You are on page 1of 9

Materials Science and Engineering A 527 (2010) 52035211

Contents lists available at ScienceDirect

Materials Science and Engineering A


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

Low-temperature superplasticity and coarsening behavior of


Ti6Al2Sn4Zr2Mo0.1Si
Chan Hee Park a , Byounggab Lee a , S.L. Semiatin b , Chong Soo Lee a,
a
b

Department of Materials Science and Engineering, Pohang University of Science and Technology, San 31, Hyoja-dong, Nam-gu, Pohang, Gyungbuk 790-784, Republic of Korea
Air Force Research Laboratory, AFRL/RXLM, Wright-Patterson AFB, OH 45433-7817, USA

a r t i c l e

i n f o

Article history:
Received 6 January 2010
Received in revised form 14 April 2010
Accepted 27 April 2010

Keywords:
Titanium alloy
Superplasticity
Dynamic globularization
Coarsening
Deformation mechanism

a b s t r a c t
The low-temperature superplasticity and dynamic coarsening behavior of Ti6Al2Sn4Zr2Mo0.1Si
were established and interpreted in the context of inelastic-deformation theory. The starting microstructure with an equiaxed-alpha particle size of 13 m was rened to 2.2 m by a treatment comprising
beta annealing/water quenching followed by warm rolling at 775 C. A series of tension and loadrelaxation tests were carried out for the coarse and ne microstructures at strain rates of 104 to 102 s1
in the temperature range of 650750 C. The ne microstructure exhibited enhanced superplasticity
(382826% elongation) compared to that of the coarse microstructure (189286% elongation); this trend
was attributed to a larger fraction of boundary sliding and lower friction stress for the ner material.
With respect to microstructure evolution, the coarsening rate of the alpha particles during deformation
was 12 times faster than that during static coarsening. Furthermore, both the static and dynamic coarsening rates for the Ti6Al2Sn4Zr2Mo0.1Si were 2.74.7 times lower than those for the Ti6Al4V,
a trend attributable to the lower diffusivity of Mo compared to that of V at a given test temperature. The
plastic ow behavior of the coarse and ne microstructures was rationalized in terms of microstructure
evolution during deformation.
2010 Elsevier B.V. All rights reserved.

1. Introduction
Alpha/beta titanium alloys are widely used for aerospace applications due to their high specic strength, excellent corrosion
resistance, and high-temperature properties [1]. For example,
Ti6Al2Sn4Zr2Mo0.1Si (Ti6242Si) has been used for jet engine
compressor disks because of its superior creep resistance and
higher service temperature (550 C) compared to the more common Ti6Al4V alloy [1,2].
With the application of net-shape processes such as isothermal
forging of billet and superplastic forming of sheet, cost savings
associated with improved material utilization can be achieved.
Despite this commercial benet, the widespread application of
such processes may be limited by processing requirements such
as high temperature (900 C) and/or low strain rate (103 s1 )
[3,4]. On the other hand, enhanced die-ll capacity/superplasticity
can be realized at lower temperatures (750 C) and/or higher
strain rates (102 s1 ) by rening the microstructure via severe
plastic deformation (SPD) methods such as high pressure torsion
(HPT), equal-channel angular extrusion (ECAE), and accumulative roll bonding (ARB) [59]. Other efforts have been made to

Corresponding author. Tel.: +82 54 279 2141; fax: +82 54 279 2399.
E-mail address: cslee@postech.ac.kr (C.S. Lee).
0921-5093/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2010.04.082

achieve enhanced superplasticity utilizing dynamic globularization of a ne lamellar microstructure, i.e., the conversion of a
transformed-beta microstructure to a ne equiaxed microstructure
[1013].
To maintain superplastic characteristics to large strains,
microstructural coarsening must be limited during deformation. In this regard, the static and dynamic coarsening
behaviors of Ti6Al4V with an ultrane-grain (UFG) microstructure have recently been interpreted in terms of a modied
LifshitzSlyozovWagner theory by Semiatin et al. [14,15]. In
related work, Sargent et al. [16] have emphasized the importance
of the alpha particle size rather than the size of alpha subgrains or
beta grains to describe the superplastic ow of the UFG Ti6Al4V.
Although this previous research provides insight into the interaction of microstructure evolution and plastic ow in alpha/beta
titanium alloys, it is unclear how solutes other than vanadium
and aluminum may affect such phenomena. The objectives of the
present work, therefore, were to demonstrate a processing method
to achieve enhanced superplasticity in another alloy in this class,
i.e., Ti6242Si, and to establish the effect of a different solute, molybdenum, whose diffusivity is noticeably slower than vanadium, on
microstructure evolution. For this purpose uniaxial and stressrelaxation tests were conducted on sheet samples. The plastic ow
observations were analyzed in the context of inelastic-deformation
theory to determine the relative contributions of grain-matrix

