Journal of Materials Science

, Volume 48, Issue 21, pp 7435–7445

Microstructure and mechanical properties of two-phase Fe30Ni20Mn20Al30. Part I: Microstructure

Authors

  • X. Wu
    • Thayer School of EngineeringDartmouth College
    • Thayer School of EngineeringDartmouth College
  • M. K. Miller
    • Materials Science and Technology DivisionOak Ridge National Laboratory
  • K. L. More
    • Materials Science and Technology DivisionOak Ridge National Laboratory
  • Z. Cai
    • X-Ray Science DivisionAdvanced Photon Source, Argonne National Laboratory
  • S. Chen
    • X-Ray Science DivisionAdvanced Photon Source, Argonne National Laboratory
Article

DOI: 10.1007/s10853-013-7558-4

Cite this article as:
Wu, X., Baker, I., Miller, M.K. et al. J Mater Sci (2013) 48: 7435. doi:10.1007/s10853-013-7558-4

Abstract

The microstructure of Fe30Ni20Mn20Al30 in both the as-cast condition and after annealing at 823 K for various times up to 72 h was characterized using transmission electron microscopy, scanning transmission electron microscopy, synchrotron-based X-ray diffraction, and atom probe tomography. The microstructure exhibited a basketweave morphology of (Mn, Fe)-rich B2-ordered (ordered b.c.c.) and (Ni, Al)-rich L21-ordered (Heusler type) phases with a lattice misfit of only 0.85 % and interfaces aligned along 〈100〉. The phase width increased from 5 nm for the as-cast alloy to 25 nm for 72 h annealed material, with no change in the elemental partitioning between the phases, with a time exponent for the coarsening kinetics of 0.19. Surprisingly, it was found that the room temperature hardness was largely independent of the phase width.

Introduction

Following the discovery of the high-yield strength (1570 MPa) and high hardness (Vickers hardness of 504 HV), nanostructured (50 nm phase width), two-phase B2 (ordered b.c.c.)/b.c.c. alloy Fe30Ni20Mn25Al25 (in at.%) [13], several high-strength b.c.c.-based two-phase alloys have been discovered in the Fe–Ni–Mn–Al system with various elemental contents ranging from 15 to 35 at.% [211], see Table 1. These alloys typically consist of either alternating b.c.c. and B2-ordered phases as noted above or alternating B2- and L21-ordered (a further ordering of the B2 structure often referred to as a Heusler phase) phases with ultrafine phase widths of 5–10 nm in the as-cast state, e.g., Fe35Ni15Mn25Al25 and Fe25Ni25Mn20Al30 [7, 8, 11], which are slightly harder (524–543 HV). All have interfaces aligned along 〈100〉. While the strengths of these alloys are comparable to those of both the strongest maraging aircraft steels and hardest bearing steels, they possess better strength-to-weight ratios, and the high aluminum content contributes to the oxidation resistance. However, at room temperature they are brittle, and after long-time aging at intermediate temperatures (~823 K) large (>1 μm long) β-Mn-structured precipitates nucleate and grow, resulting in a ~50 % increase in hardness, but even greater brittleness [1, 7].
Table 1

Summary of the phases present, phase width, and hardnesses of as-cast near-equiatomic Fe–Ni–Mn–Al alloys

Alloy (at.%)

Phases

Phase widths (nm)

Vickers hardness (HV)

Reference

Fe30Ni20Mn30Al20

f.c.c./B2

~65

454 ± 6

[10, 12]

Fe25Ni25Mn30Al20

f.c.c./B2

~85

413 ± 10

[10, 12]

Fe30Ni20Mn25Al25

b.c.c./B2

~50

504 ± 8

[1]

Fe35Ni15Mn25Al25

B2/L21

~8

543 ± 17

[7]

Fe25Ni25Mn20Al30

B2/L21

~10

524 ± 13

[11]

Fe30Ni20Mn20Al30

B2/L21

~5

514 ± 7

This paper

In addition to these b.c.c.-based, two-phase Fe–Ni–Mn–Al alloys, there are also two-phase Fe–Ni–Mn–Al alloys with microstructures with a somewhat similar appearance, although coarser (65–85 nm phase widths), but which consist of alternating f.c.c. and B2 phases, e.g., Fe30Ni20Mn30Al20 and Fe25Ni25Mn30Al20 [1012]. These alloys have lower hardness values of 413–454 HV, see Table 1.

