Content uploaded by Yakai Zhao
Author content
All content in this area was uploaded by Yakai Zhao on Oct 06, 2017
Content may be subject to copyright.
Regular article
Resistance of CoCrFeMnNi high-entropy alloy to gaseous
hydrogen embrittlement
Yakai Zhao
a
, Dong-Hyun Lee
a
,Moo-YoungSeok
a
,Jung-ALee
a
, M.P. Phaniraj
b
, Jin-Yoo Suh
b,
⁎,
Heon-Young Ha
c
, Ju-Young Kim
d
, Upadrasta Ramamurty
e
, Jae-il Jang
a,
⁎
a
Division of Materials Science and Engineering, Hanyang University, Seoul 04763, Republic of Korea
b
High Temperature Energy Materials Research Center, Korea Institute of Science and Technology, Seoul 02792, Republic of Korea
c
Ferrous Alloys Group, Korea Institute of Materials Science, Changwon 51508, Republic of Korea
d
School of Materials Science and Engineering, Ulsan National Institute of Science and Technology, Ulsan 44919, Republic of Korea
e
Department of Materials Engineering, Indian Institute of Science, Bangalore 560012, India
abstractarticle info
Article history:
Received 7 February 2017
Received in revised form 12 March 2017
Accepted 24 March 2017
Available online xxxx
The influence of hydrogen onthe mechanical behavior of the CoCrFeMnNi high-entropy alloy (HEA) was exam-
ined through tensile and nanoindentation experiments on specimens hydrogenated via gaseous and electro-
chemical methods. Results show that the HEA's resistance to gaseous hydrogen embrittlement is better than
that of two representative austenitic stainless steels, in spite of the fact that it absorbs a larger amount of hydro-
gen than the two steels. Reasons for this were discussed in terms of hydrogen-enhanced localized plasticity
mechanism and the critical amount of hydrogen required for it. These were further substantiated by additional
experiments on electrochemically charged specimens.
© 2017 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords:
High-entropy alloy
Hydrogen embrittlement
Tensile test
Nanoindentation
Thermal desorption spectroscopy
High-entropy alloys (HEAs) that comprise of three or more (nearly-)
equiatomic metallic elements have attracted significant research atten-
tion since they were first introduced about a decade back [1,2] as they
show exceptional mechanical properties, especially high strength-duc-
tility combinations at both elevated and cryogenic temperatures [3–9].
However, HEAs' susceptibility to hydrogen embrittlement (HE) [10–
14], if any, was not examined in detail. Future structural components
made of HEAs are likely to be exposed to hydrogen (H) in the working
environment of a number of potential applications. Some such exam-
ples are high-content H in the water reactor environment of nuclear
power plants [15],H
2
gas in moist environments like ambient air in
aerospace structures [16],andHstoredeitherasliquidincryogenicves-
sels for space shuttles [17,18] or as gas in high-pressure vessels for H
transportation/storage [17]. Keeping this in view, it is essential to
study the effect of H on the mechanical behavior of HEAs, which is the
purpose of the work reported in this paper.
For this, we have selected CoCrFeMnNi HEA, which is one of the
most widely investigated HEAs to date, as it crystallizes into a remark-
ably stable face-centered cubic (fcc) single phase [1,19]. It exhibits ex-
ceptional ductility and fracture toughness at cryogenic temperatures
[6,20] and good resistance to creep [7]. HE of this alloy upon gaseous
H charging was examined, since the sources of H that cause embrittle-
ment are generally gaseous in nature, and compared with that of two
austenitic stainless steels (SSs), 304 and 316L. Like HEA, these SSs also
have fcc crystal structure. Moreover, the main constituent elements in
all the three alloys are the same; they only differ in relative content in
each alloy [5]. Importantly, since the effect of H on SSs has been exten-
sively studied (e.g. Refs. [10,22–25].), well-documented data is readily
available for comparison.
