Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Materials Science & Engineering A 761 (2019) 138067

Contents lists available at ScienceDirect

Materials Science & Engineering A


journal homepage: www.elsevier.com/locate/msea

Flow behavior of diffusion bonding interface of Ti6Al4V alloy over a wide T


range of strain rates
Huiping Wua, Xifeng Lia,*, Qiongfeng Meia, Jun Chena, Guohong Wub
a
Department of Plasticity Technology, School of Materials Science and Engineering, Shanghai Jiao Tong University, Shanghai, 200030, China
b
Dongguan Hengli Die & Mold Technological Industry Development Company Ltd, Dongguan, 523460, China

ARTICLE INFO ABSTRACT

Keywords: Quasi-static and dynamic tensile tests were conducted to measure diffusion bonding (DB) interface strength of
Strain rate Ti6Al4V alloy at different strain rates. The effect of strain rate on fracture mode was analyzed by the fracto-
Johnson-cook model graphy analyses. A modified Johnson-Cook (JC) model was proposed to describe the stress-strain behavior of DB
SHPB test interface, where the strain rate hardening coefficient had a nonlinear relationship with the strain and the strain
Diffusion bonding interface strength
rate. The JC failure model with calibrated material constants was used to predict the failure strain of DB in-
Ti6Al4V alloy
terface. Using Abaqus/Explicit through a VUHARD subroutine, the modified JC model and the JC failure model
were verified by predicting load-elongation curves and failure strains at three strain rates. The models are
capable of describing the mechanical behavior of DB interface for Ti6Al4V alloy.

1. Introduction as 0.02–0.03. The compressive behavior of Ti6Al4V prepared by elec-


tron beam melting was investigated at strain rates of 1000 s−1 and
Ti6Al4V alloy provides high strength, excellent weldability, good 0.001 s−1 [7]. The stress level was higher and the fracture strain was
corrosion and heat resistance [1,2]. The diffusion bonding (DB) tech- lower for dynamic loading when compared to those for static loading.
nique has been employed to join similar or dissimilar metals in aero- Due to its simpler form and less model parameters, the Johnson-
space applications. Ti6Al4V alloy is widely used to manufacture the Cook (JC) constitutive model is widely used to describe the mechanical
important components such as aero-engine blade and disc as well as behavior of the alloy under different loading conditions [8,9]. The flow
turbine blade. In these applications, the components usually undergo behavior of Ti6Al4V at the temperatures of 700°C–-1100°C with a strain
high speed rotation or high velocity impact, and failure can easily rate of 1000 s−1 was described by JC model [10]. The excellent
happen due to the high strain rate [3]. DB interface failure is one of the agreement was found between the prediction and experiment data. A
major failure modes. Since Ti6Al4V alloy is a strain rate sensitive ma- modified JC constitutive equation was proposed to describe the dy-
terial, it is necessary to analyze its mechanical properties over a wide namic behavior of Ti6Al4V tested at temperatures of 20°C–1000 °C with
range of strain rates. In literature, the mechanical behavior of Ti6Al4V a strain rate of 1400 s−1 [11]. He et al. [12] made a comparative study
has been studied extensively at quasi-static, low and high strain rates. on the accuracy and effectiveness of the JC model, the modified JC
Quasi-static and dynamic loading tests for Ti6Al4V were carried out at model and the Arrhenius-type model in predicting the high temperature
strain rates from 10−5 s−1 to 3378 s−1 at room temperature, and the flow stress. The modified JC model and the Arrhenius-type model gave
results indicated that Ti6Al4V is a strain rate hardening material [4]. more accurate prediction of high temperature flow stress for 20CrMo
Split Hopkinson pressure bar (SHPB) tests were performed for Ti6Al4V alloy steel, since both models considered the coupled effects of strain
under strain rates from 500 s−1 to 3000 s−1 at temperatures from 20 °C rate and temperature.
to 1100 °C [5]. The flow stress was shown sensitive to both strain rate Model parameters play a key role in predicting the mechanical be-
and temperature. The strain rate sensitivity decreased with true strain havior. The chemical composition, experimental conditions and
and increased with strain rate and temperature. Tensile and shear ex- thermal treatments may result in large differences for model parameters
periments at high strain rates (400 s−1 and 1000 s−1) were conducted [13]. So far, few studies have focused on the dynamic response for DB
to determine the stress-strain curves for Ti6Al4V [6]. The strain hard- interface of Ti6Al4V alloy. To expand the usage of DB components in
ening was found low, and the strain at the necking onset was measured aerospace field, the dynamic behavior of DB interface needs to be