This paper presents a detailed characterization of the microstructure of another two-phase B2/L21 Fe–Ni–Mn–Al alloy, Fe30Ni20Mn20Al30, in both the as-cast state and after a various anneals up to 72 h at 823 K. Fe30Ni20Mn20Al30 is of particular interest, since unlike the B2/b.c.c. and B2/L21 alloys noted above, it does not produce the β-Mn-structured precipitates upon annealing. These precipitates are both deleterious to the mechanical properties, and make understanding the effects of coarsening the matrix phases on the mechanical properties difficult to comprehend. In addition to characterizing the microstructure, the room temperature hardness was measured as a function of annealing time and the results are discussed in terms of various yield strength models.

Experimental

Materials

Ingots of nominal atomic composition Fe30Ni20Mn20Al30 were produced from pieces of 99.8 % Fe, 99.95 % Ni, 99.8 % Mn, and 99.8 % Al. An additional 5 wt% Mn was added to compensate for the loss of Mn by evaporation during melting. Once weighed, the pieces were placed in a water-chilled copper crucible and arc-melted under an argon atmosphere. The ingots, ~5 cm diameter, ~50 g buttons, were flipped and re-melted twice after the initial melting to ensure a homogeneous mixture. The chemical composition of the as-cast ingots, measured using a JEOL JXA-8500F electron probe microanalyzer, was found to be Fe29.6Ni20.4Mn18.9Al31.1 (the average of five measurements), which is close to the nominal composition. The grain size of the ingots was ~200 μm.

The as-cast alloy was annealed in air at 823 K for times ranging from 10 min to 72 h, followed by air-cooling. All times are accurate to ±1 min. This annealing temperature has been shown to be optimum for increasing both the hardness and yield strength of FeNiMnAl alloys [1].

Microstructural characterization

For transmission electron microscope (TEM) examination, 3-mm diameter disks were produced by electro-discharge machining, ground to ~200 μm thickness and twin-jet electropolished in an electrolyte of 20 % nitric acid, 10 % butoxyethanol, and 70 % methanol using a Struers Tenupol 5 at a voltage of ~10 V with a current of ~40 mA at ~260 K. After electropolishing, specimens were washed alternately in ethanol and methanol for three cycles followed by a final rinse in fresh methanol. The resulting thin foils were examined using a FEI Tecnai F20 field emission gun (FEG) TEM operated at 200 kV. Bright field (BF) TEM images were acquired under two-beam conditions with a deviation parameter, s, slightly greater than zero. Dark field (DF) images were also acquired using L21 superlattice reflections. The widths of the individual phases were measured when the microstructures were viewed exactly along 〈100〉. Since the measured widths of individual phases varied, an average width was obtained using a linear intercept measurement approach with five lines drawn perpendicular to the narrowest dimension of the phases with each line having at least ten intercepts.

A Philips CM200 FEG STEM operated at 200 kV equipped with a high-angle annular dark field (HAADF) detector and an energy dispersive X-ray spectrometer (EDS) was used to collect the elemental X-ray maps. The thin specimens for STEM were produced by focused ion beam (FIB) milling in an Hitachi NB5000 FIB-SEM instrument.

X-ray diffraction (XRD) measurements were performed at the Advanced Photon Source (APS) at Argonne National Laboratory, using the hard X-ray microdiffraction facility at Beamline 2-ID-D [13]. The X-ray radiation used in this study was generated from a 7 GeV electron beam and an APS undulator A [14] in the storage ring. X-rays with energy of 10.1 keV (wavelength = 0.1228 nm) were selected by a double-crystal Si 〈111〉 monochromator. Through zone-plate focusing optics, the X-ray beam was focused down to a circular spot of 200 nm and delivered to the sample with a flux ~3 × 109 photons s−1 [13]. The sample was mounted on and its angular position was manipulated by a six-circle kappa geometry diffractometer [15]. A Rayonix Mar165 CCD detector, with 2048 × 2048 pixels and 80 μm pixel size, was mounted about 23 mm downstream of the sample to collect the diffraction signals. During data collection, the sample was continuously rotated along an axis perpendicular to the incident beam by 80° in order to capture more diffraction spots. The total counting time was 21 s. Power law subtraction was used to strip the background.

