Design and Characterization of a Novel NiAl–(Cr,Mo) Eutectic Alloy

To enhance the mechanical properties of NiAl–(Cr,Mo) in situ composites for high‐temperature structural use, achieving a regular eutectic microstructure is crucial. A novel alloy with a composition of Ni30.6Al36Cr31.4Mo2 is developed that exhibits a significantly reduced solidification interval compared to well‐studied alloys such as Ni33Al33Cr28Mo6. This advancement results from using the in‐house developed alloy design tool PyMultOpt and Calculation of Phase Diagrams analysis to minimize the solidification interval. The refined alloy demonstrates a theoretical solidification interval of 0 K. These results are verified with differential scanning calorimetry measurements at different heating rates for the new alloy and compared with the established alloy Ni33Al33Cr28Mo6. The small solidification interval leads to a uniform eutectic microstructure consisting of a (Cr,Mo) phase embedded in the NiAl matrix without primary NiAl dendrites present. This indicates that deviating from the stoichiometric NiAl composition results in a significantly lower solidification interval and thus in an actual eutectic composition of the NiAl–(Cr,Mo) system.


Introduction
The aircraft industry's need for high-temperature structural materials with exceptional physical and mechanical properties necessitates the advancement of novel alloys.12][13] Depending on the Mo concentration, the reinforcement phase's morphology changes. [11]This was first investigated for various stoichiometric Ni 33 Al 33 Cr 34Àx Mo x compositions (with x ranging from 0 to 9 at%) by Cline et al. [11] Below a molybdenum content of 0.6 at%, the (Cr,Mo) phase is present as well-aligned rods.By increasing Mo, the morphology is altered to a lamellae structure.Orientation relationship and lattice parameter mismatch analyses, performed by Schulz et al. revealed the cause of the morphology change. [14,15]For instance, whereas the NiAl lattice parameter does not change with increasing Mo content, the (Cr,Mo) lattice parameter first decreases till it reaches 0 at 0.6 at% and subsequently increases continuously.In addition, the cube-on-cube orientation between NiAl and (Cr,Mo) changes at 0.6 at% to a misorientation of 60°rotation about <111>.Besides the Mo content, the nonequilibrium solidification condition such as solidification rate influences the microstructure.[18][19] In the latter, the solidification rate can be adjusted by varying the withdrawal rate.Whittenberger et al. investigated the impact of the withdrawal rate on the microstructure and thus on the mechanical properties of a Ni 33 Al 33 Cr 33 Mo 1 alloy (all following concentrations are given in at%). [20]The authors observe for low withdrawal rates millimeter-diameter grain sizes consisting of parallel aligned micrometer-thick (Cr,Mo) and NiAl plates.By increasing the withdrawal rate, and thereby moving further away from equilibrium solidification conditions, the grain size decreases, intercellular regions form, and the (Cr,Mo) phase is arranged in a radial manner.Within the cell, the (Cr,Mo) phase is refined, while experiencing coarsening in the intercellular spaces.The difference between (Cr,Mo) phase thicknesses becomes more pronounced for withdrawal rates higher than 127 mm h À1 , leading to irregular microstructure distribution.Furthermore, in this study, it was possible to enhance the room temperature toughness to a level of 16 MPa m 0.5 for growth rates between 12.7 and 127 mm h À1 in comparison to only 6 MPa m 0.5 for binary NiAl. [20,21]Surpassing a withdrawal rate of 127 mm h À1 , room temperature toughness decreases to the level of binary NiAl.According to the authors, the microstructure is responsible for the lower mechanical properties. [20]This displays the importance of the microstructure and its regularity, wherein regularity, for example, entails uniform (Cr,Mo) phase thickness or less intercellular regions.Shang et al. reported an even higherroom temperature toughness of around 20.8 MPa m 0.5 for the Ni 33 Al 33 Cr 28 Mo 6 alloy. [22]The microstructure for the latter alloy reveals eutectic cells containing ordered NiAl and (Cr,Mo) lamellae arranged in a circular fashion. [17]The phase thickness of the strengthening (Cr,Mo)-phase increases from the cell interior to the intercellular region, which leads to an irregular microstructure.Moreover, Förner et al. found in the as-cast microstructure primary NiAl dendrites, which is a result of nonfully eutectic solidification. [23]In general, the literature presents conflicting perspectives regarding whether the Ni 33 Al 33 Cr 28 Mo 6 alloy represents a eutectic composition. [16,17,22]Nevertheless, a neareutectic material (solidification interval ΔT > 0 °C) such as Ni 33 Al 33 Cr 28 Mo 6 solidifies at high solidification rates and, consequently, under non-equilibrium conditions, in a eutectic manner.For near-eutectic materials at low solidification rates, aligning more closely to equilibrium conditions, the formation of primary dendrites is more likely.However, a truly eutectic NiAl-(Cr,Mo) composition widens the process window by reducing the dependency on a minimum required solidification rate, thus improving the processability of the material.Hence, to exploit the full potential of the NiAl-(Cr,Mo) system, a uniform, eutectic composition and microstructure is desirable.
Recent advancements in refining and optimizing the system's thermodynamic description have created new opportunities to explore novel alloy compositions. [24]Based on the work of Peng et al. and an in-house developed alloy design tool, a novel eutectic NiAl-(Cr,Mo) composite is calculated and compared to the so-believed Ni 33 Al 33 Cr 28 Mo 6 eutectic alloy. [25,26]For both alloys, the theoretical results are validated by differential scanning calorimetry (DSC) measurements.The microstructures of directionally solidified samples are investigated.