The CoCrFeMnNi HEA samples were prepared by vacuum induction
casting of nominal mixtures of the corresponding metals with purity
higher than 99 wt%. The cast ingot was hot-forged and then solution
annealed at 1100 °C for 1 h to reach an equilibrium microstructural
state. The annealed sample has a single fcc phase. The backscattered
electron (BSE) image obtained by scanning electron microscopy (SEM;
Nova NanoSEM 450, FEI Inc., Hillsboro, OR, USA) displayed in Fig. 1
shows homogeneous equiaxed grains with an average grain size of
~34 μm. For comparison purpose, solution-annealed 304 and 316L
steels were also prepared. The nominal compositions of HEA and the
two SSs are given in Table 1. All the HEA and SS specimens tested in
this study (irrespective of H charging) were always ground initially
with fine SiC papers (grit number up to 2000) and then with colloidal
silica (0.05 μm) to a mirror finish, resulting in a final thickness of
~300 μm.
Scripta Materialia 135 (2017) 54–58
⁎Corresponding authors.
E-mail addresses: jinyoo@kist.re.kr (J.-Y. Suh), jijang@hanyang.ac.kr (J. Jang).
http://dx.doi.org/10.1016/j.scriptamat.2017.03.029
1359-6462/© 2017 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Contents lists available at ScienceDirect
Scripta Materialia
journal homepage: www.elsevier.com/locate/scriptamat
Gaseous H charging (referred to as G-charging hereafter) of the HEA
and two SS specimens wasperformed in a custom-made Sieverts appa-
ratus at 300 °C under constant pressure of 15 MPa of gaseous H
2
for 72 h.
In addition, electrochemical H charging (E-charging, hereafter) of the
HEA was performed at RT using a potentiostat/galvanostat equipment
(HA-151A, Hokuto Denko, Tokyo, Japan) in 1 N H
2
SO
4
solution for
24 h under a constant current density of 100 mA/cm
2
. To minimize
the influence of outgassing, all the charged samples were immediately
immersed always into liquid nitrogen and stored until further experi-
ments, which were completed, in any case, within ~24 h after charging.
Flat dog-bone-shaped tensile samples with gauge length of 10 mm
and gauge width of 1 mm were wire cut from the annealed bulk speci-
mens. Tensile tests were carried out on these uncharged and G-charged
subsize samples using a micro-tensile tester (MTest 300, Gatan Inc.,
Pleasanton, CA, USA) at a cross-head speed of 0.1 mm/min. Nanoinden-
tation experiments were conducted using Nanoindenter-XP (formerly
MTS; now Keysight Technologies, Santa Rosa, CA, USA) with two differ-
ent three-sided pyramidal indenters (typical Berkovich and sharper
cube-corner tips) at a peak load, P
max
, of 100 mN under constant inden-
tation strain rate _
ε=(dh/dt)/h= 0.025 s
−1
where his indentation
depth and tis the time. Fractography on the tensile tested samples
and imaging of the indentation impressions was performed by SEM.
The indentation impressions were additionally profiled with an atomic
force spectroscope (AFM; XE-100, Park Systems, Suwon, Korea). For the
quantitative analysis of the H content in the charged samples, thermal
desorption spectroscopy (TDS) was performed with a quadrupole
mass spectroscope (EX0014, R-DEC Company, Tsukuba, Japan) at acon-
stant heating rate of 5 °C/min to a maximum temperature of ~800 °C.
Representative engineering stress vs. engineering strain responses
of the G-charged HEA, 304 and 316L SSs, as well as their uncharged
counterparts are displayed in Fig. 2. As expected, the ductility of both
the SSs gets reduced markedly upon charging. Such embrittlement
was absent in HEA, with the relative loss in ductility due to H charging
being merely ~ 5%. For comparison, it is ~ 61% and ~ 27% for 304 and
316L SSs respectively. It appears that G-charging also does not affect
the yield strength, flow stresses, or strain hardening rate of HEA in
any significant manner. Fig. 2b and c show both low- and high-magnifi-
cation fractographs of the tensile tested specimens of HEA in uncharged
and charged conditions, respectively. Both the fracture surface mor-
phologies are similar, and consist of dimples, indicating ductile fracture.
This observation further supports HEA's resistance to H-induced
embrittlement.
‘Is the superior embrittlementresistance of HEA due, possibly, to the
H content in it being lower than those of the two steels?’To answer this
question, the quantity of H in the G-charged HEA, 304, and 316L sam-
ples was measured by TDS, whose results are displayed in Fig. 3. In all
the three cases, the peak desorption occurs at similar temperatures.
