Intermetallics 33 (2013) 81e86

Contents lists available at SciVerse ScienceDirect

Intermetallics
journal homepage: www.elsevier.com/locate/intermet

Signicant hardening due to the formation of a sigma phase matrix in a high

entropy alloy
Ming-Hung Tsai a, *, Hao Yuan a, Guangming Cheng a, Weizong Xu a, Weiwei W. Jian a, Ming-Hao Chuang b,
Chien-Chang Juan b, An-Chou Yeh b, Su-Jien Lin b, Yuntian Zhu a
a
Department of Materials Science and Engineering, North Carolina State University, Raleigh, NC 27695, USA
b
Department of Materials Science and Engineering, National Tsing Hua University, Hsinchu 30013, Taiwan

a r t i c l e i n f o a b s t r a c t

Article history: The hardening in Al0.3CrFe1.5MnNi0.5 high-entropy alloy not only nearly triples the hardness of the
Received 14 July 2012 alloy, but also shows a quick hardening response and the absence of overaging. However, the crystal
Received in revised form structure, morphology, and composition of the hardening phase have not yet been conrmed. Here,
23 September 2012
such information regarding the hardening phase is investigated. It was found that the hardening phase
Accepted 25 September 2012
Available online 7 November 2012
is a CreMneFe ternary sigma phase. Unlike in conventional engineering alloys, the sigma phase is not
precipitated from the matrix, instead, the whole BCC matrix transforms to sigma phase almost without
changing its composition. Therefore, the hardening phenomenon is not a precipitation hardening
Keywords:
B. Age-hardening
reaction as suggested before.
B. Precipitates 2012 Elsevier Ltd. All rights reserved.
F. Electron microscopy, transmission

1. Introduction behavior is very different from that of conventional precipitation

High-entropy alloy (HEA) e an alloy with ve or more principal
elements e is a new alloy design concept proposed recently [1].
Studies show that HEAs can have high strength/hardness [1e7],
outstanding wear resistance [8], exceptional high temperature
strength/hardness [2,9,10], and very good structural and chemical
stabilities at high temperatures [11]. Due to these properties, HEAs
have great potential in many applications such as tools, molds, ring
gears and other structural materials.
Most existing works on HEAs focus on the HEAs containing Al,
Co, Cr, Cu, Fe, Ni, Ti elements [2e4,8,12e18]. Recently, a new HEA
system with attractive properties, AlxCrFe1.5MnNi0.5 (x 0.3, 0.5),
has been proposed [5]. AlxCrFe1.5MnNi0.5 has hardness between HV
300e400 in the as-cast state. However, it can be signicantly
hardened when it is aged between 600 and 900  C. For example,
hardness of Al0.3CrFe1.5MnNi0.5 (henceforth referred to as the Al0.3
alloy) can be tripled from HV 299 to HV 899 when it is aged at
750  C [19]. Although precipitation hardening was believed
responsible for the signicant increase in hardness, the hardening
* Corresponding author. Department of Materials Science and Engineering, North
Carolina State University, Room 300, Research Building II, 1001 Capability Dr.,
Raleigh, NC 27695, USA. Tel.: 1 919 515 6172; fax: 1 919 515 3419.
E-mail address: mhtsai@mx.nthu.edu.tw (M.-H. Tsai).
hardening. For example, the hardening response is very quick and
there is no overaging. It is known that diffusion in HEAs is slower
than that in conventional alloys, which is call the sluggish diffu-
sion effect [20]. However, in the Al0.3 alloy full hardening is already
achieved with a short annealing time of 2 h in some cases [5].
Moreover, after achieving full hardening, the hardness remains
almost at the same level up to 100 h aging. For example, hardness of
the Al0.3 alloy aged at 700  C for 10 h and 100 h are HV 800 and HV
805, respectively [5]. The absence of overaging at such high
temperature is not consistent with conventional age hardening
process.
The hardening phenomenon in AlxCrFe1.5MnNi0.5 is not under-
stood. In particular, key information about the precipitates is lack-
ing: the precipitates have never been observed directly, thus, the
morphology and composition of the precipitates are totally
unknown. Moreover, there is still no agreement among published
papers on the crystal structure of the precipitates. For example,
Chen et al. reported, based on XRD analysis, that the hardening
phase is a Cr5Fe6Mn8 phase [5]. However, Tsao et al., also based on
XRD analysis, reported that the hardening phase belongs to s-CrFe
[19]. This is probably due to the fact that the alloy has several
phases, making the XRD analysis somewhat difcult.
In this study, detailed transmission electron microscopy (TEM)
analysis is performed on Al0.3CrFe1.5MnNi0.5 alloy before and after
the hardening process. The crystal structure, morphology, and

