- Email: [email protected]
Linear friction welding of a near-β titanium alloy
Available online at www.sciencedirect.com
Acta Materialia 60 (2012) 770–780 www.elsevier.com/locate/actamat
Linear friction welding of a near-b titanium alloy E. Dalgaard a,b,⇑, P. Wanjara b, J. Gholipour b, X. Cao b, J.J. Jonas a b
a Materials Engineering, McGill University, 3610 University Street, Montreal, QC, Canada Aerospace Manufacturing Technology Centre, Institute for Aerospace Research, National Research Council Canada, 5145 Decelles Avenue, Montreal, QC, Canada
Received 8 April 2011; accepted 18 April 2011 Available online 16 November 2011
Abstract The linear friction welding (LFW) behaviour of near-b titanium alloy Ti–5Al–5V–5Mo–3Cr (Ti-5553) was investigated by varying the processing conditions of frequency and axial pressure. The examined mechanical properties of the welded material included microhardness and tensile properties. The maximum strains experienced by the material during LFW for each set of welding parameters were estimated based on the process parameters and then evaluated using Aramis, a three-dimensional optical deformation measurement system. The LFWed Ti-5553 was examined with electron backscatter diffraction techniques to relate the texture and phase changes to the thermomechanical conditions. Characterisation of the welds included analysis of the microstructural features of the weld region and the thermomechanically affected zone in relation to the parent material. Crown Copyright Ó 2011 Published by Elsevier Ltd. on behalf of Acta Materialia Inc. All rights reserved. Keywords: Linear friction welding; Ti-5553; Microstructure; Tensile properties
1. Introduction Beta (b) and near-b titanium alloys are of increasing interest in the aerospace industry due to their better formability and toughness as compared with the more common a+b Ti–6Al–4V alloy. High-strength metastable b alloys such as Ti-5553 have potential to replace steel as the preferred material for large components such as the landinggear truck beam on the latest generation of airframes [1]. Ti-5553 was introduced in 1997 by TIMET and has the nominal composition 5 wt.% Al, 5 wt.% V, 5 wt.% Mo, 3 wt.% Cr, with the balance being Ti. Table 1 lists some typical physical and mechanical properties of this alloy in the solution treated and aged (STA) condition. The forging behaviour of Ti-5553 is similar to that of Ti–10V–2Fe–3Al (Ti–10–2–3), though the higher b transus temperature of the former (856 °C) vs. the latter (800 °C) allows higher forging temperatures [2]. ⇑ Corresponding author at: Materials Engineering, McGill University, 3610 University Street, Montreal, QC, Canada. E-mail address: [email protected] (E. Dalgaard).
Weldability is a classic problem with Ti and its alloys. The metal rapidly reacts with atmospheric gases in the molten state, requiring protective gas shielding in order to join it successfully using fusion welding methods. The low thermal conductivity of Ti leads to longer weld times for low energy density processes such as TIG welding. The long weld times lead to slow cooling, requiring gas protection for an extended time to protect the highly reactive Ti. High energy density methods, such as laser welding or electronbeam welding, partially solve this problem by allowing a rapid welding cycle (heating, melting and cooling) through localised heat input, but still require protection (gas shielding for all areas above 350 °C or vacuum) during the time that the weld is molten. Clearly, Ti is a prime candidate for the development and application of solid-state welding methods. Linear friction welding (LFW), a solid-state process consisting of oscillating one part against a stationary part while applying an axial load (see Fig. 1), eliminates the necessity for a protective environment when welding, since the material does not reach fusion temperatures. No axial symmetry is required and the parts can have quite complex
1359-6454/$36.00 Crown Copyright Ó 2011 Published by Elsevier Ltd. on behalf of Acta Materialia Inc. All rights reserved. doi:10.1016/j.actamat.2011.04.037
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780 Table 1 Physical and tensile properties of Ti-5553 in the STA condition [2]. Average b transus, °C Density, kg m3 Tensile elastic modulus, GPa Compressive elastic modulus, GPa Ultimate tensile strength (UTS), MPa Yield strength (YS) @ 0.2%, MPa Per cent elongation at fracture (%El.), with gauge length of 4D
856 4650 112 113 1236 1174 13
Frictional Pressure
Stationary (Top) H Reciprocal Motion (Bottom) W
L
771
the materials are brought to rest after the desired shortening has been attained. Once the materials have been brought to rest and aligned, the axial pressure is increased and the weld is consolidated [14]. Interest in manufacture and/or repair of aircraft components using emerging welding technologies of importance to the aerospace industry has created an incentive for investigating the LFW of Ti-5553. This research work forms part of a larger programme on the research and technology development of LFW for the manufacture of near-a, a–b and near-b Ti alloys. Microstructure characterisation, texture analysis and mechanical property evaluation of the welds were performed as part of the LFW process development trials to define the optimum parametric window. The intent of this paper is to report on the various findings of this research work on LFWed Ti-5553. 2. Experimental procedure
a
Horizontal Fig. 1. Sample geometry and oscillation direction [6].
geometries involving curves. With heat generated directly at the interface in friction welding, a high energy density (low heat input) comparable to that developed in laser or electron-beam welding can be achieved. This, combined with the low thermal conductivity of Ti and its alloys, creates a very small heat-affected zone (HAZ) [3]. The process was first patented in 1969 [4,5], and in the early 1980s The Welding Institute (TWI) demonstrated a working LFW machine for metals. Since then the process has been developed for a variety of materials such as Ti alloys [6–8], titanium aluminides [9], and Ni-based superalloys [10–13], with a more detailed examination of the relationship between parameters and properties. Nonetheless, to the knowledge of the authors, little work has been done on the LFW of b Ti alloys. The LFW process can be divided into four distinct phases [6]. The first phase, known as the initial or contact phase, begins with contact between the two pieces in order to initiate the wear of surface asperities. The second phase, known as the transition or conditioning phase, begins when the large wear particles that were created during the first phase begin to be expelled from the interface. Frictional heat creates a soft plasticized region that is no longer able to support the applied axial load and begins to deform permanently. When moving into the third phase, known as the equilibrium or burn-off phase, the flash begins to form [14]. The axial pressure is increased and oscillation continues as in the prior phases. Frictional heat is conducted away from the interface and the plastic zone develops further, extruding material away from the interface as flash. The last phase is known as the deceleration or forge phase, where
The equipment used for welding was an MTS LFW process development system, comprised of two hydraulic actuators: the in-plane actuator that oscillates the lower work piece horizontally; and the forge actuator that applies a downward load through the top stationary workpiece. The technical specifications of the equipment are described elsewhere [6]. Table 2 shows the experimental plan used in this study as well as the calculated maximum strain rate using the equation proposed by Vairis and Frost [15], the measured welding time and the total maximum strain calculated from these two latter quantities. The baseline (BL) values of fBL (frequency) and PBL (pressure) were established based on optima reported for Ti–6Al–4V [6], since there is little in the literature regarding the friction welding of b or near-b Ti alloys and no specific parameters have been published [16]. Forging pressure was maintained at the same level as during oscillation in order to examine the effect of oscillation on the growth of the HAZ without the complication of expelling extra material during the forging phase. In general the a and b phase content of a Ti alloy is very pertinent to its properties and hence to the selection of thermomechanical processing parameters. In LFW, the welding behaviour is closely related to the flow stress; in the case of Ti-5553, since the material is already mainly b phase, the use of parameters which led to flow in the b phase of the Ti–6Al–4V material [6] was felt to be valid (since thermomechanical processing would occur in the b phase in any case). Ti-5553 material in ingot form was sectioned to obtain weld coupons of 13 mm in
Table 2 Samples prepared. Sample Frequency ID
Pressure
Estimated Welding Estimated strain time (s) max. rate (s1) strain
LFW-1 fBL PBL 3.6 LFW-2 Low (0.6fBL) High (1.5PBL) 2.2
1.3 1.6
4.8 3.5
772
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
Fig. 2. Sectioning schematic for metallographic samples extracted from welded coupons: WD, welding oscillation direction; TWD, direction transverse to WD, in the plane of welding; ND, normal to the welding plane.