For atom probe tomograph (APT) examinations, specimens were produced by a standard lift-out and annular milling procedure in the dual beam FIB–SEM system [16]. Further details of the specimen preparation and analysis technique can be found elsewhere [17, 47]. The resulting APT specimens were examined with a voltage-pulsed Cameca Instruments local electrode atom probe LEAP® 2017. All the runs were terminated by specimen fracture due to stresses imposed by the applied electrical field in the atom probe. Because of the brittle nature of the alloy, many runs were made but only a few provided usable data. The number of ions captured in the runs ranged from a low of 2,250,420 ions for the as-cast material to a high of 175,692,539 ions for the 12 h annealed specimen. The reconstructions shown have somewhat fewer ions than the total number of ions captured.

Results

Microstructure

TEM BF images, taken using the fundamental B2 (200)/L21 (400) diffraction vector, of as-cast Fe30Ni20Mn20Al30 show a very fine-scale microstructure with an average phase width of ~5 nm for each phase, see Fig. 1a. After annealing at 823 K for 30 min, the widths of the phases had coarsened to ~10 nm, see Fig. 1b, while the two phases had coarsened to ~25 nm after a 72 h anneal at 823 K and were clearly aligned along 〈100〉, see Fig. 1c. Table 2 lists the phase widths measured from BF TEM images of the as-cast Fe30Ni20Mn20Al30 and those after annealing for different times at 823 K. Note that because the phase width is less than the typical foil thickness, there will clearly be some error in the measured phase width—the quoted standard deviation is as large as 40 % of the measured width (for the as-cast specimen).
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-013-7558-4/MediaObjects/10853_2013_7558_Fig1_HTML.jpg
Fig. 1

BF TEM images, taken using the fundamental B2 (200)/L21 (400) diffraction vector shown, of Fe30Ni20Mn20Al30a as-cast, b after a 30 min anneal at 823 K, and c after a 72 h anneal at 823 K

Table 2

The composition of two phases (as determined using the APT), phase width (as determined from TEM images), Vickers hardness, and estimated yield stress of as-cast and annealed Fe30Ni20Mn20Al30

Condition

Composition (at.%)

Phase width (nm)

Hardness (HV)

Yield stress (GPa)

Phase

Fe

Ni

Mn

Al

As-cast

B2

55

10

13

22

5 ± 2

514 ± 7

1.68

L21

12

33

8

47

30 min

B2

53

5

18

24

10 ± 3

539 ± 5

1.76

L21

12

36

9

43

5 h

B2

16 ± 2

555 ± 7

1.82

L21

12 h

B2

54

9

11

26

14 ± 3

543 ± 6

1.78

L21

11

38

7

44

24 h

B2

21 ± 1

547 ± 7

1.79

L21

72 h

B2

47

7

19

27

25 ± 3

547 ± 6

1.79

L21

10

39

8

43

A [001] zone axis (ZA) selected area diffraction pattern (SADP) taken from the as-cast specimen that included both phases showed superimposed B2 and L21 reflections, see Fig. 2a. The [011] ZA SADP, see Fig. 2b, shows weak reflections, such as the L21 (\( 1\bar{1}1 \)) indicated, that are L21 superlattice reflections. All the strong reflections are both B2 and L21 reflections. Convergent beam electron diffraction (CBED) [011] ZA patterns from the individual coarsened phases after a 72 h anneal showed that the alloy consists of separate B2 (Fig. 3b) and L21 (Fig. 3c) phases in a cube-on-cube relationship, see Fig. 3a. The DF image taken using the (\( 3\bar{1}1 \)) L21 superlattice reflection at s ~ 0 shown in Fig. 3d clearly confirms the CBED information that the L21 reflections arise from a separate phase.
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-013-7558-4/MediaObjects/10853_2013_7558_Fig2_HTML.jpg
Fig. 2

(a) [001] and (b) [011] ZA SADPs for as-cast Fe30Ni20Mn20Al30. In (a), all the L21 reflections overlap the B2 reflections, whereas in (b) the weak reflections, such as the L21 (\( 1\bar{1}1 \)) indicated, are L21 superlattice reflections while all the other strong reflections are superimposed B2 and L21 reflections