0966-9795/$ e see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.intermet.2012.09.022
82 M.-H. Tsai et al. / Intermetallics 33 (2013) 81e86

composition of the hardening phase are investigated, which

revealed the phase transformation that led to the hardening
phenomenon.

2. Experimental procedures

Al0.3CrFe1.5MnNi0.5 alloy was prepared by vacuum arc melting.

Raw materials with purities higher than 99.9% were melted in an Ar
atmosphere for at least 3 times to ensure all of the raw materials
were mixed well in their liquid state. The ingots were homogenized
at 1100  C for 4 h and then cooled in air. Some of the as-
homogenized samples were then hardened at 700  C for 2 h.
Samples were ground and polished for microstructure observation
in a Hitachi S3200 scanning electron microscope (SEM). Crystal
structures were identied using an X-ray diffractometer (Rigaku
SmartLab) in the qe2q conguration. A Cu Ka radiation operated at
40 kV, 44 mA was used, and the scanning speed was either 4 or
0.2 min1. Thin-foil specimens for TEM observation were prepared
by mechanical thinning followed by ion milling. The foils were then
observed on a 200 keV TEM (JOEL JEM-2010F) with energy
dispersive X-ray spectroscopy (EDS) capability. Macro-hardness
(the overall hardness in Table 1) and micro-hardness were
measured using hardness testers (Mitutoyo HV-115 and HM-115)
under loads of 5 kg and 25 g, respectively. Micro-hardness inden-
tation was conducted in the dendritic and the inter-dendritic
regions. Size of the indented area was 5e10 mm and that of the
characteristic features is 3e8 times larger.

3. Results and discussion

Fig. 1(a) shows the XRD patterns of as-homogenized and Fig. 1. (a) The XRD patterns of the as-homogenized alloy and the annealed alloy. (b) An
example section of the XRD pattern of the annealed alloy obtained using a slow scan
annealed alloys. As-homogenized alloy is composed of a major BCC
speed of 0.2 min1.
phase and a minor FCC phase, which is in agreement with previous
reports [5,19]. After annealing, the intensity of BCC peaks decreased
signicantly. Another set of peak, which corresponds to the age-
hardening phase, appears. The intensity of FCC peaks remains
roughly unchanged. This suggests that the age-hardening phase is
converted from the BCC phase. To obtain an XRD pattern of the
hardening phase that has high signal-to-noise ratio, a slow scan
speed of 0.2 min1 was used. Fig. 1(b) shows an example section of
such pattern. It can be seen that the signal-to-noise ratio is good
enough to clearly identify all the peaks. From the pattern, the
structure of the age-hardening phase is indexed to be same as s-
CrFe (JCPDS no. 005-0708), which means it has a tetragonal
structure (space group: P42/mnm). The lattice constants are:
a 8.8090  A and c 4.5519 A. The hardening phase will henceforth
be referred to as the sigma phase.
The microstructure of the annealed alloy retained the cast
dendriteeinterdendrite structure [Fig. 2(a)]. From previous litera-
ture [5], in the as-cast alloy the dendrite region has a BCC structure
and the interdendrite has an FCC structure. Since the XRD patterns
suggest that the sigma phase is converted from the BCC phase, the
sigma phase would be expected to reside in the dendrite region.
Indeed, microhardness measurements (Table 1) taken in the
dendrite and interdendrite regions reveal that the dendrite is
hardened from HV 386 to HV 1045, while the interdendrite was not

Table 1
Hardness of the dendrite and interdendrite regions in the Al0.3CrFe1.5MnNi0.5 alloy
before and after the hardening heat treatment.

