Dendrite Structure Refinement and Mechanical Property Improvement of a Single-Crystal Superalloy

Abstract

In the present work, the effect of different casting processes on the microstructure and creep properties of the second-generation single-crystal superalloy DD419 was investigated. Under conventional production conditions and a contour-suited thermal insulation method, single-crystal rods of types A and B were fabricated, respectively. In comparison to rod type A, the solidification process of rod type B featured a 1.6-fold increase in the temperature gradient and a 32% reduction in primary dendrite spacing. The γ/γ′ eutectic in the as-cast microstructure, the residual eutectic phase, and porosity after heat treatment were also significantly reduced, resulting in the improved homogeneity of the single crystal castings. Under the testing conditions of 850 °C/650 MPa and 1050 °C/190 MPa, the stress rupture life of sample B was enhanced by 25% and 5.2%, respectively, compared to sample A. Therefore, due to dendrite structure refinement, the stress rupture life of the superalloy was evidently improved, especially at medium temperatures.

1. Introduction

Ni-based superalloys are widely used to fabricate turbine blades of aero-engines and industrial gas turbines because of their high-temperature strength, microstructural stability, and corrosion resistance [1,2,3]. The limiting components of a gas turbine are the blades in the front stages, which are under the highest gas-temperature stresses. Turbine blades are typically produced in an investment casting process, which is especially suitable for casting components with complex and near-net-shape geometry. Casting methods have been improved from conventional investment casting to directional solidification [2,3,4,5,6]. By combining the selector technology or seeding method, the single-crystal (SC) structures can be obtained to completely eliminate the detrimental grain boundaries. Currently, the SC components of superalloys are increasingly used to modify the high-temperature performance of gas turbine engines.

The mechanical properties of SC superalloys are key technical indicators for the turbine blades, directly affecting engine reliability and efficiency. Single-crystal superalloys are advanced nickel-based materials fabricated through controlled directional solidification, resulting in a macro-grain structure without grain boundaries. This structural homogeneity reduces stress-induced cracking at elevated temperatures, making them indispensable in aerospace systems, particularly within high-pressure turbine blades exposed to extreme thermal and mechanical loads. Their microstructure consists of only one grain but includes two interpenetrating phases: a γ-phase matrix hosting regularly aligned γ′-phase precipitates (Ni₃(Al,Ti)) with cuboidal morphology. These coherent precipitates function as nanoscale barriers, hindering dislocation motion and providing exceptional resistance to deformation under prolonged stress through lattice strain interactions. Alloying with high-melting-point elements such as rhenium, ruthenium, and tungsten enhances phase stability by preventing diffusion-driven coarsening of γ′ particles and avoiding the formation of brittle intermetallic compounds that weaken mechanical integrity. The synergy between microstructural design and composition optimization enables sustained performance in environments exceeding 80% of the alloy’s melting point.

According to the current research results, the mechanical properties of SC superalloys are influenced by many factors such as material composition [7,8,9,10,11,12], testing conditions (temperature and stress) [13], heat treatment processes [14,15,16], casting processes, and the metallurgical quality of the castings. The metallurgical quality of the castings includes defects such as inclusions, metallurgical flaws [17,18,19], and crystal misorientation [20,21,22,23,24,25,26,27]. In the production of superalloy components, the casting alloy is typically selected by the customers, making it hard to improve the microstructure and performance by altering the chemical composition of the alloys. The heat treatment process is generally fixed as a technical standard and cannot be changed. The testing temperature and stress for creep properties are also explicitly specified.

In the directional solidification process, the withdrawal rate is an important solidification parameter affecting dendrite growth and mainly determines the microstructure of the SC blades [28,29,30]. The most commonly used improvement method is the modification of the withdrawal rate. Lian et al. [31] stated that an increased withdrawal rate reduced the dendrite arm spacing, refined the dendritic structures, aggravated element segregation, and reduced the size and content of the spherical γ′ phase in the dendrite core and the inter-dendritic region. In a SC solidification experiment with superalloy DD9 [32], the withdrawal rate was increased from 3 mm/min to 5 mm/min. As a result, the dendrite morphology became more refined, and the size of the γ′ precipitates and the γ/γ′ eutectics decreased. However, the subsequent heat treatment and mechanical property testing were not conducted. According to reference [33], the SC turbine blades made from superalloy CMSX-4 were manufactured with variable withdrawal rates of 1 and 3 mm/min. After heat treatment, however, the highest creep resistance was obtained for the blade manufactured at a withdrawal rate of 1 mm/min. The research results with DZ22B superalloy [34] indicate that increasing the withdrawal rate can reduce the dendrite spacing. It is suggested that the withdrawal rate of the directional solidification should be varied between 6–2 mm/min to improve the production success rate. In research [35] with the liquid metal cooling (LMC) method, it was found that residual segregation was reduced, leading to an increase in creep life, more uniform microstructures, and a decrease in solidification and homogenization porosity. Structural refinement served as the underlying mechanism for this optimization, which reduced diffusion distance and increased homogenization efficiency during solution heat treatment. In research [36], the melt superheating treatment was applied to improve the structure and properties of a superalloy. As the superheating temperature rises, the secondary dendrite arm spacing and the size of γ′ particles are reduced. Still, the mass loss of the alloying elements such as Cr, Ta, and Hf becomes severe. The effect of this treatment on the mechanical property was not reported.

