Relationship between the shear moduli and

defect-induced structural relaxation of

high-entropy metallic glasses
arXiv:2503.04436v1 [cond-mat.dis-nn] 6 Mar 2025

Andrey Makarov1 , Evgenia Gonchrova1† , Jichao Qiao2† ,

Roman Konchakov1† , Vitaly Khonik1*†
1 Department of General Physics, Voronezh State Pedagogical University,
Lenin Str. 86, Voronezh, 394043, Russia.
2 School of Mechanics, Civil Engineering and Architecture, Northwestern

Polytechnical University, Xi’an, 710072, China.

*Corresponding author(s). E-mail(s): [email protected];

Contributing authors: [email protected];
[email protected]; [email protected]; [email protected];
† These authors contributed equally to this work.

Abstract
We performed high-frequency shear modulus and calorimetry measurements on
seven high-entropy metallic glasses (HEMGs) in the initial, relaxed and crys-
talline states. It is shown that the shear modulus of HEMGs is intrinsically
related with the concentration of defects responsible for structural relaxation. In
the absence of structural relaxation, temperature coefficient of shear modulus of
glass equals to that of the maternal crystal. All found regularities are governed
by a single equation.

Keywords: high-entropy metallic glasses, shear modulus, relaxation, defects

Many-component alloys can be characterized by the mixing entropy, which is a

measure of how much disorder is present in an alloy. This quantity is defined as
P
n
∆Smix = −R ci lnci where R is the universal gas constant, ci is the molar frac-
i=1
tion of the i-th element in the alloy and n is the number of constituent elements.
The alloys with a mixing entropy of Smix ≥ 1.5R are referred to as high-entropy

1
alloys [1]. They were discovered in the early 2000s and consist of five or more metal-
lic elements, each with an atomic percentage between 5% and 35%. About a decade
ago, it was discovered that high-entropy alloys can be solidified into a non-crystalline
state, forming high-entropy metallic glasses (HEMGs) [2]. It was found that HEMGs
exhibit enhanced thermal stability [3], which is manifested in higher activation ener-
gies for crystallization [4]. Since the high-mixing-entropy state results in the low excess
entropy of solid glass compared to the maternal crystal [5], HEMGs demonstrate lower
atomic mobility [4], sluggish diffusion [6] and crystallization kinetics [7] as well as slow
dynamics of homogeneous flow [8]. It was suggested that HEMGs combine the prop-
erties of both metallic glasses and high-entropy crystalline alloys [9]. It is important,
therefore, to derive a thorough understanding of the fundamental physical properties
of HEMGs. In particular, this applies to the shear elasticity, which is controlled by the
instantaneous shear modulus. Meanwhile, the instantaneous shear modulus (in prac-
tice, high-frequency shear modulus called simply shear modulus G hereafter) controls
the heights of barriers for local atomic rearrangements in different types of glasses,
including HEMGs, and, therefore, constitutes their major physical parameter [10].
Metallic glasses are produced by melting the maternal crystal and then rapidly
cooling the melt. One can raise an important question: is there any relationship
between the properties of glass and its crystalline counterpart? This in full applies to
the shear elasticity controlled by the shear moduli of glass and its maternal crystal.
However, this issue is seldom discussed in the literature [11].
On the other hand, metallic glasses are prone to spontaneous thermoactivated
changes of their properties called structural relaxation. This phenomenon is often
interpreted as a result of the changes in the system of defects – local regions with low
point symmetry [12]. These defects are considered from very different viewpoints (e.g.
Refs. [12, 13]) and structural relaxation is most often interpreted in terms of changes
of their concentration.
Meanwhile, the relation between the defect system, the shear moduli of glass and
its maternal crystal is intrinsically built into the Interstitialcy theory (IT), which was
found to provide a quantitative understanding of different relaxation phenomena in
metallic glasses (a review of the IT and its interpretation of experimental data is given
in Ref.[14]). The IT argues that the defect system of glass, the shear moduli of glass
G and its maternal crystal µ are interrelated as

G = µ exp(−αβc), (1)
where c is the concentration of defects similar to dumbbell interstitials in crystals and
responsible for reduced shear modulus of glass (diaelastic effect), β is dimensionless
shear susceptibility and dimensionless α ≈ 1 [14]. Eq.(1) after differentiation can be
rewritten as

dc(T ) d µ(T )
β = ln . (2)
dT dT G(T )

2
Table 1 High-entropy metallic glasses under investigation, their mixing entropies and the derivatives
dlnG/dT and dlnµ/dT calculated for temperature ranges with no structural relaxation.

