See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/226229199

Prediction of hardness minimum locations during natural aging in an

aluminum alloy 6061-T6 friction stir weld

Article in Journal of Materials Science · December 2009

DOI: 10.1007/s10853-009-3868-y

CITATIONS READS

31 808

4 authors:

Wanchuck Woo Hoyoung Choo

Korea Atomic Energy Research Institute (KAERI) University College London
156 PUBLICATIONS 3,348 CITATIONS 224 PUBLICATIONS 5,755 CITATIONS

SEE PROFILE SEE PROFILE

Philip J. Withers Zhiqian Feng

The University of Manchester NEOMA Business School
974 PUBLICATIONS 31,748 CITATIONS 41 PUBLICATIONS 1,358 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

3D characterisation View project

Multimodal Correlative Characterization of Damage in Materials View project

All content following this page was uploaded by Wanchuck Woo on 15 May 2014.

The user has requested enhancement of the downloaded file.

J Mater Sci (2009) 44:6302–6309
DOI 10.1007/s10853-009-3868-y

Prediction of hardness minimum locations during natural aging

in an aluminum alloy 6061-T6 friction stir weld
W. Woo Æ H. Choo Æ P. J. Withers Æ Z. Feng

Received: 6 July 2009 / Accepted: 3 September 2009 / Published online: 11 September 2009
Ó Springer Science+Business Media, LLC 2009

Abstract This study describes a method that can predict Introduction

the hardness minimum location as a function of natural
aging time in a heat-treatable 6061-T6 Al alloy plate Friction stir welding (FSW) is a solid-state joining pro-
subjected to friction stir welding (FSW). First, temperature cess utilizing a rotating tool, which consists of a pin and
distributions were simulated in the FSW plate by finite tool shoulder that apply severe plastic deformation and
element modeling. Second, to determine the natural aging frictional heating into the base material [1, 2]. The soft-
kinetics, hardness changes were measured as a function of ened material is extruded underneath and around the tool
natural aging time from a number of Al specimens that had as the tool travels along the welding line and subse-
been isothermally heat treated at different peak tempera- quently meets to form a strong metallurgical joint. FSW
tures. Finally, the simulated temperature profiles and the has been widely used in joining heat-treatable Al alloys
natural aging kinetics were correlated to predict the hard- for the transportation and aerospace sectors [3, 4]. How-
ness profiles in the FSW plate. The predicted hardness ever, microstructural changes and their resulting effect on
minimum locations are consistent with the measured mechanical properties are significant in FSW heat-treat-
hardness profiles in that the hardness moves away from the able Al alloys. It is mainly due to the growth, dissolution,
weld centerline as the aging time increases. Moreover, the and re-precipitation of the strengthening precipitates after
predicted hardness minimum is located at the similar FSW [5–9]. Thereby, extensive efforts have been made to
position of failure in cross-weld tensile samples. understand the aging behavior for FSW heat-treatable Al
alloys [10–14].
Time-dependent material property change is a critical
issue in many end applications. Several process models that
describe the aging behavior have been proposed to connect
W. Woo (&) the processing variables with microstructure and hardness/
Neutron Science Division, Korea Atomic Energy Research strength of FSW [15–18]. Frigaard et al. developed a pro-
Institute, 1045 Daedeok-daero, Yuseong-gu, Daejeon 305-353, cess model to estimate the evolution of re-precipitation and
South Korea
resultant hardness distributions in FSW AA6082-T6 and
e-mail: [email protected]
AA7108-T79 alloys [15]. Shercliff et al. suggested a
H. Choo method to predict the microstructural changes in FSW
Department of Materials Science and Engineering, based on the softening behavior of precipitates in 2xxx Al
The University of Tennessee, Knoxville, TN 37996, USA
alloys [16]. Robson and Sullivan proposed a natural aging
P. J. Withers process model considering the strength loss due to the dis-
Manchester Materials Science Center, University of Manchester, solution and coarsening of the precipitates in FSW AA7449
Grosvenor St., Manchester M1 7HS, UK [17]. Using similar approaches, Peel et al. were able to
predict the distribution of hardness in dissimilar AA6082-
Z. Feng
Materials Science and Technology Division, Oak Ridge National AA5083 FSW prior to, and subsequent to, natural aging
Laboratory, Oak Ridge, TN 37831, USA [18]. These models describe the dissolution/coarsening and

