Professional Documents
Culture Documents
01 2003 LIENERT S PDF
01 2003 LIENERT S PDF
ABSTRACT. The objective of this work strated the feasibility of FSW of steel with ergy (Refs. 5, 6) at high strain rates (Ref.
was to demonstrate the feasibility of fric- acceptable tensile properties. Based on 7) developed through interactions with
tion stir welding (FSW) for joining of mild these results, FSW of transformation- the tool. During welding, the material
steel. Defect-free welds were produced on hardenable steels, HSLA steels, and stain- along the joint is heated to a softened con-
0.25-in. plates (6.3 mm) of hot-rolled AISI less steels appears feasible. dition, transferred around the periphery
1018 mild steel at travel speeds ranging of the tool, and subsequently recoalesced
from 1 to 4 in./min (0.42 to 1.68 mm/s) Introduction along the back surface of the pin to pro-
using molybdenum-based and tungsten- duce the weld.
based alloy tools. Results for welds made Friction stir welding (FSW) is a rela- Friction stir welding of Al alloys is rel-
at 1 in./min are reported here. Axial tool tively new joining process that is presently atively well established. To date, friction
loads measured during FSW were approx- attracting considerable interest (Ref. 1). stir welds have been successfully produced
imately 4200 lbf (18.7 kN), while mea- The FSW process was developed at TWI on many of the important commercial Al
sured torques were in the range of 500 in 1991. Figure 1 shows a schematic of the alloys including the 1xxx (Ref. 8), 2xxx
in.-lb (55 Nm). Peak surface tempera- FSW process for joining of two plates in a (Refs. 911), 5xxx (Refs. 12, 13), 6xxx
tures close to 1000C (1832F) were mea- square groove configuration (Ref. 2). The (Refs. 1217), and 7xxx (Refs. 1820) fam-
sured on the workpiece and tool during process is solid-state in nature and relies ilies of alloys, as well as Al-Cu-Li alloys
welding using thermocouples and an in- on localized forging of the weld region to (Refs. 21, 22). Friction stir welding of Al
frared camera system. Comparisons be- produce the joint. The plates comprising alloys has several advantages over fusion
fore and after welding combining both the workpiece are held in compression welding processes. First, problems with
metallographic and metrology techniques and are rigidly fixtured to the machine bed solidification cracking, liquation cracking,
suggest changes in tool dimensions stem during welding. Friction stir welding uses and porosity are eliminated due to the
from both rubbing wear and deformation a nonconsumable, rotating tool that is solid-state nature of the process. Conse-
of the tool. The greatest changes in tool di- cylindrical in shape with a cylindrical pin quently, FSW permits joining of most, if
mensions occurred during the initial of smaller diameter extending from the not all, Al alloys, including those consid-
plunging stage. Microstructures of the tool shoulder. Important process parame- ered unweldable with fusion welding
welds were examined using optical and ters include the tool rpm and travel speed, processes (Ref. 23). Moreover, friction stir
scanning electron microscopy. The weld as well as the tool dimensions and the welds often exhibit improved mechanical
region displayed several microstructurally downward force on the tool. properties relative to those of fusion welds
distinct regions. Extrapolation of mea- Initially, the rotating tool is plunged on the same material (Refs. 14, 15, 21). Fi-
sured temperatures and microstructural into the joint until the shoulder contacts nally, residual stresses and distortion in
evidence suggest peak temperatures of the the top surface of the workpiece. Heating friction stir welds are typically lower than
stir zone exceeded 1100C (2012F) and is caused by rubbing of the tool faces those of fusion welds (Ref. 24). The im-
likely surpassed 1200C (2192F). Trans- against the workpiece (Refs. 3, 4) and by provements in mechanical properties,
verse tensile properties of the welds were visco-plastic dissipation of mechanical en- residual stresses, and distortion relative to
evaluated at room temperature. Welded fusion welding are thought to stem from
samples failed in regions corresponding to the lower heat input inherent with FSW.
the base metal and demonstrated yield However, no heat input data has been re-
and ultimate tensile strengths comparable KEY WORDS ported for FSW.
