Sie sind auf Seite 1von 7

Available online at www.sciencedirect.

com

Journal of Non-Crystalline Solids 354 (2008) 9941000


www.elsevier.com/locate/jnoncrysol

Temperature, frequency, and amplitude dependence


of internal friction of metallic glass
Y. Hiki
a

a,*

, M. Tanahashi b, S. Takeuchi

Department of Physics, Tokyo Institute of Technology, Emeritus, 39-3-303 Motoyoyogi, Shibuya-ku, Tokyo 151-0062, Japan
b
Department of Materials Science and Technology, Science University of Tokyo, Noda, Chiba 278-8510, Japan
Received 4 November 2006; received in revised form 2 July 2007
Available online 12 September 2007

Abstract
The internal friction Q1 and Youngs modulus E of ZrTiCuNiBe bulk metallic glass specimens were measured using a dynamical mechanical spectrometer. Temperature dependence measurements were carried out for dierently heat-treated specimens from room
temperature up to a temperature above the crystallization temperature using several vibration frequencies (0.225 Hz). Two types of Q1
peaks appeared: a high-temperature peak in the supercooled region, and a medium-temperature peak in the glassy state region. Both of
them were shown to be Arrhenius-type anelastic peaks. From the obtained results, the kinetics of movable atoms in the material were
discussed. The amplitude dependence of the internal friction was measured at a given frequency at various temperatures. Fluctuations
and rapid changes were observed in Q1 when the amplitude was gradually increased. The uctuations were especially large near the
glass transition temperature. The results were considered on the basis of the dynamics of the atoms in glass-forming materials near
the glass transition.
 2007 Elsevier B.V. All rights reserved.
PACS: 62.40.+i; 61.43.Fs; 61.43.Dq; 81.40.Jj
Keywords: Amorphous materials; Metallic glasses; Internal friction; Glass transition

1. Introduction
The investigation of metallic glasses (MGs) is actively
being carried out because of their scientic and technological importance [14]. One of the unique characteristics of
MGs is their glass-forming ability (GFA). MGs with high
GFA can be vitried even when they are cooled slowly
from the liquid state. There are a wide variety of MG alloys
composed of various elements, and GFA diers markedly
among dierent alloys. We can prepare big glassy specimens for MGs with high GFA, which are called bulk
metallic glasses (BMGs). One of the BMGs with high

Corresponding author. Tel./fax: +81 3 3469 7104.


E-mail address: hiki.tit@nifty.com (Y. Hiki).

0022-3093/$ - see front matter  2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jnoncrysol.2007.08.007

GFA is Zr-based alloys [3,4]. We can conveniently use such


alloys as a prototype for studying various properties of
glassy materials. Measuring the internal friction (IF), or
vibration mechanical loss, of materials is one of the useful
methods of studying the dynamical mechanical properties
of materials. Several investigators have studied the IF of
various Zr-based BMGs and the typical ones are mentioned here: ZrFe [5], ZrCuNiAl [68], ZrTiCu
NiAl [9,10], ZrAlNiCuNb [11], ZrTiCuNiBe
[12,13], and ZrTiCuNiBeFe [14].
The value of IF generally depends sensitively on the
temperature of the specimen, the frequency of applied
vibration, and the strain amplitude of the vibration. By
measuring the temperature and frequency dependences of
IF, we can investigate the anelastic-type mechanical loss
related to the relaxations occurring in the materials [15].

Y. Hiki et al. / Journal of Non-Crystalline Solids 354 (2008) 9941000

By studying the amplitude dependence of IF, the static-hysteresis-type mechanical loss originating from various loss
mechanisms in the materials can be investigated.
We have studied the IF of various kinds of MGs [1619].
The studies were made at low frequencies, adopting a torsion pendulum apparatus using ribbon specimens. In the
present IF study, a Zr-based BMG is chosen as the specimen material, bulk specimens are used for the experiment,
and the measurement is performed using a dynamical
mechanical spectrometer. The measurement is carried out
from room temperature up to and above the crystallization
temperature. The measurement frequency and the amplitude of vibration are widely varied. The results are analyzed and discussed on the basis of ideas of mechanical
relaxations in glasses.
2. Experimental methods
IF is measured using Dynamical Mechanical Analyzer
(DMA, type Q800, TA Instrument, Newcastle, Delaware,
USA). IF can be measured in the temperature range of
150 to 600 C, in the frequency range of 0.01200 Hz,
and in the specimen displacement amplitude range of
0.510000 lm. The deformation mode of the specimen
adopted is cantilever bending. At a denite temperature
with a denite strain amplitude of vibration, IF can quickly
be measured at several preset vibration frequencies. At the
same time, the value of Youngs modulus can also be
measured.
The specimen material adopted is the alloy Zr41.2Ti13.8
Cu12.5Ni10.0Be22.5, which is one of the BMGs with the highest GFA. The dimensions of the specimen are 17.5 mm in

