Mechanical Properties of Titanium Alloy Friction Welds

Mechanical Sciences Sector, DERA, Farnborough, UK
* TWI, Cambridge, UK

Presented at 'Titanium '99', 9th World Titanium Conference, St. Petersburg, Russia, June 1999.

Abstract

Increased performance goals for aerospace structures result in the increasing selection of high strength titanium alloys. In order to minimise cost, fabricated structures are of great interest, however, conventional fusion welding of high strength titanium alloys often leads to poor mechanical properties. Friction welding is a potential route to obtain enhanced mechanical properties in suitable geometry components. Some initial optimisation of the friction welding parameters for Ti-6Al-4V and Ti-10V-2Fe-3Al alloys has been performed and friction welds in these alloys have been produced. The effect of post weld heat treatment on the microstructure and mechanical properties has been investigated. In order to evaluate the mechanical properties of the welds both tensile and Charpy impact properties have been measured and compared with the parent metal.

1. Introduction

High strength titanium alloys are of interest for structures requiring minimum weight, especially in the aerospace industry. Along with the interest in high strength alloys, there is a growing requirement to join titanium alloy components. In the work reported here the high strength near _β alloy - Ti-10V-2Fe3-Al has been investigated, along with the commonly used Ti-6Al-4V alloy for comparative purposes. Whilst Ti-6Al-4V has been successfully joined using a variety of techniques, very little information is available for the Ti-10V-2Fe-3Al alloy. However, limited data for the TIG and electron beam (EB) welding of Ti-10V-2Fe-3Al has been published ^[1,2] but this has shown the difficulty in obtaining the same level of strength/toughness combination in the weld, as obtainable in the bulk material. For high performance applications an improved strength/toughness combination is needed and for this reason the solid state friction welding route, which has demonstrated a good balance of properties in Ti-6Al-4V ^[3] has been selected for evaluation.

2. Experimental

Test samples for friction welding were in the form of 60mm diameter cylinders, with a 20mm wall thickness and 75mm length. The material used consisted of Ti-6Al-4V (ELI grade), cut from 75mm diameter rolled bar and Ti-10V-2Fe-3Al alloy cut from 200mm diameter forged bar. All welds were made between materials of the same composition.

All welds were made using a continuous drive rotary friction welding machine (max. rotational speed = 900 rev./min, max. axial welding force = 1000 kN and rated transmission power of 75 kW). The process variables of rotation speed, welding force and axial displacement were monitored during welding. Prior to welding the test samples were located into the rotary and stationary chucks of the friction welder. Both components were then degreased using acetone. An initial investigation into the process variables had determined the appropriate welding parameters ^[4] and these were employed in this study.

Following welding some joints were subjected to heat treatment to further evaluate the mechanical properties obtained. These consisted of - Ti-6Al-4V - anneal - 730°C, 2 hours, air cool, and - Ti-10V-2Fe-3Al - solution treatment and age (STA) - 750°C, 1 hour, water quench + 500°C, 8 hours, air cool, or - solution treatment and overage (STOA) - 730°C, 1 hour, air cool + 585°C, 8 hours, air cool.

Metallographic analysis of the friction welds was undertaken, both via conventional optical microscopy and by scanning electron microscopy. In addition to the standard microscopy techniques electron back scattering diffraction analysis was also used with the Ti-10V-2Fe-3Al to assist in the phase determination and the crystallographic texture in the parent metal and at the centre of the friction weld. The mechanical properties were evaluated by bend, tensile and Charpy impact testing. Tensile and impact testing was performed on parent metal and the welds, both as welded and after heat treatment. Bend testing was performed on full length longitudinal strips from the welds and were tested in 3 point bend, using a 3T former. Standard tensile testing to BS EN 1002-1 was undertaken on bar test pieces (gauge length - 5.64mm diameter), both as welded and after heat treatment. Similarly standard Charpy impact testing was performed, with the notch cut at, and the crack propagating along, the weld.

3. Results and Discussion

Friction welds in both the Ti-6Al-4V and Ti-10V-2Fe-3Al alloys were successfully produced ( Fig.1), the flash resulting from the welding process may be clearly seen. Prior to mechanical evaluation this flash was machined off. The macroscopic structure of the friction welds for both alloys were similar to that shown for the Ti-10V-2Fe-3Al in Fig.2. This shows a central weld zone 1.9-2.1mm thick, with a heat affected zone (HAZ) extending beyond this. Similar weld thicknesses have been reported for the α/ _β Ti-17 alloy ^[5] . In contrast, weld thicknesses of ~170µm have been reported for Ti-6Al-4V and IMI 834 alloys using linear friction welding ^[6] .

Fig.1 A typical friction weld of the titanium alloy cylinders.

Fig.2 Macrograph of a Ti-10V-2Fe Al friction weld.