to those of the base metal. Welded sam- Friction stir welds on Al alloys display
ples passed bend tests to ~15% strain in Friction Stir Welding several microstructurally distinct regions
the outer fiber. The results have demon- Mild Steel including the stir zone or nugget (along
Torque the weld centerline), the heat-and-
T. J. LIENERT is a Technical Staff Member at Tool Wear deformation-affected zone (HDAZ)
Los Alamos National Laboratory, Los Alamos, Axial Load or thermomechanically affected zone
N.Mex. W. L. STELLWAG, JR., and B. B. Microstructures (TMAZ) (surrounding the stir zone), and
GRIMMETT are Application Engineers with Mechanical Properties a true heat-affected zone (HAZ) (encom-
Edison Welding Institute, Columbus, Ohio. R. W. Peak Temperatures
WARKE is an Engineer with Southwest Research passing the HDAZ) (Refs. 11, 13, 15, 16,
Institute, San Antonio, Tex. 18, 19). Microstructural evolution in the
different regions of the weld zone is
2 -S JANUARY 2003
WELDING RESEARCH
A B
Fig. 2 A Temperature cycles for thermocouples located at ~0.125 in. (3.2 mm) from the edge of the stir zone on the top surface of the 1018 steel work-
piece; B temperature cycles for thermocouples located at 0.25 and 0.38 in. (6.35 and 9.65 mm) above the tool shoulder.
tool shoulder before reaching peak tem- made at 1 in./min were evaluated by trans- of 5 to 10C/s (9 to 18F/s). It is often con-
peratures. To circumvent this problem, verse tensile testing of subscale specimens venient to characterize cooling rates ac-
thermocouples were attached to the work- at room temperature according to ASTM cording to the time for cooling between
piece in shallow grooves at several loca- E8 (Ref. 31). Tensile samples were pre- 800C and 500C (t85) (1472F to 932F)
tions on the top and bottom surfaces in pared by milling of the top and bottom (Ref. 32). Assuming cooling rates of the
this study. This method allowed the flash- surfaces to remove flashing and other sur- same order of magnitude throughout the
ing to slide over the top of the groove with- face irregularities. Face and root bend rest of the weld region, as suggested by cal-
out disturbing the thermocouples. tests were also performed on machined culations for arc welds by Adams (Ref.
Tool and workpiece temperatures were specimens of the welded samples (0.25 x 1 33), the t85 for these welds was around
also monitored using an Agema 550 x 6 in. or 6.35 x 25.4 x 152.4 mm) using a 50 to 60 s across the entire weld region.
Tracer Plus Elite Infrared System from die block with a 0.75-in. (19-mm) radius. Thermocouples attached to the circum-
FLIR systems. The camera uses a focal ference of the tool at 0.25 and 0.38 in. (6.35
plane array detector with a spectral range Results and 9.65 mm) above the shoulder read peak
of 3.6 to 5.0 microns. The array consists of temperatures of 985C (1805F) and 930C
a 320 x 240 platinum-silicon (PtSi) detec- In this study, friction stir welds were (1706F), respectively Fig. 2B. The tool
tor yielding a spatial resolution of 76,800 produced on 1018 steel at travel speeds temperatures rose during the 180-second
pixels. The camera temperature range is ranging from 1 to 4 in./min (0.42 to 1.68 plunging period and reached steady state
20C (4F) to 2000C (3632F), with a mm/s). However, results reported here soon after the welding period was begun.
measurement accuracy of 2% or 2C of pertain mainly to welds made at 1 in./min Note that the temperature plots for the tool
the temperature range. The field of view travel speed. were truncated after a short length since the
for this camera is 20 deg x 15 deg with the thermocouples were severed by motion of
standard lens. The camera produces im- Process Characterization the flashing. Fortunately, the tool tempera-
ages with 256 color levels with a user-se- ture reached steady state before the ther-
lected palette. Accurate temperature Load, torque, and temperature data mocouples were lost, thus allowing calibra-
measurement with this system requires were recorded as functions of time for sev- tion of the emissivity and monitoring of
user input of correct emissivity values. For eral defect-free welds. Conversion of the temperature throughout the rest of the weld
each weld, the emissivity was calibrated different data sets to plots as a function of with the infrared system.
against known temperatures from the distance was accomplished by multiplying Thermocouple measurements were
thermocouples placed on the tool. by the weld travel speed. confirmed by data from the infrared sys-
Metallographic samples were pro- The tool and the flashing around the tem. Peak surface temperatures of ap-
duced from welds made at 1 in./min using tool glowed a reddish-orange color during proximately 990C (1814F) were mea-
standard procedures and etched with a 2% welding suggesting peak temperatures of sured near the shoulder/workpiece
nital solution. Etched samples were exam- at least 1100C (2012F). Thermal cycles interface using the infrared system. Mea-
ined using optical microscopy (OM) with recorded from the workpiece and the tool surement of temperature at the shoul-
differential interference contrast (DIC) are presented in Figs. 2A and B, respec- der/workpiece boundary was not possible
and scanning electron microscopy (SEM) tively. Thermocouples placed approxi- during welding since this location was
with X-ray energy-dispersive spectroscopy mately 0.125 in. from the edge of the stir blocked from direct view of the infrared
(EDS). Weld microstructures were classi- zone at the top of the plate recorded peak camera by flashing. Consistent with the
fied according to a scheme developed by temperatures between 590C (1094F) thermocouple measurements, tempera-
the International Institute for Welding and 665C (1229F) Fig. 2A. Observe ture gradients of 100 to 200C/cm (540 to
(IIW) (Ref. 30). Vickers microhardness that the thermal cycles were relatively 995F/in.) were determined along the
traverses (1-mm increments) were pro- symmetric about the peak temperature length of the tool with software from the
duced across the weld regions using a and that both heating and cooling rates infrared system.