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-013-7558-4/MediaObjects/10853_2013_7558_Fig3_HTML.jpg
Fig. 3

a BF TEM image of Fe30Ni20Mn20Al30 after a 72 h anneal at 823 K taken along the [011] ZA using the fundamental B2 (200)/L21 (400) diffraction vector; b CBED pattern from the light gray matrix showing a B2-ordered structure; c CBED pattern from the dark gray plates showing a L21-ordered structure; and d DF image taken using the (\( 3\bar{1}1 \)) L21 superlattice reflection shown at s ~ 0. Note that in (b), the diffraction spots are labeled with respect to B2 lattice, while in (c) the diffraction spots are labeled with respect to the L21 lattice. There is a cube-on-cube relationship between the phases

Table 3 summarizes the TEM–EDS data from five measurements acquired from each phase (identified by diffraction data) in the 72 h annealed TEM specimen. Both phases contained Fe, Ni, Mn, and Al, but the B2-ordered phase is rich in Fe and Mn, while the L21-ordered phase is rich in Ni and Mn. Note that the fineness of the phases in shorter anneals precluded obtaining accurate EDS data from those specimens.
Table 3

Composition (at.%) of the two phases in 72 h annealed Fe30Ni20Mn20Al30 as determined using EDS in the TEM. The results are an average of the three measurements

Phase

Fe

Ni

Mn

Al

B2-ordered

32 ± 1

23 ± 1

21 ± 2

24 ± 2

L21-ordered

15 ± 1

39 ± 2

12 ± 1

34 ± 1

Note that in the SADP patterns (Fig. 2), the B2 and L21 diffraction spots are roughly coincident, indicating that the L21 lattice parameter is close to twice that of the B2 lattice parameter. Using XRD at the APS, it was possible both to resolve separate peaks from the two phases at higher angles and also observe a weak reflection L21 (111) superlattice reflection at ~21o, see Fig. 4. From the B2 (211) reflection (d spacing = 0.118 nm), the B2 lattice parameter can be calculated to be 0.289 nm. This lattice parameter is similar to that of several other B2 aluminides [1822]. Using the L21 (422) reflection (d spacing = 0.119 nm), the L21 lattice parameter can be calculated to be 0.583 nm, i.e., roughly twice the B2 lattice parameter. These lattice parameter values yield a small lattice mismatch of only 0.85 %. While the elastic strains between the phases depends on their elastic properties, the small lattice mismatch and the fact that for most cubic metals, 〈100〉 is an elastically soft direction is presumably the reason for the alignment of the interphase interfaces along 〈100〉 since this alignment will minimize the elastic strains arising from the misfit.
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-013-7558-4/MediaObjects/10853_2013_7558_Fig4_HTML.gif
Fig. 4

Synchrotron XRD pattern of as-cast Fe30Ni20Mn20Al30 powder. Note the separate B2 and L21 reflections at ~62.5o. The weak reflection at ~21o is the L21 (111) superlattice reflection

Elemental partitioning between the two phases is evident in the STEM EDS elemental maps shown in Fig. 5 from a 72 h annealed specimen. It is clear from these that Ni co-partitions with Al and that Fe co-partitions with Mn.
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-013-7558-4/MediaObjects/10853_2013_7558_Fig5_HTML.jpg
Fig. 5

EDS elemental maps of Fe30Ni20Mn20Al30 after a 72 h anneal at 823 K. The bright regions on the maps indicate elemental enrichment

APT was used to examine further the elemental partitioning between the phases. Figure 6 shows the APT iso-concentration surfaces for 50 at.% (Mn + Fe) and 50 at.% (Ni + Al) for the as-cast Fe30Ni20Mn20Al30 alloy showing the complex morphology, and the strong partitioning of the alloying elements between the two phases in the as-cast alloy, i.e., one phase is enriched in Fe and Mn and the other is enriched in Ni and Al. Unlike the TEM observations, the alignment of the phases is only weakly evident in the APT reconstructions—one cannot view the phases along 〈100〉 as in TEM images. APT iso-concentration surfaces of Fe30Ni20Mn20Al30 after annealing for different times are shown in Fig. 7. The widths of the phases determined from the reconstructions, which are shown on the figure, are roughly consistent with those determined from TEM observations (see Table 1), although for the longest annealing times, it was not possible to get an average from APT data since the size of the specimen analyzed was only a little over three times the phase width.
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-013-7558-4/MediaObjects/10853_2013_7558_Fig6_HTML.jpg
Fig. 6