The bend testing of the as welded joints showed a failure angle of 40-45° for the Ti-6Al-4V but only 5-15° for the Ti-10V-2Fe-3Al; all failures occurred at the weld/HAZ interface. The low bend angle achieved for the Ti-10V-2Fe-3Al suggests low weld/HAZ ductility.

Tensile test data for the Ti-6Al-4V alloy are shown in Fig.3 (pairs of data are available for all conditions except the parent material following an annealing treatment). Little change is seen in the strengths following welding or heat treatment. The ductility similarly shows little change after welding, indeed a small improvement may be observed. All of the fractures occurred in the bulk alloy and thus the elongation measured gives little indication of the ductility of the weld metal. This problem is frequently encountered during the mechanical assessment of welds. Notwithstanding this comment, the properties measured for the Ti-6Al-4V welds are consistent with other published work ^[3] .

Fig.3 The tensile properties of Ti-6Al-4V in both the bulk parent condition and after friction welding. Anneal - 735°C, 2 hours, air cool.

Charpy impact testing of Ti-6Al-4V showed a small enhancement in values measured, from 34-40 J in the parent materials to 43-56 J after friction welding ( Fig.4). The impact test directly examines the performance of the weld because of the position of the notch and thus obviates the difficulties seen with the tensile tests in evaluating the weld performance. The increase in impact toughness is thought to be associated with the local change in microstructure after welding. Initially the Ti-6Al-4V exhibits a relatively fine grained equiaxed α/ _β microstructure, which becomes fully transformed at the weld and HAZ and shows a typical lath-like morphology ( Fig.5). The cooling rate from welding of the weld metal is low enough to prevent martensitic α phase formation and an associated reduction in fracture toughness ^[7] . An increased fracture toughness with the change from an equiaxed to a fully transformed microstructure is consistent with previous studies ^[7] . It should be noted that relatively high impact values are found in this material because of its extra low interstitial (ELI) content.

Fig.4 Charpy impact energies for Ti-6Al-4V as bulk parent material and after friction welding. Anneal - 735°C, 2 hours, air cool.

Fig.5 Microstructure of the Ti-6Al-4V alloy in a) the as received condition and b) at the centre of the weld (as welded condition). Back scattered electron images.

Tensile evaluation of the Ti-10V-2Fe-3Al showed the significantly higher strengths achievable (UTS ~1300 MPa) with this alloy compared with the Ti-6Al-4V ( Fig.6). It should also be noted that the heat treatments used with Ti-10V-2Fe-3Al alloy have a much greater effect than with the Ti-6Al-4V. Friction welding of Ti-10V-2Fe-3Al resulted in little change in the tensile strengths and elongations, compared with the as received condition ( Fig.6). Similarly there was little difference between the bulk material and the weld following solution treatment and ageing (STA) ( Fig.6). As seen with the Ti-6Al-4V welded test pieces, fracture occurred in the bulk parent material of the friction welded Ti-10V-2Fe-3Al. The solution treatment and over ageing (STOA) heat treatment of a weld gave little increase in the strengths measured in comparison with the as received condition, however, the ductility was higher (at 15%) than both the as received (at 9-14%) and as welded (at 6-10%) conditions.

Fig.6 The tensile properties of Ti-10V-2Fe-3Al in both the bulk parent condition and after friction welding. STA - 750°C, 1 hour, water quench + 500°C, 8 hours, air cool. STOA - 730°C, 1 hour, air cool +585°C, 8 hours, air cool.

A significant decrease in the Charpy impact energies for Ti-10V-2Fe-3Al was seen after the friction welding, however, this was comparable to that obtained in the bulk material after the solution treatment and ageing ( Fig.7). The low impact energies measured in the as welded condition are consistent with the low bend radii observed for this material. Microstructural analysis via electron back scattered diffraction showed that the α/ _β phase proportions were markedly changed - from ~39% α phase and ~52% _β phase in the parent material to ~0.6% α phase and ~72% _β phase at the weld centre (Note: the phase proportion totals are less than 100% and this reflects the difficulty in indexing the diffraction patterns obtained). Alongside this difference in phase proportions there was a change in microstructural morphology ( Fig.8) from the fully transformed structure in the bulk material to one containing coarse equiaxed _β grains at the centre of the weld, with evidence of transformation products within these grains. In order to further investigate the low toughness of the Ti-10V-2Fe-3Al the crystallographic texture was determined using electron back scattered diffraction analysis. Both the α and _β phase textures were determined for the bulk parent metal from an area ~125 µm x 75 µm ( Fig.9) and because of the large grain size in this material, the texture for only 1 or 2 prior _β grains has probably been determined. However, an orientation relationship was seen between the (0001) α phase plane and the (110) _β phase plane in the bulk material ( Figs 9a and b). The texture at the centre of the weld was determined from a significantly larger area of ~400 µm x 200 µm and hence larger number of grains, than the parent metal. This larger number of grains may be partly responsible for the more diffuse poles measured ( Fig.10) but the large deformation which occurred during welding may also be expected to contribute. It should be noted that no α phase pole figure is presented because of the negligible α phase found in the weld. Despite the differences noted between the parent and weld metal measurements, some similarities can be observed, for example both regions have rotated cube orientations present. More detailed analysis of the texture changes during welding is required to fully evaluate the effect of the welding and the role of texture changes. Thus the combination of the loss of the lamellar α morphology, the increase in the _β phase volume fraction and the texture changes appear to be responsible for the low toughness in the as welded condition.