123
J Mater Sci (2009) 44:6302–6309 6303

re-precipitation behavior in FSW based on classical ther-

modynamics and diffusion [19–21].
However, numerical tracking of the precipitation
behavior requires a relatively sophisticated analysis with
unclear assumptions. Moreover, those prior process
models have limited to predict the hardness profiles
before and after natural aging. It is suspicious how the
aging response is during natural aging for the FSW heat-
treatable Al alloys. Such time-dependent model is nec-
essary and valuable aid on the mechanical evaluation of
FSW as a function of natural aging time. Subsequently,
one can determine when the material properties have
sufficiently stabilized and represent the fully natural aged
condition. It is, therefore, of practical value to ask whe-
ther a semiempirical method can be developed to provide
a predictive capability of the hardness profiles as a
function of natural aging time.
In this study, we present the results of: (1) simulated
temperature profiles in a FSW 6061-T6 Al alloy plate using
finite element (FE) modeling, (2) the natural aging kinetics
(a constitutive relationship between the temperature and
hardness at a given time) of Al 6061-T6 alloy based on a
series of hardness measurements performed as a function of Fig. 1 a Schematic of the FSW and tool geometry. The FSW 6061-
T6 Al alloy plate was meshed for the FE modeling. b Placement of
time using isothermal furnace heat-treated Al specimens,
thermocouples. Thermocouples (marked TC1–5) were inserted
and (3) predicted hardness profiles of the FSW plate under 3.2 mm below the top surface of the Al plate
the natural aging condition using the correlation between
the simulated temperature profiles and the natural aging
kinetics. In this regard, we can predict how the hardness
eight thermocouples (TC1–8), five positioned at 8, 15, 20,
minimum location changes with time, which is important
30, and 50 mm from the weld centerline on the retreating
to determine the fracture location in the FSW heat-treatable
side of the plate (Fig. 1b) and three more thermocouples
Al plate.
were positioned at 15, 20, and 30 mm on the advancing
side. Note that the retreating side is denoted as the positive
Experimental procedures y-direction (advancing side is negative) throughout the
article. Each TC was inserted 3.2 mm below the top surface
Friction stir welding of the plate.
After FSW, the sample was cut across the mid-length of
6061 Al alloy rolled plate was given a T6 (peak aged) heat the weld for microstructural characterization. Vickers
treatment, namely, solution heat treated and aged for 6 h at microhardness (HV) was measured along the y-direction of
185 °C. The nominal chemical composition in weight per- the mid-thickness on the polished surface with 0.25 mm
cent is 1.0 Mg, 0.6 Si, 0.3 Cu, and balance Al. The Al plate horizontal indent spacings using 100 g of applied load.
typically contains precipitates (GP-I, b00 , b0 , and b-Mg2Si) Note that hardness profiles were measured at about 104 s
[5, 6]. The grains in the parent plate were elongated along (4 h) and 107 s (8 months) after FSW. In addition, a total
the rolling direction being about 200 lm in diameter. The of five lines were measured across the thickness of the plate
FSW was performed along the centerline of a single to construct a hardness map. Two cross-weld tensile
(306 9 306 9 6.5 mm3) plate (Fig. 1a) using the following specimens were also prepared about 107 s after FSW to
parameters: 4.7 mm/s traveling speed, 1,250 rpm clockwise observe the fracture locations. The tensile specimens were
rotation speed, and 8,000 N compressive force using a tool prepared using EDM with the dimension of 12.5 mm wide,
with 19.05 mm shoulder diameter and 6.35 mm pin diam- 6.5 mm thick, and 50 mm long in the gage section fol-
eter of 6.23 mm length. Note that the tool was made of an lowing the ASTM E 8M-04. Tensile tests were performed
H-13 tool steel and the traveling direction of the tool was at room temperature using a Material Test System load
parallel to the rolling direction of the parent plate. The frame with hydraulic wedge grips at a constant crosshead
temperature distributions were measured during FSW using velocity providing an initial strain rate of 10-3 s-1.