5204

C.H. Park et al. / Materials Science and Engineering A 527 (2010) 52035211

deformation (GMD) and grain-boundary sliding (GBS) to the overall


behavior.
2. Internal-variable theory
The rheological model for grain-matrix deformation (GMD) and
grain-boundary sliding (GBS) is shown schematically in Fig. 1 [17].
The model demonstrates that GBS is accommodated primarily by
dislocation motion giving rise to an internal strain (a) and a nonrecoverable plastic strain (). From the model shown in Fig. 1,
relations for stress and strain rate are obtained as follows:
 = I + F

(1)

= a + + g

(2)

The variables  I and  F represent the internal stress due to a longrange interaction force among dislocations and the friction stress
due to a short-range interaction between the lattice and a dislocation, respectively; a denotes the co-rotational material time rate of
an internal strain tensor, while and g are the plastic strain rate
due to GMD and GBS, respectively. At elevated temperature,  F is
generally very small compared to  I , and a is negligible for loadrelaxation tests under a steady-state condition, thus providing a
relatively simple constitutive equation consisting of only the and
g elements. Thus, two separate constitutive equations describing
GMD and GBS can be expressed in the following form similar to
that suggested by Lee and Hart [18].

  
I

 g 
g 0

= exp

 p


=

(3)

1/Mg

temperature: 1005 C) and water quenched to obtain a martensitic microstructure with laths 0.2-m thick (Fig. 2b). The
microstructure was converted to a ne, equiaxed one via dynamic
globularization [10,1923] during warm cross-rolling at 775 C
using 10% reduction per pass to a total reduction of 80%. The rolling
process comprised two passes with a rolling direction parallel to
the length of the preform and the remaining passes parallel to the
width. The rolled sheet was water quenched following the last pass.
By this means, Ti6242Si with ne alpha particles (approximately
one-half of which were still somewhat elongated) that were separated by thin layers of the beta phase were produced (Fig. 2c). An
electron-backscatter-diffraction (EBSD) inverse-pole-gure map
for the alpha phase (showing boundaries with misorientation angle
15 ) revealed that the particle size was 2.2 m (Fig. 2d).
The low-temperature superplasticity and dynamic coarsening
behavior of Ti6242Si with the coarse and ne microstructures were
quantied via uniaxial-tension and load-relaxation tests. Tension
specimens with a gauge length of 5 mm and width of 2.5 mm
were machined parallel to the nal rolling direction of the plate.
To prevent oxidation of the surface (which can have a deleterious effect on superplasticity [24]), each specimen was coated
with Deltaglaze 151 (Acheson Colloids, Port Huron, MI). They
were tested under constant-crosshead-speed conditions using an
INSTRON 1361 screw-driven machine at homologous temperatures of 0.470.53 (650750 C) and nominal strain rates of 104
to 102 s1 . Load-relaxation tests were performed at 650750 C
using specimens with a gauge length of 25 mm. The load-time
data from the load-relaxation tests were converted to stress-vsstrain-rate plots using the relations suggested by Lee and Hart
[18], i.e.,

(4)

refer to the strength parameter and its conjuHere, * and


gate reference strain rate, respectively, for GMD, and g and g 0
represent the friction stress and its conjugate reference strain
rate, respectively, for GBS. The exponents p and Mg are the material parameters representing the characteristics of GMD and GBS,
respectively.
3. Materials and experimental procedures
The Ti6242Si alloy used in this study was supplied by RMI Titanium (Niles, OH) in the form of a disk (i.e., a billet slice) with a
diameter of 155 mm and thickness of 18 mm. Its measured chemical
composition (in wt%) was 5.55 aluminum, 1.86 tin, 4.03 zirconium,
1.90 molybdenum, 0.08 silicon, 0.06 iron, 0.13 oxygen, 0.01 carbon, 0.008 nitrogen, balance titanium. The microstructure of the
as-received material revealed a relatively coarse alpha particle size
of 13 m (Fig. 2a).
To reduce the alpha particle size, a section of the program
material measuring 100 mm width 80 mm length 18 mm thickness was beta solution treated for 1 h at 1050 C (beta-transus

Fig. 1. Rheological description of an internal-variable model consisting of grainmatrix deformation (GMD) and grain-boundary sliding (GBS).