Experimental Section
The alloy design tool PyMultOpt (in-house development) was used for the development of the eutectic alloy.It employs thermodynamic calculations based on the Calculation of Phase Diagrams (CALPHAD) method to describe alloy properties in combination with a genetic multicriteria optimization algorithm to optimize alloy compositions toward certain design goals. [27,28]o achieve a regular two-phase microstructure without primary solidified dendrites, a NiAl-(Cr,Mo) alloy should be as close to the eutectic composition as possible, which means its solidification interval should be close to zero.Thus, an alloy was developed by minimizing the difference between solidus and liquidus temperature.Furthermore, a restriction was established ensuring only the B2-NiAl and (Cr,Mo) solid solution phases existed in the solid state.For thermodynamic calculations, a database by Peng et al. was used, which provides an improved description of the eutectic trough in the Ni-Al-Cr-Mo system compared to commercially available databases. [25]All CALPHAD calculations were carried out using the Thermo-Calc software version 2017b and the TC-API by Thermo-Calc.
To analyze the microstructures of both alloys, cast ingots were prepared using arc melting under vacuum on a Bühler AM/05 machine.Therefore, nickel, aluminum, chromium, and molybdenum (each with a purity of more than 99.95%) were mixed in a crucible according to nominal composition and remelted 3 times for homogenization.Subsequently the cast ingots were directionally solidified via the Bridgman process using a withdrawal rate of 1 mm min À1 .A total of six rods-12 mm diameter and 140 mm height-were produced.The rods were cut parallel and perpendicular to the withdrawal direction and subsequently metallographically prepared.The microstructural examination was conducted with a Quanta 450 scanning electron microscope and images were further analyzed using ImageJ (National Institute of Health, Bethesda, MD, USA).The compositions of the different phases were analyzed by atom probe tomography (APT) measurements, where site-specific lift-outs were performed in a Zeiss 540 XB (Carl Zeiss AG, Germany), following the procedure described in ref. [29].The measurements were conducted in a Cameca LEAP 4000X (CAMECA Inc., Madison, USA), which was operated in laser mode to trigger field evaporation at a temperature of 45 K and a laser energy of 50 pJ.Reconstruction was carried out using Cameca's commercially available IVAS 3.6.8software.
To verify the theoretically predicted results of the solidification intervals, DSC was used.With this method, the phase transition temperatures can be identified.These temperatures corresponded to the initial and the final peak temperature T i and T f , which were the first and the last deviation from the base line of the DSC signal, see Figure 1a.Since it is not possible to reliably determine the first and last deviation in practice, the extrapolated temperatures T onset and T offset were determined in accordance with EN ISO 11357-1. [30]These temperatures were found where a constructed baseline intersects with tangents.These tangents were drawn at the curve's inflection points, which were parts of the peak's branches.The differences between T i , T f and T onset , T offset were partially compensated by the temperature calibration done the same way.Because of delay effects, the characteristic temperatures T onset and T offset were dependent on the heating rate.This mainly affects T offset , since the reference temperature reaches higher values during the reaction time with higher heating rate.To avoid further influence on T offset , the sample's weight, geometry, and contact area with the crucible were held constant.The characteristic temperatures were measured with different heating rates and then extrapolated to a heating rate of 0 K min À1 , see Figure 1b.The detected temperatures T onset ð Ṫ ¼ 0Þ and T offset ð Ṫ ¼ 0Þ corresponded to the equilibrium solidus and liquidus temperatures T sol and T liq .For the measurements, specimens were heated with different heating rates of 0.5, 1.0, 2.0, 5.0, 10, and 20 K min À1 up to 1500 °C.The device used was a NETZSCH STA 409 DSC/TG.For the arc-remelted and subsequently directionally solidified cast samples, it is expected that the solidification interval will be slightly higher due to evaporation effects.