123
6304 J Mater Sci (2009) 44:6302–6309

Isothermal furnace heat treatments the plate and the tool bottom surface and calculated based
on the torque required to rotate the circular shaft with an
The natural aging kinetics of the 6061-T6 Al alloy was axial load following:
investigated using prior isothermal heat-treated specimens. 4
A total of five specimens having the dimension of Q ¼ p2 lPNR3 ð1Þ
3
20(x) 9 50(y) 9 6.5(z) mm3 were prepared from the
as-received Al plate. The surface of each specimen was where l is the friction coefficient (taken to be 0.4 [27]), P
polished and immersed in a salt bath of the furnace. The is the constant loading pressure (28.3 MPa), N is the
peak temperature of the furnace was heated up to 250, 350, rotational speed (20.8 s-1), and R is the tool shoulder
450, 500, and 550 °C with the control of a k-type ther- radius (9.5 9 10-3 m). As a result, a first estimate of Q
mocouple. Note that each specimen experienced the peak was around 2,200 W under the current conditions.
temperature isothermally for about 10 s and water quen- Khandkar et al. determined Q using the torque (M, Nm) of
ched in about 20 s of leaving the furnace to simulate the the tool during FSW [26]. Q was given by the measured
thermal history of FSW. Hardness measurements were input torque multiplied by the rational speed (N, 1/s):
performed repeatedly after the cooling until about 106 s. It Q = 2pNM [26]. In this experiment, the average torque
established the natural aging kinetics, which is the rela- (M) was recorded as 18.8 Nm during FSW and Q was
tionship between the heat-treatment temperature and calculated as 2,100 W. Note that 13% of the heat loss was
hardness as a function of natural aging time in the 6061-T6 considered due to heat conducted through the tool during
Al alloy plate. Note that the presented hardness here is an FSW.
average of the five times measurements on the polished To confirm the Q obtained by the above numerical
surface along the mid-thickness of each specimen. prediction methods, we used an inverse thermal FE mod-
eling approach [22, 23]. The inverse modeling scheme is as
Finite element modeling follows: (i) four temperature profiles were simulated by FE
modeling on the basis of different Q values (Q = 2,200;
A three-dimensional FE model was created using a com- 2,000; 1,750; and 1,550 W), (ii) the resulting thermal
mercial FE analysis code, ABAQUS 6.7, to determine the predictions were compared to the measured temperatures
thermal evolution caused by FSW [22–30]. Note that the profiles, and (iii) finally, an appropriate Q was selected on
modeling included only the thermal heat input and ignored the basis of the best fit.
the metal flow caused by the stirring of the tool. First, the
FSW plate was meshed using a 6-node triangular prism
element (type DC3D6) and an 8-node hexahedra brick Results and discussion
element (type DC3D8) outside and underneath the tool
shoulder, respectively (Fig. 1a). The tool was not included Thermal profiles during friction stir welding
in the model. Second, the total heat input (Q) from the
FSW tool was continuously put into the nodes of the Figure 2a shows the peak temperature distributions during
meshed elements using an ABAQUS subroutine [29]. It is FSW as a function of distance from the weld centerline.
assumed that 90% of Q was imposed on the surface nodes The simulated peak temperature is the FE response at the
over the tool shoulder and 10% on the volumetric nodes of nodes located 3.2 mm below the top surface. It is evident
the tool pin [31]. The heat flow at the nodes was treated as that the best agreement between the measurements and
a heat conduction problem. Finally, when the tool moves predictions occurs when the Q is taken to be 2,000–
along the weld centerline, the temperature distributions 2,200 W. At this level of heat input, the peak temperature
were calculated for a given welding speed of FSW. The at the mid-thickness of the FSW plate is predicted to be
boundary condition of the plate surface was a convection around 480–520 °C at the centerline (y = 0 mm). It is
condition. The convective heat coefficient was selected to worth noting that the temperature range is reasonable when
be 30 W/m2°C for the top and side surface of the plate, calculating using the empirical relationship between the
which is typical for the natural convention between Al and peak temperature (T) and rotating (x)/traveling (m) speed
air, while 200 W/m2°C was used for the bottom surface parameters as shown in Eq. 2 of [1]: T ¼ Kðx=m
104 Þa 
due to the contact with the steel backing plate [22]. Tm : Following the equation, T was calculated as 490 °C
A precise determination of Q during FSW is critical when used K = 0.75, a = 0.04, and Tm (melting temper-
because the heat input and simulated temperature will be ature) = 660 °C. Figure 2b shows the comparison of the
used as a key factor to predict the hardness. A numerical measured and predicted (Q = 2,200 W) temperature pro-
heat flow model proposed by Frigaard et al. can calculate Q files as a function of time at various positions on the
[15]. In brief, Q can be generated by the friction between retreating side. Unfortunately, the thermocouple located

