Effect of Friction Time on the Mechanical and Microstructural Properties of AA6061 Joints by Continuous Drive Friction Welding

Friction welding is becoming a viable replacement of conventional joining methods. Continuous Drive Friction Welding (CDFW) is a type of friction welding used to join rods, tubes and similar shapes. Usually, the process contains a friction stage and a forging stage and the process parameters would be ticked accordingly. AA6061 is an Mg and Si aluminum alloy that is widely used in many industries. This research investigates the effect of friction time on the mechanical properties of AA6061 joints made with CDFW and the relation to the microstructure of the material and thermal profiles. It was found that AA6061 does not require a forging stage where solid joints are obtained without forging and did not fracture within the welding zones. Also, it was concluded that the process parameters are to be tailored in a way that produces a specific type of grain structure within the welding areas. Keywords-AA6061; continuous drive friction welding; tensile strength; yield strength; thermal profile; microstructure; time of friction

INTRODUCTION Friction Welding (FW) is a form of joining materials in solid phase. Some variations of FW are Friction Stir Welding (FSW), Friction Stir Spot Welding (FSSW), and Rotational Friction Welding (RFW). Continuous Drive Friction Welding (CDFW) is a form of RFW that is used to join rods and tubes of similar and dissimilar materials. Such welding techniques allow the materials to be joined without reaching the melting temperature and prevent the formation of intermetallic layers in similar material welding. When compared to conventional welding techniques, CDFW is more favorable, less costly, and more environmentally friendly. Five process parameters are affecting CDFW: rotational speed (RS), friction pressure (P f ), friction time (t f ), upset pressure (P u ), and upset time (t u ). Some studies use burn-off length (BL) instead of t f . In most cases, the friction time and pressure are much less than the upset time and pressure. The Upset stage is also sometimes referred to as the forging stage. Most studies consider a range of rotational speed, a variety of friction pressure, a range of upset pressure, and some specified times of friction or burn-off lengths. Most of the relevant literature is focused on the CDFW of Aluminum Alloy 6061 (AA6061) investigating specific process parameters and does not provide a complete relation of process parameters to mechanical properties, microstructure, and thermal profiles. Authors in [1] investigated the conditions of CDFW of some Al alloys. RS was set to 2000, 2500, and 2800rpm while P f varied between 5 and 7.64Kg, P u ranged between 17.83 and 22.93Kg, and the time of friction t f was set to 4, 7, and 10s. The type of Al alloy was not specified, and no optimum process parameters were outlined. One of the conclusions was that the lowest speed (2000rpm) produced weak welds. The mechanical properties and microstructures for joining AA6061 with AA6082 via a conventional lathe machine were studied in [2]. The study considered a friction time of 3min and an upset pressure of 10. Friction time was too long as compared to similar studies. The tensile strength of AA6061 CDFW and the related microstructure was investigated in [3] with a constant time of friction but prolonged upset times. The highest tensile strength was achieved when the conditions were set to the highest upset pressure and time. Higher upset pressures and times correspond to higher energy and required duration for the welding procedure. On the other hand, the requirement of the prolonged upset stage was not clearly supported, and macrostructure did not provide enough evidence on the conditions of the various welding parameters.
There are some studies related to similar material welding of different Al alloys or metal matrix composites. The friction welding conditions on the mechanical properties of AA7075 and AA5052 were investigated in [4,5]. It was concluded that the initial stage of the welding procedure as a critical factor in determining welding efficiency. A metal matrix composite consisting of AA6061 with 10wt% Al 2 O 3 and SiC was studied in [6]. The considered process parameters were P f , P u , BL, and RS. According to the study, the best-welded samples were obtained with 2000rpm, 2mm BL, 70MPa P f , and 130MPa P u . It was also concluded that low BL causes weak joints where high BL causes excessive flash formations. The literature about CDFW is mainly concerned with the welding of dissimilar materials. One of the main focus areas is welding Al alloys with stainless steel as employing CDFW [7][8][9][10][11][12][13]. Another investigated area is joining Al with different alloys such as copper [14], plain carbon steels [15], and titanium alloys [16]. The effect of the time of friction on the mechanical properties of titanium-stainless steel welding was also investigated [17].
The process of welding similar materials via CDFW is less complicated than dissimilar material welding. However, the authors of this research paper did not come across many studies that investigate such welding conditions. Also, since materials such as Al alloys are soft and have lower melting temperatures than ferrous alloys, the requirement of a forging stage is to be questioned. The process parameters that provide excellent mechanical properties are not clearly defined, especially when it comes to the time of friction. Some studies investigated longer times while others considered shorter periods, both with upset periods. This research aims to establish the best conditions for achieving strong AA6061 joints by CDFW. Such best conditions would be investigated in terms of mechanical properties and microstructure. Finally, the necessity of the forging stage when welding such alloys is validated.