Results and Discussion
The base alloy's ratio of Ni and Al is unity, which is the stoichiometric composition of the B2 NiAl-phase.However, according to CALPHAD calculations, Cr is highly soluble in the NiAl phase of the base alloy Ni 33 Al 33 Cr 28 Mo 6 (about 6 at% Cr at 1400 °C), while the content of Mo in NiAl is below 0.1 at%. [25]In contrast, Ni is moderately and Al greatly soluble in the (Cr,Mo) solid solution phase.Thus, in the base alloy Ni 33 Al 33 Cr 28 Mo 6 , the composition of the NiAl phase is expected to deviate from the stoichiometric composition, with an excess of Ni and therefore a deficiency of Al.The calculated Ni:Al ratio along the eutectic trough supports the latter, indicating a Ni:Al value greater than 1. [25] The solubility predictions were confirmed by atom probe measurements done by Förner et al. showing that both phases exhibit some solubility of all elements. [31]he optimized alloy's ratio of Ni and Al deviates from unity by definition.Due to the Al excess, a deviation from the stoichiometric B2 phase toward the Al-rich side is therefore expected in the modified alloy.Maintaining the equal ratio of Ni and Al limits the minimum achievable solidification range to 10 °C and thus the approximation of the global eutectic composition.Only by relaxing this constraint, the solidification range could be further minimized compared to previously investigated NiAl-(Cr,Mo) alloys.The theoretical solidification intervals for  [25] In both cases, the nickel and aluminum content was held constant at 33 at%, 33 at% and 30.6 at%, 36 at%, respectively.Only the chromium and molybdenum content was modified.The base alloy exhibits a solidification interval of 17 °C, whereas the optimized alloy shows a solidification interval of almost 0 °C.various stoichiometric NiAl-(Cr,Mo) alloys can be found in Table S1, Supportive Information.