width (W), 35 mm in height (H) and 26 mm in length (L), as illustrated in Fig. 1. Prior to welding, the contact surfaces of the samples were ground and cleaned with alcohol. LFWed samples were sectioned transverse to the oscillation direction through the weld zone (as shown in Fig. 2) for metallographic and electron backscatter diffraction (EBSD) investigation (i.e. parallel to the plane formed by the transverse and the normal directions). Conventional polishing procedures were used for optical microscopy and for preliminary preparation of the EBSD samples. Final EBSD polishing was carried out in a vibratory polisher for 12 h. It is noteworthy that the material required an extensive and sensitive polishing procedure, as electropolishing is not recommended in order to preserve the b phase. For examination of the microstructure using optical microscopy, etching was done using Kroll’s reagent. Microstructural examination was then performed using an inverted optical microscope (Olympus GX71) equipped with digital image analysis software (AnalySIS Five). Backscatter imaging and EBSD mapping were performed at 20 kV on a Hitachi S-3000N VP-SEM equipped with an
Oxford (HKL) EBSD data acquisition system (polished surface). Microhardness was measured using a Struers Duramin A300 machine with a fully automated testing cycle (stage, load, focus, measure). A load of 300 g was applied using a load cell with closed-loop circuit control, and hardness profiles were determined across the weld region using an average of three measurements for each point, with an indent interval of 0.2 mm and a dwell period of 15 s. For each weld condition, three tensile specimens having a standard sub-size geometry of 25 mm in gauge length, 6 mm in width and 4 mm in thickness were machined in accordance with ASTM E8M-01. All specimens were tested at room temperature using a 250 kN MTS 810 tensile machine equipped with an Aramis 3-D deformation measurement system. Before executing tensile testing, each sample was painted with a high-contrast random pattern of black on a white background. The functionality of the Aramis system depends on the quality of this speckle pattern. The quality of the pattern was verified before mechanical property evaluation to ensure strain recording along the entire gauge length. After examination for pattern recognition, tensile property evaluation was conducted using displacement control at a rate of 2 mm min1 up to the yield point and then 8 mm min1 up to the rupture with the Aramis system set at an acquisition rate of 2 frames per second (fps). 3. Results and discussion The microstructure of the as-received material revealed by optical microscopy and scanning electron microscopy (SEM) is shown in Fig. 3. At low magnification, extremely large b grains averaging 100–500 lm in diameter are visible. The SEM image at high magnification reveals an acicular microstructure within the b grains. One of the large b grain boundaries can be seen crossing the corner of the frame (Fig. 3b). This as-received microstructure is consistent with that expected for an ingot structure [1] without
grain boundary
(a)
(b)
Fig. 3. Parent material showing (a) a dark-field image of the large equiaxed grains (Kroll’s reagent) and (b) high magnification backscatter electron image of acicular substructure using compositional contrast (mirror polished).
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
773
Fig. 4. (a) As-welded LFW-1 showing (b) right and (c) left side flash cross-sections.
any secondary operations, such as thermomechanical processing or solutionizing and/or ageing heat treatments.
structure that is characteristic of the as-received material (Fig. 3b) can be seen clearly.
3.1. Microstructural evolution
3.2. Microtexture measurements
Visual inspection of the interface region in the LFWed Ti-5553 shows an appreciable flash from all four sides of the joint (Fig. 4a), suggesting that the weld is integral [6,14]. The flash length was found to be larger in the direction of the oscillatory movement, i.e. parallel to the specimen length as compared to that along the specimen width. However, unlike LFWed Ti–6Al–4V, which exhibited a series of ridges on the flash extruded in the direction of the reciprocating motion, the LFWed Ti-5553 flash, though rough on the outer surface, displayed no regular ripples under the process conditions examined in the present work. As the flash layer consists of plastically deformed material extruded during the welding process, the difference in the flow behaviours of the two alloys (a + b vs. near-b) is almost certainly responsible for the difference in flash morphology. While Ti–6Al–4V behaves in large part like an anisotropic hexagonal close-packed material with very limited slip systems, Ti-5553 is for the most part a body-centred cubic material with multiple slip systems activated even at room temperature, and certainly at the temperature at which flash is extruded. This leads to the very narrow spread and strong directionality of the flash flow in Ti–6Al–4V and the more vertical spread of the Ti-5553 flash. The microstructure of a welded sample is illustrated in Fig. 5. At high magnification using optical microscopy, faint indications of grain boundaries were seen, but, overall, the grains along the weld centre were not effectively revealed through chemical etching. The low etching response is attributed to the highly deformed and dynamically recrystallized microstructure in the weld zone. In Fig. 5b, the region adjacent to the weld centre is presented; here a few more equiaxed grains are visible, as well as some indication of an acicular structure in certain regions between the recrystallized grains. In the micrograph of Fig. 5c, representing the structure about 0.4 mm from the weld line, this acicular structure is clearly revealed. Finally, in Fig. 5d, 1 mm from the weld line, the undeformed acicular micro-
To reveal the details of the weld microstructure, including delineation of the grain boundaries and phase divisions, orientation data were obtained from the as-received and LFWed samples. Fig. 6 depicts the inverse pole figure and phase fraction map of the as-received material, revealing that grains are equiaxed and 100–250 lm in diameter. Also observed is that the a phase comprises approximately 3% of the surface, and by extension, volume of the material. This phase is mainly concentrated at grain boundaries. In order to improve the accuracy of the phase fraction data, points in the scan with a confidence index below 0.2 were discarded before calculation. In Fig. 7, an inverse pole figure map of sample LFW-1 can be seen. An image quality map is overlaid in order to define the grain boundary locations. Clearly, in the aswelded condition, this alloy undergoes complete dynamic recrystallization in the weld zone (200 lm wide) under the thermomechanical conditions employed, namely the combination of straining at elevated temperatures and high strain rates. Beyond the weld region, an abrupt transition to a thermomechanically affected zone (TMAZ) consisting of mixed recrystallized and deformed grains is observed, followed by a further transition to the original large b grains. While the large equiaxed and deformed grains display a variety of orientations, it can be seen that, within the recrystallized zone, the grains are predominantly oriented with their h1 1 1i directions normal to the sample surface. This represents the weld oscillation direction. This reorientation to a preferred texture may be due to the alignment during oscillation of the primary slip system in this phase, {1 1 0}h1 1 1i, to be parallel with the oscillation direction. Fig. 8 indicates an overall phase area fraction of 0.4% a in sample LFW-1. The a fraction decreases to nearly none (estimated at 0.05%) in the recrystallized weld zone. This is in contrast to the 3% a observed in the as-received material, and indicates that the fast cooling rate experienced by the material did not permit an equilibrium phase fraction to
774
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
(a)
(b)
grain boundary
(c)
(d)
Fig. 5. LFW-1 (Kroll’s reagent): (a) weld centre at high magnification, (b) 100 lm from weld centre, (c) 400 lm from weld centre, and (d) 1 mm from weld centre.