LEAP iso-concentration atom surfaces of as-cast Fe30Ni20Mn20Al30. In (a), the mauve phase contains >50 at.% (Fe + Mn); in (b), the blue phase contains >50 at.% (Ni + Al); c shows the two phases together (Color figure online)

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-013-7558-4/MediaObjects/10853_2013_7558_Fig7_HTML.jpg
Fig. 7

LEAP iso-concentration atom surfaces of Fe30Ni20Mn20Al30 annealed for 10 min, 30 min, 12 h, and 72 h at 823 K. The mauve phase contains >30 at.% Fe, while the blue phase contains >30 at.% Al (Color figure online)

The elemental compositions of the two phases in the as-cast Fe30Ni20Mn20Al30 and for the alloy after 30 min, 12 h, and 72 h anneals at 823 K were estimated from the APT data and are listed in Table 1. Note that the APT data must underrepresent the amount of Mn since the amount of Mn present in each phase is less than the composition for the alloy as a whole. The values also don’t accurately match the EDS data shown in Table 3, although the trends are the same. Thus, the APT data should be viewed as illustrating any changes in the compositions of the phases as a function of time rather than representing their absolute values. In fact, the data in Table 1 show that there is little difference in the compositions of the two phases at different annealing times.

Room temperature hardness measurements showed that Fe30Ni20Mn20Al30 had an as-cast Vickers hardness of 514 ± 7 HV, which increased slightly to 545 ± 10 HV (~6 % increase) after annealing at 823 K for 10 min, see Fig. 8. For longer annealing times up to 72 h, the hardness showed no significant changes and was essentially independent of the annealing time. Table 1 summarizes the phase width, chemistry, and the hardness of Fe30Ni20Mn20Al30 alloy after various anneals at 823 K.
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-013-7558-4/MediaObjects/10853_2013_7558_Fig8_HTML.gif
Fig. 8

Room temperature hardness (HV) as a function of annealing time (log scale) at 823 K for Fe30Ni20Mn20Al30

Discussion

Comparison with other FeNiMnAl alloys

The phase diagram for the quaternary Fe–Ni–Mn–Al system is not known and probably contains a variety of phase transformations and, hence, will exhibit a wide array of microstructures. Table 1 shows a comparison of Fe30Ni20Mn20Al30 and other near-equiatomic Fe–Ni–Mn–Al alloys that have been studied so far. The six alloys that have been investigated are listed in the order of decreasing Mn concentration and increasing Al concentration. Based on the crystal structure of the phases, these alloys can be divided into two groups: two alloys with higher Mn concentration (30 at.%) and lower Al concentration (20 at.%) showing f.c.c./B2 phases; and four alloys with lower Mn concentration (20 or 25 at.%) and higher Al concentration (25 or 30 at.%) showing b.c.c.-based ordered phases, either b.c.c/B2 or B2/L21. The phases observed are consistent with the fact that Mn stabilizes the f.c.c phase and Al stabilizes the b.c.c. phase [23]. Note that the effect of varying the Fe and Ni concentration on the phases present is not clear. Generally, the phase widths of the f.c.c./B2 alloys are larger than those of the b.c.c./B2 or B2/L21 alloys, and in turn, the phase widths of the b.c.c./B2 alloy are greater than those of the B2/L21 alloys. Surprisingly, there is no clear correlation between phase width and hardness. However, as noted in the Introduction, the presence of an f.c.c. phase appears to produce a hardness that is about ~100 HV lower than that of the alloys containing only b.c.c-based phases.

Regarding the composition of the phases, the APT (Table 1) and EDS results (Table 3; Fig. 5) show that the L21 phase is enriched in Ni and Al, while the B2 phase is enriched in Fe and Mn. While the locations of the four elements on the sublattices of the L21 and B2 crystal structures are unknown, it is worth noting that in the related alloy Fe35Ni15Mn25Al25 APT and EDS data similarly showed that the L21 phase was enriched in Ni and Al, while the B2 phase was enriched in Fe and Mn [47].

