The Influence of Mo Additions on the Microstructure and Mechanical Properties of AlCrFe2Ni2 Medium Entropy Alloys

The alloy system Al-Cr-Fe-Ni provides means for developing novel duplex materials composed of face-centered cubic (FCC) and body-centered cubic (BCC) phases with nearly equal volume fraction. We performed an alloy development study starting from the medium entropy alloy AlCrFe2Ni2 and adding small amounts of molybdenum in the following series (at.%): Al17Cr17Fe33Ni33, Al17Cr17Fe33Ni33Mo1, Al16Cr16Fe33Ni33Mo2, and Al16Cr16Fe33Ni33Mo3. We focused the research on samples with an ultrafine vermicular duplex microstructure, a unique structure requiring sufficiently high cooling rates to suppress the conventional Widmanstätten colony formation. The samples were produced by arc melting in buttons of 300 g each. We characterized the microstructure of the samples using scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDS), and electron backscatter diffraction (EBSD). The EBSD data revealed significant strain in the FCC phase resulting from the BCC→FCC phase transformation. We investigated mechanical properties of the samples by micro-indentation and 3-point bending in a miniature testing device. The test specimens were in the as-cast condition, as well as in distinct annealed conditions. Annealing treatments were carried out at 950 and 1100°C under argon. The annealing lasted from 10 min to 6 h, followed by water quenching. Prolonged annealing at 950°C of Mo-containing samples resulted in the formation of sigma-phase. Annealing at 1100°C safely avoided sigma-phase formation, while leading to a good balance between the flexural strength and ductility of these samples. Mechanical testing also included the well-established superduplex steel 1.4517 (DIN EN 10283/ASTM A890) as reference material.


INTRODUCTION
In the last two decades, research on high entropy alloys (HEAs) aimed at finding alloys that would allow obtaining single-phase equiatomic HEAs exhibiting face-centered-cubic (FCC), body-centered-cubic (BCC), and hexagonal-close-packed (HCP) structures while eliminating secondary (and tertiary) phases (Kozak et al., 2015;Steurer, 2020). However, in recent years a paradigm shift has taken place that puts the focus on dual-phase (or multi-phase) HEAs and intentionally takes advantage of their heterophase nature to achieve superior mechanical properties (Bönisch et al., 2018). Furthermore, increasing attention is given to medium entropy alloys with less than five elements, not at least with the aim of avoiding expensive and resourcecritical elements like Cobalt (Tkaczyk et al., 2018). This opens opportunities to develop novel materials for industrial applications, whenever clear development targets are set forward. The alloy system Al-(Co)-Cr-Fe-Ni is one interesting candidate for the development of dual phase materials, building on rich knowledge from the ternary subsystems Fe-Cr-Ni (Jönsson, 1995;Wróbel et al., 2015), Fe-Ni-Al (Eleno et al., 2006), etc. These subsystems host the well-known family of duplex steels (Alvarez-Armas and Degallaix-Moreuil, 2013), B2-ordered intermetallic alloys, and a series of eutectic alloys (Misra and Gibala, 1997). The alloy system Al-(Co)-Cr-Fe-Ni also offers means to exploit some features of spinodal decomposition (Langer, 1971) for microstructure tailoring. Novel dual phase materials have been reported in the system Al-(Co)-Cr-Fe-Ni (Dong et al., 2016;Abuzaid and Sehitoglu, 2018;Bönisch et al., 2018;Gwalani et al., 2019;Li Z. et al., 2020). Unique so far are vermicular duplex microstructures (Dong et al., 2016; composed of an ultrafine vermicular FCC phase intertwined with a spinodally decomposed BCC in nearly equal volume fraction. They were first reported in 2016 by Dong et al. (2016) in an AlCrFe 2 Ni 2 alloy, being termed "noodle-like" and suggested to be of eutectic origin. Similar microstructures are equally found in Co-containing alloys.
During the exploration of the Al-Cr-Fe-Ni and the Al-Co-Cr-Fe-Ni systems with respect to the addition of further alloying elements, molybdenum was investigated by Hsu et al. (2010), Zhu et al. (2010), and Dong et al. (2013), respectively. Dong et al. (2013) investigated a series of alloys AlCrFeNiMo x with the molar ratio x ranging from 0 to 1, explicitly x = 0.0, 0.2, 0.5, 0.8, 1.0. All alloys x < 0.5 displayed BCC-based microstructures. For Mo exceeding x > 0.5, the primary phase was identified as sigma phase. Compression tests for x = 0.2, corresponding to a Mo content of 4.8 at.-%, revealed fracture strength as high as 3.22 GPa and a plastic strain of 28.7%, however with a cleavage-like appearance of the fracture surface. To our best knowledge, molybdenum additions to the alloy AlCrFe 2 Ni 2 were not reported in literature, despite the increasing interested in alloys, which form dual-phase microstructures.
In this study, we selected the AlCrFe 2 Ni 2 alloy to revisit the microstructure and further develop the alloy by minor additions of molybdenum in the range from 1-3 at.-% to the baseline composition. The study encompasses four alloys, that is, Al 17 Cr 17 Fe 33 Ni 33 (Mo 0 ), Al 17 Cr 17 Fe 33 Ni 33 Mo 1 (Mo 1 ), Al 16 Cr 16 Fe 33 Ni 33 Mo 2 (Mo 2 ), and Al 16 Cr 16 Fe 33 Ni 33 Mo 3 (Mo 3 ) (at.-%). An industrial superduplex steel 1.4517 (DIN EN 10283/ASTM A890) is included as reference material. The research is embedded in a targeted European effort aiming at applications, which require a good balance between strength, ductility, and toughness along with wear and corrosion resistance in areas commonly served by duplex and superduplex steels. In this paper, we present and discuss the contribution of molybdenum to phase stability and describe the effects of annealing heat treatments on microstructure and selected mechanical properties. Main attention is given to alloys Mo 0 and Mo 2 .