Fig. 6. EBSD maps of as-received Ti-5553: (a) inverse pole figure map; (b) phase fraction map.
develop and metastable b was retained in preference to the formation of a. Once again, based on the accuracy of the TSL OIM Analysis software suggested in the literature [17,18], points with a confidence index below 0.2 were removed from the data set. It was observed that these points were nearly always indexed as a, due perhaps to the large number of lines available in the pattern, but that
manual examination of the Kikuchi pattern and the index assigned to it led to the conclusion that the indexing was erroneous. The width of the recrystallized zone in sample LFW-2 (see Figs. 9 and 10) averages 380 lm at the narrowest point (weld centre), approximately 50% wider than in LFW-1. Once again, the recrystallized zone consists almost entirely
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
HAZ
TMAZ
Weld Center
TMAZ
775
HAZ
240 µm
Fig. 7. LFW-1 image quality combined with inverse pole figure.
Fig. 8. Phase fraction map of LFW-1; points with confidence index below 0.2 removed.
TMAZ
Weld Center
Fig. 9. Inverse pole figure of LFW-2 showing one side of weld zone with adjacent deformed grains.
of grains oriented with the h1 1 1i direction normal to the surface. A close-up scan of this recrystallized region can be seen in Fig. 11. The phase fraction map confirms the earlier finding in the lower-resolution scan that there is little to no a present in these grains, as compared with the TMAZ, which contained 0.5–1 vol.% a.
In Fig. 12, b pole figures are presented for the recrystallized regions of LFW-1 and LFW-2. No a pole figures are presented due to the negligible amount of a phase observed in the material. The texture intensity of these recrystallized zones can be seen to be very high, approximately 10–15 times random. This strong texture is based on a large
776
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
Fig. 10. LFW-2 phase map; step size 2 lm; points with confidence index below 0.2 removed.
Fig. 11. High-resolution scan of equiaxed grain zone at step size 0.10 lm: (a) inverse pole figure combined with image quality map; (b) phase map.
number of small grains and is quite reliable. As expected from the inverse pole figure maps, many of these recrystallized grains are aligned along the h1 1 1i direction. Due to the very large grain size in the unrecrystallized regions of the welded specimens, the textures will need to be measured over larger scanned regions in order to be statistically significant. 3.3. Mechanical testing The microhardness profiles shown in Figs. 13 and 14 reveal that the weld region is somewhat softer than the surrounding TMAZ, which in turn is softer than the surrounding parent material; this is consistent with the micro-
structural observations of a depletion in the weldment. Sample LFW-1, welded at lower pressure and higher frequency, can be seen to display a pronounced hardness drop in the weld with a sharp drop over the range ±1.6 mm. Meanwhile LFW-2, the higher-pressure/lower-frequency sample, displays less softening in the weld centre and a slightly less abrupt drop that, however, spans a wider region as indicated by comparison with the parent metal hardness range. It is noteworthy that as the indentation spacing (200 lm to give five indentation widths from centre to centre) is comparable to the weld width, more detailed analysis of the effect of processing on the hardness distribution would require further investigation with higher-resolution methods such as nanoindentation. Nevertheless, the widths of the
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
777
Fig. 12. b Pole figures of the recrystallized zones in (a) LFW-1; (b) LFW-2. The welding direction (WD) is normal to the figure (not shown); TD indicates the direction transverse to WD, and ND the direction normal to the welding plane.
Fig. 13. Microhardness profile across the weld line for sample LFW-1 (baseline pressure and frequency).
Fig. 14. Microhardness profile across the weld line for sample LFW-2 welded at low frequency and high pressure.
softened regions in the two samples are consistent with the weld zone widths observed in the EBSD scans given the resolution of the hardness profiles. The drop in hardness is readily explained by the fact that no post-weld heat treatment was performed on these samples. The solution treatment of Ti5553 is usually carried
out below the b transus, so that some globular a remains, much of it on grain boundaries. Without this treatment, the strengthening and grain boundary pinning effects (as well as the detrimental effect on fracture toughness) of the globular a that would have formed are absent. Although an increase in strength in the weld region might have been possible due to the grain refining effects of heavy deformation combined with recrystallization at elevated temperature, this increase was not observed, perhaps obscured by the greater effect of softening due to the loss of a. Tensile testing was performed on both the as-received and the welded specimens. The as-received values in Table 3 are for the material that was not heat-treated, and are thus lower than literature values quoted for solutionized and aged Ti-5553. The large variation in the %El. observed is consistent with previous findings for a billet structure without STA, and Fanning has reported [2] lower variation in the elongation when the microstructure of Ti-5553 is homogenised using STA. As compared to the as-received tensile properties, the YS, UTS and %El. values are lower for the LFWed samples. The reduction in tensile strength of the welded samples can be explained by the presence of a softened region in the weldment with a microstructure depleted of a. In addition, there is a distinct difference in tensile properties between sample LFW-1, welded at BL axial pressure and frequency, and sample LFW-2, welded
Table 3 Tensile testing results.
