The development of high-temperature structural alloys has contributed to increasing the performance of aircraft engines and land-based gas turbines significantly. A series of improvements in alloy compositions and casting technologies in Ni-based superalloys enabled us to improve their temperature capability. Due to ever-increasing demands for sustainable air transportation and electricity production, increasing operating temperatures as well as thermal efficiency have been desired. However, the operating temperature of state-of-the-art turbine engines is approaching the incipient melting temperatures of Ni-based superalloys, preventing further increasing the operating temperature.

Nb-silicide-based composites have been considered as a potential alternative to Ni-based superalloys due to their high melting temperature (> 1600 °C) and excellent high-temperature phase stability. Bewlay et al.[1] introduced the multicomponent composites that consisted of solid solution Nb-based matrix and Nb5Si3 and/or Nb3Si and reported that these composites exhibited comparable creep performance to 2nd generation single-crystal Ni-based superalloys above 1200 °C. Although this material showed great potential as a candidate, its brittle nature and poor oxidation resistance have been major issues. In order to balance the mechanical properties and oxidation resistance, the alloying effects were studied by exploring a number of Nb–Si–X ternary systems.[2,3,4,5] Ti was found to improve intrinsic ductility and oxidation resistance, and consequently, many studies have been dedicated to investigating Nb–Si–Ti system for a better understanding of its effect on the phase equilibria and material properties. While there are a number of elements that have been considered as potential constituent elements, both theoretical and experimental studies of the ternary systems with these potential elements are quite limited. Providing information about phase equilibrium and material properties of basic ternary systems is fruitful to further accelerate the rapid discovery of new candidate compositions by combining experimental verification with advancing computational power.

In this study, three different ternary systems, Nb–Si–X (X = Zr, Ta, Al), were selected to investigate microstructure, phase stability, and high-temperature mechanical properties. Zr was selected due to the similar phase transformation behavior to Ti and could improve room temperature fracture toughness.[6] Ta was selected due to its high solubility limit to Nb, thereby expected to be effective in increasing the solid solution strengthening effect of the Nb-based matrix. Its effect on material properties will be compared to the previous works that used other refractory metals. Al was almost exclusively added to Nb-silicide-based composites to improve oxidation resistance, but not many studies about their exact effect on Nb-Si binary system. In order to explore the wide compositional space, Si and X contents were varied to change the constituent phases and their volume fractions.

The isothermal sections of ternary Nb–Si–X systems (X = Zr, Ta, Al) were calculated using Thermo-Calc software with TCHEA5 database. Two compositions were selected from (a) solid solution Nb (ssNb) + Nb5Si3 regime and (b) ssNb + Nb5Si3 + Nb3Si regime to validate the calculation and compare the material properties with or without Nb3Si phase. For the Nb–Si–Ta system, due to a lack of ssNb + Nb5Si3 + Nb3Si regime, both compositions were selected from ssNb + Nb5Si3 regime while varying Si and Ta content significantly. Selected compositions are listed in Table I as well as marked in Figure 1. The total silicide volume fraction in selected compositions was also calculated using TCHEA5.

Table I The Nominal Compositions of Selected Six Alloys
Full size table
Fig. 1
figure 1

Isothermal sections of (a) Nb–Si–Zr, (b) Nb–Si–Ta, and (c) Nb–Si–Al ternary systems at 1300 °C. Selected compositions are marked

Full size image

In order to validate the thermodynamic predictions, 20 g ingots of selected alloys were fabricated by arc melting under a flowing Ar atmosphere with a Ti getter. The ingots were homogenized at 1300 °C for 96 hours in a vacuum furnace, and then furnace cooled to room temperature. The constituent phases and microstructure of each ingot were characterized using Rigaku Smartlab X-ray diffractometer, JEOL JSM-7900F scanning electron microscopy equipped with a backscatter electron detector, energy-dispersive X-ray spectroscopy (EDS), and JEOL JSM-7200F scanning electron microscopy equipped with electron backscattered diffraction (EBSD). Microhardness was measured using a Shimadzu HMV-1 microhardness tester at room temperature. The indent was spaced every 200 µm with a load of 500 g, and the measurements were repeated once. High-temperature compression tests were performed using a Shimadzu AG-I universal testing machine equipped with a vacuum furnace. The test was initiated once the vacuum level reached at least 10−4 Pa. All tests were completed at 1300 °C with a strain rate of 10−4/s.