EXPERIMENTAL PROCEDURES
We investigated samples with an ultrafine vermicular duplex microstructure in a series of alloys AlCrFe 2 Ni 2 Mo x (molar ratio) starting from the initial composition AlCrFe 2 Ni 2 . Molybdenum additions ranged from x = 0.06 to 0.18 molar ratio, corresponding to 1-3 at.-%. Samples were produced by arc melting (Edmund Bühler AM500) of large 300 g buttons from raw elemental metals with a purity above 99.8 wt.-%. The samples were molten under 0.7 bar argon (purity 5.0) with a Ti-getter after having evacuated the chamber to 10 −5 bar followed by repeated purging with argon to prevent oxygen pickup. The samples were then flipped and remolten three times to ensure chemical homogeneity. Table 1 shows the nominal and measured alloy compositions as determined by energy dispersive x-ray spectrometry (EDS).
The arc-melted buttons were cut by wire electrical discharge machining (Excetek V400G) to produce specimens for microstructure characterization and instrumented microindentation before and after defined annealing treatments. Furthermore, from the samples, Mo 0 and Mo 2 rectangular specimens with the dimensions 30 × 5 × 1.6 mm were extracted for 3-point-bending tests and subjected to annealing heat treatments at 950 • C for 6 h followed by water quenching. Moreover, specimens from Mo 2 were annealed at 1100 • C for varying holding times, that is, 10 min to 3 h followed by water quenching. Annealing was carried out in a horizontal tube furnace (Nabertherm RHTC 80-230/15) under argon flow.
Sample preparation for microstructure analysis consisted of standard metallographic grinding and polishing steps with a final polishing step in a vibromet device (Buehler VibroMet TM ). Samples were analyzed in scanning electron microscopes (Zeiss, Gemini Ultra 55 and Gemini Leo 1550) equipped with detectors for EDS analysis (Oxford Instruments X-max 150) and electron backscatter diffraction (EBSD) (Nordlys HKL). Micro-indentation measurements were performed in an Anton Paar MHT 3 using a Vickers indenter at a loading speed of 2000 mN/min and a maximum load of 500 mN. The 3-pointbending tests were carried out in a miniature testing device (MTI SEMtester 1000 EBSD) using a constant displacement of 0.2 mm/min (strainrate ∼0.3% s −1 ) until a maximum strain of 20% or sample failure was reached. For improved statistics three to five specimen were tested per alloy composition and annealing condition.

EXPERIMENTAL RESULTS AND DISCUSSION
In section "Microstructure Characterisation, " we describe the microstructure in the arc melted condition and after annealing heat treatments with focus on morphological and crystalline features of the dual phase structure, the phase fractions, and compositions. In section "Mechanical Properties, " we present EDS measurements were performed on five randomly placed areas on the sample and the values represent the mean and standard deviation of the five measurements. the mechanical properties from micro-indentation and 3-pointbending, respectively.