995

length, 3.04.1 mm in width, and 0.931.01 mm in thickness. X-ray diraction measurement is carried out to check
the amorphous state of the specimen. An example of the
diraction pattern is shown in Fig. 1. Dierential scanning
calorimetry (DSC) is used to determine the glass transition
temperature Tg and the crystallization temperature Tx.
Examples of the DSC curves near Tg for dierent heating
rates q = dT/dt are shown in Fig. 1. From the DSC curves,
the value of Tg is determined by the conventional graphical
method: the background part at low temperatures and the
rapid-rise part near Tg are extrapolated, and Tg is determined as the intercept of the two extrapolations. Note that
Tg depends on the heating rate q. As regards Tx, when several crystallization peaks appear, the lowest-temperature
peak position is adopted as Tx. The DSC values measured
at the usual standard heating rate q = 10 K/min are
Tg = 621.8 K and Tx = 712.9 K.
3. Experimental results
3.1. Temperature and frequency dependences
In Figs. 25, the data of temperature dependences of
internal friction Q1 and Youngs modulus E measured
at several frequencies using the DMA are presented. The
measurements were performed for dierently heat-treated
specimens in the temperature range of T = 300850 K
using four measurement frequencies, f = 0.2, 1, 5, and
25 Hz. The measurement was conducted with increasing
temperature at a rate of 1 K/min, and the data were taken
every 3 K. The strain amplitude of vibration was xed at
105.

Fig. 1. Examples of X-ray diraction pattern and DSC curves for dierent heating rates.

996

Y. Hiki et al. / Journal of Non-Crystalline Solids 354 (2008) 9941000

Fig. 2. Temperature dependences of internal friction Q1 and Youngs


modulus E for four vibration frequencies in as-prepared specimen. Arrows
indicate the high-temperature internal friction peaks.

Fig. 3. Temperature dependences of internal friction Q1 and Youngs


modulus E in specimen annealed at 550 K for 20 h.

The characteristics of the obtained experimental results


are as follows.
(a) As prepared: At a low temperature, a small peak is
observed in Q1 vs. T, and a dip in E vs. T is observed

Fig. 4. Temperature dependences of internal friction Q1 and Youngs


modulus E in specimen annealed at 620 K for 20 h. Arrows indicate the
medium-temperature internal friction peaks.

Fig. 5. Temperature dependences of internal friction Q1 and Youngs


modulus E in specimen annealed at 670 K for 20 h.

near the Q1 peak position. As the temperature is


increased, Q1 increases and E decreases. No obvious
peak can be seen near Tg( = 607 K at q = 1 K/min).

Y. Hiki et al. / Journal of Non-Crystalline Solids 354 (2008) 9941000

997

A large Q1 peak and a large E dip appear at a high


temperature. The position of the Q1 peak indicated
by the arrow in the gure apparently shifts to a higher
temperature with increasing measurement frequency.
(b) Annealed at 550 K/20 h: The previously observed
low-temperature Q1 peak has disappeared. This
Q1 peak is considered to be due to an unstable state
of the as-prepared specimen. The origin of the peak
cannot be identied and is not further considered
here. Meanwhile, the high-temperature Q1 peak
apparently exists.
(c) Annealed at 620 K/20 h: In addition to the high-temperature Q1 peak, a peak can be observed in the
medium-temperature range as indicated by the
arrows. The height of the peak is not appreciable,
and the peak can be observed in this specimen state
because the background IF value is markedly reduced
here due to the annealing treatment. The position of
this peak shifts to a higher temperature as the measurement frequency is increased. The high-temperature Q1 peak still apparently exists.
(d) Annealed at 670 K/20 h: The high-temperature Q1
peak still exists. However, anomalous behaviors are
seen at the highest temperature in both Q1 vs. T
and E vs. T. Namely, values of Q1 and E jump at
a certain temperature. These behaviors can be related
to the fact that partial crystallization has already
started in the material [19].
In Fig. 6, the results of the shift of the Q1 peak position
induced by the increased measurement frequency f are
summarized. The logarithm of the experimental relaxation
time, s = 1/2pf, is plotted against the inverse of the peak
temperature, Tp1, for the cases of the high-temperature
peak (a) and for the medium-temperature peak (b). Close
correlations can apparently be seen between the two quantities in both cases.