=

P{l0 + (x P/C)}
A0 l0

P
C{l0 + (x P/C)}

(5)

(6)

in which P and P denote the load and its time derivative, respectively, A0 and l0 are the cross-sectional area and the length of the
gauge section, respectively, x is the crosshead displacement, and
C is the machine stiffness (P/x for high stiffness machine such as
INSTRON 1361).
To compare the dynamic and static coarsening behaviors of the
present material, static heat treatments were also conducted at
750 C for 6, 24, 48 and 70 h followed by water quenching.
Microstructures were examined via optical metallography
(OM), scanning electron microscopy (SEM), transmission electron
microscopy (TEM), and EBSD analysis. For OM and SEM, Krolls
reagent (5% HNO3 , 10% HF, and 85% H2 O) was used to etch specimens. Backscattered electron images (BSEI) were taken in a JEOL
high resolution SEM, JSM-7401F. Thin foils for TEM were prepared by twin-jet electropolishing at 40 C using a mixture of
5% H2 SO4 and 95% CH3 OH. TEM micrographs were obtained by
utilizing a Cs corrected HR-STEM, JF-2200 (JEOL) operating at an
accelerating voltage of 200 kV. To measure the alpha particle size,
EBSD maps for the alpha phase were constructed using TSL software with a tolerance angle of 15 . Initially, the EBSD specimens
were electropolished using a LectroPol-5 (STRUERS) operated at a
voltage of 22 V for 25 s in a solution of 410 ml methanol, 245 ml
2-butoxy ethanol, and 40 ml 60%-HClO4 . EBSD was performed in
a Helios NanoLabTM 600 scanning electron microscope with an
accelerating voltage of 30 kV and a step size of 0.08 m. The condence index of the EBSD orientation measurements was between
0.29 and 0.64.

C.H. Park et al. / Materials Science and Engineering A 527 (2010) 52035211

5205

Fig. 2. Micrographs of the Ti6242Si program material: (a) as-received microstructure (OM), (b) martensitic structure after beta annealing and water quenching (TEM), (c)
dynamically globularized microstructure (backscattered electron image; the dark phase is alpha and the lighter phase is beta), and (d) dynamically globularized microstructure
(EBSD map for the alpha phase using a tolerance angle of 15 ).

4. Results
4.1. Uniaxial-tension ow behavior and elongation to failure
True stress-vs-strain curves for the CP (coarse-particle) and
FP (ne-particle) Ti6242Si materials (e.g., results for 750 C in
Fig. 3a and b) exhibited noticeable differences in terms of their
strain-rate and temperature dependences. The curves denoted
with open symbols in Fig. 3 comprise those determined directly

from the load-stroke measurements obtained during the constantcrosshead-speed tests. A more appropriate comparison of true
material behavior was obtained by estimating the constantstrain-rate ow response by compensating for the continuously
decreasing strain rate imposed during the tension tests. For this
v)m in which
purpose, each stress data point was multiplied by (h/
is the nominal strain rate, h is the instantaneous length of the

reduced section, v is the crosshead speed, and m (= d log /d log )


is the strain-rate sensitivity. In the present work, m was esti-

Fig. 3. True stress-vs-true strain curves measured at 750 C and various strain rates for (a) CP (13 (m) and (b) FP (2.2 (m) Ti6242Si specimens. The strain-rate dependence
of the strain-rate sensitivity (m) determined from load-relaxation tests is shown in (c).

5206

C.H. Park et al. / Materials Science and Engineering A 527 (2010) 52035211

Fig. 4. Macrographs of fractured tension specimens of CP and FP Ti6242Si tested at


750 C and various strain rates.

mated from the load-relaxation data, whose strain rates span those
imposed during the uniaxial-tension tests (Fig. 3c). The corrected
(constant-strain-rate) true stress-vs-strain curves for the CP and FP
Ti6242Si materials (closed symbols in Fig. 3a and b) also showed
a noticeable dependence on strain rate. For example, the peak
stresses of the CP specimens were considerably higher than those
of the FP specimens at the same test condition, and the ow stresses
of both materials decreased with decreasing strain rate from 102
to 104 s1 . Furthermore, it is of interest to note that the FP material
exhibited considerable ow hardening at a strain rate of 104 s1 ;
this latter observation behavior will be discussed in the context of
the corresponding microstructural evolution in Section 5.3.
The elongations of the CP specimens (81286%) were considerably less than those of the FP samples (382826%) tested at the
same strain rate and temperature (Fig. 4). For instance, the elongation (expressed as a true strain) of the CP sample deformed at 750 C
and 104 s1 was 1.3, while that of the corresponding FP specimen
was 2.3. Fig. 5a summarizes the total elongations for the CP and
FP specimens tested at various temperatures and strain rates as
well as previously reported results obtained under conventional
(higher-temperature) superplastic conditions [4]. Two aspects can
be noted in this plot. First, the FP material exhibited approximately

Fig. 5. Elongation to failure of CP and FP Ti6242Si specimens tested at temperatures


between 650 and 750 C.