Microstructure Characterization
Initial Condition After Arc Melting Figure 1 -shows an overview of the microstructure in the initial "as-cast" condition of the Mo 2 sample at different magnifications, which allows identifying the multiscale structural features as follows: Figure 1A -At low magnifications, one distinguishes large dual-phase grains, which are former (parent) BCC-B2 grains. Figure 1B -Inside the grains, an ultrafine vermicular FCC phase is present (bright color), while the BCC matrix is spinodally decomposed at nanoscale. The decomposition products are BCC-B2#2 (dark) and BCC-A2# (bright Widmanstätten and vermicular FCC in this Frontier's edition, "The BCC-FCC phase transformation pathways and crystal orientation relationships in dual phase materials from Al-(Co)-Cr-Fe-Ni alloys." Molybdenum additions, even in small quantities, have a pronounced effect on the FCC phase fraction formed upon continuous cooling through the BCC→FCC phase transformation, but also on the morphological anisotropy. Figure 2 displays high magnification micrographs of the ascast microstructure in the four alloys Mo 0 through Mo 3 . The FCC phase fraction evaluated by image analysis ranges from 50% in Mo 0 , 60% Mo 1 and Mo 2 and finally 45% in Mo 3 . This is surprising, because Mo is a BCC stabilizing element. The increase of the FCC fraction from 50 to 60% seems counterintuitive but relates to the fact that Mo additions implicitly decrease the aluminum content, simply by dilution. This indicates that Mo is a less strong BCC stabilizing element compared with Al. Only above 2 at.-% Mo, for example, in Mo 3 , the BCC stabilizing effect of Mo compensates for the dilution of Al, thus leading to a decrease of the FCC phase fraction to 45%.
From Figure 2, it is perceivable that molybdenum additions lead to an increasing microstructural anisotropy, both regarding the directionality of the vermicular FCC structure but also regarding the nanoscale pattern inside the spinodally  decomposed BCC phase. The increased anisotropy is attributed to an increase of the BCC-A2 phase fraction inside the spinodal structure but likely also to an increase of the interfacial energy anisotropy. The simulation study by Vidyasagar et al. (2018) illustrates the importance of interfacial energy anisotropy during spinodal decomposition. Table 2 lists all measured quantitative data regarding the FCC phase fraction and the average thickness of the FCC vermicelli, also including the lattice strain observed by EBSD and analyzed by the MTEX software. Indeed, the EBSD analysis of the ultrafine vermicular samples in the as-cast as well as annealed conditions revealed that the FCC phase is significantly strained, containing a    high density of subgrain boundaries. The strain originates from the BCC→FCC phase transformation, which involves lattice rotation and dilatation relative to the parent BCC. Figure 3 shows the subgrain network for sample Mo 2 after annealing at 1100 • C/3 h based on the kernel average misorientation map (the average misorientation angle of a given point with all its first neighbors) plotted by MTEX. We believe that apart from classical transformation strain, the observed subgrain network relates to the growth conditions of the vermicular FCC from the initial BCC-B2 parent phase. Unlike Widmanstätten FCC plates, which grow into chemically homogeneous BCC-B2, the vermicular FCC was observed to grow into spinodally decomposing BCC-B2 (see Hecht et al. in this Frontier's edition). The compositional modulation in the parent grain is likely causing additional strain.