Fig.7 Charpy impact energies for Ti-10V-2Fe-3Al in bulk parent condition and after friction welding. STA - 750°C, 1 hour, water quench + 500°C, 8 hours, air cool. STOA - STA - 730°C, 1 hour, air cool +585°C, 8 hours, air cool.

Fig.8 Microstructure of Ti-10V-2Fe-3Al - a) bulk parent metal as received and b) centre of weld zone, as welded. Back scattered electron images - SEM.

Fig.9 Crystallographic pole figures (determined from EBSD analysis) for both the α and _β phases in the bulk parent Ti-10V-2Fe-3Al. The relevant planes are shown above each figure. T dirn - transverse direction in the original billet.

Fig.10 Crystallographic pole figures (determined from EBSD analysis) for the _β phase from the centre of the friction weld in the Ti-10V-2Fe-3Al - as welded condition. The planes presented are indicated above each figure. T dirn - transverse direction in original billet, rotational plane of welding wasparallel to the transverse plane.

Post weld heat treatment, using solution treatment and ageing (STA) of the welded Ti-10V-2Fe-3Al gave a similar toughness to the bulk material in the same condition ( Fig.7). However, the solution treatment and overage (STOA) heat treatment increased the impact toughness from ~8 J as welded to ~33 J after heat treatment. The reason for this may be seen in the microstructures shown in Fig.11. After the solution treatment and ageing very fine lath-like α phase was present in the _β grains at the centre of the weld, with coarser α phase at grain boundaries ( Fig.11a). Whilst the weld microstructure after a solution treatment and overageing heat treatment showed coarser α laths throughout the prior _β grains and again with grain boundary α present ( Fig.llb). Thus the high impact toughness after the STOA heat treatment was associated with the relatively coarse α laths present, giving a tortuous crack path. It may be seen that there has been little change in the tensile strengths from the as welded condition to those obtained after the solution treatment and over ageing heat treatment and thus the improved toughness has been achieved without sacrifice of strength. Clearly post weld heat treatment is essential with Ti-10V-2Fe-3Al to obtain the best combination of mechanical properties.

Fig.11 Microstructure of Ti-10V-2Fe-3Al friction welds, centre of the weld zone, after post weld heat treatments of a) STA - 750°C, 1 hour, water quench + 500°C, 8 hours, air cool and b) STOA - 730°C, 1 hour, aircool + 585°C, 8 hours, air cool. Optical microscopy images.

Comparison of the mechanical properties obtained in this study with alternative joining techniques shows that only diffusion bonding offers a similar balance of achievable mechanical properties after joining [8,9]. Indeed, previous work with Ti-10V-2Fe-3Al has shown that obtaining impact toughness values approaching those achieved here, or comparable to the parent metal, via the diffusion bonding route was difficult ^[9] . However, parent metal impact toughness has been measured for Ti-6Al-4V diffusion bonds ^[8] . One important feature that must be considered for structural application of friction welding is the effect of the microstructural change, from typically a fully equiaxed structure to a fully transformed condition at the weld, on the Ti-6Al-4V fatigue behaviour. Differences in the fatigue behaviour between the Ti-10V-2Fe-3Al friction welds and the bulk alloy may be less marked than the Ti-6Al-4V because the Ti-10V-2Fe-3Al alloy is generally used in the fully transformed condition, similar to the weld microstructure.

4. Conclusions

Both Ti-6Al-4V and Ti-10V-2Fe-3Al alloys may be successfully joined using rotary friction welding. The joints produced may exhibit tensile strengths and an impact toughness comparable to the bulk parent metal. Whilst post weld heat treatment may be advisable for the Ti-6Al-4V in order to stress relieve the joints, there is little measurable benefit to the tensile and impact toughness. However, post weld heat treatment is vital to the Ti-10V-2Fe-3Al in order to obtain the best balance of properties. The data here suggests that joints exhibiting a superior combination of strength and toughness than obtained by fusion welding techniques are achievable by friction welding; this combination of properties approaches that measured for diffusion bonds.

5. Acknowledgements

This research is part of TG4 (Materials and Structures) of the UK MoD Corporate Research Programme. The authors would like to thank Mr R Savage for assistance with the tensile testing and Mr T Nicholas for help with the metallographic preparation.

6. References

© Crown copyright/DERA 1999.