As-received LFW-1 LFW-2
YS (MPa)
UTS (MPa)
Uniform El. (%)
Total El. (%)
1046 ± 13 1019 ± 19 988 ± 16
1108 ± 25 1058 ± 23 1013 ± 10
7.4 ± 3.0 3.0 ± 0.5 2.0 ± 0.1
11.2 ± 6.5 4.0 ± 1.0 2.9 ± 0.9
778
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
(a)
Fracture surface WZ
Fracture surface
WZ
Fig. 15. SEM images of fracture surfaces: (a) as-received, low magnification; (b) welded, low magnification; (c) as-received, high magnification; (d) welded, high magnification.
(b)
Fig. 16. Polarized light micrographs of fractured tensile bars: (a) LFW-1 and (b) LFW-2.
at higher pressure and lower frequency, with LFW-1 displaying higher YS and UTS as well as greater elongation (see Table 3). The better mechanical performance of LFW-1 as compared to LFW-2 can be attributed to the smaller amount of material affected thermomechanically (weld zone and TMAZ), as indicated by the EBSD maps. The fracture surfaces of the welded tensile coupons revealed very large grains (see Fig. 15) that are consistent with fracture occurring in the TMAZ adjacent to the recrystallized weld zone. Both unwelded and welded samples displayed ductile fracture characteristics with some areas of shear. In fact, all welded specimens fractured in the TMAZ within 1 mm of the weld zone, as shown in Fig. 16. Conversely, as-received (unwelded) tensile specimens fractured at random locations along their gauge lengths. The consistent failure of the welded samples in the TMAZ is most likely due to a depletion in the coarse-grained microstructure, which would have a tendency to be weaker than the similarly depleted fine-grained recrystallized weld zone. An example of the strain distribution in the unwelded condition is presented in Fig. 17. This should be compared with that of the welded LFW-1 in Fig. 18. The region of strain concentration indicated as the area of highest intensity in the welded specimen corresponds to the location of the fracture that occurred immediately afterward. While it
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
779
Weld Line
TMAZ
TMAZ
Fig. 17. Strain distribution in as-received Ti-5553 just before fracture.
PM
PM
Fig. 18. Strain distribution in LFW-1 just before fracture.
can be seen that in the weld-free sample the average strain is approximately 12%, rising to about double that at the eventual fracture location, the localisation is much more pronounced in the welded specimen, with an average strain in the majority of the material of only 2%. This rises sharply to 5% just outside the TMAZ and thence steadily but rapidly to the maximum of 9.3%, almost five times the average, at the fracture location adjacent to the weld line. The resolution of the strain distribution image and the difficulty of accurately positioning the weld line (which is not altogether straight) in the centre of the specimen leads to some ambiguity in Fig. 18 as to the exact position of the fracture with respect to the weld zones. However, the fracture micrographs in Fig. 16 show clearly that the fracture occurred next to the weld line in the TMAZ. Both microhardness analysis and tensile evaluation of the welded specimens lead to the conclusion that a softened region exists at and on either side of the weld line. This soft region, depleted of the strengthening a phase, as revealed by the microstructural and textural examinations, represents an area where the strain concentrates. An object is strongest when the strain is evenly distributed over its area. Here, due to the weld zone with its corresponding a phase depletion and grain refinement, and the TMAZ with only a phase depletion, there is strain concentration caused by the inhomogeneity of the strengths of these different regions, and therefore in the flow stress; this results in a localised increase in the strain. Of future interest for welded Ti-5553 is the use of a post-weld heat treatment to promote the precipitation of the strengthening a phase. The present study indicates
that a solutionizing and ageing cycle would be beneficial in improving the properties of the weld. 4. Conclusions Near-b alloy Ti–5Al–5V–5Mo–3Cr alloy was welded using an MTS LFW process development system. Process conditions were varied in order to examine the relationship between process parameters and the microstructural, textural and mechanical properties. The following conclusions can be drawn: 1. The as-received material displayed a large equiaxed b grain structure of about 100–500 lm diameter grains. These grains contained an acicular substructure. The a phase in the as-received material comprised approximately 3% of the total, mainly concentrated at grain boundaries. 2. After welding, a very fine-grained (1–5 lm diameter) recrystallized zone was observed in the weld centre, ranging in width from 240 to 380 lm for the process conditions tested. Within this recrystallized zone, the grains were almost all oriented with their h1 1 1i directions normal to the sample surface, i.e. parallel to the weld oscillation direction. 3. The TMAZ was observed to consist of deformed large b grains with some recrystallization localised at the grain boundaries. Also, less than 1% volume fraction of a phase was observed in the TMAZ and weld zone following welding.
780
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
4. Microhardness testing revealed a softened area within ±2 mm of the weld centre line. 5. Fracture of the welded specimens during tensile testing occurred in the TMAZ within 1 mm of the recrystallized weld zone. 6. The strain in the fracture zone of the welded specimens was approximately five times the average strain in the tested specimen. By contrast, in the as-received material, the strain in the vicinity of the fracture was about double the average.
Acknowledgements The authors are grateful to Standard Aero Limited for the materials used in this study. Thanks are also due to M. Gue´rin, X. Pelletier and D. Chiriac for extensive technical assistance. References [1] Jones NG, Dashwood RJ, Dye D, Jackson M. Metall Mater Trans A 2009;40A:1944–54. [2] Fanning JC. J Mater Eng Perform 2005;14(6):788–92.
[3] Wilhelm H, Furlan R, Moloney KC. Titanium ’95: science and technology. London: Institute of Materials; 1996. p. 620–7. [4] Threadgill P. Knowledge summary, TWI.. [5] Caterpillar Tractor Co., inventor. Improvements in bonding metal parts involving their rubbing contact. UK patent GB1161800 A; 1969. [6] Wanjara P, Jahazi M. Metall Mater Trans A 2005;6A:2149–64. [7] Dalgaard E, Coghe F, Rabet L, Jahazi M, Wanjara P, Jonas JJ. Adv Mater Res 2009;89–91:124–9. [8] Dalgaard E, Wanjara P, Gholipour J, Jonas JJ. Linear friction welding of a forged near-a titanium alloy. Mater Sci Forum, in press. [9] Godfrey S, Strangwood M, Threadgill P. Linear friction welding of a gamma titanium aluminide alloy. In: SF2M/ONERA conference, Paris; 1996. [10] Mary C, Jahazi M. Adv Eng Mater 2008;10(6):573–8. [11] Chamanfar A, Jahazi M, Gholipour J, Wanjara P, Yue S. Metall Mater Trans A 2010;42:729–44. [12] Karadge M, Preuss M, Withers PJ, Bray S. Mater Sci Eng A 2008;491:446–53. [13] Wanjara P, Dalgaard E, Gholipour J, Larose J. Linear friction welding of a single crystal superalloy. Mater Sci Forum, in press. [14] Vairis A, Frost M. Mater Sci Eng 2000;A292:8–17. [15] Vairis A, Frost M. Wear 1998;217(1):117–31. [16] Attallah M, Preuss M, Bray S. Trends in welding research. In: Proceedings of the 8th international conference; 2009. p. 486–91. [17] OIM Version 3.0. Online help. Draper, UT: TSL; 2001. [18] Sharma H, van Bohemen SMC, Petrov RH, Sietsma J. Acta Mater 2010;58:2399–407.