123
J Mater Sci (2009) 44:6302–6309 6305

(a) 600
Modeling
Peak temperature ( C)
500 Measurements
Q = 2200 Watt
o

400
2000

300
1750
200

100

1550
0
-60 -40 -20 0 20 40 60
Distance from the weld centerline (mm)

(b) 500
Modeling Q = 2200 Watt
Measurements
400
Temperature ( C)

8 mm
o

15 Fig. 3 a Simulated temperature distributions in the cross-section of

300 20 the 6061-T6 Al alloy during FSW using FE modeling. b The
microstructure of the FSW 6061-T6 Al alloy with the same
30
temperature contour shown in (a). The four typical regions are
200 marked on the cross-section: the base material (BM), heat-affected
zone (HAZ), thermomechanically affected zone (TMAZ), and dynamic
100
recrystallized zone (DXZ)
50

0
0 20 40 60 80 100 120
Time (sec)
Softening and natural aging kinetics

Fig. 2 a Simulated peak temperature distributions as a function of Figure 4a shows the hardness changes as a function of
distance from the weld centerline in four different cases (Q = 2,200; natural aging time measured from the isothermally heat-
2,000; 1,750; and 1,550 W) during FSW. Measured peak tempera-
tures were marked at seven different locations. b Measured and treated 6061-T6 Al alloy specimens exposed at different
simulated temperature variations as a function of time during FSW at temperatures from 250 to 550 °C. In the case of the 550 °C
8, 15, 20, 30, and 50 mm from the weld centerline on the retreating heat treatment, there is a significant decrease in hardness at
side room temperature compared to the initial hardness
(110 HV) and increases up to about 75 HV after 106 s of
natural aging. In contrast, for the sample exposed to
8 mm from the weld centerline broke as the tool passed 350 °C, the loss of hardness is less marked and it recovers
over it and so the peak temperature is not available. much more slowly to a lower level (*70 HV).
Overall, the 2,200-W curve shows good agreement with the Several studies suggest that significant dissolution and
experimental measurements. reversion of fine precipitates (b00 ) will occur with increas-
Figure 3a shows the predicted temperature distributions ing prior exposure temperature [15–18]. This explains the
when the tool is coincident with the viewed cross-section of more loss of hardness and recovery of the hardness during
the plate. Unsurprisingly, the highest temperatures are natural aging when exposed at the higher temperature as
developed near the top surface due to the contact with the shown in Fig. 4a; whereas, the limited softening and
tool shoulder and gradually decrease away from the weld recovery capabilities of the specimens exposed to the low
centerline. A difference of about 140 °C was simulated temperatures are due to incomplete dissolution of harden-
through the thickness of the plate at the centerline, i.e., ing precipitates and the formation of nonhardening pre-
about 590 °C at the top surface and 450 °C at the bottom. In cipitates [6, 12]. Thus, it has been reported that a very
Fig. 3b, the simulated temperature profiles are superim- complex series of growth, dissolution, and re-precipitation
posed on the microstructure linking the peak temperatures phenomena occur for the whole sequence of precipitates
to the characteristic regions. It shows that the dynamic re- during softening and hardness recovery [15–21].
crystallized zone (DXZ) experienced about 480–550 °C, Rather than simulate this whole natural aging process,
the thermomechanically affected zone (TMAZ) about 430– the softening and natural aging kinetics of the heat-treat-
480 °C, and the heat-affected zone (HAZ) less than 430 °C. able Al alloys can be empirically modeled using the

123
6306 J Mater Sci (2009) 44:6302–6309

Prediction and measurements of the hardness profiles

Hardness variations of the FSW heat-treatable Al alloy

strongly depend on the natural aging time and/or heat input
[13, 15]. Table 1 summarizes the predicted hardness vari-
ations at t = 1, 104, and 107 s using the AT and BT. Note
that the selected peak temperatures are shown in Fig. 3a,
and the hardness was calculated using Eqs. 2 and 3 at
different natural aging times. The hardness data at 1 s gives
an impression of the hardness immediately after welding
even though the FSW may not be cooled instantaneously.
The predicted hardness variations were presented on the
cross-section of the FSW (Fig. 5a). Note that the hardness
variations were marked on the thermal profiles derived
from the FE simulation of the 2,200-W heat-input case as