*
Corresponding author.
E-mail address: lixifeng@sjtu.edu.cn (X. Li).

https://doi.org/10.1016/j.msea.2019.138067
Received 29 April 2019; Received in revised form 21 June 2019; Accepted 23 June 2019
Available online 25 June 2019
0921-5093/ © 2019 Elsevier B.V. All rights reserved.
H. Wu, et al. Materials Science & Engineering A 761 (2019) 138067

Fig. 2. Tensile sample sprayed with speckles.

Fig. 1. Schematic sketch of the samples: (a) DB, (b) tensile test (unit: mm).
Fig. 3. SHPB test system: (1) incident bar, (2) transmission bar, (3) strain
gauges, (4) tested specimen.
explored. This work aims to investigate the mechanical properties of DB
interface of Ti6Al4V alloy over a wide range of strain rates, especially
to 1700 s−1 were investigated by using the SHPB testing system at room
for dynamic loading condition. A modified JC model is established for
temperature as shown in Fig. 3. The specimen was installed between
predicting DB interface strength, and the JC failure model is used to
incident bar and transmission bar by screw thread connection. The
predict failure strain.
fracture surface of tensile samples was observed by JSM7600F scanning
electron microscopy (SEM).
In order to measure the fracture strains at different triaxiality ratios,
2. Experiments
tensile tests on the specimens with different notch radii and shear tests
were conducted at a strain rate of 0.0001 s−1. These specimen geo-
As-received Ti6Al4V alloy has a chemical composition (wt%) of
metries are shown in Fig. 4.
5.98 Al, 4.25 V, 0.23 Fe, 0.23 Si and balance Ti. Ti6Al4V alloy block
was cut into rectangular samples with 80 mm × 80 mm × 25 mm by a
spark discharge machine. The bonding surfaces were ground progres- 3. Experimental results
sively using SiC sandpapers of 600, 800, 1200, 1500 grit and cleaned in
acetone. DB tests of Ti6Al4V were carried out in vacuum hot-pressing 3.1. Mechanical behavior of DB interface
furnace with a pressure of 2.0 × 10−3 Pa and a heating rate of 10 °C/
min. The DB temperature, pressure and holding time were set as 900 °C, Optical microstructure of as-received and bonded alloy are shown in
3 MPa and 30 min, respectively. The bonded interface of Ti6Al4V alloy
was ground with SiC sandpapers of 800 and 1500 grit, polished with
SiO2 emulsion and corroded for 6 s in etching solution consisting of
2 vol% HF+4 vol% HNO3+94 vol% H2O. Then it was observed by Zeiss
Axio Imager. M2m optical microscope (OM). Fig. 1 shows the schematic
sketch of DB sample and tensile ones. To ensure the DB interface in the
middle of tensile sample, one end of DB sample was firstly line-cut, and
then it was ground until the distance between this end and the DB in-
terface is 15.5 mm. The DB interface is located at the minimum cross
section area to favor the crack occurrence at the interface.
The tensile tests for DB interface of Ti6Al4V alloy were conducted at
different strain rates. Three specimens were repeated for each condi-
tion. Quasi-static (10−4 s−1) and low strain-rate (10−2 s−1 and 0.1 s−1)
tensile tests were conducted on universal testing machine at room
temperature. The specimens were sprayed with speckles (Fig. 2). And
their strain was measured by digital image correlation (DIC) tech-
nology. The tensile samples of as-received alloy were also tested at Fig. 4. Specimen geometry for measuring the fracture strain (green line in-
quasi-static and low strain rates for comparison. dicates DB interface, unit: mm). (For interpretation of the references to colour
The mechanical behaviors at high strain rates ranging from 500 s−1 in this figure legend, the reader is referred to the Web version of this article.)

2
H. Wu, et al. Materials Science & Engineering A 761 (2019) 138067

Fig. 5. Optical microstructure of as-received alloy and DB interface: (a) as-re-


ceived alloy, (b) DB interface.