Figure 1 shows the calculated isothermal sections of Nb–Si–X (X = Zr, Ta, Al) at 1300 °C. Narrow solid solution Nb (ssNb) fields were predicted in all three fields. ssNb + Nb5Si3 phase fields also exist in all systems with 2 to 37 at. pct Si. Nb–Si–Zr exhibits a very narrow ssNb + Nb5Si3 phase field with up to 2 at. pct Zr, whereas it is relatively wider in Nb–Si–Al with up to 10 at. pct Al, then becomes significantly wider in Nb–Si–Ta with up to 50 at. pct Ta. Nb–Si–Zr and Nb–Si–Al exhibit wide ssNb + Nb5Si3 + Nb3Si field, while it is absent in Nb–Si–Ta.

Table II shows the crystal structures of identified phases and lattice parameters obtained by XRD. The indexed XRD spectra are available as supplementary information (see Electronic Supplementary Figure S1). In addition to the matrix BCC Nb phase, three different Nb5Si3-type silicides, α-Nb5Si3 (tetragonal, D8l), β-Nb5Si3 (tetragonal, D8m), and γ-Nb5Si3 (hexagonal, D88), are identified. The primary Nb5Si3-type is detected as β-Nb5Si3 in Ta-B, whereas it is α-Nb5Si3 in all other alloys. γ-Nb5Si3 is only detected in Zr-B, Ta-A, and Al-B alloys. In addition, a Nb3Si-type silicide with a tetragonal structure is also detected in all alloys. Regardless of element X, the lattice parameter of ssNb increases as element X concentration increases. Comparing Zr-A and Zr-B, and Al-A and Al-B, the lattice parameter of α-Nb5Si3 of a increases but c decreases as Si content increases and element X content decreases thereby c/a ratio decreases. α-Nb5Si3 lattice exhibits two systematically non-equivalent Nb and Si positions: (a) a close-packed plane (Nb-I, Si-I), and (b) a less close-packed plane (Nb-II, Si-II). Shi et al.[7] calculated the formation of α-Nb5Si3 by introducing third elements and predicted that Zr preferentially occupied Nb-II sites, while Al preferentially occupied Si-II sites to minimize the lattice expansion. Our result indicates that introducing alloying elements to less close-packed layers results in an increase of a and a decrease of c. The Nb3Si phase exhibited the same tendency as its lattice parameter of a increases but c decreases as Si content increases and element X content decreases. This is consistent with the experimental observation by Sanker et al.,[8] who also reported the increase of a and decrease of c in Nb3Si by Zr addition. Ta-A and Ta-B also exhibit the change in the lattice parameters of Nb3Si; however, a decreases as c increases as Ta content increases. This is consistent with Sankar et al.’s findings and prior DFT calculations.[7,9]

Table II The Crystal Structures and Lattice Parameters of Identified Phases by X-ray Diffraction Analysis
Full size table

Overall, the phase equilibrium predictions made by TCHEA5 are in good agreement with the experimental results. The constituent phases of Zr-A, Ta-B, and Al-A alloys were predicted as ssNb and Nb5Si3, which are consistent with experimental results. Ta-A, however, contains Nb3Si phase in addition to the predicted two phases. The constituent phases of Zr-B and Al-B were predicted as ssNb, Nb5Si3, and Nb3Si, which are also consistent with experimental results. Ta is known to preferentially occupy the Nb site and stabilize Nb3Si,[10] so having Nb3Si in addition to ssNb and Nb5Si3 is expected in Ta-A and Ta-B. Nb–Si–Ta ternary system was not critically or tentatively assessed in the utilized database.[11] Thus, it is reasonable that the calculation did not accurately predict equilibrium phases. Li et al.[5] experimentally investigated the phase equilibrium of Nb–Si–Ta system and reported that tantalum silicide phases would likely form by increasing both Si and Ta contents, but this information is not yet included. Nb–Si–Zr was also not assessed in this database, but unlike Nb–Si–Ta, the equilibrium prediction was accurate. Nb–Si–Ti system was critically assessed in TCHEA5, and Nb–Si–Zr exhibited similar constituent phases to Nb–Si–Ti at given equilibrium temperatures. It would allow us to make accurate equilibrium predictions without assessed information.

The results of phase field modeling and phase identification are further validated by SEM and EBSD analysis. Figure 2 shows the microstructure of six alloys after homogenizing treatment.