Annealed Condition After Isothermal Heat Treatments
Annealing heat treatments were carried out at 950 • C for 6 h mainly to verify the propensity of the Mo 0 and Mo 2 alloys to FIGURE 6 | Element distribution from EDS mapping in the sample Mo 2 for (A) as-cast condition and (B) after annealing at 1100 • C for 3 h. Noteworthy is the sluggish partitioning of molybdenum in the as-cast condition, which is recovered after heat treatment.
form the brittle sigma phase (tetragonal, space group no.136). EBSD analysis shows the absence of the sigma phase in Mo 0 , while in alloy Mo 2 about 17% sigma-phase is present with a composition around Al = 6 ± 2, Cr = 35 ± 2, Fe = 39 ± 1, Ni = 13 ± 2 and Mo = 7 ± 1 (at.-%). where former BCC-A2 was present ( Figure 4B). Sigma phase formation can be successfully prevented, even in sample Mo 2 , after annealing at 1100 • C up to 3 h followed by water quenching. Such heat treatment parameters are common practice for duplex steels, even for thick section castings, to avoid sigma phase formation (Fritz, 2014;Llorca-Isern et al., 2017). A small series of annealing treatments were performed for Mo 2 at 1100 • C with distinct holding durations of 10 min, 1, 2, and 3 h, followed by water quenching. The series aimed  at verifying the coarsening of the ultrafine vermicular FCC phase and of the spinodal structure. Figures 5A,B illustrates the effects of coarsening by comparing the as-cast ( Figure 5A) and annealed microstructures after 1100 • C/3 h ( Figure 5B). The comparison shows that the FCC phase thickens considerably, without changes in the phase fraction (see Table 2). The average thickness increases from 0.5 µm in the as-cast state to 1.3 µm in the annealed state. Coarsening is associated with an increasing contiguity of the FCC network. Figures 5C,D displays the size distribution of the FCC phase measured by the line intercept method along horizontal and vertical lines. The most dramatic coarsening is seen in the spinodal structure, which vanishes nearly completely, leaving behind few globulitic BCC-A2 particles. Without having analyzed the BCC-A2 fraction quantitatively, we qualitatively see a decrease of the BCC-A2 fraction, which is in agreement with the thermodynamics of the system. By heat treatment, the subgrain network inside the FCC phase is not changed significantly. We do not observe significant changes of the intergranular misorientation or of the low angle grain boundary distribution. Table 2 summaries the morphological and crystallographic features of the FCC phase for Mo 0 and Mo 2 for all heat treatment conditions. Due to coarsening, it was feasible to measure the chemical composition of the individual phases in Mo 0 and Mo 2 by EDS. Table 3 summarizes the results. As can be seen, molybdenum partitions mainly to the BCC-A2 phase, while being <0.5 at.% in the BCC-B2 phase. The FCC phase contains 2.8 at.% Mo. EDS maps provided in Figure 6 show the elemental partitioning between the phases for Mo 2 in the as-cast and annealed 1100 • C/3 h condition. An interesting and noteworthy observation is that molybdenum partitioning occurs only during heat treatment. The as-cast sample (Figure 6, left column) shows a sluggish Mo-partitioning, even though the other elements are clearly partitioned. This is likely due to kinetic effects, which are commonly termed as "solute trapping, " being associated to limited diffusion of Mo in BCC-B2. After heat treatment (Figure 6, right column), molybdenum partitioning is complete, and mainly to the BCC-A2 and the FCC-A1 phases. The Mocontent of the BCC-B2 phase is below 0.5 at.-%.

Mechanical Properties
The basic mechanical properties investigated on small samples in the as-cast and annealed states relate to results from micro-indentation and 3-point-bending, as presented below FIGURE 8 | (A) Flexural stress-strain curves in as-cast conditions: AlCrFe 2 Ni 2 (Mo 0 ) in red, AlCrFe 2 Ni 2 Mo 0.1 (Mo 2 ) in magenta, superduplex steel reference 1.4517 in blue, (B) after annealing at 950 • C/6 h: Mo 0 in red, Mo 2 in magenta, (C) Mo 2 after annealing at 1100 • C for 10 min (yellow), 1 h (green), 2 h (black), and 3 h (light blue). Illustration of the spread shown in (D) for Mo 2 1100 • C/1 h and the reference superduplex steel. Table 4 for the Mo 0 and Mo 2 in all tested conditions.

Micro-Indentation
Figures 7A,B provide the Vickers hardness (HV0.05) and the indentation modulus (E it ) as function of the molybdenum content for samples in the as-cast state along with the values of the FCC phase fraction. The Vickers hardness increases markedly with increasing Molybdenum content from 326 ± 6 HV0.05 in Mo 0 to 418 ± 7 HV0.05 in Mo 3 , independent of the BCC/FCC phase ratio. The origin of this increase may relate to increased density and bulk modulus of the alloy as was reported by Ledbetter et al. for the Fe-Cr-Ni system (Ledbetter and Austin, 1988;Ledbetter and Kim, 1988). However, we remind the microstructure observations (compare Figure 2), which showed that Mo increases the BCC-A2 phase fraction inside the spinodal BCC, leading to a plate-like morphology. At this stage, the quantitative analysis of the BCC-A2 fraction is still pending. As shown in Figure 7B, the indentation modulus increases only slightly from 190 to 196 GPa.
The evolution of hardness and indentation modulus for the annealed samples (Mo 0 and Mo 2 ) are listed in Table 4 along with results from 3-point-bending. As expected, upon annealing the hardness decreases with increasing temperature and/or annealing time. The lowest hardness of around 315 HV0.05 is reached in Mo 2 after annealing at 1100 • C for 3 h, which is still significantly higher than the hardness of the reference superduplex steel 1.4517 with 185 HV0.05.