Acta Materialia 60 (2012) 770–780 www.elsevier.com/locate/actamat
Linear friction welding of a near-b titanium alloy E. Dalgaard a,b,⇑, P. Wanjara b, J. Gholipour b, X. Cao b, J.J. Jonas a b
a Materials Engineering, McGill University, 3610 University Street, Montreal, QC, Canada Aerospace Manufacturing Technology Centre, Institute for Aerospace Research, National Research Council Canada, 5145 Decelles Avenue, Montreal, QC, Canada
Received 8 April 2011; accepted 18 April 2011 Available online 16 November 2011
Abstract The linear friction welding (LFW) behaviour of near-b titanium alloy Ti–5Al–5V–5Mo–3Cr (Ti-5553) was investigated by varying the processing conditions of frequency and axial pressure. The examined mechanical properties of the welded material included microhardness and tensile properties. The maximum strains experienced by the material during LFW for each set of welding parameters were estimated based on the process parameters and then evaluated using Aramis, a three-dimensional optical deformation measurement system. The LFWed Ti-5553 was examined with electron backscatter diffraction techniques to relate the texture and phase changes to the thermomechanical conditions. Characterisation of the welds included analysis of the microstructural features of the weld region and the thermomechanically affected zone in relation to the parent material. Crown Copyright Ó 2011 Published by Elsevier Ltd. on behalf of Acta Materialia Inc. All rights reserved. Keywords: Linear friction welding; Ti-5553; Microstructure; Tensile properties
1. Introduction Beta (b) and near-b titanium alloys are of increasing interest in the aerospace industry due to their better formability and toughness as compared with the more common a+b Ti–6Al–4V alloy. High-strength metastable b alloys such as Ti-5553 have potential to replace steel as the preferred material for large components such as the landinggear truck beam on the latest generation of airframes [1]. Ti-5553 was introduced in 1997 by TIMET and has the nominal composition 5 wt.% Al, 5 wt.% V, 5 wt.% Mo, 3 wt.% Cr, with the balance being Ti. Table 1 lists some typical physical and mechanical properties of this alloy in the solution treated and aged (STA) condition. The forging behaviour of Ti-5553 is similar to that of Ti–10V–2Fe–3Al (Ti–10–2–3), though the higher b transus temperature of the former (856 °C) vs. the latter (800 °C) allows higher forging temperatures [2]. ⇑ Corresponding author at: Materials Engineering, McGill University, 3610 University Street, Montreal, QC, Canada. E-mail address: [email protected] (E. Dalgaard).
Weldability is a classic problem with Ti and its alloys. The metal rapidly reacts with atmospheric gases in the molten state, requiring protective gas shielding in order to join it successfully using fusion welding methods. The low thermal conductivity of Ti leads to longer weld times for low energy density processes such as TIG welding. The long weld times lead to slow cooling, requiring gas protection for an extended time to protect the highly reactive Ti. High energy density methods, such as laser welding or electronbeam welding, partially solve this problem by allowing a rapid welding cycle (heating, melting and cooling) through localised heat input, but still require protection (gas shielding for all areas above 350 °C or vacuum) during the time that the weld is molten. Clearly, Ti is a prime candidate for the development and application of solid-state welding methods. Linear friction welding (LFW), a solid-state process consisting of oscillating one part against a stationary part while applying an axial load (see Fig. 1), eliminates the necessity for a protective environment when welding, since the material does not reach fusion temperatures. No axial symmetry is required and the parts can have quite complex
1359-6454/$36.00 Crown Copyright Ó 2011 Published by Elsevier Ltd. on behalf of Acta Materialia Inc. All rights reserved. doi:10.1016/j.actamat.2011.04.037
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780 Table 1 Physical and tensile properties of Ti-5553 in the STA condition [2]. Average b transus, °C Density, kg m3 Tensile elastic modulus, GPa Compressive elastic modulus, GPa Ultimate tensile strength (UTS), MPa Yield strength (YS) @ 0.2%, MPa Per cent elongation at fracture (%El.), with gauge length of 4D
856 4650 112 113 1236 1174 13
Frictional Pressure
Stationary (Top) H Reciprocal Motion (Bottom) W
L
771
the materials are brought to rest after the desired shortening has been attained. Once the materials have been brought to rest and aligned, the axial pressure is increased and the weld is consolidated [14]. Interest in manufacture and/or repair of aircraft components using emerging welding technologies of importance to the aerospace industry has created an incentive for investigating the LFW of Ti-5553. This research work forms part of a larger programme on the research and technology development of LFW for the manufacture of near-a, a–b and near-b Ti alloys. Microstructure characterisation, texture analysis and mechanical property evaluation of the welds were performed as part of the LFW process development trials to define the optimum parametric window. The intent of this paper is to report on the various findings of this research work on LFWed Ti-5553. 2. Experimental procedure
a
Horizontal Fig. 1. Sample geometry and oscillation direction [6].