Fig. 2
figure 2

SEM micrographs of (a) Zr-A, (b) Zr-B, (c) Ta-A, (d) Ta-B, (e) Al-A, and (f) Al-B alloys after homogenized at 1300 °C for 96 h

Full size image

In Group A alloys (Zr-A, Ta-A, Al-A), ssNb appears as large proeutectic islands in dendritic form. ssNb is also observed in the fine eutectic colonies along with Nb5Si3 and Nb3Si phases. Interlamellar spacing is 200–500 nm in Zr-A, 200–600 nm in Ta-A, and 500–800 nm in Al-B. In Zr-A and Al-A, dark areas are observed near ssNb islands. In these areas, ssNb and α-Nb5Si3 are detected in Zr-A, whereas they consist of ssNb, Al-rich α-Nb5Si3, and Nb3Si in Al-A.

Group B alloys (Zr-B, Ta-B, Al-B) exhibited different morphologies of ssNb and silicide phases compared to Group A alloys. In Zr-B, ssNb appears as islands of various sizes, from nano-scale to over 10 µm. α-Nb5Si3 is observed as primary silicide islands. Additionally, the eutectic mixture of ssNb, α-Nb5Si3, γ-Nb5Si3, and Nb3Si is detected. Zr-rich (α,γ)-Nb5Si3 appears as darker areas, as shown in Figure 2(b). The same structural trends are also observed in Al-B. The β-Nb5Si3 exhibits distorted cuboidal morphology in Ta-B, as shown in Figure 2(d). Distorted cuboidal primary silicides appear in both Zr-B and Ta-B, but the edge length of silicides in Ta-B is approximately twice as large as the ones in Zr-B.

The eutectic mixture of ssNb, (α,γ)-Nb5Si3, and Nb3Si is detected in fine lamella areas in all Group A alloys. Shi et al.[7] calculated the formation energy of α-Nb5Si3 and reported it would be significantly high with Al addition compared to Zr or Ta addition. Additionally, Tang et al.[12] suggested that Al prefers to occupy ssNb phase rather than α-Nb5Si3. Thus, Al is less likely to dissolve into α-Nb5Si3, thereby forming Al-rich ssNb phase. Both Al-A and Al-B exhibit ultrafine eutectic mixtures of ssNb and silicide phases, which is consistent with the findings from Kashyap et al.[13]

Figure 3 shows EBSD phase maps obtained at the area with fine eutectic mixtures of ssNb and silicides. The primary silicide phases in these areas are identified as α-Nb5Si3 in Zr-A, Zr-B, Ta-A, Al-A, and Al-B, whereas it is β-Nb5Si3 in Ta-B. These results are consistent with XRD results. Although γ-Nb5Si3 is identified, its volume fraction is significantly small compared to other phases: 0.8, 0.8, and 2.7 pct in Zr-B, Ta-A, and Al-B, respectively. Figure 4 shows EBSD phase maps obtained at ssNb/silicide boundaries located in fine eutectic mixture regions. In these regions, Nb3Si, α-Nb5Si3, and γ-Nb5Si3 are detected. This confirms that secondary and third silicide phases are finely distributed and tend to form at ssNb/silicide boundaries. Fine γ-Nb5Si3 with a scale up to 200 nm is found at ssNb islands near ssNb/Nb3Si boundaries in Zr-B, whereas it is broadly distributed in Nb3Si + α-Nb5Si3 regime near ssNb in Ta-A and Al-B as shown in Figures 4(a), (b), and (e).

Fig. 3
figure 3

EBSD phase maps of (a) Zr-A, (b) Zr-B, (c) Ta-A, (d) Ta-B, (e) Al-A, and (f) Al-B alloys obtained at eutectic mixture of ssNb and silicide phases. Yellow represents ssNb, red represents α-Nb5Si3, pink represents β-Nb5Si3, green represents γ-Nb5Si3, and blue represents Nb3Si (Color figure online)

Full size image
Fig. 4
figure 4

EBSD phase maps obtained at ssNb/primary silicide interfaces: (a) Zr-B, (b) Ta-A, (c) Ta-B, (d) Al-A, and (e) Al-B alloys obtained at eutectic mixture of ssNb and silicide phases. Yellow represents ssNb, red represents α-Nb5Si3, pink represents β-Nb5Si3, green represents γ-Nb5Si3, and blue represents Nb3Si (Color figure online)