Table 1 Hardness variations as a function of the natural aging time

FSW peak AT (HV) BT (HV/ln s) HV after aging time of:
temp. (°C)
1s 104 s 107 s

540 43 3.1 43 69 88
500 48 1.4 48 61 70
450 53 0.7 53 60 66
400 59 0.4 59 63 65
350 68 0.3 68 71 72
310 79 0.2 79 81 82
260 94 0.1 94 96 97
The initial hardness constant (AT) and the aging kinetics (BT) were
calculated using Eq. 3, and the natural aging hardness (HV) was
Fig. 4 a Hardness variation as a function of natural aging time after
obtained from Eq. 2 for a given peak temperature and aging time
heat treatments of 6061-T6 Al alloy samples at various temperatures
raging from 250 to 550 °C. b Arrhenius plot for the initial hardness
(AT) and aging kinetics (BT) as a function of inverse temperature (1/T)

relationship between hardness (HV) and natural aging time

(t, s) for a given heat-treatment temperature (T, K) [32, 33]:
HV ¼ AT þ BT  ln t ð2Þ
where AT is the initial hardness at 1 s and BT is the slope
coefficient (HV/ln s) of natural aging. Figure 4b shows the
initial hardness constant (AT) and the aging kinetics (BT) as
a function of heat-treatment temperature. It was obtained
from the softening and hardening results as shown in
Fig. 4a. Linear and exponential fitting results provide the
relationships:
1
AT ¼ 3:7 þ 25600 ; ln BT ¼ 63 expð1600=TÞ  2:0
T
ð3Þ
Consequently, the constitutive relationship has been
Fig. 5 a Predicted hardness variations as a function of natural aging
achieved correlating the hardness (HV), heat-treatment
time (nominally 1, 104, 107 s) in 2,200-W heat-input case. b Mapping
temperature (T), and aging time (t) for the Al 6061-T6 of the hardness using the measurements across the FSW specimen
alloy in Eq. 2. 107 s after FSW

123
J Mater Sci (2009) 44:6302–6309 6307

shown in Fig. 3a. The peak temperature experienced HAZ TMAZ DXZ TMAZ HAZ
110
gradual increase toward the centerline and the predicted Predictions Measurements
initial hardness falls to 43 HV near the weld centerline. 7
10 s 7
10 s
100
Upon natural aging it is predicted to recover to 69 HV at 4
10 s 4
10 s
104 s and 88 HV at 107 s, while the hardness beyond 90
1s
Hardness Completed natural aging
15 mm from the centerline remains essentially unchanged.

Hardness (Hv)
minimum (MG model [20])

This trend in hardness evolution causes the minimum 80

hardness location to move with time. To validate the
hardness predictions, the hardness map was measured at 70
107 s after FSW (Fig. 5b). Note that the yellow-colored
regions indicate where the hardness is below 70 HV. It 60
shows that the hardness minimum is located at about
9–10 mm from the centerline in good agreement to the 50 Immediately
after welding
simulated one. It is also the case of the two other heat-input (MG model [20])
40
cases (Q = 2,000 and 1,750 W) presented in Fig. 6. By -2
-20 --15 -10 -5 0 5 10
1 15
1 20
comparing the three heat-input cases, it clearly shows that Distance from the weld centerline (mm)
the hardness minimum moves outward; the higher the heat
Fig. 7 The measured and predicted hardness profiles at 1, 104, and
input the longer is the natural aging time. 107 s after FSW. These are compared to the hardness profiles
Since the hardening precipitates primarily govern the predicted by the MG process model in [20]
strength of heat-treatable Al alloys, the location of the
hardness minimum is a potential fracture location during
transverse tensile loading in the FSW heat-treatable Al These changes of the hardness minimum location are
alloys [30, 34]. Figure 7 compares between the hardness related to the different natural aging responses in each
predictions and measurements profiled along the mid- characteristic region of FSW. The DXZ is the region
thickness of the cross-section in FSW. Note that the pre- softened the most due to the complete dissolution of pre-
dicted hardness within ±4 mm was profiled using the cipitates at the highest peak temperature (over 480 °C)
maximum temperature considering significant metal flow during FSW. Subsequently, the DXZ has the highest
within the DXZ. The predicted result shows that the capacity for re-precipitation and increases of the hardness
hardness minimum is changed from 8 mm at 104 s to when it fully natural aged [13, 35]. Similarly, the hardness
10 mm at 107 s after FSW. It is consistent with the hard- in the TMAZ was increased under natural aging, although
ness measurement results. Moreover, cross-weld tensile it is not as much as that of the DXZ. As a result, the
tests on samples at 107 s showed the fracture locations to hardness minimum location moves outward the HAZ,
be about 10 mm from the weld centerline on the advancing which is observed in the current experimental and predic-
side in accordance with the current predictions. tion results (Fig. 7).