geometries involving curves. With heat generated directly at the interface in friction welding, a high energy density (low heat input) comparable to that developed in laser or electron-beam welding can be achieved. This, combined with the low thermal conductivity of Ti and its alloys, creates a very small heat-affected zone (HAZ) [3]. The process was first patented in 1969 [4,5], and in the early 1980s The Welding Institute (TWI) demonstrated a working LFW machine for metals. Since then the process has been developed for a variety of materials such as Ti alloys [6–8], titanium aluminides [9], and Ni-based superalloys [10–13], with a more detailed examination of the relationship between parameters and properties. Nonetheless, to the knowledge of the authors, little work has been done on the LFW of b Ti alloys. The LFW process can be divided into four distinct phases [6]. The first phase, known as the initial or contact phase, begins with contact between the two pieces in order to initiate the wear of surface asperities. The second phase, known as the transition or conditioning phase, begins when the large wear particles that were created during the first phase begin to be expelled from the interface. Frictional heat creates a soft plasticized region that is no longer able to support the applied axial load and begins to deform permanently. When moving into the third phase, known as the equilibrium or burn-off phase, the flash begins to form [14]. The axial pressure is increased and oscillation continues as in the prior phases. Frictional heat is conducted away from the interface and the plastic zone develops further, extruding material away from the interface as flash. The last phase is known as the deceleration or forge phase, where
The equipment used for welding was an MTS LFW process development system, comprised of two hydraulic actuators: the in-plane actuator that oscillates the lower work piece horizontally; and the forge actuator that applies a downward load through the top stationary workpiece. The technical specifications of the equipment are described elsewhere [6]. Table 2 shows the experimental plan used in this study as well as the calculated maximum strain rate using the equation proposed by Vairis and Frost [15], the measured welding time and the total maximum strain calculated from these two latter quantities. The baseline (BL) values of fBL (frequency) and PBL (pressure) were established based on optima reported for Ti–6Al–4V [6], since there is little in the literature regarding the friction welding of b or near-b Ti alloys and no specific parameters have been published [16]. Forging pressure was maintained at the same level as during oscillation in order to examine the effect of oscillation on the growth of the HAZ without the complication of expelling extra material during the forging phase. In general the a and b phase content of a Ti alloy is very pertinent to its properties and hence to the selection of thermomechanical processing parameters. In LFW, the welding behaviour is closely related to the flow stress; in the case of Ti-5553, since the material is already mainly b phase, the use of parameters which led to flow in the b phase of the Ti–6Al–4V material [6] was felt to be valid (since thermomechanical processing would occur in the b phase in any case). Ti-5553 material in ingot form was sectioned to obtain weld coupons of 13 mm in
Table 2 Samples prepared. Sample Frequency ID
Pressure
Estimated Welding Estimated strain time (s) max. rate (s1) strain
LFW-1 fBL PBL 3.6 LFW-2 Low (0.6fBL) High (1.5PBL) 2.2
1.3 1.6
4.8 3.5
772
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
Fig. 2. Sectioning schematic for metallographic samples extracted from welded coupons: WD, welding oscillation direction; TWD, direction transverse to WD, in the plane of welding; ND, normal to the welding plane.
width (W), 35 mm in height (H) and 26 mm in length (L), as illustrated in Fig. 1. Prior to welding, the contact surfaces of the samples were ground and cleaned with alcohol. LFWed samples were sectioned transverse to the oscillation direction through the weld zone (as shown in Fig. 2) for metallographic and electron backscatter diffraction (EBSD) investigation (i.e. parallel to the plane formed by the transverse and the normal directions). Conventional polishing procedures were used for optical microscopy and for preliminary preparation of the EBSD samples. Final EBSD polishing was carried out in a vibratory polisher for 12 h. It is noteworthy that the material required an extensive and sensitive polishing procedure, as electropolishing is not recommended in order to preserve the b phase. For examination of the microstructure using optical microscopy, etching was done using Kroll’s reagent. Microstructural examination was then performed using an inverted optical microscope (Olympus GX71) equipped with digital image analysis software (AnalySIS Five). Backscatter imaging and EBSD mapping were performed at 20 kV on a Hitachi S-3000N VP-SEM equipped with an
Oxford (HKL) EBSD data acquisition system (polished surface). Microhardness was measured using a Struers Duramin A300 machine with a fully automated testing cycle (stage, load, focus, measure). A load of 300 g was applied using a load cell with closed-loop circuit control, and hardness profiles were determined across the weld region using an average of three measurements for each point, with an indent interval of 0.2 mm and a dwell period of 15 s. For each weld condition, three tensile specimens having a standard sub-size geometry of 25 mm in gauge length, 6 mm in width and 4 mm in thickness were machined in accordance with ASTM E8M-01. All specimens were tested at room temperature using a 250 kN MTS 810 tensile machine equipped with an Aramis 3-D deformation measurement system. Before executing tensile testing, each sample was painted with a high-contrast random pattern of black on a white background. The functionality of the Aramis system depends on the quality of this speckle pattern. The quality of the pattern was verified before mechanical property evaluation to ensure strain recording along the entire gauge length. After examination for pattern recognition, tensile property evaluation was conducted using displacement control at a rate of 2 mm min1 up to the yield point and then 8 mm min1 up to the rupture with the Aramis system set at an acquisition rate of 2 frames per second (fps). 3. Results and discussion The microstructure of the as-received material revealed by optical microscopy and scanning electron microscopy (SEM) is shown in Fig. 3. At low magnification, extremely large b grains averaging 100–500 lm in diameter are visible. The SEM image at high magnification reveals an acicular microstructure within the b grains. One of the large b grain boundaries can be seen crossing the corner of the frame (Fig. 3b). This as-received microstructure is consistent with that expected for an ingot structure [1] without
grain boundary
(a)
(b)
Fig. 3. Parent material showing (a) a dark-field image of the large equiaxed grains (Kroll’s reagent) and (b) high magnification backscatter electron image of acicular substructure using compositional contrast (mirror polished).
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
773
Fig. 4. (a) As-welded LFW-1 showing (b) right and (c) left side flash cross-sections.
any secondary operations, such as thermomechanical processing or solutionizing and/or ageing heat treatments.
structure that is characteristic of the as-received material (Fig. 3b) can be seen clearly.