No Composition (at.%) Smix /R dlnG/dT (×10−4 K−1 ) dlnµ/dT (×10−4 K−1 )

1 Zr31.6 Cu37.8 Hf13.4 Al8.7 Ag8.4 1.42 −2.69 ± 0.03 −2.54 ± 0.04
2 Zr31 Ti27 Be26 Cu10 Ni6 1.46 −2.44 ± 0.05 −2.26 ± 0.04
3 (Ti37.31 Zr22.75 Be26.39 Al4.59 Cu9 )94 Co6 1.56 −2.34 ± 0.05 −2.11 ± 0.05
4 Ti20 Zr20 Hf20 Be20 Cu20 1.61 −2.57 ± 0.08 −2.32 ± 0.03
5 Ti20 Zr20 Hf20 Be20 Ni20 1.61 −2.62 ± 0.07 −2.47 ± 0.08
6 Zr35 Cu25 Hf13 Al11 Ag8 Ni8 1.63 −2.46 ± 0.03 −2.36 ± 0.03
7 Zr35 Hf17.5 Al12.5 Ni12 Cu10 Co7.5 Ti5.5 1.77 −2.29 ± 0.06 −2.28 ± 0.04

It follows from this equation that if c = const and, therefore, structural relaxation is
absent then the right-hand side must be zero. In this case one can write down that

d lnG(T ) d lnµ(T )
= . (3)
dT dT
This equation shows that temperature coefficients of the shear moduli in the glassy
and crystalline states in the absence of structural relaxation should be equal, i.e.
G−1 dG/dT = µ−1 dµ/dT .
On the other hand, in the presence of structural relaxation, the defect concentration
c below the glass transition temperature Tg decreases, i.e. dc/dT < 0, and, therefore,
d µ(T )
the derivative dT ln G(T ) in Eq.(2) should be negative. Conversely, since c increases
d µ(T )
above Tg , one should obtain the inequality dT ln G(T ) > 0 in this case. All these
predictions on the defect concentration, shear moduli of glass and maternal crystal as
well as the validity of Eq.(1) are verified in the present work.
We studied 7 HEMGs listed in Table 1 with the mixing entropies 1.42 ≤ Smix /R ≤
1.77. The glasses were obtained by melt quenching into a copper mold and X-ray ver-
ified to be fully amorphous. Differential scanning calorimetry (DSC) was performed
with a Hitachi DSC 7020 instrument in high-purity nitrogen atmosphere using 50-70
mg samples. A crystallized sample of nearly the same mass was placed into the refer-
ence cell, so that the instrument measured the difference in the heat flow between the
glassy and crystalline samples referred to as the differential heat flow ∆W hereafter.
Shear modulus was measured by the electromagnetic transformation method. In
this method, transverse resonant vibrations (f = 400 − 600 kHz) of a sample (5 ×
5 × 2 mm3 ) are produced due to Lorentz interaction of external magnetic field with
surface current excited by a coil [15]. The shear modulus was calculated as G(T ) =
2
Grt f f(T
2
)
, where frt and Grt are the vibration frequency and shear modulus at room
rt
temperature, respectively. The error of the determination of temperature changes of G
was about 5 ppm near room temperature and about 100 ppm near the glass transition
temperature Tg . Room temperature shear modulus was determined using resonant
ultrasound spectroscopy with a precision of 1-2%.