Full size image

Both Zr-A and Zr-B exhibited α-Nb5Si3 as a primary Nb5Si3-type, as shown in Figures 3(a) and (b). It was suggested that Zr tends to occupy the α-Nb5Si3 phase as well as work as α-Nb5Si3 stabilizer,[7,12] which is in agreement with our observations. Even though Nb3Si formation was not predicted in Group A alloys and Ta-B, it is identified in all six alloys by XRD analysis. However, in EBSD mappings, its volume fraction is significantly different by alloy composition; 0.2 pct (Zr-A), 8.0 pct (Zr-B), 19.2 pct (Ta-A), 0.4 pct (Ta-B), 0.5 pct (Al-A), and 13.0 pct (Al-B). It was reported that the decomposition of Nb3Si to ssNb and α-Nb5Si3 accelerated with more than 2 at. pct addition of Zr due to the change in a/c ratio of lattice parameters of Nb3Si, which ultimately made it unstable.[8] Zr-A has a similar nominal composition to their alloys and barely exhibited Nb3Si phase, indicating most of Nb3Si was decomposed to ssNb and α-Nb5Si3. Conversely, Zr-B has a higher fraction of Nb3Si with higher Zr content than Zr-A. It has been revealed that the dissolution rate of Nb3Si is quite slow, even at a temperature above 0.9Tm.[14] Our result suggested that the decomposition acceleration effect of Zr on dissolution kinetics to high silicide fraction alloy is limited even with a high Zr concentration, and therefore, the homogenizing treatment with longer duration and higher temperature is needed in order to complete the decomposition.

Table III shows the Vickers microhardness of six alloys. Regardless of the third element, Group B alloy exhibited higher hardness than Group A alloy. In both groups, Zr alloy exhibited the lowest hardness, whereas Ta alloy exhibited the highest. Figure 5(a) shows the stress–strain curves obtained by high-temperature compression tests. The compression test for Ta-B and Al-B was not able to be performed due to its brittleness, which caused the failure of sample preparation. The 0.2 pct proof stress of each alloy is 419 MPa (Zr-B), 159 MPa (Zr-A), 132 MPa (Ta-A), and 64 MPa (Al-A) in descending order. The estimated yield strength at room temperature and 0.2 pct proof stress at 1300 °C are compared in Figure 5(b). The yield strength was estimated by \({\sigma }_{\text{y}}\approx {H}_{\text{v}}/3\), where Hv is the Vickers microhardness. Although Zr-A exhibits the lowest hardness among Group A alloys, 0.2 pct proof stress at 1300 °C is 20 pct and 148 pct higher than Ta-A and Al-A, respectively. In addition, while Ta-A exhibits a 15 pct higher yield strength than Zr-B at room temperature, its 0.2 pct proof stress is less than 1/3 of Zr-B. It is remarkable that Zr addition can effectively improve the high-temperature strength while remaining ductile at room temperature.

Table III Vickers Microhardness of All Six Alloys
Full size table
Fig. 5
figure 5

(a) True stress–True strain curves of Zr-A, Zr-B, Ta-A, and Al-A performed at 1300 °C with strain rate of 10−4/s, and (b) Comparison between the estmated yield strength of room temperature and 0.2 pct proof stress at 1300 °C

Full size image

Generally, lamella structures have more ssNb/silicide interfaces that may cause more rapid diffusion of elements as temperature increases; therefore, Group A alloys tend to exhibit lower 0.2 pct proof stress than Group B alloys. Regarding the effect of alloying elements on the mechanical properties, it was predicted that Zr and Al tend to decrease the overall strength of α-Nb5Si3 phase, whereas Ta tends to increase the strain resistance at 0 K.[7,12] In our results, the hardness of Ta-A is 66 pct higher than Zr-A and 46 pct higher than Al-A, which is consistent with these predictions. Comparing Zr-A and Zr-B, the 0.2 pct proof stress of Zr-B is 163 pct higher. Zr-B exhibits large primary α-Nb5Si3 islands. They are considered to be effective obstacles for mobile dislocations traveling through the ssNb phase. High-temperature deformation of Nb-silicide-based alloys is dominated by the plastic deformation of ssNb, and silicide phases disrupt further deformation.[15,16]

Zr-B and Ta-A exhibit softening behavior after reaching peak stress, while it is absent in Zr-A and Al-A, which might be triggered by continuous dynamic recrystallization (CDRX) in ssNb and silicide phases, activated by external applied stress and high-temperature environment.[17] Furthermore, Zr-A exhibited a serration in stress–strain curves, known as Portevin–Le Chatelier (PLC) effect. This occurs due to the interaction between solutes and mobile dislocations.[18] Serration is classified into five categories based on the different characteristics. In the present work, it exhibited Type D, which is represented by a plateau and then a sudden increase in stress while the strain rate kept increasing. This behavior indicating that the inhomogeneous local deformation occurs due to the PLC effect. Further investigation of deformed microstructure is needed for the verification of the effect of microstructure evolution on above deformation behavior as well as to determine the deformation mechanisms.