Microstructure and Mechanical Properties of CMSX-4 Single Crystals Prepared by Additive Manufacturing

Polished samples were used for the hardness tests. The tests were performed according to HV1 with a computer-controlled low-load micro-indentation tester M-400G from Leco with 15-seconds loading time for each indentation. The mean hardness was determined from three parallel measurement series in building direction with a distance of 0.3 mm. The hardness was determined with the software Hardest from the diagonals of the indentations.

The tensile tests were performed on an Instron 4505 machine with a 100 kN load cell and a strain rate of 0.83 × 10⁻³/s. For the thermal tensile tests, a furnace for temperatures up to 1250 °C was applied with three PtRh10 pct-Pt thermocouples for temperature measurement and control. In order to achieve a homogenous temperature distribution within the samples, the tensile tests were started 15 minutes after reaching the test temperature. At room temperature, the sensor MFA25 (gauge length: 25 mm) was used for strain measurement. At high temperatures, the strain had to be determined from the traverse movement since the mounting of the sensor was not possible. Sample geometry was chosen according to DIN 50125-B10 × 30 with M10 threads. The samples were machined from the 〈001〉 oriented round samples. Thus, the surrounding polycrystalline (PX) shell and the grain selection zone were removed. The sample diameter in the gauge length was reduced from 6 to 5 mm in order to guarantee a single crystal in the test specimens.

Miniature LCF specimens have been machined from SX CMSX-4^® SEBM cylinders in the as-built and heat-treated state and from the heat-treated reference cast material ERBO/1C (CMSX-4^® type, described in Reference 19). The LCF specimens have a cylindrical measurement length of 6 mm and a corresponding diameter of 2 mm. They were precisely oriented with the Laue technique in combination with spark erosion cutting such that the specimen axis was parallel to the dendrite growth direction ([001] direction). Load-controlled LCF tests have been carried out until rupture at a temperature of 950 °C, with a frequency of 0.25 Hz (triangular signal shape) in tension-tension loading \( \left( {R - {\text{ratio}}\frac{{\sigma_{\hbox{min} } }}{{\sigma_{\hbox{max} } }} = \, 0.62 - 0.65} \right) \). In between LCF tests with different mean stress levels (550 and 590 MPa, respectively) the stress amplitude was kept constant at 130 MPa. The strain was measured indirectly with a mechanical extensometer placed on the fixture. ERBO/1C and CMSX-4^® SEBM specimens have been tested with the same parameters in order to compare the results directly.

Miniature tensile creep specimens[20] with the longitudinal axis parallel to the buildup direction, i.e., 〈100〉 direction, have been machined out of the reference cast material ERBO/1C and as-built/HT SX SEBM CMSX-4^®. The creep specimens with dog bone geometry have a measurement length of 9 mm and a thickness of 2 mm. The experiments were performed on a creep testing machine from Denison-Mayes-Group, type TC 20 Mark II with three-zone resistance furnace up to 1200 °C. Temperature control is based on up to three Typ S PtRh/Pt-thermocouples with PtRh-shell. As-built and fully heat-treated samples were tested at 850 °C/600 MPa and 1050 °C/160 MPa.

Figure 5 shows stress-strain curves of SX CMSX-4^® SEBM specimens at RT, 650 °C, 800 °C, and 950 °C.

The specimens were tested in the as-built and HT state. Yield strength as well as ductility increases with increasing temperature up to 800 °C. With increasing temperature, the strength decreases whereas the fracture strain further increases. This well-known behavior results from a mechanism change from cutting of the γ’ precipitates by dislocations, resulting in stacking faults and maximum strength, to a climb-assisted bypassing process.[23] In addition and especially for 800 °C, the stress-strain curves show an upper yield point. Beyond this yield point, the stress decreases and eventually increases again due to strain hardening effects. This behavior results from collective dislocation movement as observed before.[24] Favorable orientated glide systems get activated when the critical shear stress is reached. Within the SX, concurrent dislocation movement is possible due to the lack of grain boundaries. As a consequence, the stress decreases due to geometrical softening. Subsequently, dislocations pile up again, other glide systems are activated and multiple slip occurs. In this state, dislocation movement is at more or less constant stress level, if the γ’-precipitation structure is homogenous as within the HT samples.