2.5 times larger total elongations compared to those for CP material in the temperature range of 650750 C. Second, despite the
difference in phase volume fractions, the total elongations for FP
specimens deformed at 750 C were similar to those for the conventional material tested at the much higher temperature of 940 C
[4]. The high elongation of the FP Ti6242Si can be attributed to
the operation of phase-boundary sliding even at 750 C, as will be
discussed in Section 5.1. Such a conclusion is supported by values
of the strain-rate sensitivity (m) of the CP and FP materials estimated from load-relaxation tests. The FP microstructure showed
substantially higher ms at both test temperatures and each strain
rate (Fig. 3c). Because maximum elongation is usually found for
test conditions having the highest m value, it is not surprising to
observe the largest total elongation for the FP microstructure for
tests at 750 C, 104 s1 , i.e., 826%.
4.2. Load-relaxation behavior
The stress-vs-strain-rate data obtained from the load-relaxation
tests (data points, Fig. 6) were used to quantify and interpret

Fig. 6. Comparison of experimental ow stress-vs-strain-rate data obtained by load-relaxation tests () and predicted curves based on the internal-variable theory of
inelastic deformation (lines). The alpha particle sizes and test temperatures were (a) 13 (m and 650 C, (b) 13 (m and 750 C, (c) 2.2 (m and 650 C, and (d) 2.2 (m and 750 C,
respectively.

C.H. Park et al. / Materials Science and Engineering A 527 (2010) 52035211

5207

Table 1
Constitutive parameters for CP and FP Ti6Al2Sn4Zr2Mo0.1Si materials determined from load-relaxation tests.
Material

Temperature ( C)

GMD parameters

GBS parameters

log *

log

log g

log g 0

Mg

CP (d = 13 m)

650
750

3.01
2.91

4.46
2.01

0.15
0.15

1.99
1.04

7.02
6.30

0.5
0.5

FP (d = 2.2 m)

650
750

3.32
3.01

1.72
1.30

0.15
0.15

1.38
0.85

6.25
6.21

0.5
0.5

the relative contributions of grain-matrix deformation (GMD) and


grain-boundary sliding (GBS) during the low-temperature plastic
ow of both the CP and FP Ti6242Si. With increasing temperature, the curves tended to shift toward a higher-strain-rate and
lower-stress region. Although both CP and FP specimens exhibited
similar behavior, the ow stress for FP material decreased more
steeply with decreasing strain rate, thus implying higher-strainrate sensitivity in comparison to that for CP specimens at the same
temperature.
Based on the internal-variable theory of inelastic deformation,
predictions of the stress-vs-strain-rate response assuming GBS,
GMD, and GBS + GMD were plotted and compared to the measurements in Fig. 6. The constitutive (tting) parameters used in the
present work are listed in Table 1. The shapes of all of the tted
lines for the CP specimen (Fig. 6a and b) were noticeably different
from those for the FP specimen (Fig. 6c and d). The experimental data for the CP specimens at both 650 and 750 C were well
t by assuming only GMD was activated (Eq. (3)). On the other
hand, the experimental results for FP specimens were not t well
by the predicted trend based solely on GMD. Differences between
the measurements and the tted lines for GMD were most noticeable at intermediate strain rates (105 s1 103 s1 ) (Fig. 6c
and d). In this strain-rate range, a straight line or a slightly concave
downward curve was observed, implying that another deformation
mode (in addition to GMD) may have also been operative. In this
regard, previous research on other metallic systems has suggested
that such curvature is mainly due to the operation of GBS [17,25].
Therefore, it is reasonable to assume that GBS is also the dominant
mode for the FP specimens in the present work.
4.3. Coarsening behavior
Typical microstructures of tension specimens deformed at
750 C, 104 s1 for four different times are shown in Fig. 7ad.
These microstructures indicate signicant coarsening of the size of
the alpha particles compared to the initial microstructure (Fig. 2c
and d). Specically, the initial alpha particle size (2.2 m) increased
to 4.0, 5.1, 5.6 and 6.3 m after 5, 10, 15 and 21 h, respectively,
under dynamic test conditions. By contrast, during static annealing
at the same temperature, only moderate coarsening of the alpha
phase was observed (Fig. 7eh), i.e., particle sizes of 2.6, 3.1, 3.7
and 4.2 m were developed after 6, 24, 48 and 70 h, respectively.
5. Discussion
5.1. Quantitative analysis of deformation mechanisms
More detailed examination of the load-relaxation behavior provides insight into the reason for enhanced superplasticity in the
ne, dynamically globularized microstructure in comparison to the
behavior for the conventional (coarser) material. Specically, the
results in Table 1 reveal that the value of Mg (the material parameter representing the characteristics of GBS) is identical (i.e., 0.5)
for both the CP and FP Ti6242Si materials. This is in accord with
the earlier efforts reporting values of Mg for two-phase and quasi-