Fig. 6. Analysis of high-temperature (a) and medium-temperature (b)


internal friction peaks. Arrhenius t Eq. (2) is applied for (a) and (b). The
VTF t Eq. (1) is also applied for (a). Fitted results are represented by the
solid and broken curves.

3.2. Amplitude dependence


The dependence of IF on the strain amplitude of vibration was studied at a denite vibration frequency at various
temperatures. The frequency of vibration was xed at 1 Hz,
and the as-prepared specimen was used for the measurement. The temperature of the specimen is increased with
a rate of 1 K/min to a certain temperature and kept constant. The strain amplitude A is increased from 105 to
102 and the Q1 value is measured. Then the temperature
is increased to another constant temperature and the Q1
vs. A measurement is repeated. Each measurement run is
completed in 30 min. Examples of the results taken at three
temperatures are shown in Fig. 7. The features of obtained
results are complicated, especially in the high-amplitude
region.
Thus, we intend to perform the amplitude dependence
experiment at lower strain amplitude levels. The results
of such experiments are shown in Fig. 8, where the ampli-

Fig. 7. Dependence of internal friction Q1 on vibration amplitude A in


high-amplitude levels measured at dierent temperatures.

tude range is A = 105104. The measurements were carried out at successively higher temperatures, 300, 550, . . .,
800 K, and then again at lower temperatures 590 K and

998

Y. Hiki et al. / Journal of Non-Crystalline Solids 354 (2008) 9941000

out an IF study similar to ours for the alloy Zr46.75Ti8.25


Cu7.5Ni10.0Be27.5 [13]. They also observed a high-temperature IF peak, and from the peak shift, the parameter values
B = 9440 K and T0 = 352 K were obtained. From these
two experimental results, the following two points can be
noted. (i) The obtained parameter values are very dierent
even though the tested materials are similar in their compositions. This is not so well understood when the IF peak is
considered to be associated with the relaxation arising from
the overall motions of all the atoms in the material,
namely, the hydrodynamic relaxation. (ii) In both experiments, the obtained T0 value seems to be unreasonable.
At this temperature, the relaxation time diverges (see Eq.
(1)), the hydrodynamic behavior cannot exist above this
temperature, and thus, the T0 value should not be lower
than Tg(620 K). After considering these two points, the
VTF formalism for explaining the origin of the IF peak
seems to be unsuitable.
Now both the IF peak observed at high temperatures
and the peak observed at medium temperature are considered to be due to relaxations associated with the local
motions of atoms or defects in the material. Furthermore,
the motion is assumed to be of the thermal activation type.
Then, the Arrhenius-type relaxation time can be taken:
s s0 expE=k B T ;

Fig. 8. Dependence of internal friction Q1 on vibration amplitude A in


low-amplitude levels measured at dierent temperatures. Measurements
are successively made at temperatures from 300 K to 800 K, and then
again at lower temperatures 590 and 640 K.

640 K. It can be seen that the Q1 vs. A behavior is apparently altered as the measurement temperature is changed.
4. Discussion
4.1. IF peaks
Consider rst the high-temperature IF peak. This peak
appears in the supercooled region of the material (Tp > Tg).
In this temperature region, the viscoelastic relaxation in the
glass-forming material is liquid like, or hydrodynamic, and
here, the VogelTammannFulcher relation is applied for
the relaxation time [20]:
s A exp B=T  T 0 ;

where A, B, and T0 are constants. The origin of the observed high-temperature IF peak is for the moment considered to be due to such a relaxation. The log s vs. Tp1 data
in Fig. 6(a) are tted to the VTF formula (Eq. (1)) where T
is to be read Tp. The broken curve in the gure shows the
tted one. The tted parameters are log A = 10.3,
B = 13025 K, and T0 = 108 K. Other authors have carried

where s0 is the pre-exponential factor, kB is the Boltzmann


constant, and E is the activation energy. Results of data ttings are shown by the solid lines for the high- and medium-temperature peaks in Figs. 6(a) and (b), respectively.
The tted parameters are
high-temperature peak : log s0 11:9; E 1:55 eV;
medium-temperature peak : log s0 19:8; E 2:05 eV:
3
It is well known that even in glassy materials, the Snoektype anelasticity can appear. For example, a low-frequency
IF peak is observed in hydrogenated MG at a temperature
lower than room temperature [21]. Hydrogen atoms are
frequently contained in typical MG specimens due to technical reasons, and a low-temperature H-peak is frequently
observed. The peak always appears at low temperatures.
In the present case of the high-temperature IF peak, the
value of log s0 nearly coincides with that for the hopping of
a single atom, log s0 = 12 (Debye frequency 1012 Hz).
Namely, the element contributing to the anelasticity may
be a single atom, and the stress-induced ordering [15] of
the atoms can be the origin of the anelasticity. In the present material, the Be atom is the smallest and can move
most easily. There have been studies of the diusion of
Be in the material using the backscattering technique
[22,23]. The result showed that the activation energy for
the diusion is around 1.1 eV [23]. From the result of Eq.
(3), it is considered that the high-temperature IF peak
can be related to the motion of Be atoms. The conclusion
is, however, still not denite.