500-g load and a 15-s dwell time. Mechan- were rather slow. Average cooling rates Representative load and torque curves
ical properties of samples from welds near peak temperature were on the order are presented in Fig. 3A and B. The
WELDING JOURNAL 3 -S
WELDING RESEARCH
A B
Fig. 3 A Load and torque curves during plunge stage for FSW of 1018 steel; B load and torque curves during weld stage for FSW of 1018 steel.
plunging period lasted almost three min- Torque on the tool averaged around sions occurred during the initial plunging
utes Fig. 3A. During plunging of the 500 in.-lb (55 N.m) during the welding pe- stage. Measurements showed the tip of
tool, the load rose initially when the pin riod Fig. 3B. No appreciable change the pin mushroomed similar to a resis-
was in contact with the cold workpiece was found in the torque with different tance-welding electrode and lost approxi-
and quickly fell as the material was soft- travel speeds. The torque curve showed mately 0.010 in. (0.25 mm) of its original
ened due to heating. The softened mate- high-frequency variations similar to those 0.245 in. (6.2 mm) of length during the
rial was displaced until the pin again en- of the load curve but none of the low- plunging stage and a total of 0.012 in. (0.30
countered colder material, and the frequency fluctuations. Fast-Fourier mm) for the entire weld.
process then repeated. Initial load curves Transform results showed a number of
exhibited several spikes up to about dominant frequencies including those cor- Microstructural Characterization
2750 pounds of force (lbf) (12.2 kN) dur- responding to the tool rpm and its har-
ing the first two to three minutes of the monics, as well as 60-Hz frequencies Features of the base metal microstruc-
plunging period prior to welding. The picked up from nearby electrical equip- ture are shown in Fig. 4A and B. The as-
large spike in load up to 7600 lbf (33.8 kN) ment. Several other frequencies including received base metal exhibited a mi-
occurring approximately 175 s into the those corresponding to the rapid fluctua- crostructure of equiaxed ferrite grains
plunge cycle corresponded to the point in tions mentioned above were also identi- approximately 20 to 30 m in diameter
time when the shoulder contacted the top fied. The source of these frequencies is and smaller grains of fine pearlite, as seen
surface of the plate. The torque value rose still under investigation. While many of in Fig. 4A. Fig. 4B is an SEM micrograph
slowly during the last minute of plunging the frequencies are likely spurious, some showing the fine lamellar spacing in a
and reached steady state when shoulder of them may be related to physical pearlite colony.
contact occurred. processes occurring during FSW. Identifi- A low-magnification optical micro-
Average axial loads in the range of cation of these frequencies may lead to graph of the weld region is presented in
4200 lbf (18.7 kN) were recorded during further process insight such as those found Fig. 5. The weld area displayed several mi-
the welding period for welds made at a in machining studies (Refs. 36, 37). crostructurally distinct regions including
travel speed of 1 in./min Fig. 3B. Aver- Torque and load values reported here the stir zone along the weld centerline, the
age tool pressures calculated using the for FSW of mild steel can be compared to heat-affected zone (HAZ) surrounding
tool load and shoulder area were approx- those for FSW of other materials, such as the stir zone, and the base metal. Unlike
imately 11.5 ksi (79.2 MPa). Note this Ti and Al alloys (Ref. 38). For example, welds on Al alloys, no distinct HDAZ was
value is very close to the flow stress re- the average torque for FSW of mild steel seen in the steel welds. Note that flow lines
ported for mild steel at 1100C (2012F) was lower than that for FSW of a Ti-6 Al- and other indications of deformation that
(Refs. 25, 34). Increased tool loads were 4 V alloy (675 in.-lb). In addition, it is in- persist to room temperature in friction stir
observed at faster travel speeds. Higher- teresting to note that the load and torque welds on Al alloys, such as those found in
frequency fluctuations of 750 lbf (3.3 values found for steel were comparable to the HDAZ (Ref. 13), are not found in sim-
kN) were observed in the load curve. those determined for FSW of a 7075-T6 Al ilar welds on mild steel apparently due to
These rapid fluctuations are similar to alloy. However, the steel values were the allotropic transformations experi-
those seen by tribologists during friction greater than those for FSW of 5083-O, enced during cooling. Hence, the pres-
testing (Ref. 35) and may correspond to 6082-T6, and 6061-T6 Al alloys. ence of any HDAZ has likely been ob-
discrete stick-slip events that occurred Comparisons of tool dimensions were scured in the steel welds by these
during welding. Low-frequency undula- made before and after welding using an transformations.