3
 An example of shear modulus measurements is presented in Fig.1(a), which gives
temperature dependences of G in the initial state, after relaxation performed by heat-
ing into the supercooled liquid state (i.e. into the range Tg < T < Tx , where Tx is the
crystallization onset temperatures) and in fully crystalline (maternal) state. In the ini-
tial state, G first linearly decreases with temperature due to the anharmonicity, next
starts to increase due to exothermal relaxation (see DSC scan in Fig.1(b)) at a temper-
ini
ature Tsr ≈ 490 K, then begins to fall above Tg ≈ 701 K and finally rapidly increases
upon crystallization onset at Tx ≈ 759 K. In the relaxed state, room-temperature value
of G is increased by ≈ 7% and the upward G-rise because of exothermal structural
relaxation is absent. Instead, heating results in endothermal relaxation (see Fig.1(b))
and shear modulus starts to fall below purely anharmonic decrease at a temperature
rel
Tsr ≈ 575 K. Above Tg , temperature behaviour of relaxed sample repeats that of the
initial specimen. Crystallization leads to an increase of G by ≈ 24% and G smoothly
decreases with temperature upon heating.
Figure 1(c) gives the defect concentration c of the same glass in the initial and
relaxed states as a function of temperature calculated using Eq.(1) with G(T ) and
µ(T ) dependences shown in Fig.1(a) and a typical shear susceptibility β = 20. In the
initial state, c ≈ 0.012 and is nearly constant up to the temperature of exothermal
ini
structural relaxation onset Tsr , next c decreases by about one tenth near Tg and
rapidly increases above Tg . After relaxation, c is decreased by about one third at room
temperature and remains constant upon heating up to the beginning of endothermal
rel
structural relaxation at Tsr , which results in an increase of the shear modulus up to
T ≈ Tg . Above Tg , temperature dependences of G in the initial and relaxed states are
close to each other. Similar data were obtained for other HEMGs.
We can now check the validity of Eq.(2). This is done in Fig.2, which gives the
d µ(T )
derivatives dT ln G(T ) for six HEMGs in the initial and relaxed states. The plots are
quite similar and can be commented using the data on Zr31.6 Cu37.8 Hf13.4 Al8.7 Ag8.4
ini
(Fig.2(a)). Below the onset of exothermal relaxation of the initial glass at Tsr ≈ 490
K, this derivative is zero. Above this temperature, the derivative becomes negative
reflecting a decrease of the defect concentration (see Fig.1(c)), in accordance with
Eq.(2). Above Tg , the defect concentration rapidly increases due to endothermal relax-
d µ(T )
ation in the supercooled liquid state (Fig.1(b)) so that dT ln G(T ) sharply increases
d µ(T )
with temperature, in line with Eq.(2). After relaxation, dT ln G(T ) remains close to zero
rel
up the beginning of endothermal relaxation at Tsr ≈ 575 K (Fig.2(b)) and increases
at higher temperatures that follows the growing defect concentration c. Entering the
supercooled liquid state at T > Tg results in a strong endothermal flow (Fig.2(b))
d µ(T )
leading to a rapid increase of both c and dT ln G(T ) , which is reflected by Eq.(2). Sim-
ilar interconnection between the heat effects, defect concentration c and derivative
d µ(T )
dT ln G(T ) is observed for other HEMGs.
In the absence of structural relaxation (i.e. if dc/dT = 0), Eq.(3) should be obeyed.
This prediction is verified in Fig.3, which gives the interdepedence between the deriva-
tives dlnG/dT and dlnµ/dT , which are calculated for the temperature ranges of 300
ini
K< T < Tsr (in these ranges G(T ) and µ(T ) are linear in temperature) for all indi-
cated HEMGs. It is seen that the slope of the linear fit for this interdependence is close

4
 dG dµ
to unity that means the equality of temperature coefficients GdT and µdT of the shear
moduli for glass and maternal crystal in the absence of structural relaxation. At that,
any dependence of these derivatives on the mixing entropy ∆Smix is not detected.
One can conclude, thus, that the shear moduli G of HEMGs are intrinsically related
with the defect concentration c: relaxation increase of G originates from an exothermal
decrease of c. Conversely, a decrease of G due to structural relaxation takes place
because of endothermal rise of the defect concentration. The latter process is strongly
pronounced in the supercooled liquid state above Tg . If thermoactivated structural
relaxation is absent, then temperature coefficients of the shear moduli in the glassy
and crystalline states are equal.
All aforementioned regularities originate from Eq. (1), which constitutes the main
hypothesis of the Interstitialcy theory [14]. The present experiments and data analysis,
therefore, provide further support for this theory.
Acknowledgements. This work was supported by the Russian Science Foundation
under the Grant No. 23-12-00162.
Conflict of interests. The authors declare no competing interests.

References
[1] 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., 2004, vol.6, pp. 299-303.

[2] A. Takeuchi, N. Chen, T. Wada, Y. Yokoyama, H. Kato, A. Inoue, J.W. Yeh:

Intermetallics, 2021, vol.19, pp. 1546-1554.

[3] H. Luan, X. Zhang, H. Ding, F. Zhang, J.H. Luan, Z.B. Jiao, Y.-C. Yang, H. Bu,
R. Wang, J. Gu, C. Shao, Q. Yu, Y. Shao, Q. Zeng, N. Chen, C.T. Liu, K.-F.
Yao: Nat. Comm., 2022, vol.13, p. 2183.

[4] C.J. Chen, R. Xu, B.J. Yin, Y.Z. He, J.Y. Zhang, P. Zhang, B.L. Shen:
Intermetallics, 2023, vol.157, p. 107887.

[5] G.V. Afonin, J.C. Qiao, A.S. Makarov, R.A. Konchakov, E.V. Goncharova, N.P.
Kobelev, V.A. Khonik: Appl. Phys. Lett., 2024, vol.124, p.151905.