Comparison between a prior and the current predictions

For comparison, we simulate the hardness profiles for the

current case (Q = 2,200 W) using the Myhr and Grong’s
(MG) model [20] (Fig. 7). In brief, the MG method theo-
retically calculates the reaction kinetics of the precipitates
in fractions under the softening and hardening condi-
tions, and predicts the hardness profile using the following
equation [20]:
HV ¼ HVmin þ ðHVmax  HVmin Þa ð4Þ
where HVmax (110 HV) is the initial hardness of the base
metal, and HVmin (45 HV) is the hardness in the absence of
the precipitates. Note that the fraction of precipitates (a)
under the dissolution or hardening condition was adopted
from [20]. Overall, it is a good agreement between the
Fig. 6 Predicted hardness minimum locations as a function of natural
aging time (1, 103, 104, and 107 s) under three different total heat hardness profiles simulated using the MG model and cur-
input cases (Q = 2,200; 2,000; and 1,750 W) rent method (Fig. 7). There is a slight discrepancy between

123
6308 J Mater Sci (2009) 44:6302–6309

the two hardness profiles at the same aging time. It could Laboratory (ORNL), managed by UT-Battelle, LLC for the U.S.
be attributed to the different peak temperature and sub- Department of Energy under Contract No. DE-AC05-00OR22725.
WW was supported by Nuclear Research and Development Program
sequent hardness aging responses in each location of the of the Korea Science and Engineering Foundation funded by the
current method, whereas the GM model simultaneously Korean government. PJW is grateful to the EPSRC Lightweight
switches the hardness profile from the softening to recovery alloys portfolio grant for financial support. The authors would like to
curve at a point where the hardness has the same value thank B. Lovell, S. A. David, C. J. Rawn, and A. Frederick for their
help. WW is especially grateful to David Richards for help with the
[20]. In the end, it should be mentioned that the current modeling during his visit to Manchester.
method enables us to simulate the hardness profiles and
minimum locations throughout the natural aging in FSW
Al 6061-T6 alloy, while the MG model can simulate the
hardness immediately after the welding and after the References
completion of the natural aging process (Fig. 7).
1. Mishra RS, Ma ZY (2005) Mater Sci Eng R 50:1
2. Threadgill PL, Leonard AJ, Shercliff HR, Withers PJ (2009) Int
Mater Rev 54:49
Conclusions 3. Thomas WM, Nicholas ED (1997) Mater Des 18:4
4. Smith IJ, Lord DDR (2008) In: Seventh international symposium
Hardness variations during natural aging were predicted in on friction stir welding, TWI, Japan
5. Murr LE, Liu G, Mcclure JC (1998) J Mater Sci 33:1243
a FSW of the heat-treatable 6061-T6 Al alloy using a FE 10.1023/A:1004385928163
model of the thermal history and the natural aging kinetics 6. Sato YS, Kokawa H, Enomoto M, Jogan S (1999) Metall Mater
obtained by isothermal furnace heat treatments. Trans A 30:2429
7. Jata KV, Sankaran KK, Ruschau JJ (2000) Metall Mater Trans A
1. In situ temperature measurement data during FSW was 31:2181
used to validate the FE modeling. The peak temper- 8. Sutton MA, Yang B, Reynolds AP, Taylor R (2002) Mater Sci
Eng A 323:160
ature profiles were calculated for each characteristic
9. Su JQ, Nelson TW, Mishra R, Mahoney M (2003) Acta Mater