Fig. 7. The engineering stress-strain curves for bonded Ti6Al4V alloy at dif-
ferent strain rates.

discontinuous yielding is defined as the ultimate tensile stress. When


the strain rate increases from 0.1 s−1 to 500 s−1, the strain rate hard-
ening obviously happens and the ultimate tensile stress increases from
1186 MPa to 1426 MPa. However, the dynamic strength improvement
of bonded Ti6Al4V alloy ceases when the strain rate is above 750 s−1.
When the strain rate is higher than 750 s−1, a significant yield stress
drop occurs. The curves can be divided into four stages, including
elastic deformation derived from the impact of incident bar and sub-
sequent pre-yield microplastic deformation, small drop of yield stress,
yield propagation expressed as yield plateau, and the final work hard-
Fig. 6. The engineering stress-strain curves for as-received and bonded Ti6Al4V ening and thermal softening [14]. It should be noted that the maximum
alloys at quasi-static and low strain rates. stress is achieved before the yield plateau appears. In the fourth stage, a
short-period strain hardening is observed, and then the stress increases
Fig. 5. As-received alloy consists of long strip and equiaxed α phases as to an ultimate value. Finally, the stress slightly decreases with the in-
well as lamellar β phase. After DB test, long strip α phase almost dis- creasing strain.
appears. Equiaxed α phase and intergranular β phase along the The mechanical properties for bonded Ti6Al4V alloy are listed in
boundaries of α phase exist in bonded alloy. No obvious voids are ob- Table 2. The uniform elongation is defined as the elongation at the
served at DB interface, which indicates good DB quality. point of maximum force. It corresponds to ultimate tensile strength. The
The engineering stress-strain curves of as-received and bonded al- uniform elongation at high strain rates is smaller than those at quasi-
loys at quasi-static and low strain rates are shown in Fig. 6. Their me- static and low strain rates. This results from the thermal softening ef-
chanical properties are also listed in Table 1. It can be found that both fect. The fracture strains at different strain rates are calculated by using
as-received and bonded alloys have positive strain rate sensitivity at f = ln(A 0 / Af ) , where A0 and Af are the cross-sectional areas of spe-
room temperature. Furthermore, the stress of bonded alloy is slightly cimen before test and after fracture, respectively. Here, A0 and Af are
lower than that of as-received alloy. Meanwhile, bonded alloy shows calculated according to ASTM E8/E8M − 13 (Standard test methods for
more obvious plastic stage. When tested at 0.0001 s−1 and 0.01 s−1, tension testing of metallic materials).
bonded alloy indicates lower strain-hardening rate than as-received
alloy. While tested at 0.1 s−1 for both alloys, the strain hardening stops 3.2. Fracture surface morphology
earlier. Then the stress starts to decrease slowly when the strain exceeds
4%. For as-received alloy tested at quasi-static and low strain rates, all
The engineering stress-strain curves for bonded alloy at quasi-static, samples fractured due to the propagation of an inclined crack or a
low and high strain rates are shown in Fig. 7. The maximum stress after transverse crack at early stage followed by a slant crack in the final
stage, as shown in Fig. 8 (a) and Fig. 8 (b). The plastic fibrous zone

Table 1 Table 2
Mechanical properties for as-received and bonded Ti6Al4V alloys at quasi-static Mechanical properties for bonded Ti6Al4V alloy at different strain rates.
and low strain rates.
Material Strain Ultimate tensile Uniform Fracture strain
Material Strain Yield stress Ultimate Fracture elongation rate (s 1) stress (MPa) elongation (%)
rate 0.2 (MPa) tensile stress
(s 1) (MPa) Bonded alloy 0.0001 1115 ± 6 5.22 ± 0.05 0.335 ± 0.014
0.01 1164 ± 8 3.97 ± 0.03 0.349 ± 0.012
As-received alloy 0.0001 1023 ± 6 1130 ± 5 0.119 ± 0.001 0.1 1186 ± 9 4.03 ± 0.06 0.353 ± 0.004
0.01 1114 ± 9 1179 ± 6 0.143 ± 0.003 500 1426 ± 4 2.88 ± 0.12 0.312 ± 0.010
0.1 1178 ± 5 1207 ± 5 0.111 ± 0.002 750 1447 ± 3 2.88 ± 0.06 0.279 ± 0.003
Bonded alloy 0.0001 1013 ± 7 1115 ± 6 0.174 ± 0.002 1000 1434 ± 1 3.49 ± 0.24 0.271 ± 0.006
0.01 1103 ± 12 1164 ± 8 0.162 ± 0.006 1500 1422 ± 8 3.91 ± 0.19 0.260 ± 0.002
0.1 1105 ± 16 1186 ± 9 0.171 ± 0.002 1700 1427 ± 17 2.97 ± 0.14 0.271 ± 0.003