II. MATERIALS AND METHODS
AA6061 is a precipitation-hardened aluminum alloy containing Mg and Si. It has good mechanical properties and usually good weldability. The chemical composition of AA6061 in weight percentages is given in Table I. AA6061 rods (d=10 and l=65mm) were welded via CDFW process. CDFW was achieved using a conventional lathe machine. Friction and upset pressure were achieved via a hydraulic mechanism that was fabricated in the lab and attached to the lathe machine ( Figure 1). The hydraulic mechanism controls and maintains the friction pressure on the fixed part side while the rotating part is fixed to the chuck of the lathe machine. The two parts to be welded are fixed on the two parts of the machine, then centered and brought together with a slight clearance left in between. The welding process starts by operating the lathe machine to start the rotating motion, and then the friction pressure is encaged and applied through the hydraulic mechanism. A timer controls the duration of the welding process and stops both rotational motion and pressure application accordingly. The welding process was conducted according to the parameters shown in Table II. RS and P f values were taken from [1][2][3]6], which obtained good welding samples. Also, identifying the requirements of an upset stage was investigated by not including any forging in the welded samples and comparing the mechanical properties to non-welded samples. As such, the time of friction (t f ) was the only variable since it was desired to understand the effect of welding duration on the welding zones and microstructures. The lower limit was set to 4s due to the current machine capability: it was not possible to achieve welded samples in less than 4s. The welding temperature was measured using an Infrared Dual Laser Point Thermometer (operational range of -50 to 800°C, spatial accuracy of 1mm and time accuracy of 0.5s). It was calibrated to AA6061 before actual temperature measurement. The temperature measurement was video recorded and then analyzed and plotted on a PC. Five samples were acquired for each set of parameter conditions. Three samples were used for tensile testing and two were used for evaluating the welding zones and microstructure. The welded samples are shown in Figure 2. Tensile testing was carried out using a WDW-20 computer-controlled electromechanical universal tensile testing machine according to ASTM: E8 at a rate of 0.5mm/min. The samples were machined to satisfy the tensile test standards and were then tensile tested. Figure 3 shows the tensile sample dimensions. The remaining samples were cut longitudinally, using a precision cutting machine (Pico 75), ground, polished, and etched as per the requirements for such materials [3,18]. Macroscopic graphs were obtained via a digital microscope, while microscopic micrographs were achieved by SEM.

III. RESULTS
The results of the experiments include the analysis of the stress-strain curves for all welded samples as a function of friction time. The welding temperature at the interface was measured for all considered times of friction. Finally, macro and microstructure investigation were conducted.

A. Tensile Properties
The corresponding yield and tensile strength values for AA6061 were 76MPa, and 130MPa respectively. Figure 4 shows the average engineering stress-strain (σ-ε) curves as a function of the time of friction (t f ) or the welding time since no forging was involved. From Figure 4, the σ-ε curves for t f =4, 5, and 6s seem to have relatively close σ y and UTS values. The corresponding σ y (0.02% ε) and UTS varied between 65 and 80MPa and between 117 and 125MPa respectively. This variation is illustrated in Figure 5. It can be shown that both σ y and UTS do not vary much, corresponding to the friction times of 4, 5, and 6s. The samples welded at 5s had a less plastic deformation curve since each curve represents an average of three curves. One of the samples welded at 5s had poor tensile properties, hence, on average, samples welded at 5s had less plastic deformation. The other two samples welded at 5s exhibited good σ-ε curves and ductile fractures. Figure 6 shows the fracture surfaces for the samples fabricated with those friction times. Based on fracture surfaces shown in Figure 6, all samples fabricated at t f =4, 5, and 6s (except the 5s sample mentioned earlier) fractured in a ductile manner, and the fracture occurred within the base metal outside of the welding zones. On the other hand, the σ-ε curves for the higher friction or welding time, i.e., t f equal to 7, 8, 9, and 10s show different indications. It can be seen from Figure 5 that σ y and UTS for these welding times are much less than the short-time welded samples. σ y and UTS for the CDFW sample at t f =7s were 69MPa and 103MPa respectively on average with a substantial variation in UTS between 124 and 86MPa. The large variation in UTS was because one sample exhibited ductile failure, and the failure location was outside the welded area, while the other two samples exhibited brittle failure that was initiated from the interface between the welded surfaces as shown in Figure 6. All the remaining samples for welding times 8, 9, and 10s failed in a brittle manner, hence the deficient tensile properties shown in Figures 4 and 5. When the friction time was set to 8s, σ y and UTS values were found to be about 55MPa and 83MPa, respectively (on average). A large variation in UTS was observed, as indicated in Figure 5. The variation in UTS, in this case, was because two samples failed at the interface with very little material mixing (similar to the failure surfaces shown in Figure 7), while one sample failure showed signs of moderate material joining as shown in Figure 6. As the friction time increased to 9s, σ y and UTS were measured at 57 and 73MPa, respectively (on average) with a little variation in UTS. Finally, σ y and UTS were measured at 47 and 57MPa, respectively (on average) when the welding time was set to 10s. All welded samples fabricated at 9 and 10s failed at the welding interface with very little material mixing, as shown in Figure 8.