Great efforts have been made to increase the production yield of the SC superalloy components. However, even in the qualified SC products, the as-cast microstructure can still vary significantly due to the differences in casting process conditions, which, in turn, affect the subsequent heat-treated microstructure and the mechanical properties. Therefore, optimizing the casting process to improve both the as-cast and heat-treated microstructures can effectively enhance the mechanical properties of SC superalloys, thereby improving the efficiency and reliability of aero-engines and gas turbines.

2. Materials and Methods

The alloy used in this study was a second-generation SC superalloy DD419, and its chemical composition is listed in

Table 1. Chemical composition of alloy DD419 (wt.%), remeasured by the authors.

Cr	Co	Mo	W	Re	Al	Ti	Ta	Hf	Ni
6.44	9.57	0.61	6.38	2.94	5.60	1.02	6.47	0.11	Bal.

. The supplier of the alloy was Jiangsu Longda Superalloy Co., Ltd., Wuxi, China.

To fabricate the SC samples, two same wax clusters were assembled with 10 wax rods, each with a diameter of 15 mm and length of 180 mm. The lower part of each rod was connected to a spiral crystal selector and mounted around a central pillar on a base plate, thus forming a ring-shaped wax cluster. Using the conventional lost-wax process, two ceramic molds were prepared for the casting experiments (experiments A and B) in a vacuum directional solidification furnace (ALD, Germany, Model: VIM-IC/DS/SC). For both experiments, the mold preheating temperature and pouring temperature were set at 1550 °C, and the withdrawal speed was maintained at 3.5 mm/min. The only difference between the two experiments was in the insulation condition between the heating room and the cooling zone of the furnace during the solidification process. Experiment A was conducted under conventional production conditions (Figure 1a), while in experiment B, special measures were taken to improve the insulation condition between the heating and cooling zones in the furnace (Figure 1b).

In experiment B, as shown in Figure 1b, insulating materials were used to fill the space between the rods and the central pillar of the mold cluster (internal insulation). In addition, a contour-suited insulation baffle conforming the external contour of the mold cluster was added to the conventional circular baffle at the bottom of the heating zone. This design eliminated the through-hole region in the center of the mold cluster and reduced the gap between the outer surface of the mold and the insulation baffle. The goal of these contour-suited insulation measures was to create effective thermal insulation between the furnace’s heating and cooling zones, enhance the temperature gradient during solidification, and refine the dendrite structure of the SC castings by reducing the dendrite spacing. It was measured that under conventional conditions (experiment A), the temperature gradient was approximately 1.9 K/mm, while in experiment B, with contour-suited insulation measures, the optimized temperature gradient increased to 4.9 K/mm.

After casting experiments A and B, the solidified samples were knocked out of the ceramic mold and separated appropriately from the casting cluster. The samples were then macro-etched to confirm that each sample consisted of only one grain, without any stray grains or other defects. The crystal orientation of the samples was measured using a Laue instrument (Proto, LaSalle, ON, Canada, Model: LAUE COS). The samples with misorientations smaller than 5 degrees were selected for further investigation.

A portion of each set of SC rods (A and B) was used for as-cast microstructural analysis, while the remaining rods underwent the following heat treatment process: the solution heat treatment at 1305 °C for 4 h, the primary aging treatment at 1140 °C for 6 h, and the secondary aging treatment at 870 °C for 16 h. Cross-section samples were extracted from the bottom (Position 1), middle (Position 2), and top (Position 3) of both as-cast and heat-treated rods (A and B) using an electric discharge machine (SUZHOU BAOMA, Suzhou, China, Model: BM T400C). After polishing and grinding, the samples were chemically etched with a solution of HCl (20 mL), FeCl₃ (10 g), and H₂O (20 mL). The as-cast and heat-treated microstructures were observed and analyzed using a metallographic microscope (NIKON, Tokyo, Japan, Model: MM-400) to determine the dendrite spacing (λ), micro-porosity (ρ), as-cast γ/γ′ eutectic, and residual eutectic fraction. Under as-cast conditions, the dendrite spacing was measured in a view field of 2.5 mm × 2.5 mm in the rod center, and the most severe eutectic and its content were measured in a 1 mm × 1.25 mm view field for each sample. In the fully heat-treated state, the average residual eutectic structure and micro-porosity were measured in a 2.5 mm × 2.5 mm view field at the center of each sample’s cross-section, while the most severe localized residual eutectic and micro-porosity were measured in a 1 mm × 1.25 mm view field.