3.1. Microstructural evolution
3.2. Microtexture measurements
Visual inspection of the interface region in the LFWed Ti-5553 shows an appreciable flash from all four sides of the joint (Fig. 4a), suggesting that the weld is integral [6,14]. The flash length was found to be larger in the direction of the oscillatory movement, i.e. parallel to the specimen length as compared to that along the specimen width. However, unlike LFWed Ti–6Al–4V, which exhibited a series of ridges on the flash extruded in the direction of the reciprocating motion, the LFWed Ti-5553 flash, though rough on the outer surface, displayed no regular ripples under the process conditions examined in the present work. As the flash layer consists of plastically deformed material extruded during the welding process, the difference in the flow behaviours of the two alloys (a + b vs. near-b) is almost certainly responsible for the difference in flash morphology. While Ti–6Al–4V behaves in large part like an anisotropic hexagonal close-packed material with very limited slip systems, Ti-5553 is for the most part a body-centred cubic material with multiple slip systems activated even at room temperature, and certainly at the temperature at which flash is extruded. This leads to the very narrow spread and strong directionality of the flash flow in Ti–6Al–4V and the more vertical spread of the Ti-5553 flash. The microstructure of a welded sample is illustrated in Fig. 5. At high magnification using optical microscopy, faint indications of grain boundaries were seen, but, overall, the grains along the weld centre were not effectively revealed through chemical etching. The low etching response is attributed to the highly deformed and dynamically recrystallized microstructure in the weld zone. In Fig. 5b, the region adjacent to the weld centre is presented; here a few more equiaxed grains are visible, as well as some indication of an acicular structure in certain regions between the recrystallized grains. In the micrograph of Fig. 5c, representing the structure about 0.4 mm from the weld line, this acicular structure is clearly revealed. Finally, in Fig. 5d, 1 mm from the weld line, the undeformed acicular micro-
To reveal the details of the weld microstructure, including delineation of the grain boundaries and phase divisions, orientation data were obtained from the as-received and LFWed samples. Fig. 6 depicts the inverse pole figure and phase fraction map of the as-received material, revealing that grains are equiaxed and 100–250 lm in diameter. Also observed is that the a phase comprises approximately 3% of the surface, and by extension, volume of the material. This phase is mainly concentrated at grain boundaries. In order to improve the accuracy of the phase fraction data, points in the scan with a confidence index below 0.2 were discarded before calculation. In Fig. 7, an inverse pole figure map of sample LFW-1 can be seen. An image quality map is overlaid in order to define the grain boundary locations. Clearly, in the aswelded condition, this alloy undergoes complete dynamic recrystallization in the weld zone (200 lm wide) under the thermomechanical conditions employed, namely the combination of straining at elevated temperatures and high strain rates. Beyond the weld region, an abrupt transition to a thermomechanically affected zone (TMAZ) consisting of mixed recrystallized and deformed grains is observed, followed by a further transition to the original large b grains. While the large equiaxed and deformed grains display a variety of orientations, it can be seen that, within the recrystallized zone, the grains are predominantly oriented with their h1 1 1i directions normal to the sample surface. This represents the weld oscillation direction. This reorientation to a preferred texture may be due to the alignment during oscillation of the primary slip system in this phase, {1 1 0}h1 1 1i, to be parallel with the oscillation direction. Fig. 8 indicates an overall phase area fraction of 0.4% a in sample LFW-1. The a fraction decreases to nearly none (estimated at 0.05%) in the recrystallized weld zone. This is in contrast to the 3% a observed in the as-received material, and indicates that the fast cooling rate experienced by the material did not permit an equilibrium phase fraction to
774
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
(a)
(b)
grain boundary
(c)
(d)
Fig. 5. LFW-1 (Kroll’s reagent): (a) weld centre at high magnification, (b) 100 lm from weld centre, (c) 400 lm from weld centre, and (d) 1 mm from weld centre.
Fig. 6. EBSD maps of as-received Ti-5553: (a) inverse pole figure map; (b) phase fraction map.
develop and metastable b was retained in preference to the formation of a. Once again, based on the accuracy of the TSL OIM Analysis software suggested in the literature [17,18], points with a confidence index below 0.2 were removed from the data set. It was observed that these points were nearly always indexed as a, due perhaps to the large number of lines available in the pattern, but that
manual examination of the Kikuchi pattern and the index assigned to it led to the conclusion that the indexing was erroneous. The width of the recrystallized zone in sample LFW-2 (see Figs. 9 and 10) averages 380 lm at the narrowest point (weld centre), approximately 50% wider than in LFW-1. Once again, the recrystallized zone consists almost entirely
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
HAZ
TMAZ
Weld Center
TMAZ
775
HAZ
240 µm
Fig. 7. LFW-1 image quality combined with inverse pole figure.
Fig. 8. Phase fraction map of LFW-1; points with confidence index below 0.2 removed.
TMAZ
Weld Center
Fig. 9. Inverse pole figure of LFW-2 showing one side of weld zone with adjacent deformed grains.
of grains oriented with the h1 1 1i direction normal to the surface. A close-up scan of this recrystallized region can be seen in Fig. 11. The phase fraction map confirms the earlier finding in the lower-resolution scan that there is little to no a present in these grains, as compared with the TMAZ, which contained 0.5–1 vol.% a.
In Fig. 12, b pole figures are presented for the recrystallized regions of LFW-1 and LFW-2. No a pole figures are presented due to the negligible amount of a phase observed in the material. The texture intensity of these recrystallized zones can be seen to be very high, approximately 10–15 times random. This strong texture is based on a large
776
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
Fig. 10. LFW-2 phase map; step size 2 lm; points with confidence index below 0.2 removed.
Fig. 11. High-resolution scan of equiaxed grain zone at step size 0.10 lm: (a) inverse pole figure combined with image quality map; (b) phase map.
number of small grains and is quite reliable. As expected from the inverse pole figure maps, many of these recrystallized grains are aligned along the h1 1 1i direction. Due to the very large grain size in the unrecrystallized regions of the welded specimens, the textures will need to be measured over larger scanned regions in order to be statistically significant. 3.3. Mechanical testing The microhardness profiles shown in Figs. 13 and 14 reveal that the weld region is somewhat softer than the surrounding TMAZ, which in turn is softer than the surrounding parent material; this is consistent with the micro-
structural observations of a depletion in the weldment. Sample LFW-1, welded at lower pressure and higher frequency, can be seen to display a pronounced hardness drop in the weld with a sharp drop over the range ±1.6 mm. Meanwhile LFW-2, the higher-pressure/lower-frequency sample, displays less softening in the weld centre and a slightly less abrupt drop that, however, spans a wider region as indicated by comparison with the parent metal hardness range. It is noteworthy that as the indentation spacing (200 lm to give five indentation widths from centre to centre) is comparable to the weld width, more detailed analysis of the effect of processing on the hardness distribution would require further investigation with higher-resolution methods such as nanoindentation. Nevertheless, the widths of the
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
777
Fig. 12. b Pole figures of the recrystallized zones in (a) LFW-1; (b) LFW-2. The welding direction (WD) is normal to the figure (not shown); TD indicates the direction transverse to WD, and ND the direction normal to the welding plane.
Fig. 13. Microhardness profile across the weld line for sample LFW-1 (baseline pressure and frequency).