3
H. Wu, et al. Materials Science & Engineering A 761 (2019) 138067

Fig. 9. Fracture surfaces of bonded alloy at different strain rates: (a) 0.01 s−1,
(b) 0.1 s−1, (c) 500 s−1, (d) 1500 s−1, (e) 1700 s−1.

along the shear stress direction in the shear lip zone. The area of plastic
fibrous zone decreases and the area of shear lip zone increases with the
increase of strain rate.
The top views of fracture surfaces for bonded samples tested at
0.0001 s−1 and 1000 s−1 are shown in Fig. 10 and Fig. 11. For quasi-
static loaded sample (0.0001 s−1), lots of dimples and subdimples are
seen in both fibrous and shear lip regions, which indicates the ductile
fracture. When the sample loaded at 1000 s−1, the obvious difference
can be observed. In Fig. 11, the ductile fracture occurs in Zone A. The
dimples in Zone A are deeper than those loaded at 0.0001 s−1, as shown
in Figs. 10(d) and Fig. 11(d). The dimples, microvoids, inclusions and
cleavage planes are found in Zone B as shown in Fig. 11(b), which in-
dicate a mixed fracture characteristics. Large percentage of cleavage
Fig. 8. Fracture surfaces of as-received alloy after loading at quasi-static and planes is observed in zone C (shear lip zone), which shows cleavage
low strain rates: (a) 0.0001s−1, (b) 0.1 s−1, (c) 0.0001 s−1, (d) zone B in (c), (e) fracture mode. Overall, the fracture mechanism changes from ductile
zone A in (c), (f) 0.1 s−1, (g) zone A in (f), (h) zone B in (f). fracture mode to a mixture of ductile and cleavage fracture with the
increase of strain rate. The similar phenomenon has been reported by
shows an elliptical profile (white marked circle). It indicates that the
crack initiates in the center and then propagates to both sides under
shear stress. The top views of fracture surface tested at 0.0001 s−1 and
0.1 s−1 are shown in Fig. 8 (c) and (f). Large quantities of cleavage
planes are observed in zone A on the fracture surface as shown in
Fig. 8(e) and (g). Instead, the crack propagation zone (zone B) is
dominated by equiaxed dimples as revealed in Fig. 8(d) and (h).
Overall, the fracture surfaces are characterized with microvoids, dim-
ples and cleavage planes. Consequently, the mixed-mode ductile and
cleavage fracture happened for as-received alloy tested at quasi-static
and low strain rates.
The fracture surfaces of bonded alloy after testing at different strain
rates are shown in Fig. 9. The bonded alloy tested below 1700 s−1
exhibits a typical cup and cone fracture with plastic fibrous zone and
shear lip zone. While the slant fracture exists in the sample tested at
1700 s−1. The cup and cone fracture in this work includes two stages.
The microvoids firstly initiate at the inclusions or dislocation pile-ups,
and then the coalescence of microvoids leads to a transverse crack in Fig. 10. Fracture surface of bonded sample after loading at 0.0001 s−1: (a)
the center of tested sample. In the second stage, the crack propagates overall view, (b) zone B, (c) zone C, (d) zone A.

4
H. Wu, et al. Materials Science & Engineering A 761 (2019) 138067

Fig. 11. Fracture surface of bonded sample after loading at 1000 s−1: (a)
overall view, (b) zone B, (c) zone C, (d) zone A.
Fig. 12. Relation curve of ln( A) and ln at the reference strain rate and
temperature.
other scholars [15,16]. Moreover, the void distribution is homogenous
in fracture surface of the sample loaded at 0.0001 s−1 (Fig. 10). While
the voids are heterogeneously distributed in the sample loaded at 1000 = [ak (A 0 + )n + 1 a A1 + A2 1 exp A3 + A 4 2 1 + C ln
s−1 (Fig. 11). The volume fraction of local void in white marked circle 0

is significantly more than other zones. (4)