The heat-treated SC rods (A and B) with well-aligned primary crystal orientations (misorientation < 5 degrees) were selected for mechanical property testing. The upper sections of these SC rods were cut and used for stress rupture property testing with the electronic universal testing machine (WANCE, Shenzhen, China, Model: ETM105D). A total of 12 testings were carried out under two conditions: 850 °C at 650 MPa and 1050 °C at 190 MPa. Specifically, 3 A-rods and 3 B-rods were tested under each condition, respectively. The samples were cylindrical threaded specimens with a length of 66 mm, a gauge section diameter of 4.5 mm, and a gauge length of 23 mm.

3. Results and Discussion

3.1. As-Cast Microstructure

Under as-cast conditions, the most severe eutectic phase structure and its content were measured in a 1 mm × 1.25 mm view field for each sample. Figure 2 shows the worst view fields of as-cast microstructures at different heights of rods A and B. In Figure 2(a1–a3), the cross-sections of the bottom, middle, and top parts of rod A are represented, respectively. Correspondingly, Figure 2(b1–b3) show the cross-sections from the bottom to the top of rod B. The as-cast microstructure is mainly composed of cross-shaped γ-phase dendrites (dark regions) and γ/γ′ eutectic islands (bright white regions) with irregular shapes between the dendrites. It can be observed that the dendrite structure in rod A is relatively coarser than that in rod B.

The primary dendrite spacing (λ) at each cross-section of the two types of rods was measured using the unit area method, and the results are shown in Figure 3. Rod A was fabricated under normal production conditions with a relatively lower temperature gradient, resulting in a coarser dendrite structure. Furthermore, as the height increased, the dendrite spacing also increased, from 348 μm at the bottom to 452 μm at the top. In rod B, which was fabricated with modified thermal insulation measures, significantly smaller dendrite spacings were measured. Moreover, as the height of rod B increased, the dendrite spacing became even smaller.

As can be evaluated in Figure 3, the average dendrite spacing of sample A was 391 μm, 1.48 times that of rod B (266 μm). At the top of the rods, the dendrite spacing of sample A3 was 1.86 times that of sample B3. This indicates that the use of the contour-suited insulation measures effectively reduced the dendrite spacing in the SC castings. Moreover, the reduction in dendrite spacing was more pronounced toward the upper part of the castings.

The volume fraction of γ/γ′ eutectic (f_E) at each cross-section of the as-cast SC rods A and B (Figure 2) was measured and is summarized in Figure 4. As the solidification process progressed from the bottom to the top, the eutectic fraction in rod A showed an increasing trend, which corresponds to the coarsening trend of the dendrite structure in rod A (Figure 2). In contrast, the eutectic fraction in rod B gradually decreased, which also mirrored the change in dendrite spacing in rod B (Figure 2). The reduction in eutectic fraction was more pronounced towards the rod top. At the top of rod B (B3), the eutectic fraction (f_E = 7.43%) was only 70% of that at section A3 (f_E = 10.61%).

As shown in Figure 2, the dendrite structure in the as-cast rod A was coarser than that in rod B, leading to the larger size of γ/γ′ eutectic islands. Due to the refinement in dendrite structure, the fraction and the size of eutectic islands in as-cast rod B was reduced.

3.2. Heat-Treated Microstructure

Single-crystal rods A and B were subjected to heat treatment under the same conditions. Cross-section samples were taken from the bottom, middle, and top of the rods. The worst residual eutectic microstructures (1 mm × 1.25 mm view field) are shown in Figure 5. Although the dendrite structure became somewhat blurred after high-temperature heat treatment, it can still be distinguished. The positions and spacings of the dendrites remained unchanged, but the size and the fraction of the γ/γ′ eutectic (bright white phase) between the dendrites was significantly reduced compared to the as-cast microstructure (Figure 2).