tions of 250 lbf (1.1 kN) were also optical comparator. The metrology data, EDS results show the light etching fea-
found. These fluctuations occurred over combined with results of microstructural tures near the bottom center of the stir zone
distances of approximately 1 in. (2.54 cm) analysis of the tools and workpiece, sug- are Mo-enriched regions. Note that the fea-
and were attributed to the growth and gest changes in tool dimensions stem from tures were not discrete pieces of Mo that
expulsion of flashing around the end of both rubbing wear and deformation of the broke off from the tool. Rather, they ap-
the tool. tool. The greatest changes in tool dimen- peared very similar to composite structures
4 -S JANUARY 2003
WELDING RESEARCH
A B
Fig. 4 Micrographs of the 1018 steel base metal. A Optical micrograph showing equiaxed grains of ferrite (F) as the major phase and smaller grains of
fine pearlite (P) as the minor constituent; B SEM micrograph showing the fine spacing of the lamellae of the pearlite.
WELDING JOURNAL 5 -S
WELDING RESEARCH
ing a uniform flux) as the product of flux ing FSW. Hence, Fig. 9 serves only to il-
and the cross-sectional area of the tool. lustrate schematically the effective peak
These calculations suggest the power ab- temperatures experienced in the different
sorbed by the tool ranged from 300 to 600 regions of the HAZ.
W (1.025 x 103 to 2.050 x 103 Btu/h) for As shown in Fig. 9, the microstructures
FSW of mild steel during steady state. of the regions of the HAZ can be spatially
Fig. 5 Optical micrograph of a transverse sec- It is also important to note that torque correlated to peak temperatures on the
tion of a friction stir weld on 1018 steel. The stir and weld rpm data also permit estimation phase diagram, with effective peak tem-
zone, HAZ, and base metal regions can be dis- of total weld power and heat input for peratures decreasing with distance from
tinguished.
FSW. Comparisons can then be made with the stir zone. Material in the grain-
other welding processes such as gas metal coarsened region experienced the highest
rial in contact with the shoulder. arc welding (GMAW). Once the total temperatures in the HAZ. The HDAZ, if
Microstructural evidence is also con- power and power lost to the tool are it ever existed, was likely contained within
sistent with peak temperatures above known, the process efficiency can be ap- the area that became the grain-coarsened
1100C (2012F) in the workpiece. Bear in proximated. Initial estimates suggest a region. Microstructural and experimental
mind that a grain-coarsened region was process efficiency of approximately 0.75 evidence suggest peak temperatures were
observed in the HAZ immediately adja- for the conditions described in this paper. well above the effective A3 temperature
cent to the stir zone. As discussed in a later Further details of calculations of power for this alloy during welding, thus allowing
section of this paper, formation of a grain- and heat input for FSW of mild steel, com- some austenite grain growth. Evidence for
coarsened region in the HAZ of welds in parisons with estimated GMAW values, the existence of a HDAZ was probably lost
this alloy requires temperatures of at least and estimations of process efficiency for during decomposition of the austenite
1100C (2012F). Moreover, tempera- FSW of steel will be discussed in a subse- grains on cooling. The grain-coarsened re-
tures within the stir zone are believed to quent paper by the authors. gion contained phases similar to those of
be greater than those of the HAZ. In fact, the stir zone but somewhat smaller in scale
extrapolation of the gradients in mi- Microstructural Evolution in the HAZ due to the lower peak temperatures.
crostructure in the HAZ to the Fe-Fe3C Microstructural results indicate peak
phase diagram suggests peak tempera- With the possible exception of an temperatures encountered in the grain-
tures in the stir zone above 1200C HDAZ, the HAZ experiences little or no refined region fell just above the effective
(2192F). deformation during FSW and conse- A3 temperature. The decomposition of
Note that knowledge of the thermal quently is similar to the HAZ found in arc austenite to ferrite and pearlite on cooling
gradient along the tool allows approxima- welds. Hence, microstructural evolution promoted a finer grain size in this region.
tion of the heat flux lost along the length in the various regions of the HAZ of fric- The microstructure of this region con-
of the tool using Fouriers Law (Ref. 43) tion stir welds on 1018 steel alloys can be tained fine grains of pearlite and proeu-
(assuming radiative and convective losses approximated with the aid of a section of tectoid ferrite.