where a, k, A0, n, A1, A2, A3, A4 are material constants. The data at
4. Model and verification reference strain rate of 0.0001s−1 are selected to determine the mate-
rial constants. The true stress-plastic strain curve is obtained and fitted
4.1. Modified JC model and JC failure model by equation (4) as shown in Fig. 13. The material constants a, k, A0, n,
A1, A2, A3 and A4 are determined as 0.8649, 1160.13, 0.00247,
The original JC model is expressed as: 0.07302, 2722.43, 419.58, 7.04 and 131.13, respectively. It can accu-
m
rately describe the stress increase and saturation with increasing the
n T Tr strain.
= A+B 1 + C ln 1
0 Tm Tr (1) To calculate the strain-rate hardening coefficient C, equation (4) is
transformed as follows.
where σ, ε, , 0 , Tr, Tm, A, B, n, C and m denote the von Mises equivalent
flow stress, equivalent plastic strain, strain rate, reference strain rate, 1 = C ln
ak (A 0 + )n + (1 a)(A1 + A2 (1 exp( A3 + A 4 2 ))) (5)
reference temperature, melting temperature, yield stress at reference
0

temperature and strain rate, strain hardening coefficient, strain hard- Substituting the stress data at different strains and strain rates, the rela-
ening exponent, strain-rate hardening coefficient and thermal softening tion between /(ak (A 0 + )n + (1 a)(A1 + A2 (1 exp ( A3 + A 4 2)))) 1
coefficient, respectively. and ln( / 0 ) is obtained as plotted in Fig. 14.
The tensile and SHPB tests were carried out at room temperature, Apparently, the C value (the curve slope in Fig. 14) varies at dif-
which is the reference temperature (20 °C). Thus the thermal softening ferent plastic strains and strain rates, which is a function of strain and
effect is ignored, and equation (1) can be transformed as below. strain rate. equation (6) is used to fit C values. And the coefficients of
B0 ~ B6 are determined by using MATLAB software as 0.1434,
= (A + B n) 1 + C ln −0.0446, 0.0995, −0.1196, 0.0302, −0.00048 and −0.0082, re-
0 (2) spectively. The experiment and simulation C values at different strains
−1 and strain rates (0.01 s−1, 0.1 s−1, 500 s−1, 1000 s−1 and 1500 s−1)
When the reference strain rate is 0.0001s , the strain rate hard-
ening influence is ignored. equation (2) is further converted to
= (A + B n) . By taking the natural logarithm of both sides, equation
(3) is obtained as:
ln( A) = ln B + n ln (3)
−1
Fig. 12 shows the relation curve of ln( A)~ ln at 0.0001 s and
20 °C. A nonlinear relation exists between ln( A) and ln . Appar-
ently, the original JC model cannot show accurately the material be-
havior at reference strain rate and temperature with a constant of n.
From the stress-strain curves (Fig. 7) at different strain rates, it can
be found that the stress slowly rises with the increasing strain, and the
hardening is weak. Low strain hardening is obvious, especially for high
strain rates. Hence a saturated model needs to be considered. A mod-
ified JC model is proposed to improve the prediction accuracy as shown
in equation (4). The strain hardening term (A + B n) in original JC
model is replaced by a mixture of swift model and modified voce model,
i.e., ak (A0 + )n + (1 a)[A1 + A2 [1 exp( A3 + A 4 2 )]]. A weight Fig. 13. The true stress-plastic strain curve at reference strain rate and tem-
coefficient a is used and given by: perature.

5
H. Wu, et al. Materials Science & Engineering A 761 (2019) 138067

Fig. 14. Relationship between /(ak (A0 + )n + (1 a)(A1 + A2 (1 exp


( A3 + A 4 2 )))) 1 and.ln( / 0)

Fig. 16. FE model of tensile test.

were simulated by using Abaqus/Explicit software through a VUHARD.