This, in turn, indicates that the sites within the fcc lattice, which H oc-
cupies, may be similar in all the three alloys. They correspond to inter-
stitial lattice sites and weak H trapping defects such as dislocations
[26,27]. Importantly, Fig. 3 shows that the desorbed H content from
HEA (~76.5 ppm) is higher than that from either of the SSs. It also is
higher than most literature values of SSs exposed to higher H
2
pressure
than that in the present study [28–30]. Since no new phase was found
through X-ray diffraction analysis after H charging of the HEA [21],
such high H solubility may be attributed to the relatively high content
of Cr and Mn in HEA compared to in other austenitic SSs, since the
two elements were found to be crucial in influencing H solubility [22,
31]. In addition, the lattice distortion in HEA can also be one of the con-
tributing factors for the enhanced solubility as demonstrated for the
solid state H storage application of HEA [32]. On this basis, we conclude
that the absence of HE in HEA is not due to lack of H and seek an alter-
native explanation for it.
Two widely-accepted micromechanismsfor HE in SSs are the follow-
ing: (a) stress-induced austenite-to-martensite transformation [24,28]
and (b) H-enhanced localized plasticity (HELP) [11,33,34].Since
stress-induced transformation was not observed to occur in the
CoCrFeMnNi HEA [6,19,20], the former mechanism can be excluded. In
HELP, the dissolved H enhances local dislocation mobility and slip pla-
narity [25,33], which in turn promotes cleavage fracture. Therefore, HE
through HELP is more pronounced in materials that are already prone
to planar slip [33,34]. Systematic transmission electron microscopy by
George and coworkers [20,35] do indeed show that the CoCrFeMnNi
HEA deforms primarily through planar slip.Therefore, it is natural to an-
ticipate that the HELP mechanism will be of significance in this HEA. But
planar slip is only a necessary condition for HELP. For it to occur, suffi-
ciently high and local H content is also essential [36,37].Thisisespecial-
ly important when H diffusivity is sluggish, which is the case in for fcc
metals and alloys [26]. In contrast, and for example, fast diffusion of H
to highly-stressed regions in body-centered cubic (bcc) metals am-
plifies the effect of H on the plastic deformation in them. That high
local H content is necessary for HELP in fcc metals is further supported
by the observation that SSs such as 310 and some 316 steels with high
Ni equivalent, which also do not undergo strain-induced martensitic
transformation, are not prone to HE in H
2
gas atmosphere [22,29,34,
38], but are usually susceptible to HE upon E-charging [23,36,37].This
is because cross-slip dominates plasticity in such SSs and thus HE can
occur if and only if the H content in the SS upon charging is extremely
high, which is possible in E-charging [30,39].
From the above discussion, it is reasonable to conclude that the G-
charged CoCrFeMnNi HEA did not get embrittled possibly because the
local H content upon charging is below the threshold level required
for triggering HELP. To elaborate this further through experiments, the
HEA was exposed to a much harsher H environment through the E-
charging process. Subsequently, nanoindentation was utilized to evalu-
ate and compare the mechanical responses of uncharged, G-charged,
and E-charged samples. Representative nanoindentation load–displace-
ment (P-h) curves, obtained by using both Berkovich and cube-corner
indenters, are provided in Fig. 4a. The hardness values, which were
Fig. 1. SEM-BSE image showing the equiaxed grain structure with grain size ~34 μmof
solution-annealed CoCrFeMnNi HEA.
Table 1
Nominal chemical compositions (wt%) of CoCrFeMnNi HEA (transferred from nominal
20 at.% for each element), 304, and 316 SSs.
Co Cr Fe Mn Ni C Si Mo V N Cu
HEA 21.0 18.6 19.9 19.6 20.9
304 0.01 17.74 Bal. 1.05 7.83 0.05 0.56 0.20 0.01 0.006
316L 0.19 16.50 Bal. 1.29 9.77 0.02 0.50 2.07 0.10 0.02 0.27
55Y. Zhao et al. / Scripta Materialia 135 (2017) 54–58
estimated from them by the Oliver-Pharr method [40], are summarized
in Table 2. In agreement with the tensile test results, no significant
change in the hardness is noted upon G-charging. In contrast, substan-
tial enhancement in the hardness (by ~63% in Berkovich and ~37% in
cube corner) occurs upon E-charging.