Hardness (HV)

Overall Dendrite Interdendrite

As-homogenized 317  9 386  5 261  15
Fig. 2. (a) SEM images of the annealed alloy. (b) High-magnication image of the
Annealed (700  C, 2 h) 840  23 1045  30 258  29
dendrite region in (a).
 M.-H. Tsai et al. / Intermetallics 33 (2013) 81e86 83

Table 2 the interdendrite is rich in Ni. Al, Mn, and Fe do not show clear
Compositions (in at.%) of the dendrite and the interdendrite regions in the annealed preference between the two regions.
alloy as determined by SEM-EDS. The errors are the standard deviation.
To directly observe the morphology of the sigma phase, the
Al Cr Mn Fe Ni dendrite of the annealed alloy was observed with TEM (Fig. 3). The
Dendrite 6.8  0.6 23.2  0.5 23.8  0.2 36.1  0.6 10.2  0.8 dendrite contains numerous spherical particles whose sizes ranges
Interdendrite 7.1  0.0 15.7  0.0 26.3  0.2 35.4  0.3 15.5  0.2 between 150 and 300 nm. This concurs with our observation in
SEM. The particles do have an ordered BCC structure (inset of Fig. 3)
and their composition is rich in Ni, Al, and Mn. Since these
precipitates have a crystal structure identical to NiAl and the ratio
of Ni:(Al Mn) is also near 1:1 (Table 3), it is suggested that this
phase is a Ni(Al, Mn) phase. It is also seen in Fig. 3 that there are
some smaller (<50 nm) spherical precipitates located between the
large ones. These small precipitates have similar composition and
identical crystal structure to the large ones. Besides these spherical
precipitates, no other precipitate/2nd phase can be observed in the
dendrite. Therefore, the only candidate for the sigma phase seems
to be the matrix. This means that the whole matrix has transformed
into the sigma phase. Indeed, the selected area diffraction (SAD)
patterns taken from different major zone axes (Fig. 4) reveal that
the matrix has a tetragonal structure same as s-CrFe. Thus, it
appears that the formation of sigma phase is not through precipi-
tation from a supersaturated solution, which is totally different
from the previous expectations (i.e., sigma phase is in the form of
very ne precipitates [5,19]). Meanwhile, the composition of the
sigma phase is also determined (Table 3). The sigma phase is almost
a CreMneFe ternary phase with a composition ratio Cr:Mn:Fe close
to 3:2:4. Another interesting nding is that all the spherical
precipitates in the same grain have identical crystallographic
Fig. 3. TEM image of the dendrite region in the annealed alloy. Inset shows the SAD orientation.
pattern of the precipitates. To further understand how the sigma phase was formed, the
dendrite of the as-homogenized alloy was also investigated by TEM.
Fig. 5(a) shows the bright-eld (BF) image of the as-homogenized
hardened at all. Thus, the main contributor to the age hardening is alloy and the corresponding overall SAD pattern. The overall SAD
in the dendrite region and this region will be our main target of pattern shows only one set of ordered BCC structure (inset of
investigation. High magnication images of the dendrite region Fig. 5(a)). Dark-eld (DF) images corresponding to the BF image
[Fig. 2(b)] show that there are spherical particles distributed in the [Fig. 5(b)] were formed using the superlattice spots. It is seen that
dendrite. It was suggested in previous literatures that these parti- only the precipitates embedded in the matrix have the ordered
cles are (Ni, Al)-rich precipitates with an ordered BCC (B2) structure structure. Similar to the annealed alloy, there are large (150e
[5,19]. We will examine these precipitates by TEM later. Addition- 400 nm) and small (<50 nm) precipitates, and their structure and
ally, EDS results (Table 2) show that the dendrite is rich in Cr and orientation are the same. Therefore, the dendrite of the as-

Table 3
Compositions (in at.%) of the matrix and the precipitates in the dendrite regions of the as-homogenized and annealed alloys as determined by TEM-EDS. The errors are the
standard deviation.