Y. Hiki et al. / Journal of Non-Crystalline Solids 354 (2008) 9941000

In the case of the medium-temperature IF peak, the


obtained value of log s0 is smaller and the value of E is larger than those for the high-temperature IF peak (Eq. (3)).
It has been shown by one of the present authors [24,25] that
two kinds of relaxations of the thermal activation type can
be considered: the simple relaxation and the complex
relaxation. In a thermal activation process, relaxing elements jump over an energy barrier. In the simple or usual
relaxation, the jumps of relaxing elements occur independently. In the complex relaxation, jumps of several relaxing elements occur cooperatively and simultaneously.
Many authors have carried out theoretical, computational,
and experimental studies of such cooperative motions [24].
We have developed a phenomenological theory concerning
this problem, and the obtained conclusion to be cited here
is as follows. In a complex relaxation, where n relaxing elements jump cooperatively, the following relation holds:
Eobs nE;

where Eobs is the activation energy observed in the complex


relaxation, and E is the activation energy for the relaxation
of a single relaxing element composing the complex relaxation. It has also been shown that the value of logs0 is usually smaller in the case of the complex relaxation than in
the case of the simple relaxation [24,25]. Thus, the IF peak
at medium temperature is considered to be due to a kind of
complex relaxation. However, what kind of group of atoms
is contributing to the complex relaxation is not clear. For a
group of Be atoms, which is the most probable candidate,
the observed activation energy must be larger than 1.55 2
(atomic pair) = 3.1 eV. The observed activation energy value (2.05 eV) is meaningfully smaller than this. The origin
of the medium-temperature IF peak is thus still not certain.
4.2. Amplitude dependence
For the result shown in Fig. 7, a unied interpretation
cannot be given. In this case, the strain amplitude is rather
large. The irregularity seen in Q1 vs. A can be explained as
follows. The application of large stress to the specimen can
suddenly produce a severe modication of the internal state
of the specimen, and IF is markedly changed. Note here
that such a high-amplitude IF measurement provides a
kind of experimental method for studying the behavior of
glassy materials under strong applied forces. More precise
experimental studies should be made in the future.
Now, the experimental results shown in Fig. 8 are considered. When a hysteresis exists between the applied stress
and the induced strain in a material, a static mechanical
loss proportional to the amount of the hysteresis occurs.
When a vibration stress is applied to the material, IF due
to the hysteresis occurs. The GranatoLucke mechanism
is a typical example of this kind of loss [26]. In a solid containing dislocations pinned by weak pinning, the depinning
caused by the increased applied stress results in irreversible
strain in the solid and a hysteresis occurs. In such a hyster-

999

esis-type loss, IF increases monotonically with the strain


amplitude.
In the present result shown in Fig. 8, no apparent monotonic increase in IF with increasing strain is seen (T = 300,
550 K). However, it is seen that there are uctuations in
Q1 vs. A, and the uctuations are most prominent at
T = 620 and 670 K. Note that these temperatures are near
the glass transition Tg (620 K). After the temperature is
increased to 800 K and the Q1 vs. A measurements are
repeated at lower temperatures (T = 590, 640 K), the uctuations are seen to be extremely diminished at these temperatures, and these temperatures are near Tg.
It is assumed that in the specimen material, an internal
stress exists, and the stress is uctuating with time. In the
present IF experiment, the strain is increased with time at
a constant rate, while the internal stress is uctuating with
time. Thus, a hysteresis can exist between the strain and the
stress, IF due to the hysteresis occurs, and Q1 vs. A shows
apparent uctuations. The uctuation of the internal stress
mentioned above is considered to be due to irregular thermal motions of atoms in the glassy material, which can be
very severe near Tg [20]. These severe uctuations near Tg
should be diminished when the specimen is heated above
Tx (713 K) and the material is crystallized. These anticipated behaviors are actually observed in the experimental
result.
5. Conclusion
Internal friction measurements have been carried out for
a metallic alloy glass with high glass-forming ability,
Zr41.2Ti13.8Cu12.5Ni10.0Be22.5, using dynamical mechanical
analyzer. The temperature dependence of the internal friction was measured from room temperature up to a temperature above the crystallization temperature using several
measurement frequencies. Two kinds of internal friction
peaks apparently appeared: a high-temperature peak in
the supercooled region and a medium-temperature peak
in the glassy state region. Both of them were explained as
Arrhenius-type anelastic peaks. The activation energy
and the pre-exponential factor for the two activation processes were determined. On the basis of the obtained
numerical results, it was considered that the high-temperature peak was due to a hopping of a single atom (presumably the Be atom) and the medium-temperature peak was
due to a cooperative hopping movement of atoms. Such
a study is eective for investigating the kinetics of movable
atoms in glass-forming materials. The amplitude dependence of the internal friction was measured at a denite frequency at various temperatures. The measurement was
mainly made at low strain amplitude levels. There was no
tendency of the internal friction to increase monotonically
with increasing amplitude. Fluctuations were seen in the
internal friction value when the amplitude was increased.
The uctuations were very severe near the glass transition
temperature, and were reduced after the specimen was crystallized. These results were explained by considering the