Observations of the surface morphology around the cube-corner
nanoindentations, displayed in Fig. 4b, provide further support to the
H effects on the indentation-induced plasticity. Cube-corner indenter,
being sharper, induces higher stresses and strains underneath the in-
denter. This, in turn, leads to pronounced impression morphologies
[41,42]. The morphologies around the indents in the uncharged and
G-charged specimens appear similar, with a large number of slip-traces.
These not only confirm the planar nature of slip in the HEA examined,
but also affirm the earlier observation that G-charging does not affect
the plasticity in the HEA in any significantmanner. In contrast, the num-
ber of slip steps aroundthe indents made on the E-charged sample is far
smaller, confirming a significantly reduced plasticity upon E-charging. A
comparison of the AFM profiles, displayed in Fig. 4b, indicates that both
the height and spacing of the slip steps around nanoindentations in the
E-charged sample are larger than those in the uncharged one. This ob-
servation further supports the notion of reduced cross-slip and thus
more enhanced slip planarity in the former [25]. These observations
provide evidence for the reduced plastic deformation susceptibility
and enhanced slip planarity by E-charging vis-à-vis G-charging.
At this juncture, it is interestingto note that the overall H content in
the E-charged CoCrFeMnNi HEA specimen, which we have reported in a
Fig. 2. Results from tensile tests. (a) Engineering stress–engineering strain curves (with the numbers showing ductility loss, el
loss
, due to G-charging). The low- and high-magnification
SEM images shows the fracture surfaces of (b) uncharged and (c) G-charged HEA.
56 Y. Zhao et al. / Scripta Materialia 135 (2017) 54–58
recent study [21], is ~45.0 wppm, i.e., less than that in the G-charged
specimen of the present study. Then, the homogeneity of H concentra-
tions in both cases comes into focus. In the G-charged sample, it is rea-
sonable to expect a homogeneous distribution through the thickness of
the sample since the time (72 h) - temperature (300 °C) combination of
chargingis sufficient for H diffusion throughout the specimen [37,43].In
contrast, E-charging was carried out at room temperature (RT), where
the hydrogen diffusivity is much lower, and for a shorter duration
(24 h). Note that the temperature dependence of hydrogen diffusivity
is significant for fcc metals due to the large activation barrier energy
for diffusion. Therefore, a steep gradient in the H concentration in the
E-charged sample is possible. It, as a function of the distance from the
surface, x, can be estimated by [43]:
Cx;tc
ðÞ¼C01−erf x
2ffiffiffiffiffiffiffiffiffiffi
DHtc
p
;ð1Þ
where t
c
is the charging time, and D
H
is H diffusivity in HEA. Given the
similarity in the composition and the crystal structure, D
H
is assumed
to be the same as that of 300-series SS (~ 3.17 × 10
−16
m
2
/s at RT
[31]). The concentration profiles of H in the samples, estimated using
Eq. (1), are shown in Fig. 4c for both G- and E-charged samples. We
see from it that H is mostly concentrated near the surface in the latter
(see [21,44] for details). At the surface, Cis as high as ~ 1143 wppm,
which can conceivably trigger HELP. Since nanoindentation probes the
surface properties, a marked increase hardness values of the E-charged
HEA specimen was observed [21,25,45].
In summary, the influence of gaseous and electrochemical H charg-
ing on the mechanical behavior of the CoCrFeMnNi HEA was investigat-
ed through tensile and nanoindentation experiments. The results show
that G-charged HEA, in spite of its strong ability to absorb H, is resistant
to embrittlement whereas marked ductility losses were noted for 304
and 316L SSs subjected to identical charging conditions. Through nano-
indentation experiments on both G- and E-charged specimens, we rea-
son that the former fails to embrittle HEA because the H content upon
charging is below the threshold required for triggering the H-enhanced
localized plasticity mechanism.
The work at Hanyang University was supported by the National
Research Foundation of Korea (NRF) grants funded by the Ministry of
Science, ICT & Future Planning (MSIP) (No. 2014M2A8A1030385 and
No. 2015R1A5A1037627). The work at KIST was supported by the
Convergence Agenda Program (CAP) of the Korea Research Council of
Fundamental Science and Technology. The authors wish to thank
Mr. Han-Jin Kim (KIST) and Mr. Chang-Geun Lee (KIMS) for their
valuable supports with experiments.