single-phase alloys as being 0.5 and 1.0, respectively [9,17,26]. In


addition, Table 1 shows that the strength parameter * increased
with decreasing particle size at the same temperature. More importantly, the values of g (the friction stress for GBS) are lower in FP
specimens than in CP specimens for a given temperature. This indicates that the friction stress decreased with decreasing particle size
in superplastic region (intermediate strain-rate conditions showing higher elongations and m values as compared to high and low
strain rate conditions) and that GBS can operate more readily for
material with the FP microstructure.
Internal-variable model estimates of the relative contribution of
GMD and GBS to the overall strain/strain rate are shown in Fig. 8.
For both the CP and FP specimens, the GBS fraction increased with
temperature from 650 to 750 C. Moreover, the fraction of strain
accommodated by GBS in FP specimens was 3550%, while it was
only 220% in CP specimens. These predictions of the fraction
of GBS provide at least a qualitative explanation for the observed
trends in m values and total elongation.
5.2. Dynamic and static coarsening kinetics
The effect of different solutes on the coarsening kinetics of
alpha/beta titanium alloys is of great interest. The coarsening of a
precipitate dispersion of innitesimal volume fraction was treated
initially in the classical work of Lifshitz, Slyozov, and Wagner
[2729]. The LifshitzSlyozovWagner theory yields the following
relation:
n
r n r 0
= K(t t0 )

(7)

in which r and r 0 denote the average particle radius at time (t)


and the initial time (t0 ), respectively, K is the rate constant (dependent on phase composition, interface energy, etc.), and n is the
coarsening exponent indicating the relevant mechanism (n = 2 for
interface-reaction-controlled kinetics and n = 3 for bulk-diffusioncontrolled behavior). Recently, it was shown that Eq. (7) with
n = 3 describes well the coarsening kinetics of Ti6Al4V [14,15].
In these past efforts, reasonable predictions of the dynamic and
static rate constants were made using suitable values of the input
parameters and an adjustment for the nite volume fraction of
alpha at processing temperatures in the hot- and warm-working
regimes.
Fig. 9 shows the present coarsening data for Ti6242Si tted with
different values of n (n = 2, 3, and 4). The best t corresponds to n = 3
for both dynamic and static conditions. Thus, the coarsening of the
alpha particles in Ti6242Si is most likely controlled by diffusion
through the beta matrix. The present results are consistent with
the earlier observations for Ti6Al4V with a UFG microstructure
[1416], for which the growth kinetics of alpha phase were also
controlled by bulk diffusion.
3 vs t t plots in Fig. 9)
The values of K (the slope of r 3 r 0
0
are summarized in Table 2 for both Ti6242Si and Ti6Al4V. The
dynamic coarsening rate for Ti6242Si (1.43 m3 /h) was approximately 12 times higher than the static value (0.12 m3 /h). In light
of the previous results and interpretation for Ti6Al4V [15,16],
this increase can be rationalized on the basis of enhanced diffusion

5208

C.H. Park et al. / Materials Science and Engineering A 527 (2010) 52035211

Fig. 7. EBSD inverse-pole-gure maps (tolerance angle = 15 ) for FP specimens deformed at 750 C and 104 s1 for (a) 5 h, (b) 10 h, (c) 15 h and (d) 21 h. The four EBSD maps
at the bottom are for FP specimens statically annealed at 750 C for (e) 6 h, (f) 24 h, (g) 48 h and (h) 70 h, respectively.

associated with the generation of a dislocation substructure in the


beta phase during deformation.
As shown in Table 2, the coarsening-rate constants K of
Ti6242Si are 2.74.7 times lower than those of Ti6Al4V at similar temperatures in the range 750775 C [30]. In this regard, the

diffusion-controlled coarsening and growth of alpha particles in


alpha/beta titanium alloys have been shown to be controlled by the
properties of the solute with the lowest diffusivity [1416,31], i.e.,
Mo and V solutes in Ti6242Si and Ti6Al4V, respectively. The analysis can be made more quantitative using the following relationship

Table 2
Comparison of the values of the coarsening-rate constant K for alpha particles in Ti6Al2Sn4Zr2Mo0.1Si and Ti6Al4V.
Material

Temperature ( C)

Static K (m3 /h)

Dynamic K (m3 /h)

D of rate-limiting solute (m2 /s)

Ti6242Si
Ti64

750
775

0.12
0.56 [30]

1.43
3.9 [30]

0.72 103 , D of Mo from Eq. (9)


4.5 103 , D of V from Eq. (10)