Region Al Cr Mn Fe Ni
As-homo. Matrix 1.9  0.3 30.1  0.8 22.2  0.3 43.1  0.4 2.6  0.5
Precipitates 24.7  0.6 2.5  0.1 22.6  0.7 6.2  0.5 44.1  0.5
Annealed Matrix 1.7  1.2 32.1  1.6 21.3  1.0 41.4  2.2 3.5  1.6
Precipitates 27.5  1.3 2.3  0.4 18.7  0.8 5.4  0.5 46.0  1.1

Fig. 4. SAD patterns of the sigma phase (the matrix in the dendrite region) taken from different zone axes. (a) [001], (b) [100], (c) [110].
84 M.-H. Tsai et al. / Intermetallics 33 (2013) 81e86
Fig. 5. TEM images and SAD patterns of the as-homogenized alloy. (a) TEM bright-eld image. (b) TEM dark-eld image corresponding to (a). (c) SAD pattern of the matrix. (d) SAD
pattern of the precipitates. Inset of (a) shows the overall SAD pattern of (a).
homogenized alloy is composed of a matrix that has BCC structure
and some precipitates that have B2 structure. The matrix and the
precipitates have very similar lattice parameter and identical
crystallographic orientation, so that their SAD patterns overlap
with each other and look like one. This can be seen clearly if the
SAD pattern of the matrix and that of the precipitates were taken
separately and compared [Fig. 5(c) and (d)]. Such co-existence of
one BCC and one B2 phases with almost identical lattice constant
and orientation has been reported in AlCoCrCuFeNi alloy [12], but
the morphology of the phases in AlCoCrCuFeNi are very different
from the present case. EDS results (Table 3) show that the precip-
itates are rich in Ni, Al and Mn and the matrix is rich in Fe, Cr, and
Mn. The compositions of the two phases are very similar to their
counterparts in the annealed alloy. For example, the element
having the maximum concentration variation during annealing in
the matrix phase is Cr, and the change is 2 at.%. This indicates that
there is very little composition change associated with the hard-
ening phenomenon.
The microstructure and composition analysis of the alloys
before and after annealing reveals that the hardening phenomenon
is not due to precipitation hardening. Instead, it is due to the
transformation of the BCC matrix to the hard CreMneFe sigma
phase. This can be understood from the CreFeeMn ternary phase
diagram [21]. Although there is no phase diagram information for
1100  C, it can still be judged from the 1000 and 1200  C isotherms
that the equilibrium phase for the matrix composition at 1100  C is
BCC. Moreover, the equilibrium phase at 700  C (Fig. 6) is the sigma
phase. Therefore, the BCC phase which is stable at homogenization
temperature was preserved to room temperature in the air cooling
process. During the annealing at 700  C, it transformed to sigma
phase. The stability of BCC over sigma phase at high temperature is
because solid solution phases have higher entropy, which becomes
increasingly important as temperature increases. This has been
shown in a number of binary and ternary sigma-forming systems
[22e25], although similar thermodynamic modeling for the Cre
FeeMn system has not been conducted so far. Such high-
temperature stability of solution phases is particularly important
in high-entropy alloys, whose congurational entropy is signi-
cantly increased [1,26,27].
Sigma phase is an intermetallic compound that shows up in many
different binary and ternary alloy systems. It is probably the inter-
metallic phase having the broadest range of existence [28]. Sigma
phase lacks multiple slip systems, and thus is hard and brittle. It is
usually regarded as a detrimental phase for common engineering
alloys because it offers an ideal morphology for initiation and prop-
agation of fracture [29,30]. Interestingly, the formation of sigma
phase in Al0.3CrFe1.5MnNi0.5 is quite different from that in commer-
cial engineering alloys. In engineering alloys sigma phase is usually
precipitated from a matrix. In the present alloy, however, sigma
phase itself becomes the matrix. For high temperature applications
this can be a merit because sigma phase is more stable than FCC, BCC,
or HCP structures over a wide temperature range [28]. Therefore, the
alloy can have a very stable microstructure and thus a prolonged
service life at high temperature, and might be ideal for applications
with minimal requirement for ductility. The Al0.3CrFe1.5MnNi0.5 alloy
thus provides an entirely new frontier for future alloy design.
 M.-H. Tsai et al. / Intermetallics 33 (2013) 81e86 85