are negligible). Estimates of the flux, using the Fe-Fe3C phase diagram shown in Fig. The intercritical region was character-
a temperature-compensated thermal con- 9. The vertical line in the phase diagram ized by a bimodal distribution of ferrite
ductivity for the tool material and assum- indicates the composition of a 0.18 wt-% grain sizes. Material in the intercritical re-
ing one-dimensional heat flow, are on the C alloy. Note that use of a binary phase di- gion was exposed to peak temperatures in
order of 100 to 200 W/cm2 (315 x 103 to agram to model transformation behavior the two-phase (ferrite and austenite) re-
630 x 103 Btu/hft2). The magnitude of the in a multicomponent alloy such as 1018 in- gion between the effective A1 and A3 tem-
heat flux lost to the tool is useful in estab- troduces some uncertainties with respect peratures. This thermal cycle caused
lishing boundary conditions for modeling to the position of phase boundaries. transformation of the pearlite to austenite
of heat flow during FSW. Moreover, use of the equilibrium diagram on heating (with the reverse reaction on
In addition, the power dissipated in does not capture the time-dependent na- cooling) and resulted in refinement of the
heating the tool can be estimated (assum- ture of transformations experienced dur- pearlite colonies and some of the ferrite
A B
Fig. 6 Optical micrographs of the stir zone. A Center of stir zone: grain boundary ferrite [PF(G)], ferrite-carbide aggregate [FC], and ferrite with aligned
second phase [FS(A)]; B top surface of stir zone: finer ferrite.
6 -S JANUARY 2003
WELDING RESEARCH
A B
C D
Fig. 7 Optical micrographs of the different regions of the HAZ. A Grain-coarsened region: grain boundary ferrite [PF(G)] and ferrite with aligned [FS(A)]
and nonaligned second phases [FS(NA)]; B grain-refined region: pearlite (black) and proeutectoid ferrite (white); C intercritical region: pearlite (black)
and ferrite (white); D spheroidized carbide region: pearlite (black) and ferrite (white).
grains (Ref. 40). Microstructural observa- the HAZ. While the HAZ experiences a curring discontinuously (nucleation of
tions imply the material in the region of thermal cycle only, the stir zone (and any new grains and growth by motion of high-
spheroidized carbides experienced suffi- HDAZ) experiences both thermal and angle grain boundaries) or continuously
cient time below the A1 temperature to mechanical cycles. Hence, effects of the (increasing misorientation of existing sub-
allow partial homogenization and spher- combined thermomechanical cycle must grain boundaries due to absorption of dis-
oidization of the cementite. be considered in a discussion of mi- locations, also described as extended re-
The large microstructural features of crostructural development of the stir covery) (Ref. 45). The operative
the grain-coarsened HAZ and the pres- zone. mechanism is dependent on the stress (or
ence of pearlite in the grain-refined region The thermomechanical cycle of the stir strain) and strain rate histories, as well as
attest to the slow cooling rates estimated zone of mild steel involves hot working the stacking fault energy (SFE) of the ma-
for the HAZ. In fact, the microstructures processes. As discussed earlier, the im- terial (Ref. 25). For a given material,
of the grain-coarsened and grain-refined portant parameters that control mi- higher stresses (or strains) and strain rates
regions are consistent with those pre- crostructural evolution are the total strain, tend to promote discontinuous recrystal-
dicted by the continuous-cooling transfor- the strain rate, and the temperature cy- lization. Moreover, materials with low
mation diagram developed for arc welds cles. Temperature measurements and mi- SFE, such as mild steels in the austenitic
on a similar steel (Ref. 44). In agreement crostructural evidence indicate peak tem- phase field, generally undergo discontinu-
with the observed microstructures and peratures in the stir zone exceed 1100C ous dynamic recrystallization since the
cooling rates, the diagram predicts the (2012F). At 1100C (2012F), the mild dislocations tend to dissociate making re-
presence of mainly ferrite and pearlite steel is austenitic (FCC). Total strains dur- covery (involving cross-slip and climb
with some intermediate product for a ing FSW are thought to exceed 100 (=1), mechanisms) more difficult. Al alloys,
t85 of 60 s, such as that experienced in and strain rates have been estimated at with their high SFE, however, tend to un-
this work. 102 s1 (Ref. 7). dergo continuous dynamic recrystalliza-
Several restorative processes may op- tion (Ref. 22).
Microstructural Evolution in the erate solely or in combination during hot Under the conditions of FSW encoun-
Stir Zone working including dynamic recovery, dy- tered
. in this work (T>1100C and
namic recrystallization, and metadynamic 102s1), the material in the stir zone of
Microstructural evolution in the stir recrystallization (Ref. 25). Dynamic re- mild steel is expected to undergo discon-
zone is not as easy to interpret as that in crystallization may also be defined as oc- tinuous dynamic recrystallization (Refs.