1000

Y. Hiki et al. / Journal of Non-Crystalline Solids 354 (2008) 9941000

thermal motion of atoms in glassy materials. Such a study


is eective for investigating the dynamics of atoms composing the glasses, especially near the glass transition
temperature.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]

A. Inoue, Mater. Trans. JIM 36 (1995) 866.


A. Inoue, Acta Mater. 48 (2000) 279.
W.L. Johnson, Mater. Sci. Forum 225/227 (1996) 35.
W.L. Johnson, MRS Bull. (October) (1999) 42.
O. Yoshinari, Y. Iwata, M. Yamada, K. Tanaka, J. Physique III 6
(1996) C8613.
R. Scarfone, H-R. Sinning, J. Alloys Comp. 310 (2000) 229.
K. Schroter, E. Donth, J. Non-Cryst. Solids 307/310 (2002) 270.
T. Ichitsubo, S. Kai, H. Ogi, M. Hirao, K. Tanaka, Scripta Mat. 49
(2003) 267.
J.M. Pelletier, J. Perez, J.L. Soubeyroux, J. Non-Cryst. Solids 274
(2000) 301.
T.A.M. Aboki, M.L. Masse, A. Dezellus, P. Ochin, R. Portier, Mat.
Sci. Eng. A 370 (2004) 330.
B. Zhang, F.Q. Zu, K. Zhen, J.P. Shui, P. Wen, J. Phys.: Condens.
Matter 14 (2002) 7461.
D.N. Perera, A.P. Tsai, J. Phys. D: Appl. Phys. 33 (2000) 1937.

[13] P. Wen, D.Q. Zhao, M.X. Pan, W.H. Wang, J.P. Shui, Y.P. Sun,
Intermetallics 12 (2004) 1245.
[14] Q. Wang, J.M. Pelletier, Y.D. Dong, Y.F. Ji, Mat. Sci. Eng. A 370
(2004) 316.
[15] A.S. Nowick, B.S. Berry, Anelastic Relaxation in Crystalline Solids,
Academic, NY, 1972.
[16] Y. Hiki, T. Yagi, A. Aida, S. Takeuchi, J. Alloys Comp. 355 (2003)
42.
[17] Y. Hiki, T. Aida, S. Takeuchi, in: Proceedings of the 3rd International
Symposium on Slow Dynamics in Complex Systems, American
Institute of Physics, NY, 2004, p. 661.
[18] Y. Hiki, T. Yagi, T. Aida, S. Takeuchi, Mat. Sci. Eng. A 370 (2004)
302.
[19] Y. Hiki, M. Tanahashi, S. Takeuchi, Mat. Sci. Eng. A 442 (2006)
287.
[20] I. Gutzow, J. Schmelzer, The Vitreous State, Springer, Berlin, 1995.
[21] T. Yagi, T. Imai, R. Tamura, S. Takeuchi, Mat. Sci. Eng. A 370
(2004) 264.
[22] U. Geyer, S. Schneider, W.L. Johnson, Y. Qiu, T.A. Tombrello,
M.-P. Macht, Phys. Rev. Lett. 75 (1995) 2364.
[23] U. Geyer, W.L. Johnson, S. Schneider, Y. Qui, T.A. Tombrello,
M.-P. Macht, Appl. Phys. Lett. 69 (1996) 2492.
[24] Y. Kogure, Y. Hiki, J. Appl. Phys. 88 (2000) 582.
[25] Y. Hiki, Y. Kogure, Rec. Res. Devel. Non-Cryst. Solids 3 (2003) 199.
[26] A.V. Granato, K. Lucke, J. Appl. Phys. 27 (1956) 583, 789.

Das könnte Ihnen auch gefallen