Fig. 3. H desorption curves obtainedfrom TDS measurements of CoCrFeMnNi HEA,304 SS,
and 316L SS samples that were G-charged under the exactly same condition.
Fig. 4. G-charging vs. E-charging of CoCrFeMnNi HEA. (a) Representative load-
displacement curves of uncharged, G-charged, and E-charged specimens by
nanoindentation tests with Berkovich a nd cube-corner indenters. (b) Representative
SEM images of cube-corner indentation impressions with AFM profiles showing the slip
steps for uncharged and E-charged specimens. (c) Estimated H content distributions
along specimen thickness of G- and E-charged HEAs with the inset showing an enlarged
view of the curves within 30 μm near surface.
Table 2
Hardness values of uncharged, G-charged, and E-charged CoCrFeMnNi HEAs estimated
from nanoindentation tests with Berkovich and cube-corner indenters.
Condition of HEA Nanoindentation hardness (GPa)
Berkovich Cube-corner
Uncharged 2.69 ± 0.06 2.55 ± 0.10
G-charged 2.74 ± 0.09 2.54 ± 0.07
E-charged 4.39 ± 0.20 3.49 ± 0.06
57Y. Zhao et al. / Scripta Materialia 135 (2017) 54–58
References
[1] B. Cantor, I.T.H. Chang, P. Knight, A. Vincent, Mater. Sci. Eng. A 375–377 (2004)
213–218.
[2] J.-W. Yeh,S.-K. Chen, S.-J. Lin, J.-Y. Gan, T.-S. Chin, T.-T. Shun, C.-H. Tsau, S.-Y. Chang,
Adv. Eng. Mater. 6 (2004) 299–303.
[3] Y. Zhang,T.T. Zuo, Z. Tang, M.C. Gao, K.A. Dahmen, P.K. Liaw, Z.P. Lu, Prog. Mater. Sci.
61 (2014) 1–93.
[4] Y.F. Ye, Q. Wang, J. Lu, C.T. Liu, Y. Yang, Mater. Today 19 (2016) 349–362.
[5] D.B. Miracle, O.N. Senkov, Acta Mater. 122 (2017) 448–511.
[6] B. Gludovatz, A. Hohenwarter, D. Catoor, E.H. Chang, E.P. George, R.O. Ritchie, Sci-
ence 345 (2014) 1153–1158.
[7] D.-H. Lee, M.-Y. Seok, Y. Zhao, I.-C. Choi, J. He, Z. Lu, J.-Y. Suh, U. Ramamurty, M.
Kawasaki, T.G. Langdon, J.-i. Jang, Acta Mater. 109 (2016) 314–322.
[8] V. Ocelik, N. Janssen, S.N. Smith, J.Th.M. De Hosson, JOM 68 (2016) 1810–1818.
[9] J.C. Rao, V. Ocelik, D. Vainchtein, Z. Tang, P.K. Liaw, J.Th.M. De Hosson, Mater. Lett.
176 (2016) 29–32.
[10] M.R. Louthan, G.R. Caskey, J.A. Do novan, D.E. Rawl , Mater. Sci. Eng. 10 (1972)
357–368.
[11] H.K. Birnbaum, P. Sofronis, Mater. Sci. Eng. 176 (1994) 191–202.
[12] I.M. Robertson, P. Sofronis, A. Nagao, M.L. Martin, S. Wang, D.W. Gross, K.E. Nygren,
Metall. Mater. Trans. A 26 (2 015) 2323–2341.
[13] A. Barnoush, H. Vehoff, Acta Mater. 58 (2010) 5274–5285.
[14] Y. Zhao, I.-C. Choi, M.-Y. Seok, M.-H. Kim, D.-H. Kim, U. Ramamurty, J.-Y. Suh, J.-i.
Jang, Acta Mater. 78 (2014) 213–221.
[15] R.S. Dutta, J. Nucl. Mater. 393 (2009) 343–349.
[16] G. Hénaff, G. Odemer, A. Tonneau-Morel, Int. J. Fatigue 29 (2007) 1927–1940.
[17] H. Barthélémy, Int. J. Hydrog. Energy 37 (2012) 17364–17372.
[18] R.R. Boyer, Mater. Sci. Eng. A 213 (1996) 103–114.