[6] Y.J. Duan, L.T. Zhang, J.C. Qiao, Y.-J. Wang, Y. Yang, T. Wada, H. Kato,
J.M. Pelletier, E. Pineda, D. Crespo: Phys. Rev. Lett., 2022, vol.129, p. 175501.

[7] M. Yang, X.J. Liu, Y. Wu, H. Wang, X.Z. Wang, P.Z. Lu: Mater. Res. Lett.,
2018. vol.6, p.495-500.

[8] A.S. Makarov, J.B. Cui, J.C. Qiao, G.V. Afonin, N.P. Kobelev, V.A. Khonik:
Intermetallics, 2024, v.175, p.108478.

[9] Yu Chen, Z.-W. Dai, J.-Z. Jiang: J. Alloys Compd., 2021, vol.866, p.158852.

5
[10] J.C. Dyre: Rev. Mod. Phys., 2006, vol.78, pp. 953-972.

[11] Yu.P. Mitrofanov, N.P. Kobelev, V.A. Khonik: Phys. Sol. State, 2019, vol.61,
pp. 962-968.

[12] Y.Q. Cheng, E. Ma, Prog. Mater. Sci., 2011, vol.56, pp.379-473.

[13] S.T. Liu, F.X. Li, M.Z. Li, W.H. Wang: Sci. Rep., 2017, vol.7, p.11558.

[14] N.P. Kobelev, V.A. Khonik: Physics–Uspekhi, 2023, vol.66, pp.673-690.

[15] A.N. Vasil’ev, Yu.P. Gaidukov: Sov. Phys. Usp., 1983, vol. 26, pp.952-973.

6
 0.012

defect concentration c
g
ini
c T
sr
rel
T T
sr x

0.008

0.004 initial

relaxed

0.000
DW (W/kg)

150 b

initial
differential heat flow

100
5
relaxed

Tg

50 10

400 500 600 700

exo
0
endo

45 Zr Cu
31.6 37.8
Hf
13.4
Al
8.7
Ag
8.4
(GPa)

initial

relaxed 3 K/min
Shear modulus

40 crystalline

rel
T
sr
T
g

T
x
35

a ini
T
sr

300 400 500 600 700

temperature (K)

Fig. 1 Temperature dependences of the shear modulus (a), differential heat flow ∆W (b) and defect
concentration (c) c of glassy Zr31.6 Cu37.8 Hf13.4 Al8.7 Ag8.4 in the initial and relaxed states. The defect
concentration c is calculated using Eq.(1) with β = 20. Temperatures of structural relaxation onset
ini ) and relaxed state (T rel ) together with the calorimetric glass transition tem-
in the initial state (Tsr sr
perature Tg and crystallization onset temperature Tx are indicated. The sold lines give corresponding
linear approximations.

7
 12 6
Zr Cu Hf Al Ag Zr Ti Be Cu Ni
31.6 37.8 13.4 8.7 8.4 31 27 26 10 6

8 Tg initial
3
initial 3 K/min relaxed
relaxed
4
Trel
sr
0

a Tinisr b
-3

)
-4

-1
)
-1

K
(Ti Zr Be Al Cu ) Co Ti Zr Hf Be Cu
37.31 22.75 26.39 4.59 9 94 6 20 20 20 20 20
12
K

-4
-4

m/G)/dT ( ×10
3 initial
m/G)/dT ( ×10

initial 8
relaxed
relaxed
4
0

c d
-3

dln(
-4
dln(

6 Ti Zr Hf Be Ni 12
20 20 20 20 20 Zr Cu Hf Al Ag Ni
35 25 13 11 8 8

4 9
initial

2 relaxed initial 6
relaxed
0 3

-2 0

-4 e f
-3

400 500 600 700 400 500 600 700

temperature (K)

Fig. 2 Temperature dependences of the derivative dln(µ/G)/dT for indicated HEMGs in the initial
and relaxed states.

2.2
)
-1
K

7
-4

3
2.4
×10

2
dlnG/dT (

4
5
2.6

2.8
2.6 2.4 2.2 2.0

dln m/dT ( ×10 -4

K
-1
)

Fig. 3 Interdependence of the derivatives dlnG/dT and dlnµ/dT for the glassy and crystalline
states. The former derivative is calculated for low temperature regions with no structural relaxation
(i.e. below the temperatures denoted as Tsr ini in Fig.1(a)). The numbers give HEMGs compositions

according to Table 1. The slope of the linear fit is 0.91 ± 0.18.