WELDING JOURNAL 7 -S
WELDING RESEARCH
Fig. 8 Microhardness traverse for friction stir welded 1018 mild steel. Fig. 9 Schematic illustration of the different regions of the HAZ of a
Hardness readings are spatially correlated to the different regions of the friction stir weld on mild steel and their relation to effective temperatures
weld. in the Fe-Fe3C binary system.
25, 34), especially in the region near the Criteria for selection of new materials occurred during the initial plunging stage.
tool shoulder, which experiences greater and designs are clearly needed. Of course, 4) The weld region displayed several
and more prolonged strain relative to the the ideal tool material should be stronger microstructurally distinct regions includ-
rest of the stir zone. Consistent with the and harder than the workpiece material at ing the stir zone (along the weld center-
principles of recrystallization (Ref. 46), the welding temperatures. However, other re- line), a grain-coarsened region (surround-
greater strain near the surface gives rise to quirements for successful tool materials ing the stir zone), a grain-refined region
a finer recrystallized grain size. However, are not well understood. For example, the (encompassing the grain-coarsened re-
the slow cooling rates experienced in the effects of the relationships between the gion), an intercritical region, and a sub-
stir zone after passage of the tool are also thermophysical properties of the tool and critical region containing partially spher-
expected to allow considerable time at the workpiece on heat flow and tool life oidized carbides. The different regions
temperature for grain growth to occur must be studied. Furthermore, the metal- develop in accord with the local thermo-
throughout the remainder of the stir zone, lurgical compatibility between the tool mechanical cycle.
thereby yielding the coarser microstruc- and workpiece, and subsequent effects on 5) Welded samples failed in regions
tures observed away from the top surface. tool life, must be better understood. For corresponding to the base metal and
instance, tool/workpiece alloy combina- demonstrated yield and ultimate tensile
Tool Life Issues tions that form solid solutions may pro- strengths comparable to those of the base
mote alloying of the surface of the tool and metal. Welded samples passed bend tests
As discussed earlier, wear and defor- subsequent local changes in tool proper- to 15% strain in the outer fiber.
mation of the tool, especially the pin, oc- ties that cause a type of high-temperature 6) Results of this study have demon-
curs during FSW of mild steel using cur- corrosive wear to occur. strated the feasibility of FSW of steel with-
rent tool materials. While the wear and out loss of tensile properties. Based on
deformation of the tool do not appear to Conclusions these results, FSW of transformation-
impact tensile properties greatly, there is hardenable steels, HSLA steels, and stain-
considerable concern regarding tool life. 1) Defect-free welds were produced on less steels may be feasible.
In fact, further advances in FSW of high- of 0.25-in.-thick (6.35-mm) mild steel with
temperature materials such as steels, Ti al- FSW over a range of travel speeds from 1 Acknowledgments
loys, and Ni alloys rely on finding tool ma- to 4 in./min (0.42 to 1.68 mm/s). Tool loads
terials, tool designs, and processing strate- during FSW of mild steel at 1 in./min (0.42 The authors wish to thank D. S.
gies to extend tool life. mm/s) were approximately 4200 lbf (18.7 LaPolla and B. Sikora of EWI for their as-
In this work, the greatest amount of wear kN), while measured torques were in the sistance in this work. The authors also ex-
and deformation was found to occur during range of 500 in.-lb (55 N.m). tend their appreciation to Dr. M. Q. John-
the plunging stage. This phenomenon likely 2) Peak surface temperatures close to son and Dr. J. E. Gould, both of EWI, for
occurs due to the high load spikes during 1000C (1832F) were measured on the their careful reviews of this manuscript
plunging that evidently derive from the tool above the shoulder during FSW using and for helpful discussions throughout the
greater flow stress of the cold workpiece thermocouples and an infrared camera course of this work. Finally, the authors
material. Strategies involving preheating system. Extrapolation of measured tem- wish to acknowledge the support of the
the plunge area and maintaining slow peratures and microstructural evidence EWI Cooperative Research Program
plunge rates, as well as use of a partial-pen- suggest peak temperatures of the stir zone (CRP) in funding this effort.
etration, partial-diameter hole, act to limit exceeded 1100C (2012F) and likely sur-
wear and deformation during plunging. passed 1200C (2192F). References
Moreover, an oversize pin can be utilized to 3) Comparisons before and after weld-
allow for material loss during plunging, ing combining both metallographic and 1. Thomas, W. M., et al. 1991. International
yielding a proper sized pin for the welding metrology techniques suggest changes in Patent Application No. PCT/GB92/02203 and
stage. However, these strategies do not tool dimensions resulted from both rub- GB Application No. 9125978.8.
completely resolve the problems, and more bing wear and deformation of the tool. 2. Gould, J. E. , Feng, Z., and Ditzel, P. 1996.