DSC Verification
DSC measurements were executed to verify the theoretical solidification intervals of both alloys.The extrapolated onset and offset temperatures T onset ð Ṫ ¼ 0Þ and T offset ð Ṫ ¼ 0Þ and therefore the equilibrium solidus and liquidus temperatures T sol and T liq for both alloys are listed in Table 1.
As anticipated from the CALPHAD calculation, the solidification interval for the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 is significantly lower at around 5 °C than the one for the base alloy Ni 33 Al 33 Cr 28 Mo 6 at around 26 °C.In comparison to the theoretical solidification interval of 17 °C for the base alloy and nearly 0 °C for the optimized alloy, the experimental values are slightly higher.However, considering the inherent inaccuracies of both the experimental method and the CALPHAD calculations, it can be assumed that the thermodynamic simulation and the experimental data are in good agreement with each other.Furthermore, it should be noted that using T offset for finding the melting interval always results in values that are slightly too high.The actual end of the phase transformation is somewhere in between the peak temperature T p (see Figure 1) and the offset temperature T offset , since the temperatures of the samples and the reference have to equalize again after the end of the reaction.,32] Shang et al. attributed the formation of cellular eutectic microstructures to small inevitable impurity elements and molybdenum enrichment at the solidus-liquidus interface. [22]Cell size measurements exhibit an average cell size of 36.6 AE 6.0 μm.At the cell boundaries, the (Cr,Mo) lamellae are coarsened.Within the cells, the lamellae thickness is on average 240 AE 120 nm.This value increases nearly by a factor of 5 to 1201 AE 512 nm at the cell boundaries.The higher lamellae thickness is a result of a decrease in the thermal gradient and solidification rate at the cell boundaries. [17]herefore, diffusion is enhanced and the lamellae growth is more pronounced.It should be noted that the wide variation in lamellae thickness arises from the 2D analysis of circular arranged (Cr,Mo) lamellae.

As-Cast Microstructure
In general, the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 (see Figure 4b) reveals a more regular microstructure with alternating eutectic NiAl matrix and (Cr,Mo) lamellae.In contrast to the base alloy, the coarsening of the (Cr,Mo) lamellae at the cell boundaries is much less pronounced.The average lamella spacing for the center as well as the cell boundary region are 299 AE 90 nm and 870 AE 362 nm, respectively.Therefore, the difference between the cell interior and the cell boundary decreased for the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 .On the contrary the average cell size doubled to a value of 69.6 AE 4.9 μm.Considering the non-equilibrium solidification during the Bridgman process, the thermal gradient and actual liquidus temperature at the solidusliquidus interface determine the level of undercooling and therefore, the driving force behind solidification. [33]Given that the thermal gradient is similar for both castings, the actual liquidus temperature is key to the microstructural differences observed.The base alloy Ni 33 Al 33 Cr 28 Mo 6 is expected to exhibit higher undercooling due to a higher actual liquidus temperature because it is a non-eutectic alloy (with a theoretical solidification  interval > 0 °C).Conversely, the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 (with a theoretical solidification interval = 0 °C) has a lower actual liquidus temperature, resulting in a reduced driving force and, consequently, a coarser cell structure.
Figure 5 displays the microstructure of the optimized alloy cast ingot at a withdrawal distance of 65 mm.It can be clearly seen that no coarsening of the lamellae occurs at the cell boundaries.This morphology results from a more uniform, nearly planar eutectic growth during solidification due to stable process conditions such as a nearly constant thermal gradient and solidification rate.The (Cr,Mo) lamellae thickness measures 523 AE 110 nm and is comparable to the mean value of the cell interior and intercellular region in Figure 4b.Such fully uniform microstructure regions are not detected in the base alloy Ni 33 Al 33 Cr 28 Mo 6 , regardless of the withdrawal distance.One might expect that a regularly distributed (Cr,Mo) phase, such as that shown in Figure 5, could potentially enhance ductility.Coarsened intercellular regions exhibit larger areas of pure NiAl phase, in which the ductility at room temperature is poor.By eliminating these coarsened intercellular regions, fewer of these weak points would be present, and therefore, a higher ductility could be expected.Nevertheless, the increase in ductility requires experimental validation.Around 42.2 AE 1.3 vol% of the base alloy consists of (Cr,Mo) phase, which is in good agreement with the experimental data by Gombola et al. [14] By reducing the Mo content in the (Cr,Mo) solid solution phase, a decrease in (Cr,Mo) phase fraction is reported. [14]The identical trend is determined for the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 with a (Cr,Mo) phase fraction of 39.2 AE 1.6 vol%.
As mentioned before, it is only possible to achieve an optimized alloy with a small solidification interval of less than  2a. [31]For the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 the NiAl phase exhibits a non-stoichiometric composition with an Al excess, resulting in a Ni:Al ratio of 0.90.Because of the higher amount of Al, constitutional vacancies in the Ni sublattice are generated, which can influence the mechanical properties. [6,34]Cr and Mo are barely present with 0.3 AE 0.2 at% and 0.1 AE 0.1 at% in the intermetallic NiAl phase due to its strong order tendency.For the regular eutectic regions, Cr and Mo are dissolved in the NiAl matrix, whereas for the intercellular coarser regions in Figure 4b (Cr,Mo) nano-precipitates  are present.Förner et al. reported a similar solubility of both Cr and Mo in the directionally solidified ingots and additive manufactured samples of Ni 33 Al 33 Cr 28 Mo 6 . [23,31]The Ni:Al ratio changes to 1.05, which means an excess of Ni in the NiAl phase.Consequently, changing the Ni:Al ratio within the NiAl phase does not influence the solubility of Cr and Mo.As expected, the bright (Cr,Mo) phase consists predominantly of 84.1 AE 0.5 at% Cr and 5.4 AE 0.2 at% Mo.Whilst Al is highly soluble with 10.3 AE 0.5 at%, Ni is nearly not present in the (Cr,Mo) solid solution phase.This elevated solubility of Al is in good agreement with the calculated results at 1000 °C, see Table 2b.
Investigations of the NiAl-(Cr,Mo) system with stoichiometric composition of the NiAl phase have shown that with increasing molybdenum content in the (Cr,Mo) phase, the solidification interval increases until the NiAl-Mo ternary is reached. [14]his strongly suggests that alloys with a stoichiometric composition of NiAl and a mixed content of Mo and Cr deviate from the actual eutectic composition.The results of this study have shown that to accurately calculate the eutectic composition with variable Cr and Mo content, the system must be considered quaternary rather than quasi-ternary.Accordingly, for changing Mo and Cr ratios, the ratio and fraction of Ni and Al must be recalculated in order to achieve a truly eutectic composition.