Fig. 6. 700  C isotherm of CreFeeMn system [21]. Composition of the CreFeeMn BCC/sigma phase (ignoring Al and Ni) is marked with a red square. (For interpretation of the
references to color in this gure legend, the reader is referred to the web version of this article.)

Our analysis also explains the following questions: 1. Why is 4. Conclusions

the hardening process so fast? 2. Why is there no overaging in
the hardening curve? 3. Why do all the precipitates in the
annealed alloy have the same crystallographic orientation?
Firstly, the hardening transformation is fast because there is very
little diffusion involved e compositions of the BCC phase and
sigma phase are quite similar. Therefore, the phase trans-
formation probably involves only local rearrangement of the
atoms. Secondly, the absence of overaging is simply because this
process is not age hardening. Thus, there is no precipitate
coarsening and associated softening. Thirdly, the orientations of
the precipitates are the same because the orientation was
inherited from the as-homogenized state. In the as-homogenized
alloy all the Ni(Al, Mn) precipitates have the same orientation e
identical to that of the matrix. During aging the matrix changes
its crystal structure. However, since the B2 precipitates do not
change their composition or crystal structure, it is reasonable
that they retain their orientation, too. The only noticeable
change in the precipitates is that they changed their shape
slightly.
The hardening phase transformation that occurs in the Al0.3Cr-
Fe1.5MnNi0.5 high-entropy alloy at 700  C was investigated. It was
found that the hardening phase locates in the dendrite region and is
a CreMneFe ternary sigma phase with a tetragonal structure. The
dendrite of the as-homogenized alloy is composed of a matrix that
has BCC structure and some Ni(Al, Mn) precipitates that have B2
structure. The matrix and the precipitates have very similar lattice
parameter and identical crystallographic orientation. During
annealing, the BCC matrix transforms into the sigma phase almost
without altering its composition. In contrast, for the Ni(Al, Mn)
precipitates, their composition and structure both remain
unchanged during annealing. Therefore, the hardening phase
transformation in the Al0.3CrFe1.5MnNi0.5 HEA is not the expected
precipitation hardening reaction, but a phase transformation from
BCC to tetragonal that does not involve much diffusion. The exis-
tence of sigma phase not as common precipitate form but as
a matrix may open a new design route for thermodynamically
stable high temperature alloys.
86 M.-H. Tsai et al. / Intermetallics 33 (2013) 81e86
Acknowledgments
M. H. Tsai gratefully thanks the National Science Council of
Taiwan for the nancial support from the Postdoctoral Research