work is needed in this area. The greatest changes in tool dimensions Preliminary modeling of the friction stir weld-
8 -S JANUARY 2003
WELDING RESEARCH
ing process. Proceedings of ICAWT, pp. 225237. Computational Mechanics Publica- Revision 2, 1998. Guide to the Light Microscope
297310. EWI, Columbus, Ohio. tions. Examination of Ferritic Steel Weld Metals. IIW.
3. Kong, H. S., and Ashby, M. F. 1991. Fric- 15. Lienert, T. J., Grylls, R. J., Gould, J. E., 31. ASTM, 2000. ASTM Standard E 8-00:
tion heating maps and their applications. MRS and Fraser, H. L. 1998. Deformation mi- Standard test methods for tension testing of
Bulletin 16(10): 4148. crostructures in friction stir welds on 6061- metallic materials. Annual Book of ASTM Stan-
4. Feng, Z., Gould, J. E., and Lienert, T. J. T651. Proceedings of Hot Deformation of Alu- dards. Vol. 03.01.
1998. A heat flow model for friction stir weld- minum Alloys, pp. 159167. TMS. 32. Easterling, K. 1983. Introduction to the
ing of aluminum alloys. Proceedings of Hot De- 16. Murr, L. E., Liu, G., and McClure, J. C. Physical Metallurgy of Welding. pp. 2326. Lon-
formation of Aluminum Alloys. pp. 149158. 1998. A TEM study of precipitation and related don, U.K.: Butterworths & Co.
TMS. microstructures in friction stir welded 6061 alu- 33. Adams, C. M. 1958. Cooling rates and
5. Gao, Y., and Wagoner, R. H. 1987. A sim- minum. Journal of Materials Science 33: peak temperatures in fusion welding. Welding
plified model for heat generation during the 12431251. Journal 37(5): 210-s to 215-s.
uniaxial tensile test. Metallurgical Transactions 17. Sato, Y. S., Kokawa, H., Enomoto, M., 34. Hughes, K. E., Nair, K. D., and Sellars,
18A: 10011009. and Jogan, S. 1999. Microstructural evolution C. M. 1974. Temperature and flow stress during
6. Braga, H. C., and Barbosa, R. A. 1992. of 6063 during friction stir welding. Metallurgi- hot extrusion of steel. Metals Technology 1(4):
Simulation of the increase in temperature due cal and Materials Transactions 30A: 24292437. 161169.
to adiabatic heating in hot deformation 18. Rhodes, C. G., Mahoney, M. W., Bingel, 35. ASM Handbook, Vol. 18, Friction, lubri-
processes. Proceedings of 47th Brazilian Associ- W. H., Spurling, R. A., and Bampton, C. C., cation and wear technology, p. 48. Materials
ation of Metallurgy and Materials (ABM) Annual 1997. Effects of friction stir welding on mi- Park, Ohio: ASM International.
Conference pp. 441457. ABM. crostructure of 7075 aluminum. Scripta Materi- 36. M. N. Hamdan and A. E. Bayoumi, 1989.
7. Nunes, A. C., Jr., Bernstein, E. L., and ala 36(1): 6975. An approach to study the effects of tool geom-
McClure, J. C. 2000. A rotating plug model for 19. Mahoney, M. W., Rhodes, C. G., Flintoff, etry on the primary chatter vibration in orthog-
friction stir welding. Submitted for publication J. G., Bingel, W. H., and Spurling, R. A. 1998. onal cutting. Journal of Sound and Vibration
in the Welding Journal. Properties of friction stir welded 7075 T651 alu- 128(3): 451469.
8. Murr, L. E., Liu, G., and McClure, J. C. minum. Metallurgical and Materials Transactions 37. Bayoumi, A. E., Barnwal, S., and Hut-
1997. Dynamic recrystallization in friction stir 29A: 19551964. ton, D. V. 1993. Prediction of flank wear and en-
welding of aluminum alloy 1100. Journal of Ma- 20. Jata, K. V., Sankaran, K. K., and gagements from force measurements in end
terials Science 16: 18011803. Ruschau, J. J. 2000. Friction-stir welding effects milling operations. Wear 170: 255266.
9. Christner, B. K., and Sylva, G. D. 1996. on microstructure and fatigue of aluminum 38. Lienert, T. J., and Stellwag, W. L., Jr.