For the heat-treated SC rods A and B, the average residual eutectic fractions f_H (2.5 mm × 2.5 mm view field) on the bottom, middle, and top cross-sections were measured and is summarized in Figure 6. It can be observed that the residual eutectic fraction f_H in both rods A and B decreased with the solidification height. The average residual eutectic fraction in rod A was approximately 1.39%, while in rod B it was only around 0.17%, nearly one order of magnitude lower than that of rod A.

It should be noted that the values of residual eutectic fraction shown in Figure 6 are the average ones on each cross-section of the heat-treated rods. In reality, the distribution of residual eutectic in the matrix was very uneven Therefore, for each cross-section of SC rods A and B, the worst view field (1 mm × 1.25 mm) containing the most residual eutectic was selected, The correspondingly measured values for residual eutectic fraction f_W are shown in Figure 7. For rod A, the f_W-values measured in the worst view fields of the three sections exceeded 4.0%, showing an increasing trend with the rod height. In rod B, the corresponding f_W-values measured in the worst view fields were only one-third to one-fifth of those in sample A. In comparison with Figure 6, it is clear that the local eutectic fraction f_W in the worst view field was approximately 3–7 times higher than the average one f_H on the entire cross-section.

In addition to the residual eutectic fraction f_H on each cross-section, the sizes i.e., the areas of individual residual eutectic islands in heat-treated rods A and B, were also quantified. Similar eutectic patterns were observed across different height sections of the same sample. Figure 8a,b present the statistical results of the residual eutectic sizes on the middle cross-section of rods A and B, respectively. It can be observed that the total number of residual eutectic islands in samples A and B were 176 and 110, respectively, with a relatively small difference. However, the maximum size of the residual eutectic islands was measured to be 3269 μm² for sample A and only 431 μm² for sample B. The average sizes of the residual eutectic islands in samples A and B were 413 μm² and 88.3 μm², respectively. This indicates that the implementation of the improved insulation measures for experiment B not only reduced the residual eutectic content in the matrix after heat treatment, but also significantly decreased both the average and maximum sizes of the eutectic islands. This has important implications for enhancing the mechanical properties of the material, as larger residual eutectic islands cause greater damage to the material’s mechanical performance.

The main purpose of solution heat treatment for superalloys is to dissolve the coarse γ′/γ eutectic structure, which is characterized by the transformation of the coarse γ′ phase into the γ matrix. This process is essentially the homogenization of alloying elements, achieved through atomic diffusion mechanisms. Key factors influencing the residual eutectic content after solution treatment are solution temperature and time. The higher the solution temperature and the longer the time, the more sufficient the diffusion of alloying elements, resulting in a lower residual eutectic content. The homogenization time t for alloying elements can be expressed as [37]:

t = L²/(4π²D)

(1)

where L is the characteristic diffusion distance, equivalent to half of the primary dendrite spacing λ in dendrite structures, and D is the diffusion coefficient of alloying elements in the matrix. From Equation (1), it can be observed that the homogenization time t during solution treatment is proportional to the square of the characteristic diffusion distance L. When dendrite spacing λ is reduced, the homogenization time can be significantly shortened. In this experiment, the average dendrite spacing of sample A was 1.48 times that of sample B. To achieve the same solution effect, the homogenization time for sample A needed to be 2.2 times longer than that of sample B. Since the same solution heat treatment temperature and time were used in this experiment, the degree of homogenization in sample A was far inferior to that of sample B, leading to more and larger residual eutectic islands in sample A.

Micro-porosity was examined on the polished cross-sections without etching. In Figure 9(a1–a3,b1–b3) the micro-porosity images of the bottom, middle, and top cross-sections of samples A and B are shown, respectively. It can be observed that the micro-porosity area in sample A was slightly larger than that in sample B.

The micro-porosity area of the three cross-sections of both types of samples was measured. The ratio of the total micro-porosity area to the total area of the view fields was defined as the porosity ratio ρ. The distribution of the ρ-values of the cross-sections of both samples is shown in Figure 10. The porosity ratios of sample A showed little variation, with an average value of 0.26%. In contrast, the average porosity rate of the three cross-sections of sample B was 0.17%, which was better than that of sample A.

The worst view field of 1 mm × 1.25 mm with the most porosity was selected on each cross-section of SC samples A and B, and the results are shown in Figure 11. It can be observed that the worst view fields in sample A had more severe local porosity than those in sample B.