Summary and Conclusion
This study introduces a novel eutectic NiAl-(Cr,Mo) alloy with a composition of Ni 30.6 Al 36 Cr 31.4Mo 2 .The developed alloy exhibits promising microstructural features with a high potential to overcome current limiting factors such as low fracture toughness at room temperature.The following conclusion can be drawn.1) Thermodynamic calculation shows a solidification interval of 17 °C for the base alloy Ni 33 Al 33 Cr 28 Mo 6 .The optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 exhibits a theoretical solidification interval of nearly 0 °C.2) DSC measurements verify both calculated solidification intervals and therefore, confirm the near eutectic character of the base alloy and the eutectic composition of the novel NiAl-(Cr,Mo) alloy.3) Microstructural investigations of both alloys highlight the more regular microstructure with less pronounced coarse intercellular regions and the potential for the optimized alloy to completely eliminate the lamellae thickness differences within a cell.4) The chemical compositions of both phases display the necessity to treat the Ni-Al-Cr-Mo system as a quaternary instead of a quasi-ternary system, if one is aiming for a eutectic composition.In addition, because of the mutual solubility, a nominal stoichiometric NiAl concentration does not result in stoichiometric composition.By allowing a nominal off-stoichiometric composition of NiAl, the resulting alloys exhibit a smaller solidification interval than 10 °C and therefore a eutectic character.
Table 2. a) Phase-specific composition of the (Cr,Mo) phase and the NiAl phase in the cell center using APT.The results for the base alloy Ni 33 Al 33 Cr 28 Mo 6 are reported by Förner et al. for directionally solidified samples with an identical withdrawal rate. [31]b) Calculated chemical compositions for each phase at 1000 °C.
3.1.PyMultOpt OptimizationThe minimization of the solidification interval based on thermodynamic calculations yielded a composition of Ni 30.6 Al 36 Cr 31.4Mo 2 (at%).It should be noted that this final composition is rounded to 0.1 at%, since lower tolerances are hardly achievable during alloy production.In Figure 2, the solidus and liquidus temperature curves, T sol and T liq , respectively, are presented for a) the base alloy Ni 33 Al 33 Cr 28 Mo 6 and b) the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 .The minimum solidification interval for the base alloy Ni 33 Al 33 Cr 28 Mo 6 amounts to 17 °C between T liq = 1450 °C and T sol = 1433 °C.As a result of the optimization, the minimum solidification interval for the Ni 30.6 Al 36 Cr 31.4Mo 2 alloy is theoretically just above 0 °C.

