Optimisation of Inertia Friction Welding Steel to 6061 Aluminium

Optimisation of Inertia Friction Welding Steel to 6061 Aluminium J.L. Rutherford, P.B. Prangnell School of Materials, University of Manchester, U.K. 1 Introduction There is a growing need to join aluminum to other materials in automotive manufacturing to reduce fuel consumption through the substitution of light alloys for steel and to allow the use of combinations of different types of materials in more weight efficient hybrid structures. The successful welding of materials like steel to aluminum alloys with fusion processes is very difficult due to the high levels of brittle intermetallic phases that can form [1,2]. In this context, friction welding methods are particularly attractive because the welds can be produced in the solid state with a lower heat input. In addition, friction welding is far more efficient than alternative laser or arc welding techniques. While rotational friction welding is restricted in its application to axi-symmetric geometries, it is still widely used for components like steering shafts, valve stems and drive trains [3,4]. There are two main rotational friction welding methods, direct drive and inertia welding. In direct drive friction welding one part is rotated by a motor and the welding process typically involves two stages - a low pressure constant rotation rate friction stage, where heat is generated, and a higher pressure upsetting stage, where the weld is formed. In comparison, inertia friction welding (IFW) is a simpler process where the required energy is stored in a flywheel [4]. In this case the rotating part is forced against the stationary section, under a constant pressure, while simultaneously cutting the drive to the flywheel, whereupon rapid deceleration occurs and upsetting naturally takes place when the material becomes hot enough to plastically flow. With IFW there are only three adjustable parameters, the applied pressure (P), rotational speed (ω), and moment of inertia of the flywheel (I). Both processes rely on the expulsion of the original contacting surfaces as flash during the upsetting stage to remove any surface contamination and oxide, in order to form a metallurgical bond, by achieving intimate contact between the two deforming surfaces under conditions of high temperature and pressure [4]. However, when welding aluminum to steel, because the temperature is limited by the low melting point of aluminum the steel does not flow [5-7]. This restricts oxide removal by flash formation to the softer aluminum part and the original steel surface is largely undeformed, although it has been claimed some very localized wear/deformation can occur at the interface [8]. Nevertheless, several papers have reported that direct drive friction welding can still give joint strengths of higher than 80% of the aluminum parent in dissimilar steelaluminum welds [5-8]. In this work control of the level of interfacial reaction has proved critical, with a compromise required between the minimum friction time for sufficient heat generation to achieve bonding, and the short weld times necessary to restrict the thickness of the interface layer and give better strength [5,6,8]. To date no comparable research has been published for inertia welding, which can produce welds with a shorter cycle time and thus potentially reduce the level of interfacial reaction. Furthermore, despite the fact the steel component does not deform in the process, little attention has been given to the effect of the condition of the prior steel surface, in terms of cleanli-

ness and roughness on bond formation. Here we will attempt to address these issues by investigating optimization of the IFW process, for a typical steel to Al-alloy combination, in terms of the effect of the welding parameters and steel surface condition on the weld strengths obtainable and their relationship to the bonding mechanisms at the interface. 2 Experimental Inertia welding was performed between 6061-T6 aluminum alloy and C45 carbon steel 18mm diameter bars using a Production Technology Inc. M120 inertia welder. For the standard welding procedure both weld surfaces were cleaned immediately prior to welding by abrasion with 120 grit SiC paper and washing in solvent. To determine the optimum conditions a broad spectrum of parameters were used. The applied pressure was varied from P = 60 to 300 MPa, flywheel moment of inertia I = 0.07-0.36 kg m 2, and rotational speeds between ω = 1,225 and 2,225 rpm, giving a kinetic energy range of ~ 0.6 to 4.3 kj. Weld tensile strengths were tested using 100 mm long welded bars with threaded ends, after flash removal, using a cross head speed of 0.01mm per second. Once the optimum welding conditions were determined the effect of roughness of the steel surface was studied by grinding immediately prior to welding using different SiC paper grades, bench marked against a metallographically polished surface. The effect of exposure of the cleaned steel surface was also studied by leaving the samples for different times prior to welding after grinding. Post weld heat treatments were also applied to determine the influence of interfacial reaction by increasing the thermal exposure using different hold times at the 6061 alloys solution treatment temperature (530 C). The degree of bonding for all of the weld conditions was determined by using image analysis from the area of aluminum left adhered to the steel surface on the fractured tensile samples, after etching to darken the unbonded clean steel surface. This was compared to measurements of the relative length of interface that was void, or crevasse, free and appeared to be fully adhered in metallographic sections by SEM analysis. However, with this second approach it was impossible to tell if there was a metallurgical or kissing bond between the adherends. SEM analysis of the metallographic sections was also used to measure the level of interface roughness using the mean normal distance from a straight datum line along the interface [9]. 3 Results and Disscussion An example of the pressure, ram displacement (upset), and rotational speed decay curve, against time, for a full inertia weld cycle, is shown in Fig. 1, along with metallographic sections though a weld produced under optimum conditions (ω = 2225 RPM, P = 170 MPa, I = 0.07kg m 2 ). It can be seen in Fig. 1a that the flywheel starts to decelerate when the pressure is applied and the ram displacement initially accelerates as the material temperature rises. At the end of the cycle the rotation stops abruptly when there is insufficient torque remaining to deform the cooling welded parts. The whole weld cycle is very short (~ 0.5 sec). Although a friction stage cannot be unambiguously distinguished, as in direct drive friction welding, the initial change in slope of the ram displacement curve is taken as indicative of the end of rubbing friction, where the parts are heating up prior to substantial upsetting occurring (Fig. 1a).