The JC failure model and damage evolution approach based on fracture
energy were used to delete failed elements. The simulated load-elon-
gation curves and failure strains were compared with the experimental
Fig. 15. Comparison of C values by experiment and simulation at different ones. In order to reduce computational time, a quarter model of tensile
strains and strain rates (blue data points and three dimentinal surface indicate sample was used with x-plane and z-plane as the symmetrical planes
experimental and simulated C values, respectively). (For interpretation of the and y-axis as tensile direction. The sample with 25088 eight-node linear
references to colour in this figure legend, the reader is referred to the Web brick elements is shown in Fig. 16. One end of specimen is under fixed
version of this article.)
constraint and the other end is tensile loaded with constant velocity.
The tensile velocities corresponding to strain rates of 0.1 s−1, 500 s−1
are plotted in three dimensional space as shown in Fig. 15. It is clear and 1000 s−1 are 0.13 mm/s, 500 mm/s and 1500 mm/s, respectively.
that equation (6) shows good prediction for the C value at different The load-elongation curves by simulation and experiment for 0.1
plastic strains and strain rates. s−1, 500 s−1 and 1000 s−1 are shown in Fig. 17. The simulated max-
1/2
imum load is consistent with the experimental one tested at 0.1 s−1 as
C = B0 + B1 + B2 + B3 [ln( / 0)]1/2 + B4 ln( / 0 ) + B5 [ln( / 0 )]2
shown in Fig. 17 (a). While tested at 500 s−1 and 1000 s−1, the si-
+ B6 ln( / 0 ) (6) mulated load-elongation curves are slightly higher than the experi-
mental ones. The maximum loads measured by simulation and experi-
The damage may initiate and develop during dynamic loading. The
ment are 10.30 kN and 10.05 kN when loaded at 500 s−1, respectively.
damage and fracture criterion of bonded Ti6Al4V alloy is based on
For strain rate of 1000 s−1, the maximum loads measured by simulation
Johnson-cook failure model, which is expressed as [17]:
and experiment are 10.41 kN and 10.14 kN, separately. The prediction
error is below 2.66%.
f = (D1 + D2 exp(D3 )) 1 + D4 ln Fig. 18 shows the equivalent plastic strain distributions at fracture
0 (7)
by simulation at three strain rates. The predicted fracture strains at 0.1
where εf is failure strain; D1, D2, D3, and D4 are material constants; is s−1, 500 s−1 and 1000 s−1 are 0.344, 0.308 and 0.271, respectively.
triaxiality ratio. The thermal softening effect is ignored. To determine They are close to those measured by tests, which are listed in Table 2.
D1, D2 and D3, the tensile tests on specimens with different notch radii The prediction error of fracture strain is below 3.64%. When the da-
were carried out at 0.0001 s−1 and 20 °C. was calculated by FE ana- mage parameter D = ( p/ f ) reaches unity, the failure elements are
lysis. Its average value is selected since varies during tensile loading. deleted, where p is the increment of equivalent plastic strain. The
D4 was determined by different failure strains at high strain rates. D1, maximum effective plastic strain occurs in the center of specimen,
D2, D3 and D4 are 0.05, 0.2097, 0.5116 and −0.0202, respectively. where the damage initiates.

4.2. Model verification 5. Discussion

To verify the modified JC model, tensile tests at different strain rates The strain rate hardening and thermal softening are two competitive

6
H. Wu, et al. Materials Science & Engineering A 761 (2019) 138067

The reason for the yield stress drop before the yield plateau ap-
pearance can be explained from the viewpoint of dislocation movement
as follows. Firstly, the relation between dislocation density , average
dislocation movement velocity v and plastic strain rate p is expressed
in equation (8). The average dislocation velocity v depends on applied
stress as shown in equation (9) [14]:
p = bv (8)

v m (9)