Figure 2 .
Figure 2. Calculated liquidus and solidus curves based on the database developed by Peng et al. for a) the base alloy Ni 33 Al 33 Cr 28 Mo 6 and b) the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 .[25]In both cases, the nickel and aluminum content was held constant at 33 at%, 33 at% and 30.6 at%, 36 at%, respectively.Only the chromium and molybdenum content was modified.The base alloy exhibits a solidification interval of 17 °C, whereas the optimized alloy shows a solidification interval of almost 0 °C.

Figure 1 .
Figure 1.a) Schematic DSC curve with the characteristic temperatures: initial temperature T i , peak temperature T p , final temperature T f , onset temperature T onset , and offset temperature T offset .b) By extrapolating T onset and T offset from various heating rates, the equilibrium solidus and liquidus temperature T sol and T liq can be determined.These correspond to the temperatures at a heating rate of 0 K min À1 .

Figure 3
depicts the onset and offset temperatures T onset and T offset for various heating rates for a) the base alloy Ni 33 Al 33 Cr 28 Mo 6 and b) the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 .

Figure 4
Figure 4 depicts the microstructure of both alloys at a withdrawal distance of 30 mm.The base alloy Ni 33 Al 33 Cr 28 Mo 6 (Figure 4a) reveals non-uniform cellular eutectic cells consisting of irregular bright (Cr,Mo) lamellae embedded in the darker NiAl matrix.The cellular growth of lamellae is consistent with previous findings on this alloy composition.[14][15][16]21,32] Shang t al. attributed the formation of cellular eutectic microstructures to small inevitable impurity elements and molybdenum enrichment at the

Figure 3 .
Figure 3. DSC measurements at various heating rates for a) the base alloy Ni 33 Al 33 Cr 28 Mo 6 and b) the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 .The equilibrium solidus and liquidus temperatures T sol and T liq are determined by extrapolating T onset and T offset to 0 K min À1 .
10 °C by deviating from a ratio of unity for Ni and Al.APT measurements for the base alloy Ni 33 Al 33 Cr 28 Mo 6 by Förner et al. and for the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 support this statement by revealing the individual element solubility in each phase, see Table

Figure 4 .
Figure 4. Microstructure of the as-cast alloys-perpendicular to the withdrawal direction.In backscattered electrons mode, the NiAl phase appears dark, whereas the Cr,Mo phase appears bright.a) The base alloy Ni 33 Al 33 Cr 28 Mo 6 exhibits a cellular eutectic microstructure with (Cr,Mo) lamellae embedded in a NiAl matrix.b) The optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 shows a more regular microstructure with alternating NiAl and (Cr,Mo) phase lamellae.

Figure 5 .
Figure 5. Microstructure of the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 showing regions with a completely regular eutectic microstructure without coarsened (Cr,Mo) lamellae at the cell boundaries.The measured lamellae thickness in these regions corresponds to the mean value of the cell interior and intercellular region, cf.Figure 4b.

Table 1 .
Solidus and liquidus temperature T sol and T liq and the solidification interval for the base alloy Ni 33 Al 33 Cr 28 Mo 6 and the optimized alloy Ni 30.6 Al 36 Cr 31.4Mo 2 .The theoretical solidification interval was calculated with CALPHAD.