The homogenous γ’ size distribution within the HT specimens also leads to a higher yield strength compared to the as-built state. Figure 6 shows yield strength and ultimate tensile strength of SX SEBM CMSX-4® samples in the as-built and HT state as a function of temperature. The shaded region indicates reference values for cast and heat-treated CMSX-4® for various γ’ sizes.[25] The yield strength and ultimate strength show a pronounced anomaly similar to the conventional cast material. After heat treatment, the yield strength slightly increases compared to the as-built state, whereas the ultimate tensile strength remains unchanged. This behavior can be ascribed to the inhomogeneous γ’ size distribution and differences of the γ/γ’ morphology within dendritic and interdendritic areas in the as-built state, see Figure 3. In the HT state, the γ’ size is homogeneous all over the sample (see Figure 4) and therefore a large scattering in mechanical properties is barely detectable, as expected.

Figure 6 also shows that the strength values of HT SX SEBM CMSX-4® samples are comparable to those measured for conventional cast material.[25] In addition, SX SEBM CMSX-4® virtually shows the same static strength as the heat-treated cast material but already in the as-built condition. There is no significant difference between the SEBM as-built and HT samples. This behavior results from the pronounced in situ heat treatment characteristic for SEBM supported by the fine microstructure due to rapid solidification. That is, as-built SEBM samples are already partly homogenized and aged. The γ’ size within the dendritic area of the as-built samples varies between 70 nm (near the top) and 500 nm (at the bottom) with about 70 vol pct of γ’ phase (for all positions) determined by the linear intercept method. The γ’ size of the HT samples is about 450 nm. Obviously, the γ’ size has no strong influence on the (HT) tensile properties.

Figure 7 shows the mean stress during LCF testing plotted versus the cycles to failure. It can be seen that the HT CMSX-4^® SEBM specimens achieve a lifetime as high or even higher than the specimens from the cast and heat-treated reference material ERBO/1C. In contrast, the as-built SEBM specimen fails much earlier than the specimens from heat-treated SEBM and cast material.

Table III lists the results of the LCF tests. The heat-treated SX CMSX-4^® SEBM specimens yield almost twice as high values for elongation (16.8 to 17.1 pct) and contraction (16.4 to 18.4 pct) at fracture compared to ERBO/1C (about 10 pct elongation and 11 pct contraction at fracture). The SEBM as-built sample that showed the lowest fatigue lifetime reached the highest elongation and contraction strains (18.7 and 21.6 pct, respectively).

In Figure 8, the accumulated strain and the evolution of the plastic strain amplitude are shown as a function of cycle number. The strain evolution of the SEBM HT and the cast HT material is similar, while the SEBM as-built material shows much faster elongation. The failure strain of both as-built and HT SEBM specimens is comparable and remarkably higher than that of the cast HT specimen. The evolution of plastic strain amplitude depicts an initial decrease for all three material variants until a more or less expressed minimum. This behavior, known as cyclic hardening, can be associated to initiation and moving of dislocations until the dislocation density becomes high enough to constrain further dislocation movement.[26] After reaching the minimum strain amplitude, further increase in the strain amplitude occurs due to simultaneous diffusion assisted deformation processes and onset of damage evolution such as crack formation and crack propagation. The change rate of the strain amplitude is linked to the effective deformation and damage mechanisms.

In Figures 9(a) through (c), exemplary fracture surfaces of cast, as-built SEBM and heat-treated SEBM material are shown, and in Figures 9(d) through (f) the characteristic porosity within the different materials is shown. Microscopic investigations by SEM revealed that porosity played a major role in crack initiation in both, SEBM and cast material. Since the solidification microstructure of SEBM material is about two orders of magnitudes smaller as in the cast material, also the size and spacing of pores are much smaller in SEBM material. In longitudinal sections of heat-treated cast material pores are already obvious in the processed material. The mean pore size of the investigated heat-treated cast material is about 63 µm², as determined in Reference 27. After LCF fracture, fatigue crack initiation at almost each pore of the cast material was observed, with the cracks linking in the fracture plane. In SEBM material, fine interdendritic porosity with a pore size between 1 and 10 µm² occurs in the entire material. Crack initiation and crack linking at these pores are only observed in the fracture plane.