Phase coarsening

Upon annealing, phase coarsening occurred, as shown by both TEM and APT observations. Unlike B2/L21 NiAl–Ni2AlTi alloys, which have a similar structure after casting, [2428], there did not appear to be a loss of coherency upon annealing and there was no evidence of the formation of misfit dislocations at the interphase interfaces. The latter features are presumably because the lattice misfit is low, i.e., <1 %.

There are several well-established diffusion mechanisms that can explain isothermal coarsening kinetics. In addition to lattice diffusion, dislocation or pipe diffusion, which has a much smaller activation energy compared with lattice diffusion and becomes significant at low temperatures, [29] can occur. However, very few dislocations were observed in TEM images of Fe30Ni20Mn20Al30 Thus, this mechanism is unlikely. The Lifshitz, Slyozov [30] and Wagner [31] or LSW theory predicts that the average particle radius, r, grows with annealing time, t, according to the equation \( (r^{3} - r_{0}^{3} )^{1/3} = kt^{1/3} \), where in the present case, \( r_{0} \) is the as-cast phase size, and k is a coarsening constant that depends on a chemical diffusion coefficient, interfacial free energy, and other parameters. Nevertheless, there are reservations in adopting this theory. First, strictly speaking, the LSW theory applies specifically to spherical precipitates in a supersaturated solution. Second, the ideal limit of zero volume fraction of the precipitates can never be approached in the as-cast and annealed two-phase Fe30Ni20Mn20Al30 alloys. Another model, based on trans-interface diffusion-controlled coarsening, developed by Ardell and Ozolins [32], is based on the coarsening of Ni–Al alloys consisting of ordered Ni3Al precipitates with coherent interfaces in a disordered Ni–Al matrix. This theory predicts that the average radius grows linearly with the square root of the annealing time, i.e., \( (r^{3} - r_{0}^{3} )^{1/3} = kt^{1/2} \).

Figure 9 is a plot of \( (r^{3} - r_{0}^{3} )^{1/3} \) versus annealing time on a log–log scale. The coarsening exponent was found to be 0.19, i.e., less than either of the above two models predicted. This may be because of the small lattice mismatch (0.85 %) between the phases. For Fe30Ni20Mn20Al3, even after a 72 h anneal at 823 K the B2/L21 interfaces remain coherent, no interfacial dislocations were observed and there was no significant change in the chemistry of the phases as their size increased. Observations on B2/L21 two-phase alloys in some other alloy systems showed both that interfacial coherency was lost after long anneals and that the lattice misfit was accommodated by the generation of a 〈100〉 interfacial dislocations [25, 26, 33, 34]. The degree of mismatch between the two phases in these studies, as determined by the interfacial dislocation spacing, was 1–2.5 % [25, 32]. The origin of the low coarsening exponent remains unclear.
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-013-7558-4/MediaObjects/10853_2013_7558_Fig9_HTML.gif
Fig. 9

Plot of (r3 – r03)1/3 versus annealing time on a log–log scale, showing that the coarsening exponent is ~0.19

Mechanical properties

An unusual feature of the room temperature mechanical properties of Fe30Ni20Mn20Al30 is that the hardness is largely independent of phase width over the range 5–25 nm. Of course, the Vickers hardness is a complex parameter that is approximately equal to the flow stress at 8 % strain, i.e., it is equal to the yield strength plus the increase in strength due to work hardening after 8 % strain. However, for an ideal elastic–plastic metal, which is often a reasonable model for alloys with very high yield strength [35], the yield strength can be reasonably taken to be the hardness/3 [36].

Different deformation mechanisms produce different dependencies of the yield strength on particle size (radius), r, or phase width. At small particle sizes, particle cutting normally occurs producing a yield strength dependency of r0.5, while particle looping, which typically occurs for larger particles, produces a yield strength dependency of 1/r. Obviously, neither applies here.

The microstructures observed here have a similar appearance to some formed by spinodal decomposition [37]. Several models have been developed to describe the relationship between the compositional amplitude, A, defined as the chemical difference between the two phases, and the wavelength, λ, defined as twice the phase width, and the yield strength for spinodal-type microstructures. It is worth noting that these models are focused on the early stages of spinodal coarsening where the compositions of the phases change as a function of time and the compositional profile changes from a sinusoidal variation with distance to a more rectangular profile [3843]. These models, which are based on different dislocation interactions with the microstructure, yield a variety of dependencies of the strength on λ, i.e., λ [38], 1/λ [39, 40], λ2/3 [41], or \( \lambda^{ - 2/3} \) [38]. Again none of these models can apply to the alloy studied here.