AP = s /l0 (10)
where b, l0, s and AP denote the burger vector, original gauge length,
speed constant and applied strain rate, respectively. m is a constant
depending on the crystal purity, temperature and dislocation type.
When an isolated condition is assumed, the sample with low dislocation
density is stretched at high constant speed. Then the stress rises sharply
in the elastic stage. However, the dislocations hardly move and p is
almost zero. The dislocations rapidly multiply near to the upper
yielding point. The dislocation movement means that the plastic de-
formation happens. The stress increase stops at the upper yielding
point, where p equals to AP . However, the dislocation density further
increases with increasing the strain. This can result in an unbalance
between p and AP , i.e., p > AP . Then the stress drops and the dis-
locations slow down to keep two strain rates match again. Conse-
quently, rapid dislocation multiplication at high strain rate contributes
to the stress drop.
Due to more uniform and equiaxed microstructure in comparison
with as-received alloy (Fig. 5), bonded alloy shows more obvious plastic
stage. During plastic deformation stage, the microvoids undergo nu-
cleation, growth and aggregation. Finally, the fracture occurs and the
dimples appear on the fracture surface. The dimple fracture is a high-
energy absorption process [2,20]. From the fracture surface observa-
tions for bonded alloy, the ductile fracture area accounts for a major
fraction at quasi-static and low strain rates. It means that more energy is
absorbed during dimple formation. Consequently, the samples loaded at
quasi-static and low strain rates have a better plasticity. The area de-
crease of plastic fibrous zone can explain the plasticity reduction with
increasing strain rate. Moreover, when loaded at high strain rate, local
void amount in white marked circle (Fig. 11) is significantly more than
that in other zones, which may result in earlier void coalescence.
Hence, the localized void distribution in dynamically loaded sample
leads to a low plasticity [15].
The modified JC model includes a saturated term which considers
the low strain hardening degree for bonded Ti6Al4V alloy. Moreover,
from the stress-strain curves and processing experiment data, the strain
rate hardening coefficient varies at different plastic strains and strain
rates. Thus C is expressed with a function of strain and strain rate. Since
the saturated term together with the coupling effects of strain rate and
strain on the flow behavior is considered, good agreements are
achieved for the load-elongation curves at three strain rates.
Furthermore, the material constant values of D1, D2, D3, and D4 for JC
failure model are reasonable, which are obtained through tensile tests
Fig. 17. Load-elongation curves by simulation and experiment for tensile tests on the specimens with three different notch radii and shear tests. The
at different strain rates: (a) 0.1 s−1, (b) 500 s−1, (c) 1000 s−1. predicted failure strains for all three strain rates also agree with ex-
perimental ones. The modified JC model and JC failure model have
mechanisms during tensile loading. Under the high strain rate condi- high prediction accuracy.
tion, the accumulated heat converted from approximately 90% plastic
deformation work cannot dissipate in a short time, which causes the 6. Conclusions
temperature increase and material softening [2]. When the strain rate
exceeds 500 s−1, the thermal softening effect strengthens. While the In this paper, the tensile tests of bonded Ti6Al4V alloy are con-
strain rate and strain hardening effects are overwhelmed. Apparently, ducted over a wide range of strain rates of 0.0001–1700 s−1. The main
the increase of strain rate enhances the yield and ultimate strengths, conclusions are drawn as below:
which results from the strain rate hardening. When loading at high
strain rate above 750 s−1, the thermal softening plays a dominant role (1) When testing at quasi-static and low strain rates, the yield and ul-
and impedes the dynamic strength improvement. Similar conclusion is timate strengths of bonded interfaces are slightly lower than those
also found in other literature [18,19]. of as-received alloy. A lower strain hardening and better plasticity

7
H. Wu, et al. Materials Science & Engineering A 761 (2019) 138067

Fig. 18. The predicted equivalent plastic strain distributions at fracture under different strain rates: (a) 0.1 s−1, (b) 500 s−1, (c) 1000 s−1.

