early stage in starter block; however, with the optimization of grain orientation, competitive grain growth is increasingly weak and the geometry shape gradually dominates to optimize grain orientation at the later stage in starter block. (2) The modes of grain selection are associated with the initial angles of spiral parts. When the initial angle is large enough (>50 ), the grain orientation could be optimized by competitive grain growth; when the initial angle is small (<50 ), due to the variation of thermal condition and the instable grain growth state in spiral part, single crystal could only be selected by geometry shape but the orientation cannot be optimized. (3) The simulation results accord well with the experimental ones. Acknowledgements This work was financially supported by the National Basic Research Program (973 Program) of China under Grant No. CB63 (CB6316), the National Natural Science Foundation of China (NSFC) under Grant No. 508061, No. 509304, No. 571165 and the fund of the State Key Laboratory of Solidification Processing in NWPU under Grant No. SKLSP1112. The authors are grateful for those supports. REFERENCES Fig. 11 Predicted thermal condition at the time when the liquidus approaches and reaches the spiral parts with (a1), (b1) ; (a2), (b2) 50 ; (a3), (b3) 70 initial angles According to the discussion above, for the spiral part, when the initial angle is large enough (>50 ), the grain orientation could be optimized by competitive grain growth; when the initial angle is small (<50 ), due to the variation of thermal condition and the instable grain growth state in spiral part, single crystal could only be selected by geometry shape but the orientation cannot be optimized. 5. Conclusions [1 ] M. Cell, D.N. Duhl and A.F. Giamei: Superalloys, 1980, 5. [2 ] A.I. Epishin and G. Nolze: Crystallogr. Rep., 06, 51, 7. [3 ] H.J. Dai, J.C. Gebelin, M. Newell, R.C. Reed, N. D Souza, P.D. Brown and H.B. Dong: Superalloys, 08, 367. [4 ] P. Carter, D.C. Cox, C.A. Gandin and R.C. Reed: Mater. Sci. Eng. A, 00, 280, 233. [5 ] Q.Y. Xu, B.C. Liu, Z.J. Liang, J.R. Li, S.Z. Liu and H.L. Yuan. Mater. Sci. Forum, 06, 508, 111. [6 ] H. Esaka, M. Tamura and K. Shinozuka: Mater. Trans., 03, 44, 829. [7 ] H. Esaka, K. Shinozuka and M. Tamura. Mater. Sci. Eng. A, 05, 413 414, 151. [8 ] D. Pan, Q.Y. Xu, B.C. Liu, J.R. Li, S.Z. Liu, H.L. Yuan and H.P. Jin. JOM,, 62, 30. [9 ] X.B. Meng, J.G. Li, T. Jin, X.F. Sun, C.B. Sun and Z.Q. Hu: J. Mater. Sci. Technol., 11, 27, 118. [] D. Walton and B. Chalmers: Trans. Metall. Soc. AIME, 1959, 215, 447. [11] C.A. Gandin and M. Rappaz: Acta Mater., 1994, 42, 2233. [12] M. Rappaz and C.A. Gandin: Acta Metall. Mater., 1993, 41, 345. [13] C.A. Gandin, J.L. Desbiolles, M. Rappaz and Ph. Thevoz: Mater. Trans. A, 1999, 30, 3153. [14] Y.Z. Zhou, A. Volek and N.R. Green: Acta Mater., 08, 56, 2631. [15] Y.Z. Zhou and N.R. Green: in Proc 11th Int. Symp. on Superalloys 08, Champion, PA, USA, 08, 317 324.