C.H. Park et al. / Materials Science and Engineering A 527 (2010) 52035211

5209

rection factors, the difference in the coarsening rates is most likely


due to differences in the diffusivity of the rate-limiting solute (Mo in
Ti6242Si and V in Ti6Al4V) in the beta matrix. Previous diffusioncouple measurements [31,34] provide the following dependence of
diffusivity on temperature for Mo and V in beta titanium:

DMo (m2 /s) = 52, 500 exp

DV (m2 /s) = 77, 000 exp

Fig. 8. Relative contribution of GMD and GBS to the overall deformation for CP and
FP Ti6242Si at temperatures of 650 and 750 C and a strain rate of 104 s1 .

for the rate constant K in Eq. (7) [29,32,33]:


K=

8{f ()D C (1 C )VM }


9{RT (C C )2 (1 + ln r/ ln C )}

(8)

Here, f() denotes the functional dependence of the rate constant


on the volume fraction of the alpha phase, D is the diffusivity of
the rate-limiting solute in the beta matrix,  is the alphabeta
interface energy, C and C are the equilibrium concentrations (in
atomic-fraction terms) in the beta and alpha phase, respectively, of
the rate-limiting solute, VM is molar volume of the alpha phase, R is
the gas constant, T is the absolute temperature, and r is the activity
coefcient of the rate-limiting solute in the beta matrix.
According to earlier research, the alpha volume fractions in
Ti6Al4V and Ti6242Si are similar (8090%) at T 750775 C [4],
thus suggesting that the observed differences in the coarsening rate
of the two alloys are not associated with a volume-fraction effect.
In addition, the concentration factor, C (1 C )/(C C )2 , of the
present alloy (11.40) is similar to that of Ti6Al4V (i.e., 10.07)
at 730 C [16,31]. Because the molar volume is similar in the two
alloys and assuming little difference in the thermodynamic cor-

 18, 520 
T (K)

 17, 460 
T (K)

(9)

(10)

Eqs. (9) and (10) indicate that the diffusivity of the Mo solute in the
beta matrix for the Ti6242Si is approximately one-sixth that of V in
Ti6Al4V alloy at the same temperature (Table 2). Thus, the lower
dynamic and static coarsening rates of the Ti6242Si compared to
those of Ti6Al4V can be explained at least qualitatively to be a
result of differences in the diffusivity of the rate-limiting solute.
5.3. Effect of microstructure evolution on plastic ow
The 104 s1 true stresstrue strain curves adjusted for a constant strain rate (closed symbols in Fig. 3a and b) exhibited near
steady-state ow for CP samples and ow hardening for FP specimens. These observations can be rationalized in the context
of recent research on the superplastic ow behavior of twophase Ti alloys [30]. In this former work, the instantaneous ow
response was related to the evolving microstructure using the
BirdMukherjeeDorn generalized constitutive equation [35], viz.,
=

 ADGb    n  b p
kT

(11)

in which A is a constant, D is a diffusivity, G is the shear modulus, b


is the burgers vector, k is Boltzmanns constant, T is absolute temperature,  is the ow stress, n (=1/m) is the stress exponent, d is
the alpha particle size, and p is the grain size exponent. At a specic temperature and strain rate, Eq. (11) can be simplied to the

Fig. 9. The coarsening behavior of FP specimen under dynamic ( = 104 s1 ) and static conditions at 750 C. The experimental data were t with n equal to (a) 2, (b) 3,
or (c) 4.

5210

C.H. Park et al. / Materials Science and Engineering A 527 (2010) 52035211

following relationship:

=
0

 d p/n

1. The as-received microstructure with an equiaxed-alpha particle


size of 13 m (CP) can be rened to 2.2 m (FP) by warm
working of a martensitic microstructure at 775 C.
2. Due to its ner microstructure and higher m values, the FP specimens have considerably higher total elongations compared to
those of CP specimens at the same temperature in the range of
650750 C.
3. Internal-variable theory reveals that the enhancement of superplasticity in the rened microstructure is attributable to a higher
fraction (3550%) of GBS and lower friction stress ( g ) compared to the coarse microstructure.
4. The dynamic coarsening of the alpha phase in Ti6242Si is
12 times faster than that of static coarsening. Both static
and dynamic coarsening kinetics are controlled by bulk diffusion. The coarsening rate of the Ti6242Si is much lower (24
times) than that of the Ti6Al4V at the same test temperature primarily due to the lower diffusivity of the rate-limiting
solute.
5. The ow behavior of FP specimens shows a noticeable dependence on strain rate at 750 C due to the competition between
dynamic spheroidization and dynamic coarsening. Flow hardening due to dynamic coarsening dominates at 104 s1 while
ow softening due to dynamic spheroidization (and deformation
heating) prevails at 103 and 102 s1 .