Interestingly, models developed by Dahlgren [42] and Kato et al. [43], which consider the internal coherency stress to be the main factor controlling the strength, find that the yield strength is independent of λ and only proportional to A. As evident in Table 1, upon annealing, the phases in Fe30Ni20Mn20Al30 showed no significant change in composition, while the phase width coarsened continuously. After an initial increase for short annealing times, the hardness also shows no significant change. The reason for the initial ~6 % increase in hardness is unclear, but could result from an increase in the sharpness of the interphase interfaces. The strength of Fe30Ni20Mn20Al30 seems to be independent of the phase width and may depend on the phase amplitude. Some models that predict a dependence of the yield strength on λ [3841] consider modulus variation arising from compositional fluctuations. However, for polycrystalline B2 compounds, the elastic modulus is not strongly dependent on the composition [4446]. Thus, one might expect that the modulus variation in Fe30Ni20Mn20Al30 to be quite small. So the strengthening would only depend on the internal coherency stress, which depends on the phase amplitude, as predicted in the models developed by Dahlgren [42] and Kato et al. [43]. Thus, no changes in strength would be expected in Fe30Ni20Mn20Al30 upon annealing since the phase compositions do not change. While the lack of change in hardness as the phase width changes in Fe30Ni20Mn20Al30 remains unexplained, similar behavior has also been observed in the related B2/L21 alloy Fe25Ni25Mn20Al30.

Conclusions

An investigation conducted on the microstructure and room temperature hardness of Fe30Ni20Mn20Al30 showed that
  1. (1)

    The microstructure of as-cast Fe30Ni20Mn20Al30 consists of (Fe, Mn)-rich B2 and (Ni, Al)-rich L21 phases roughly aligned along the 〈100〉 direction with a lattice mismatch of 0.85 %.

     
  2. (2)

    Upon annealing at 823 K, the phase width increased from ~5 nm in the as-cast state to ~25 nm after a 72 h anneal, while no significant change was observed in the composition of the phases.

     
  3. (3)

    A plot of \( (r^{3} - r_{0}^{3} )^{1/3} \)versus t, where \( r_{0} \) is the as-cast particle size and r is the particle size at time t, revealed an unusual coarsening exponent of 0.19.

     
  4. (4)

    The hardness of the as-cast alloy was 514 ± 7 HV. A ~6 % increase to 545 ± 10 HV was observed after annealing at 823 K for 10 min. However, the hardness showed no significant change upon subsequent annealing for times up to 72 h, i.e., the strength is independent of the phase width.

     

Highlights

  • The microstructure of two-phase Fe30Ni20Mn20Al30 has been characterized using TEM, STEM, APT and XRD.

  • The as-cast alloy exhibits a basketweave morphology consisting of B2/L21 phase aligned along <100>.

  • Annealing at 823 K for 72 h increased the phase width from 5 nm in the as-cast alloy to 25 nm in the annealed material.

  • The exponent related to the coarsening kinetics was determined to be ~0.2.

  • The coarsening of the microstructure produced little change in room temperature hardness.

Acknowledgements

This research was supported by the US Department of Energy (DOE), Office of Basic Energy Sciences grant DE-FG02-07ER46392 (X. W and I. B). Research was supported ORNL’s ShaRE User Facility, which is sponsored by the Scientific User Facilities Division, Office of Basic Energy Sciences, U.S. Department of Energy (M. K. M. and K. L. M.). Use of the Advanced Photon Source, an Office of Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science by Argonne National Laboratory, was supported by the U.S. DOE under Contract No. DE-AC02-06CH11357 (Z. C. and S. C.). The authors gratefully acknowledge K. F. Russell for technical assistance. Prof. Paul Munroe of the University of New South Wales is thanked for the electron probe microanalyzer measurements. We would also like to thanks the reviewers for their numerous useful comments. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing official policies, either expressed or implied of the DOE or the U.S. Government.

Copyright information

© Springer Science+Business Media New York 2013