In Fig. 1b it can be seen that the grain structure in the aluminum part shows evidence of a large localized plastic strain near the interface, with compression and reorientation of the parent grain structure. The deformation zone and HAZ are wider near the edge of the bar. In comparison no deformation, or HAZ, was observed in the steel side of the weld in any of the samples analyzed by SEM. (a) (b) Fig. 1. (a) Example pressure, ram displacement (upset), and rotational speed decay curves, against time along with metallographic sections showing the deformation in the Al part (b) produced under optimum conditions (ω = 2225 RPM, P = 170 MPa, I = 0.07 kg m 2 ). (a) (b) Fig. 2. Weld tensile strengths (a) and weld cycle time (b) vs pressure, for different combinations of inertia and rotation speed; 4.3 kj, 0.16 kg m 2, 2225 rpm; 3 kj, 0.36 kg m 2, 1225 rpm; 1.9 kj, 0.07 kg m 2, 2225 rpm; 1.3 kj, 0.16 kg m 2, 1471 rpm; 0.6 kj, 0.07 kg m 2, 1225rpm. 2.1 Effect of Process Parameters In Fig. 2a the weld tensile strengths are shown plotted against pressure for different combinations of inertia and rotation speed, for welds produced with the standard roughness and cleaning treatment. It can be seen that all the curves show an optimum pressure which gives maximum weld strength. The welds made with too high an energy (e.g. >2.5 kj) produced an excessive upset unless low pressures were used and had a lower peak strength. Equally, welds produced with too low an energy showed poor bonding (0.6 kj), due to a lower weld temperature and too little flash being ejected. Optimum conditions were found for an energy of ~ 1.5-2 kj and a ram pressure in the range 150-200 MPa. For too low a pressure bonding is poorer, which suggests that a high pressure is needed for intimate contact between the adherends. Too

high a pressure results in a large upset with more energy being dissipated in flash formation and a lower temperature reached at the interface with a corresponding lower bond strength. However, increasing pressure also reduces the weld time (Fig. 2b). Encouragingly, under optimized conditions (ω = 2225 RPM, P = 170 MPa, I = 0.07 kg m 2 ) the maximum strength achieved was 258 MPa, compared to the aluminum alloy yield strength of 305 MPa, giving a joint efficiency of 84%. A similar percentage area of adhered aluminum was found on the surface of the steel part after fracture (see below 2.2). Using a lower moment of inertia compensated for by a higher rotation rate was found to tend to give an improvement in optimum bond strength (Fig. 2a). A lower inertia results in more rapid deceleration of the flywheel and a shorter weld time, which could potentially reduce the extent of interfacial reaction. However, the dominant effect on the level of reaction is probably the peak temperature reached during welding. The effect of the welding conditions on the weld cycle time (defined in Fig. 1a) can be seen in Fig. 2b, where all the weld times can be seen to reduce with higher pressures, which increases the rate of energy transfer during the rubbing stage and the rate upsetting occurs. From Fig. 2b it can also be seen that welds with a higher flywheel energy take longer and, reducing the fly wheel moment of inertia tends to reduce the total weld cycle increasing the rate energy is transferred to the work piece. Al Steel 10 µm Fig. 3. Backscatter SEM image of a rough interface between 6061 and C45 steel produced under optimum welding conditions with a pre-ground steel surface using 120 grit SiC paper. Fig. 4. Effect of prolonged heat treatment at 530 C post welding and atmospheric exposure of cleaned steel surface prior to welding on tensile strength of samples produced under optimum conditions. Despite the influence of a shorter weld time on an improved bond strength, no obvious reaction layer could be resolved by SEM in any of the IFW weld sections (Fig. 3). In similar Al- Steel direct drive friction weld interface reaction layers have been noted to only become visible by SEM with friction times greater than ~ 0.5 Sec [3], where longer friction times give greater weld temperatures. However, from TEM investigation Yamamoto et al. [8] has noted a linear relationship between thickness of the reaction layer, and friction time between carbon steel and various Al-alloys. In this work the layer was found to contain Fe 2 Al 5, Fe 4 Al 3 and Fe 2 Al 5, depending on the friction tine and alloy combination. Yamamoto et al. [8] also demonstrated a clear relationship between reaction layer thickness and weld strength. A thin probably discontinuous reaction layer would thus be expected even for short weld times [8,10]. To see if a reaction layer was contributing to a poorer weld strength samples produced under optimum conditions were post-weld heat treated at their solution treatment temperature for prolonged periods (Fig. 4). No change in strength was seen for up to half an hour at