Three-Point Bending Tests
Three-point-bending tests of miniature specimens 30 × 5 × 1.6 mm reveal the potential of the alloy Mo 2 for applications as structural material compared to the reference superduplex steel. Furthermore, the tests also shed light on means of balancing strength and ductility by annealing treatments. The main results are shown in Figure 8 as follows: Figure 8A compares the as-cast Mo 0 and Mo 2 samples to the as received and heat treated (1150 • C for 3 h) superduplex steel 1.4517. Both alloys show pronounced and similar strain hardening, with the Mo-containing Mo 2 being slightly stronger than the Mo-free Mo 0 . Both materials outpace the superduplex steel in terms of maximum flexural strength, however at the expense of ductility. The ductility is none the less considerably high, reaching about 15% flexural strain at fracture. Figure 8B compares the Mo 0 and Mo 2 in the heat-treated condition 950 • C/6 h and reveals the embrittlement by sigma phase in the Mo 2 sample. Figure 8C compares distinct annealing durations at 1100 • C for Mo 2 samples and shows that the yield strength decreases with increasing annealing durations, while the ductility increases significantly to flexural stains above 20%. The flexural UTS remains as high as 2000 MPa compared to 1600 MPa of the superduplex reference steel, while the yield stress falls down to the level of the reference steel. In all above figures, the plotted stress-strain curves are averaged over 3 to 5 tested specimens. Figure 8D illustrates the spread, taking as representative example all tested Mo 2 specimens annealed at 1100 • C/1 h as well as the reference steel. All specimens reach the flexural strain of 20% being the limit of the testing device.
Further, the fracture surface of one as-cast Mo 2 specimen, failed during bending, was inspected in SEM as well as in a metallographic section plane placed perpendicular to the fracture surface. We identified both, inter-and intragranular fracture paths, as shown in Figure 9. In Figure 9A, we outline the grain boundaries to highlight that a part of the fracture path follows a grain boundary. Grain boundaries are decorated by a thin layer of FCC phase. Figure 9B displays a SEM-SE view of the fractures specimen with labels indicating the tensile and compression side. Finally, Figure 9C shows a high magnification of the fracture surface including a region of intergranular fracture on the left side and a region with intragranular fracture on the right side.
Taken together, the bending results clearly prove a remarkable strain hardening capability of the investigated alloys, which is indeed one of the most prominent properties of HEAs as already discussed by Wang and Misra (2011) and Basu et al. (2018). Current work is directed toward better understanding the strain partitioning between BCC and FCC phases and identify the mechanisms behind strain hardening. Furthermore, the vermicular and highly contiguous microstructure is expected to show attractive fatigue properties, which are currently investigated. The hardness values, which outpass the superduplex steel by far, may improve the wear resistance, while calling for a careful investigation of the impact toughness.

SUMMARY AND CONCLUSION
We have investigated the effects of minor additions of molybdenum to the medium entropy alloy AlCrFe 2 Ni 2 regarding the microstructure and basic mechanical properties with focus on annealing heat treatments. The alloy develops an ultrafine vermicular microstructure composed of 60% FCC and 40% BCC, the BCC being spinodally decomposed into BCC-A2 + BCC-B2. The vermicular microstructure displays a high contiguity, the phases forming a neatly entangled network. The following main conclusions are drawn: • The ultrafine vermicular microstructure formed in the Mo-free baseline alloy AlCrFe 2 Ni 2 is retained for Mo additions up to 3 at.%; however the directionality of the FCC phase increases, as does the anisotropy of the spinodal pattern. The latter is attributed to an increasing fraction of the BCC-A2 phase but may also relate to stronger interfacial anisotropy between the product phases of the spinodal decomposition. • The FCC phase in the ultrafine vermicular microstructure displays significant strain and a network of low angle grain boundaries. The average intragranular misorientation does not change significantly during annealing treatments. • Alloy Mo 2 with 2 at.% Mo is prone to sigma formation and related embrittlement. Annealing heat treatments must exceed the stability limit of the sigma phase, that is, ∼1000 • C. At 950 • C the sigma fraction is as high as 17%.
Annealing at 1100 • C safely avoids sigma formation. • Alloy Mo 2 with 2 at.% Mo shows promising mechanical properties, which can be balanced by annealing at 1100 • C.
With increasing annealing duration, the ductility (flexural strain) increases at the expense of the yield strength. The ultimate flexural strength remains nearly unchanged at ∼2000 MPa, well exceeding the strength of the reference superduplex steel 1.4517. Like all investigated medium and HEAs, Mo 2 shows a remarkable strain hardening behavior, expected to prove beneficial to the wear resistance. • An annealing treatment at 1100 • C for the duration of 30 min to 1 h with subsequent water quenching is recommended.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.