Figure 12a,b, respectively, show the area distribution statistics of all micro-porosity holes in samples A and B. The number of porosity holes in samples A and B were 456 and 342, respectively. The average area of the porosity hole in samples A and B was 110.9 and 92.0 μm². The size distribution of porosity holes in sample A was more dispersed, with more large-sized porosity holes. The area of the largest porosity hole in sample A was 1066 μm², while that in sample B was only 892 μm². There were six porosity holes with areas exceeding 600 μm², and three holes exceeding 900 μm² in sample A. In contrast, sample B had only one porosity hole with an area over 600 μm² and none exceeding 900 μm². This indicates that due to the larger-sized porosity holes in sample A, the mechanical properties were more significantly compromised.

3.3. Stress Rupture Property

After solution and aging heat treatment under the same conditions, samples A and B were cut from the upper section to prepare test samples for stress rupture property testing at 850 °C/650 MPa and 1050 °C/190 MPa. In total, twelve testings were carried out, in which three A-rods and three B-rods were tested under each condition, respectively. The crystal misorientation of the samples were less than 5 degrees, and the measured stress rupture life is shown in

Table 2. The stress rupture lives of samples A and B measured under different testing conditions.

Stress rupture properties (h)	Samples	A			Ave.	B			Ave.
	850 °C/650 MPa	117.3	108.5	120.0	115.3	153.1	137.0	141.8	144.0
	1050 °C/190 MPa	86.0	88.2	93.1	89.1	99.2	94.3	87.7	93.7

. Under the test condition of 850 °C/650 MPa, the mean stress rupture lives of sample A and sample B were 115.3 h and 144.0 h, respectively, with sample B showing a 25% improvement over sample A. Under the test condition of 1050 °C/190 MPa, the mean stress rupture lives of sample A and sample B were 89.1 h and 93.7 h, respectively, with sample B showing a 5.2% improvement over sample A. This indicates that refining the dendrite structure can improve the stress rupture property of SC castings, but the improvement in high-temperature stress rupture property is not as significant as that at mid-range temperatures.

Table 3. Summary of the main measurement results for samples A and B.

Description			Rod A				Rod B
			Bottom	Middle	Top	Ave.	Bottom	Middle	Top	Ave.
Temperature gradient: G (K/mm)			1.9				4.9
Dendrite spacing: λ (μm) (Figure 2)			348	374	452	391	287	269	243	266
As-cast eutectic fraction: fE (%) (Figure 3)			9.50	9.80	10.61	9.97	9.10	7.67	7.43	8.20
After heat treatment	Residual eutectic fraction	Average: fH (%) (Figure 6)	1.87	1.16	1.12	1.39	0.22	0.16	0.14	0.17
		Worst: fW (%) (Figure 7)	4.0	4.8	5.5	4.8	1.3	1.2	1.1	1.2
	Micro-porosity ratio ρ(%)	Average: (Figure 10)	0.28	0.26	0.23	0.27	0.22	0.15	0.13	0.17
		Worst: (Figure 11)	0.43	0.30	0.26	0.33	0.30	0.25	0.24	0.26

summarizes the main measurement results of this study. Compared to sample A prepared under conventional production conditions, sample B with contour-suited insulation measures exhibited better performance across all indicators. This was primarily reflected in the significant refinement of the dendrite structure, with a marked reduction in dendrite spacing. Both the as-cast eutectic and the residual eutectic content after heat treatment were significantly reduced In particular, the reduction in large-sized residual eutectic and porosity was notable, as it improved the uniformity of the microstructure and the mechanical properties of the sample, especially the mid-temperature creep performance (Table 2).

4. Conclusions

• Single-crystal rods of types A and B were fabricated under conventional production conditions and with contour-suited insulation measures, respectively. The temperature gradients during the solidification process were measured as 1.9 K/mm and 4.9 K/mm for rod types A and B, and the corresponding dendrite spacings were reduced from 391 μm to 266 μm. This demonstrates the effective role of the contour-suited insulation in refining the dendrite structure.
• The average γ/γ′ eutectic fraction in the as-cast structure of rods A and B was 9.97% and 8.20%, respectively, indicating the effect of dendrite structure refinement on the eutectic reduction in the cast structure.
• After heat treatment, the average residual eutectic fraction in rods A and B was 1.39% and 0.17%, with the latter being only 15% of the former.
• After heat treatment, the average porosity in rods A and B was 0.26% and 0.17%, respectively. Moreover, the number of large porosity holes in rod A was much greater than that in rod B, causing more significant damage to the material’s performance.
• Under the test conditions of 850 °C/650 MPa and 1050 °C/190 MPa, the mean stress rupture lives of rod B were improved by 25% and 5.2%, respectively, compared to rod A. This suggests that the stress rupture life of the superalloy was evidently improved by dendrite structure refinement, especially at medium temperatures.