[19] F. Otto, N.L. Hanold, E.P. George, Intermetallics 54 (2014) 39–48.
[20] F. Otto, A.Dlouhý, Ch. Somsen, H. Bei, G. Eggeler, E.P. George, Acta Mater. 61 (2013)
5743–5755.
[21] Y. Zhao, D.-H. Lee, J.-A. Lee, W.-J. Kim, H.N. Han, U. Ramamurty, J.-A. Suh, J.-i. Jang,
Int. J. Hydrog. Energy (2017) http://dx.doi.org/10.1016/j.ijhydene.2017.02.061.
[22] G.R. Caskey, in: R.A. Oriani, J.P. Hirth, M. Smialowski (Eds.), Hydrogen Degradation
of Ferrous Alloys, Noyes Publications, Park Ridge, NJ 1985, pp. 822–862.
[23] M.B. Whiteman, A.R. Troiano, Corrosion 21 (1965) 53–56.
[24] G. Han, J. He, S. Fukuyama, K. Yokogawa, Acta Mater. 46 (1998) 4559–4570.
[25] K.A. Nibur, D.F. Bahr, B.P. Somerday, Acta Mater. 54 (2006) 2677–2684.
[26] A. Turnbull, in: R.P. Gangloff, B.P. Somerday (Eds.), Gaseous Hydrogen Embrittle-
ment of Materials in Energy Technologies, Woodhead Publishing Ltd., Cambridge
2012, pp. 89–128.
[27] J.H. Ryu, Y.S. Chun, C.S. Lee, H.K.D.H. Bhadeshia, D.W. Suh, Acta Mater. 60 (2012)
4085–4092.
[28] L. Zhang, Z. Li, J. Zheng, Y. Zhao, P. Xu, C. Zhao, X. Li, Int. J. Hydrog. Energy 38 (2013)
8208–8214.
[29] C. San Marchi, T. Michler, K.A. Nibur, B.P. Somerday, Int. J. Hydrog. Energy 35 (2010)
9736–9745.
[30] A.-M. Brass, J. Chêne, Corros. Sci. 48 (2006) 3222–3242.
[31] C. San Marchi, B.P. Somerday, S.L. Robinson, Int. J. Hydrog. Energy 32 (2007)
100–116.
[32] M. Sahlberg, D. Karlsson, C. Zlotea, U. Jansson, Sci. Rep. 6 (2016) 36770.
[33] C. San Marchi, B.P. Somerday, X. Tang, G.H.Schiroky, Int. J. Hydrog. Energy 33 (2008)
889–904.
[34] T. Michler, C. San Marchi, J. Naumann, S. Weber, M. Martin, Int. J. Hydrog. Energy 37
(2012) 16231–16246.
[35] G. Laplanche, A. Lostka, O.M. Horst, G. Eggeler, E.P. George, Acta Mater. 118 (2016)
152–163.
[36] D.G. Ulmer, C.J. Altstetter, Acta Metall. Mater. 39 (1991) 1237–1248.
[37] D.P. Abraham, C.J. Altstetter, Metall. Mater. Trans. A 26 (1995) 2849–2858.
[38] L. Zhang, M. Wen, M. Imade, S. Fukuyama, K. Yokogawa, Acta Mater. 56 (2008)
3414–3421.
[39] Y. Zhao, M.-Y. Seok, I.-C. Choi, Y.-H. Lee, S.-J. Park, U. Ramamurty, J.-Y. Suh, J.-i. Jang,
Scr. Mater. 107 (2015) 46–49.
[40] W.C. Oliver, G.M. Pharr, J. Mater. Res. 7 (1992) 1564–1583.
[41] Y. Zhao, I.-C. Choi, M.-Y. Seok, U. Ramamurty, J.-Y. Suh, J.-i. Jang, Scr. Mater. 93
(2014) 56–59.
[42] J.-i. Jang, G.M. Pharr, Acta Mater. 56 (2008) 4458–4469.
[43] J. Crank, The Mathematics of Diffusion, second ed. Oxford University Press, London,
1975.
[44] A.E. Pontini, J.D. Hermida, Scr. Mater. 37 (1997) 1831–1837.
[45] K.R. Morasch, D.F. Bahr, Scr. Mater. 45 (2001) 839–845.
58 Y. Zhao et al. / Scripta Materialia 135 (2017) 54–58