Cracks associated with initiation at pores are propagating mainly in fracture mode I, perpendicular to the load. The fracture mirror at the individual pores displays a square shape, which results from preferred in plane crack propagation in 〈110〉 directions and has been described for creep and fatigue fracture.[28,29] The fracture surfaces of all specimens revealed areas of mode I fracture initiated by porosity and fracture along crystallographic {111} planes. The ratio between mode I fracture and fracture along slip planes was about 0.5 in both heat-treated specimens and about 0.25 in the as-built SEBM specimen. Further, in contrast to both heat-treated specimens, the as-built specimen revealed in the length section multiple oxidized cracks starting at the surface. The susceptibility of the as-built material for surface cracks can be associated to the build process, adding the material layer by layer leaving some microstructural inhomogeneities between the layers. The thermal treatment has an homogenization effect. That is why less surface cracks are observed in the heat-treated SEBM material after fatigue fracture. However, on the fracture surface of as-built specimens, traces of oxidized surface cracks were found. With this information, the evolution of the strain amplitude in Figure 8 can be further interpreted. The early onset of accelerated increase in strain amplitude observed for the as-built specimen is probably connected to the propagation of surface cracks. The high stress intensity at the surface crack tips is likely to activate macroscopic slip and crack propagation along {111} slip planes, resulting in early specimen fracture. Surface cracks are less frequent and shorter in heat-treated LCF specimens after fracture. Both, heat-treated SEBM and cast specimens displayed a long life time and similar portions of pore-assisted mode I fracture. Differences are the size of pores and pore-connecting cracks as found on the fracture surface and the observation that in length sections cracks are only found in the cast specimen with the larger pores. Obviously, crack initiation at pores started later in the HT SEBM specimen than in the cast specimen. This conclusion is also enhanced by the evolution of the strain amplitude, which is slightly faster for the cast specimen (see slope of the curves after strain amplitude minimum and onset of accelerated increase of strain amplitude in Figure 8(b)) . The reason for the higher ductility of the SEBM material (as-built and heat-treated) compared to the heat-treated cast material is not yet clear. Thus, further investigations of dislocation formation and evolution by TEM are envisaged.

Figure 10 shows two sets of creep curves plotted as strain ε as a function of time t. In the low-temperature high-stress creep regime, the creep curves of the as-built and the solution annealed SEBM material differ strongly. However, after solution annealing, the SEBM data show a very similar creep behavior to the cast material ERBO/1C. The rupture strain of SEBM material (close to 20 pct) is larger than that of the cast material (close to 15 pct). In the high-temperature low-stress regime, the differences between as-built and solution annealed SEBM materials are less pronounced. After solution annealing, the SEBM material shows a very similar rupture strain (close to 20 pct) like the cast material and a slightly better creep resistance.

Figure 11 shows the γ/γ’ microstructure in the as-built and HT condition after creep at 850 °C/600 MPa and 1050 °C/160 MPa. After creep, the inhomogeneous γ/γ’ microstructure of the as-built condition is still obvious. Caron et al.[30] have shown that the γ/γ’ size has a strong influence on creep properties for temperatures between 760 °C and 850 °C. Decreasing the γ’ size from 340 to 220 nm strongly reduces the time to fracture. Thus, the small γ’ size within the dendritic regions may explain the high creep rates. Nevertheless, the as-built state shows different particle sizes, small γ’ particles within the dendrites but larger ones within the interdendritic regions. The influence of the γ’ particles within the interdendritic regions on the creep behavior is yet not clear.

At 1050 °C, the creep curves show no significant differences between as-built and HT samples. Again, this is in accordance with results from Caron et al. [30] where the γ’ size has no strong influence for temperatures between 980 °C and 1050 °C due to coarsening as a result of rafting perpendicular to the load direction.