(12)

d0

in which  0 and d0 denote the initial ow stress and alpha particle


size, respectively.
For the CP specimen deformed to failure at 750 C, 104 s1 (time
7.4 h), the alpha particle diameter at the end of deformation was
estimated to be 13.2 m using Eq. (7) (with K = 1.43 m3 /h, per
Table 2, and r0 = 6.5 m). The measured alpha particle diameter
after deformation at the same condition was about 13.4 m, close
to the estimated one. Taking the values of p and n as 2 [36] and 2.94
(Fig. 3c), respectively, and inserting these into Eq. (12), the ow
stress at end of the deformation was estimated to be almost the
same as the initial ow stress (i.e., / 0 1.01). Thus, the observation of near steady-state ow for FP samples deformed at 104 s1
appears as expected.
In comparison to the CP behavior, the ow behavior of the FP
specimens was more complex. Because approximately one-half
of the alpha particles were elongated (aspect ratio 1:2) along
the transverse direction in the initial condition (Fig. 2c and d),
ow softening due to dynamic spheroidization occurs simultaneously with ow hardening due to dynamic coarsening for the FP
material. Assuming that mass transfer occurring during dynamic
spheroidization is controlled by volume diffusion, the time ( vd )
to complete dynamic spheroidization can be estimated from the
following equation [37]:

3 [0.3287/3 (1 + 1 0.7634/3 ) ]
vd
=

4[((2(1 + ))/(3(0.5 0.5731/3 )) + (0.51/3 + 0.6652/3 )/(3(0.143 + 0.9341/3 ))]
=
 =

w
u

+ 0.5

u3 RT
D CF  VM

(13a)

(13b)

(13c)

In Eqs. (13a)(13c), CF is a concentration factor (=11.40 at 730 C


[31]),  is the alpha/beta interface energy (=0.4 J/m2 [37]), and
VM is the molar volume (=1.044 105 m3 /mol [37]). For the elongated particles in the FP specimen, the measured average lengths of
major (w) and minor (u) axes are 3.1 and 1.7 m, respectively. The
diffusion coefcient under dynamic condition (D ) is 10 times
higher than that under static conditions (as discussed in Section
5.2); hence, its value was set as 0.0072 m2 /s.
From Eqs. (13a)(13c), the estimated time for dynamic
spheroidization of the FP microstructure at 750 C is 1.24 h, which
corresponds to a true strain of 0.4 at 104 s1 . This result indicates
that the mitigation of ow hardening by dynamic spheroidization/ow softening would be signicant only during the initial
stage of deformation (i.e., strains 0.4). At strains greater than 0.4,
hardening due to dynamic coarsening prevails at 104 s1 , as was
observed.
For deformation at a strain rate of 103 or 102 s1 , the time
for dynamic spheroidization is close to or much less than that for
failure, respectively, resulting in ow softening, as observed. Deformation heating would also tend to lead to an increment in ow
softening at these higher strain rates, especially at 102 s1 .
6. Summary
The low-temperature superplasticity and coarsening kinetics of
Ti6242Si with a ne microstructure developed by beta annealing
and warm working were investigated. The following conclusions
were drawn from this work:

Acknowledgement
This work was supported by a grant from the US Air Force Ofce
of Scientic Research and its Asian Ofce of Aerospace Research
and Development (Dr. J.P. Singh, program manager).
References
[1] R. Boyer, G. Welsch, E.W. Collings, Materials Properties Handbook, Titanium
Alloys, ASM International, Materials Park, OH, 1994, pp. 453748.
[2] J.C. Williams, E.A. Starke, Acta Mater. 51 (2003) 57755799.
[3] J.W. Edington, K.N. Melton, C.P. Cutler, Prog. Mater. Sci. 21 (1976) 61170.
[4] M.T. Cope, D.R. Evetts, N. Ridley, J. Mater. Sci. 21 (1986) 40034008.
[5] R.Z. Valiev, T.G. Langdon, Prog. Mater. Sci. 51 (2006) 881981.
[6] S.C. Yoon, A.V. Nagasekhar, H.S. Kim, Met. Mater. Int. 15 (2009) 215219.
[7] G.A. Salishchev, R.M. Galeyev, O.R. Valiakhmetov, R.V. Saullin, R.Y. Lutfullin,
O.N. Senkov, F.H. Froes, O.A. Kaibyshev, J. Mater. Process. Technol. 116 (2001)
265268.
[8] Y. Saito, H. Utsunomiya, N. Tsuji, T. Sakai, Acta Mater. 47 (1999) 579583.
[9] Y.G. Ko, C.S. Lee, D.H. Shin, S.L. Semiatin, Metall. Mater. Trans. 37A (2006)
381391.
[10] C.H. Park, Y.G. Ko, J.-W. Park, C.S. Lee, Mater. Sci. Eng. A 496 (2008) 150158.
[11] J.H. Kim, N.S. Reddy, J.T. Yeom, J.K. Hong, C.S. Lee, N.-K. Park, Met. Mater. Int. 15
(2009) 427437.
[12] H Inagaki, Z. Metallkd. 86 (1995) 643650.
[13] S.V. Zherebstov, G.A. Salishchev, R.M. Galeyev, O.R. Valiakhmetov, S.L. Semiatin,
in: M.J. Zehetbauer, R.Z. Valiev (Eds.), Proc. Second International Conference
on Nanomaterials by Severe Plastic Deformation: Fundamentals-ProcessingApplications, Wiley-VCH, Weinheim, Germany, 2004, pp. 835840.
[14] S.L. Semiatin, B.C. Kirby, G.A. Salishchev, Metall. Mater. Trans. 35A (2004)
28092819.
[15] S.L. Semiatin, M.W. Corbett, P.N. Fagin, G.A. Salishchev, C.S. Lee, Metall. Mater.
Trans. 37A (2006) 11251136.
[16] G.A. Sargent, A.P. Zane, P.N. Fagin, A.K. Ghosh, S.L. Semiatin, Metall. Mater. Trans.
39A (2008) 29492964.
[17] T.K. Ha, Y.W. Chang, Acta Mater. 46 (1998) 27412749.
[18] D. Lee, E.W. Hart, Metall. Trans. 2 (1971) 12451248.

C.H. Park et al. / Materials Science and Engineering A 527 (2010) 52035211
[19] S.L. Semiatin, V. Seetharaman, I. Weiss, Mater. Sci. Eng. A 263 (1999)
257271.
[20] T. Seshacharyulu, S.C. Medeiros, W.G. Frazier, Y.V.R.K. Prasad, Mater. Sci. Eng.
A325 (2002) 112125.
[21] E.B. Shell, S.L. Semiatin, Metall. Mater. Trans. 30A (1999) 32193229.
[22] J.H. Kim, S.L. Semiatin, C.S. Lee, Acta Mater. 51 (2003) 56135626.
[23] C.H. Park, K.-T. Park, D.H. Shin, C.S. Lee, Mater. Trans. 49 (2008) 2196
2200.
[24] S.N. Patankar, Y.T. Kwang, T.M. Jen, J. Mater. Process. Technol. 112 (2001) 2428.
[25] J.S. Kim, Y.W. Chang, C.S. Lee, Metall. Mater. Trans. 29A (1998) 217226.
[26] J.E. Park, S.L. Semiatin, C.S. Lee, Y.W. Chang, Mater. Sci. Eng. A410A411 (2005)
124129.
[27] I.M. Lifshitz, V.V. Slyozov, J. Phys. Chem. Solids 19 (1961) 3550.
[28] C. Wagner, Z. Elektrochem. 65 (1961) 581594.
[29] J.W. Martin, R.D. Doherty, B. Cantor, Stability of Microstructure in Metallic
Systems, second ed., Cambridge University Press, Cambridge, 1997.

5211

[30] S.L. Semiatin, P.N. Fagin, J.F. Betten, A. Zane, A.K. Ghosh, G.A. Sargent, Metall.
Mater. Trans. 41A (2010) 499512.
[31] S.L. Semiatin, T.M. Lehner, J.D. Miller, R.D. Doherty, D.U. Furrer, Metall. Mater.
Trans. 38A (2007) 910921.
[32] H.A. Calderon, P.W. Voorhees, J.L. Murray, G. Kostorz, Acta Metall. Mater. 42
(1994) 9911000.
[33] R.D. Doherty, in: R.W. Cahn, P. Haasen (Eds.), Physical Metallurgy, NorthHolland, Amsterdam, 1996, pp. 13631506.
[34] S.L. Semiatin, T.M. Brown, T.A. Goff, P.N. Fagin, D.R. Barker, R.E. Turner, J.M.
Murry, J.D. Miller, F. Zhang, Metall. Mater. Trans. 35A (2004) 30153018.
[35] J.E. Bird, A.K. Mukherjee, J.E. Dorn, in: D.G. Brandon, A. Rosen (Eds.), Quantitative Relation Between Microstructure and Properties, Israel Universities Press,
Jerusalem, Israel, 1969, pp. 255342.
[36] C.H. Park, B. Lee, C.S. Lee, Trans. Mater. Process. 18 (2009) 544549.
[37] S.L. Semiatin, N. Stefansson, R.D. Doherty, Metall. Mater. Trans. 36A (2005)
13721376.

You might also like