are shown for bonded alloy, for which the DB interface shows References
ductile fracture.
(2) The strain-rate hardening can be obviously observed at strain rate [1] P. Verleysen, J. Peirs, Quasi-static and high strain rate fracture behavior of Ti6Al4V,
of below 1000 s−1. The yield and ultimate strengths of DB interface Int. J. Impact Eng. 108 (2017) 370–388.
[2] T.F. Zhou, J.J. Wu, J.T. Che, Y. Wang, X.B. Wang, Dynamic shear characteristics of
have little change with further increasing strain rate. The strain rate titanium alloy Ti-6Al-4V at large strain rates by the split Hopkinson pressure bar
sensitivity decreases with the increasing strain rate. A lower plas- test, Int. J. Impact Eng. 109 (2017) 167–177.
ticity exists under dynamic loading condition. A mixture of ductile [3] J.Q. Tan, M. Zhan, S. Liu, T. Huang, J. Guo, H. Yang, A modified Johnson-Cook
model for tensile flow behaviors of 7050-T7451 aluminum alloy at high strain rates,
and cleavage fracture occurs at higher strain rates. Mater. Sci. Eng. A 631 (2015) 214–219.
(3) A modified JC model is proposed to predict the dynamic flow be- [4] A.S. Khan, Y.S. Suh, R. Kazmi, Quasi-static and dynamic loading responses and
havior of bonded Ti6Al4V alloy. A saturated term is included in the constitutive modelling of titanium alloys, Int. J. Plast. 20 (2004) 2233–2248.
[5] W.S. Lee, C.F. Lin, Plastic deformation and fracture behavior of Ti6Al4V alloy
model. The strain rate hardening coefficient is a function of strain loaded with high strain rate under various temperatures, Mater. Sci. Eng. A 241
and strain rate. The modified JC model and JC failure model pro- (1998) 48–59.
vide good simulation results under different strain rates. The pre- [6] J. Peirs, P. Verleysen, W. Van Paepegem, J. Degrieck, Determining the stress-strain
behavior at large strains from high strain rate tensile and shear experiments, Int. J.
diction error of the maximum loads and fracture strains are below
Impact Eng. 38 (2011) 406–415.
2.66% and 3.64%, respectively. [7] A. Mohammadhosseini, S.H. Masood, D. Fraser, M. Jahedi, Dynamic compressive
behavior of Ti-6Al-4V alloy processed by electron beam melting under high strain
rate loading, Adv. Manuf. 3 (2015) 232–243.
Acknowledgments [8] G.R. Johnson, W.H. Cook, A constitutive model and data for metals subjected to
large strains, high strain rates and high temperatures, Proc. Seventh Int. Symp.
This work was supported by Aeronautical Science Foundation of Ballist. (1983) 541–547.
[9] H.W. Meyer, D.S. Kleponis, Modeling the high strain rate behavior of titanium
China (2016ZE57008), the SAST-SJTU fund (USCAST2016-20) and the undergoing ballistic impact and penetration, Int. J. Impact Eng. 26 (2001) 509–521.
National Natural Science Foundation of China (No. 51875350). We are [10] W.S. Lee, C.F. Lin, High-temperature deformation behavior of Ti6Al4V alloy eval-
grateful for these financial supports. uated by high strain-rate compression tests, J. Mater. Process. Technol. 75 (1998)

8
H. Wu, et al. Materials Science & Engineering A 761 (2019) 138067

127–136. [16] A.R. Khalifeh, A. Dehghan Banaraki, H. Danesh Manesh, M. Dehghan Banaraki,
[11] S. Seo, O. Min, H. Yang, Constitutive equation for Ti-6Al-4V at high temperatures Investigating of the tensile mechanical properties of structural steels at high strain
measured using the SHPB technique, Int. J. Impact Eng. 31 (2005) 735–754. rates, Mater. Sci. Eng. A 712 (2018) 232–239.
[12] A. He, G.L. Xie, H.L. Zhang, X.T. Wang, A comparative study on Johnson-Cook, [17] G.R. Johnson, W.H. Cook, Fracture characteristics of three metals subjected to
modified Johnson-Cook and Arrhenius-type constitutive models to predict the high various strains, strain rates, temperatures and pressures, Eng. Fract. Mech. 21
temperature flow stress in 20CrMo alloy steel, Mater. Des. 52 (2013) 677–685. (1985) 31–48.
[13] F. Ducobu, E. Rivière-Lorphèvre, E. Filippi, On the importance of the choice of the [18] S.S. Yu, Y.B. Lu, Y. Cai, The strain-rate effect of engineering materials and its unified
parameters of the Johnson-Cook constitutive model and their influence on the re- model, Lat. Am. J. Solid. Struct. 10 (2013) 833–844.
sults of a Ti6Al4V orthogonal cutting model, Int. J. Mech. Sci. 122 (2017) 143–155. [19] Z.G. Tu, Y. Lu, Evaluation of typical concrete material models used in hydrocodes
[14] D. Hull, D.J. Bacon, Introduction to Dislocations, Butterworth-heinemann, 2001. for high dynamic response simulations, Int. J. Impact Eng. 36 (2009) 132–146.
[15] Chandan Pandey, M.M. Mahapatra, Pradeep Kumar, N. Saini, Effect of strain rate [20] S. Rizal, H. Homma, Dimple fracture under short pulse loading, Int. J. Impact Eng.
and notch geometry on tensile properties and fracture mechanism of creep strength 24 (2000) 69–83.
enhanced ferritic P91 steel, J. Nucl. Mater. 498 (2018) 176–186.

You might also like