Friction stir weld development for aerospace alloy 7050-T7451. Metallurgical and Materials 2001. Determination of load, torque, and tool
applications. Proceedings of ICAWT 1996, pp. Transactions 31A: 21812192. temperature during friction stir welding of alu-
311320. EWI, Columbus, Ohio. 21. Kinchen, D. G., Li, Z., and Adams, G. P. minum alloys. Abstracts of Papers from the 2001
10. Li, Z. X., Arbegast, W. J., and Hartley, P. 1999. Mechanical properties of friction stir AWS Convention, pp. 152155. Miami, Fla:
J. 1998. Microstructure characterization and welds in Al-Li 2195-T8. Proceedings of the First American Welding Society.
stress corrosion evaluation of friction stir International Conference on Friction Stir Weld- 39. Benjamin, J. S., and Volin, T. E. 1974.
welded Al 2195 and Al 2219 alloys. Proceedings ing. Session 9, Paper 2. Cambridge, The mechanism of mechanical alloying. Met.
of the Fifth International Conference on Trends in England:TWI. Trans. 5A: 19291934.
Welding Research, pp. 568573. Materials Park, 22. Jata, K. V., and Semiatin, S. L. 2000. 40. Benjamin, J. S. 1976. Mechanical alloy-
Ohio: ASM International. Continuous dynamic recrystallization during ing. Scientific American 234(5): 4048.
11. Strangwood, M., Berry, J. E., Cleugh, D. friction stir welding of high-strength aluminum 41. Easterling, K. 1983. Introduction to the
P., Leonard, A. J., and Threadgill, P. L. 1999. alloys. Scripta Materiala 43: 743749. Physical Metallurgy of Welding. pp. 141143.
Characterization of the thermo-mechanical ef- 23. Dawes, C. J., and Thomas, W. M. 1996. London, U.K.: Butterworths & Co.
fects on microstructural development in fric- Friction stir process welds aluminum alloys. 42. Kuo, S. 1987. Welding Metallurgy. pp.
tion stir welded age hardening aluminium- Welding Journal 75 (3): 4145. 326329.New York, N.Y.: John Wiley and Sons.
based alloys. Proceedings of the First 24. Wang, X. L., Feng, Z., David, S. A., 43. Holman, J. P. 1976. Heat Transfer. pp.
International Conference on Friction Stir Weld- Spooner, S., and Hubbard, C. R. 2000. Neutron 111. New York, N.Y.: McGraw-Hill.
ing. Session 11, Paper 3. Cambridge, England: diffraction study of residual stresses in friction 44. Linnert, G. E. 1967. Welding Metallurgy:
TWI. stir welds. Submitted for publication in Pro- Carbon and Alloy Steels. 3rd ed., Vol. 2. pp.
12. Reynolds, A. P. 1998. Mechanical and ceedings of The International Conference on 312325. Miami, Fla: American Welding Soci-
corrosion performance of GTA and friction stir Residual Stresses-6. ety. Excerpted from M. Inagaki and H.
welded aluminum for tailor welded blanks: al- 25. Courtney, T. H. 1990. Mechanical Be- Sekiguchi, 1960. Continuous cooling transfor-
loys 5454 and 6061. Proceedings of the Fifth In- havior of Materials. pp. 309317. New York. mation diagrams of steels for welding and their
ternational Conference on Trends in Welding Re- N.Y.: McGraw-Hill. applications. Transactions of the National Re-
search, pp. 563567. Materials Park, Ohio: 26. EWI Insights. Vol. 11, no. 4, 1998. search Institute for Metals (Japan) 2(2):
ASM International. 27. Lienert, T. J., and Gould, J. E. 1999. Fric- 102125.
13. Karlsson, L., Svensson, L.-E., and Lars- tion stir welding of mild steel. Proceedings of the 45. Humphries, F. J. 1997. A unified theory
son, H. 1998. Characteristics of friction stir First International Conference on Friction Stir of recovery, recrystallization and grain growth
welded aluminum alloys. Proceedings of the Welding. Session 10, Paper 3. Cambridge Eng- based on the stability and growth of cellular mi-
Fifth International Conference on Trends in land: TWI. crostructures. Acta Materiala 45(10):
Welding Research. pp. 574579. Materials Park, 28. Thomas, W. M.,Threadgill, P. L. , and 42314240.
Ohio: ASM International. Nicholas, E. D. 1999. Feasibility of friction stir 46. Shewmon, P. G. 1969. Transformations in
14. Haagensen, P. J., Midling, O. T., and welding of steel. Science and Technology of Metals. 1st ed. pp. 9397, New York, N.Y.: Mc-
Ranes, M. 1995. Fatigue performance of fric- Welding and Joining 4(6): 365372. Graw-Hill.
tion stir butt welds in a 6000 series aluminum 29. Occupational Safety and Health Admin-
alloy. Proceedings of the International Confer- istration, 2001. Occupational Exposure to Hexa-
ence on Computer Methods and Experimental valent Chromium. OSHA RIN: 1218-AB45.
Methods for Surface Treatment Effects II, pp. 30. IIW Document IX-1533-88 IXJ-123-87
WELDING JOURNAL 9 -S