530 C, but for longer exposures there was a progressive decline to zero strength after 48 hours. This implies the reaction layer is not that significant in IFW due to the short cycle time, or more rapid kinetics are seen under welding conditions, and suggests higher resolution studies are needed to better understand this behavior. Fig. 5. Effect of surface roughness on weld strength and the corresponding area of adhered aluminum found on the fracture surfaces and measured along weld cross sections prior to tensile testing, for welds produced under the standard optimum conditions. 2.2 Effect of Surface Condition on Weld Strength When dissimilar welds are produced the surface condition of the non-deforming steel part is clearly important, as any oxide and contamination can only be removed by abrasion/reaction with the softer Al alloy and is not directly expelled in the flash. To investigate the importance of this variable the influence of different levels of surface roughness and corrosion by atmospheric exposure were studied using the standard optimum welding conditions. In Fig. 5 the effect of surface roughness on weld strength is shown, along with the area of adhered aluminum found on the fracture surfaces and measured along weld cross sections. It can be seen, that for the normal 120 grit SiC grinding treatment adopted, there is an increase in weld strength by 100 MPa relative to that seen for a polished finish on the steel bonding surface, which gave a weld strength of only ~ 150 MPa. There is a corresponding large reduction in the area of aluminum adhered to the steel surface after fracture, which decreases from ~ 80 to 30% on reducing the interface roughness to a polished finish (Fig. 5). Measurements of the relative adhered interface length from SEM cross sections (that was free of interfacial pores or cracks) did not decrease as significantly with reduced roughness suggesting that there are areas where a kissing bond occurs that can not be distinguished from regions of metallurgical bonding in the weld cross sections. Bonding on a polished surface therefore only occurs in patches, presumably where the oxide was originally thinner or has been disrupted locally by the welding process. The role of surface roughness is thus clearly very important in friction welding when one body is non-deforming. Although some reentrant angles can be seen in Fig. 3, such a large effect cannot be attributed simply to mechanical locking. The increase in interfacial area will contribute to a strength increase, but it is also possible that the slip sticking conditions are different at a rough surface and this influences the metal flow of the softer material at the interface during the bonding stage of the weld cycle. To investigate the influence of surface contamination, conventionally prepared steel weld samples were left for different times exposed to atmospheric conditions, after the usual surface preparation before welding. The results of this exercise are shown in Fig. 4. From this data it can be seen that this also causes a substantial loss of weld strength, due to the development of a more complete and thicker oxide layer on the initially clean steel surface. Indeed after 24 hrs exposure the weld strength has reduced by half. This effect further illustrates the

limited capacity of the soft aluminum part to clean contaminants from the undeforming steel surface, which would normally be removed in the flash in a conventional weld [4]. 4 Conclusions The influence of process and interface variables on the performance of aluminum 6061 alloy to C45 carbon steel inertia rotational friction welds has been studied. Under optimized conditions bond strengths of 84% of the parent 6061 alloy s proof stress were achieved. This corresponded to a weld energy of ~ 2 kj, a low flywheel inertia and an applied pressure of 170 MPa. The bond strength was seen to improve with lower flywheel inertias and shorter weld times. No evidence of interfacial reaction could be resolved by SEM analysis, owing to the very short weld cycles possible with this process. Because the weld temperatures were restricted by the softer Al material, the steel part did not deform, which prevented oxide and contaminates being expelled from the steel surface during flash formation. As a result the strength of the welds was found to be extremely sensitive to the condition of the steel surface, increasing dramatically with surface roughness. Minimizing the extent of oxidation of the steel surface prior to welding was also shown to be of vital importance to the weld strength. 5 Acknowledgements This research was funded through the University of Manchester EPSRC Light alloys Portfolio Partnership (EP/D029201/1). In addition the authors would like to thank Chris Heason of Corus, Swinden Labs UK, for the provison of materials. 7 References [1] Katayama, Laser Welding Tech., 2002, 50 69-73. [2] Peyre, P. Sierra, G. Deschaux-Beaume, F. Stuart, D., Fras, G., Mat. Sci. Eng., 2007, 444A, 327-338. [3] Spindler, D.E., Weld. J, 1994, 37-42. [4] Wang, K.K., Welding Research Council Bulletin, 2004, No.204, p. 1-22. [5] Fukumoto, S., Tsubakino, H., Okita, K., Aritoshi, M., Tomita, T., 1999, 15, 1080-1086. [6] Yokoyama, T. Ogawa, K., Welding Int., 2003, 17, 514-523 [7] Kato, K., Tokisue, H, Welding Int., 2004, 18, 861-8672. [8] Yamamoto, N. Takahashi, M., Artoshi, A., Ikeuchi, K. Mat. Sci. Forum, 2007, 539-543, 3865-3871. [9] Yu, X., Wang, L., Int. J Machine Tools & Man., 1999, 39, 459-469l. [10] Hirose, A., Imaeda, H. Kondo, M, Kobayashi, K.F., Mat. Sci. Forum, 2007, 539-543, Part 4, 3888-3893.