Abroad Program. The authors appreciate the help in TEM analysis
from Dr. Amar S. Kumbhar of CHANL.
References
[1] Yeh JW, Chen SK, Lin SJ, Gan JY, Chin TS, Shun TT, et al. Adv Eng Mater 2004;6:
299e303.
[2] Tong CJ, Chen MR, Chen SK, Yeh JW, Shun TT, Lin SJ, et al. Metall Mater Trans A
2005;36A:1263e71.
[3] Zhou YJ, Zhang Y, Wang YL, Chen GL. Appl Phys Lett 2007;90:181904.
[4] Wen LH, Kou HC, Li JS, Chang H, Xue XY, Zhou L. Intermetallics 2009;17:266e9.
[5] Chen ST, Tang WY, Kuo YF, Chen SY, Tsau CH, Shun TT, et al. Mater Sci Eng A e
Struct Mater Prop Microstruct Process 2010;527:5818e25.
[6] Zhu JM, Fu HM, Zhang HF, Wang AM, Li H, Hu ZQ. Mater Sci Eng A e Struct
Mater Prop Microstruct Process 2010;527:7210e4.
[7] Zhu JM, Fu HM, Zhang HF, Wang AM, Li H, Hu ZQ. Mater Sci Eng A e Struct
Mater Prop Microstruct Process 2010;527:6975e9.
[8] Chuang MH, Tsai MH, Wang WR, Lin SJ, Yeh JW. Acta Mater 2011;59:6308e17.
[9] Hsu CY, Juan CC, Wang WR, Sheu TS, Yeh JW, Chen SK. Mater Sci Eng A e
Struct Mater Prop Microstruct Process 2011;528:3581e8.
[10] Senkov ON, Wilks GB, Scott JM, Miracle DB. Intermetallics 2011;19:698e706.
[11] Tsai MH, Wang CW, Tsai CW, Shen WJ, Yeh JW, Gan JY, et al. J Electrochem Soc
2011;158:H1161e5.
[12] Tong CJ, Chen YL, Chen SK, Yeh JW, Shun TT, Tsau CH, et al. Metall Mater Trans
A 2005;36A:881e93.
[13] Tsai CW, Chen YL, Tsai MH, Yeh JW, Shun TT, Chen SK. J Alloys Compd 2009;
486:427e35.
[14] Liu ZY, Guo S, Liu XJ, Ye JC, Yang Y, Wang XL, et al. Scr Mater 2011;64:868e71.
[15] Singh S, Wanderka N, Murty BS, Glatzel U, Banhart J. Acta Mater 2011;59:
182e90.
[16] Zhang KB, Fu ZY. Intermetallics 2012;22:24e32.
[17] Lin CM, Tsai HL. Intermetallics 2011;19:288e94.
[18] Wang WR, Wang WL, Wang SC, Tsai YC, Lai CH, Yeh JW. Intermetallics 2012;
26:44e51.
[19] Tsao LC, Chen CS, Chu CP. Mater Des 2012;36:854e8.
[20] Yeh JW. Ann Chim e Sci Mat 2006;31:633e48.
[21] Raynor GV, Rivlin VG. Phase equilibria in iron ternary alloys. London, UK: The
Institute of Metals; 1988. p. 288e99.
[22] Chvatalova K, Houserova J, Sob M, Vrestal J. J Alloys Compd 2004;378:
71e4.
[23] Havrankova J, Vrestal J, Wang LG, Sob M. Phys Rev B 2001;63.
[24] Lin JC, Chuang YY, Hsieh KC, Chang YA. Calphad e Computer Coupling of
Phase Diagrams and Thermochemistry 1987;11:73e81.
[25] Andersson JO, Sundman B. Calphad e Computer Coupling of Phase Diagrams
and Thermochemistry 1987;11:83e92.
[26] Ng C, Guo S, Luan J, Shi S, Liu CT. Intermetallics 2012;31:165e72.
[27] Yeh JW, Chen SK, Gan JY, Lin SJ, Chin TS, Shun TT, et al. Metall Mater Trans A
2004;35A:2533e6.
[28] Joubert JM. Prog Mater Sci 2008;53:528e83.
[29] Kington AV, Noble FW. Mater Sci Eng A e Struct Mater Prop Microstruct
Process 1991;138:259e66.
[30] Chastell DJ, Flewitt PEJ. Mater Sci Eng 1979;38:153e62.