region in the FSW: DXZ about 480–550 °C, TMAZ 53:713
about 430–480 °C, and HAZ less than 430 °C for the 10. Genevois C, Deschamps A, Denquin A, Doisneau-Cottignies B
current conditions. (2005) Acta Mater 53:2447
11. Kamp N, Sullivan A, Tomasi R, Robson JD (2006) Acta Mater
2. The hardness measurements were performed as a
54:2003
function of natural aging time using a number of 12. Dumont M, Steuwer A, Deschamps A, Peel M, Withers PJ (2006)
isothermally heat-treated Al alloy samples in a furnace Acta Mater 54:4793
at various temperatures. The results provide the initial 13. Woo W, Choo H, Brown DW, Feng Z (2007) Metall Mater Trans
A 38:69
hardness constant (AT) due to the softening and
14. Simar A, Bréchet Y, de Meester B, Denquin A, Pardoen T (2007)
logarithmic aging response (BT) of strengthening pre- Acta Mater 55:6133
cipitates in 6061-T6 Al alloy: AT = -3.7 ? 25,600 15. Frigaard Ø, Grong Ø, Midling OT (2001) Metall Mater Trans A
(1/T), ln BT = 63exp(-1600/T) - 2.0, where T is the 32:1189
16. Shercliff HR, Russell MJ, Taylor A, Dickerson TL (2005) Méc
heat-treatment temperature (K).
Ind 6:25
3. The simulated temperature profiles in the FSW plate 17. Robson JD, Sullivan A (2006) Mater Sci Technol 22:146
and the natural aging kinetics were correlated to 18. Peel MJ, Steuwer A, Withers PJ (2006) Metall Mater Trans A
predict the hardness variations as a function of aging 37:2195
19. Shercliff HR, Ashby MF (1990) Acta Metall Mater 38:1789
time after FSW. The simulated profiles for the 2,200-
20. Myhr OR, Grong Ø (1991) Acta Metall Mater 39:2693
W heat-input case show that the hardness minimum 21. Bjørneklett BI, Grong Ø, Myhr OR, Kluken AO (1999) Metall
location was changed from the TMAZ (8 mm) at 104 s Mater Trans A 30:2667
to the HAZ (10 mm) at 107 s. It is a good agreement 22. Chao YJ, Qi X (1998) J Mater Process Manuf Sci 7:215
23. Chao YJ, Qi X, Tang W (2003) Trans ASME 125:138
with the hardness measurements and fracture locations
24. Ulysse P (2003) Int J Mach Tools Manuf 42:1549
in the tensile tests of FSW. 25. Chen CM, Kovacevic R (2003) Int J Mach Tools Manuf 43:1319
4. The current model successfully predicted the complete 26. Khandkar MZH, Khan JA, Reynolds AP (2003) Sci Technol
time-dependent hardness minimum locations at the Weld Join 8:165
27. Song M, Kovacevic R (2004) Proc Inst Mech Eng Part B: J Eng
three different heat-inputs cases (1,750; 2,000; and
Manuf 218:17
2,200 W) in FSW Al 6061-T6 alloy. 28. Nandan R, Roy GG, Debroy T (2006) Metall Mater Trans A
37:1247
Acknowledgements This work was supported by the NSF Inter- 29. Richards DG, Prangnell PB, Withers PJ, Williams SW, Wescott
national Materials Institutes (IMI) Program under contract DMR- A, Oliver EC (2006) Mater Sci Forum 524–525:71
0231320. This research was sponsored by the Laboratory Directed 30. Feng Z, Wang XL, David SA, Sklad PS (2007) Sci Technol Weld
Research and Development program of Oak Ridge National Join 12:348

123
J Mater Sci (2009) 44:6302–6309 6309

31. Schmidt H, Hattel J, Wert J (2004) Model Simul Mater Sci Eng 34. Nelson TW, Steel RJ, Arbegast WJ (2003) Sci Technol Weld Join
12:143 8:283
32. Esmaeili S, Lloyd DJ, Poole WJ (2003) Acta Mater 51:3467 35. Linton VM, Ripley MI (2008) Acta Mater 56:4319
33. Mrówka-Nowotnik G, Sieniawski J (2005) J Mater Process
Technol 162:367

123

View publication stats