Fig. 14. Microhardness profile across the weld line for sample LFW-2 welded at low frequency and high pressure.
softened regions in the two samples are consistent with the weld zone widths observed in the EBSD scans given the resolution of the hardness profiles. The drop in hardness is readily explained by the fact that no post-weld heat treatment was performed on these samples. The solution treatment of Ti5553 is usually carried
out below the b transus, so that some globular a remains, much of it on grain boundaries. Without this treatment, the strengthening and grain boundary pinning effects (as well as the detrimental effect on fracture toughness) of the globular a that would have formed are absent. Although an increase in strength in the weld region might have been possible due to the grain refining effects of heavy deformation combined with recrystallization at elevated temperature, this increase was not observed, perhaps obscured by the greater effect of softening due to the loss of a. Tensile testing was performed on both the as-received and the welded specimens. The as-received values in Table 3 are for the material that was not heat-treated, and are thus lower than literature values quoted for solutionized and aged Ti-5553. The large variation in the %El. observed is consistent with previous findings for a billet structure without STA, and Fanning has reported [2] lower variation in the elongation when the microstructure of Ti-5553 is homogenised using STA. As compared to the as-received tensile properties, the YS, UTS and %El. values are lower for the LFWed samples. The reduction in tensile strength of the welded samples can be explained by the presence of a softened region in the weldment with a microstructure depleted of a. In addition, there is a distinct difference in tensile properties between sample LFW-1, welded at BL axial pressure and frequency, and sample LFW-2, welded
Table 3 Tensile testing results.
As-received LFW-1 LFW-2
YS (MPa)
UTS (MPa)
Uniform El. (%)
Total El. (%)
1046 ± 13 1019 ± 19 988 ± 16
1108 ± 25 1058 ± 23 1013 ± 10
7.4 ± 3.0 3.0 ± 0.5 2.0 ± 0.1
11.2 ± 6.5 4.0 ± 1.0 2.9 ± 0.9
778
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
(a)
Fracture surface WZ
Fracture surface
WZ
Fig. 15. SEM images of fracture surfaces: (a) as-received, low magnification; (b) welded, low magnification; (c) as-received, high magnification; (d) welded, high magnification.
(b)
Fig. 16. Polarized light micrographs of fractured tensile bars: (a) LFW-1 and (b) LFW-2.
at higher pressure and lower frequency, with LFW-1 displaying higher YS and UTS as well as greater elongation (see Table 3). The better mechanical performance of LFW-1 as compared to LFW-2 can be attributed to the smaller amount of material affected thermomechanically (weld zone and TMAZ), as indicated by the EBSD maps. The fracture surfaces of the welded tensile coupons revealed very large grains (see Fig. 15) that are consistent with fracture occurring in the TMAZ adjacent to the recrystallized weld zone. Both unwelded and welded samples displayed ductile fracture characteristics with some areas of shear. In fact, all welded specimens fractured in the TMAZ within 1 mm of the weld zone, as shown in Fig. 16. Conversely, as-received (unwelded) tensile specimens fractured at random locations along their gauge lengths. The consistent failure of the welded samples in the TMAZ is most likely due to a depletion in the coarse-grained microstructure, which would have a tendency to be weaker than the similarly depleted fine-grained recrystallized weld zone. An example of the strain distribution in the unwelded condition is presented in Fig. 17. This should be compared with that of the welded LFW-1 in Fig. 18. The region of strain concentration indicated as the area of highest intensity in the welded specimen corresponds to the location of the fracture that occurred immediately afterward. While it
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
779
Weld Line
TMAZ
TMAZ
Fig. 17. Strain distribution in as-received Ti-5553 just before fracture.
PM
PM
Fig. 18. Strain distribution in LFW-1 just before fracture.
can be seen that in the weld-free sample the average strain is approximately 12%, rising to about double that at the eventual fracture location, the localisation is much more pronounced in the welded specimen, with an average strain in the majority of the material of only 2%. This rises sharply to 5% just outside the TMAZ and thence steadily but rapidly to the maximum of 9.3%, almost five times the average, at the fracture location adjacent to the weld line. The resolution of the strain distribution image and the difficulty of accurately positioning the weld line (which is not altogether straight) in the centre of the specimen leads to some ambiguity in Fig. 18 as to the exact position of the fracture with respect to the weld zones. However, the fracture micrographs in Fig. 16 show clearly that the fracture occurred next to the weld line in the TMAZ. Both microhardness analysis and tensile evaluation of the welded specimens lead to the conclusion that a softened region exists at and on either side of the weld line. This soft region, depleted of the strengthening a phase, as revealed by the microstructural and textural examinations, represents an area where the strain concentrates. An object is strongest when the strain is evenly distributed over its area. Here, due to the weld zone with its corresponding a phase depletion and grain refinement, and the TMAZ with only a phase depletion, there is strain concentration caused by the inhomogeneity of the strengths of these different regions, and therefore in the flow stress; this results in a localised increase in the strain. Of future interest for welded Ti-5553 is the use of a post-weld heat treatment to promote the precipitation of the strengthening a phase. The present study indicates
that a solutionizing and ageing cycle would be beneficial in improving the properties of the weld. 4. Conclusions Near-b alloy Ti–5Al–5V–5Mo–3Cr alloy was welded using an MTS LFW process development system. Process conditions were varied in order to examine the relationship between process parameters and the microstructural, textural and mechanical properties. The following conclusions can be drawn: 1. The as-received material displayed a large equiaxed b grain structure of about 100–500 lm diameter grains. These grains contained an acicular substructure. The a phase in the as-received material comprised approximately 3% of the total, mainly concentrated at grain boundaries. 2. After welding, a very fine-grained (1–5 lm diameter) recrystallized zone was observed in the weld centre, ranging in width from 240 to 380 lm for the process conditions tested. Within this recrystallized zone, the grains were almost all oriented with their h1 1 1i directions normal to the sample surface, i.e. parallel to the weld oscillation direction. 3. The TMAZ was observed to consist of deformed large b grains with some recrystallization localised at the grain boundaries. Also, less than 1% volume fraction of a phase was observed in the TMAZ and weld zone following welding.
780
E. Dalgaard et al. / Acta Materialia 60 (2012) 770–780
4. Microhardness testing revealed a softened area within ±2 mm of the weld centre line. 5. Fracture of the welded specimens during tensile testing occurred in the TMAZ within 1 mm of the recrystallized weld zone. 6. The strain in the fracture zone of the welded specimens was approximately five times the average strain in the tested specimen. By contrast, in the as-received material, the strain in the vicinity of the fracture was about double the average.
Acknowledgements The authors are grateful to Standard Aero Limited for the materials used in this study. Thanks are also due to M. Gue´rin, X. Pelletier and D. Chiriac for extensive technical assistance. References [1] Jones NG, Dashwood RJ, Dye D, Jackson M. Metall Mater Trans A 2009;40A:1944–54. [2] Fanning JC. J Mater Eng Perform 2005;14(6):788–92.
[3] Wilhelm H, Furlan R, Moloney KC. Titanium ’95: science and technology. London: Institute of Materials; 1996. p. 620–7. [4] Threadgill P. Knowledge summary, TWI.