B. Thermal Profiles
The thermal profile of the welding procedure during CDFW was investigated by recording the welding temperature during the welding procedure. Figure 9 shows the variation of temperature as a function of time for the various welding times considered in this work. The data were used for constructing the temperature curves of the average of the three measurements at each welding time. Temperature change per second was video-recorded during the experiments. The maximum average temperature (highest peak) seems to be relatively close for all samples except for the sample welded at 10s. Also, the average peak temperature for all samples occurred between 3 and 4s, as shown in Figure 9. The average maximum temperature (welding times: 4-9s) varied between 175 and 185°C while it was 152°C for samples welded at 10s. Meanwhile, the cooling curves for higher friction times seemed to decrease for 2-3 seconds, increase again to a temperature lower than the maximum temperature obtained before, and then continuously decrease until reaching room temperature. The profiles of the thermal curves shown in Figure 9 indicate that a friction time of 4s may be adequate for welding AA6061 samples (with the considered process parameters). Any further increase in welding time affects slightly the maximum temperature but causes a second increase in temperature that might be critical on the mechanical properties of the welded joint.

C. Microscopy
The welded joints were investigated further with a digital microscope and SEM. Figure 10  The samples welded at t f equal to 9 or 10s were not considered anymore since such welding times produce inferior welds for CDFW AA6061 as shown earlier. Referring to the illustrations shown in Figure 10, the sample welded at 4s exhibited minimal flash and the total thermally affected area seems to be relatively large. On the other hand, the 8s welded sample exhibited a large flash, a less total thermal affected area, and a dark region surrounding the welded area inductive of undergoing higher thermal cycles. The samples welded at 6, and 7s fall between the conditions for the 4, and 8s welded samples. In these samples, the total thermally affected area starts decreasing, and the formation of the dark area surrounding the welding area starts becoming visible. Lastly, the width of the central part of the welded area seems to be similar across all samples. To compare the microstructures of the welded samples shown in Figure 10, zoomed images were obtained with the digital microscope as shown in Figure 11 and SEM micrographs, as shown in Figure 12. In order to highlight the effect of welding time on microstructure, the welding times of 4 and 8s were selected. ImageJ was used to further analyze these images by reducing them to black and white, as shown in Figure 11. The 4s welded sample seemed to have a combination of large and small grains while the 8s sample seemed to have almost uniform grain size. The grain size distribution appears to become more consistent as welding time increases. Another observation from Figure 11 is that shorter welding times seem to produce finer grains in the near-flash area (upper areas) as compared to longer welding times. The grain size distribution for the 4s welded sample seems to increase gradually from the near-flash area to the center. Such a feature is not distinctive for the 8s welded sample shown on the bottom right illustration in Figure 11. Scanning Electron Microscopy was performed on 4s and 8s welded samples to further investigate the evidence of microstructure differences. Figure 12 shows the SEM micrographs for the welded samples at 4s (upper), and 8s (lower). We can see that there is a clear difference in the microstructure and comparing the SEM micrographs outlines some of these differences. The grains at 8s welding time were almost the same sizes and grown to relatively uniform sizes, while the 4s sample had a range of non-uniformly sized grains. Furthermore, the 4s welded sample showed many tiny white particles that could be inductive of nucleation of new grains while the 8s welded sample has no such feature. IV. DISCUSSION CDFW of AA6061 seems to depend on the condition of the resulting microstructure. The process parameters considered in this study are shared among the studies related to CDFW of AA6061, yet the results about thermal cycles and short welding times are innovative. Samples welded with less than 6s of friction time without any upset had superior tensile properties than samples welded at longer welding times with similar conditions. The tensile properties of the welded joint decrease significantly as welding time increases, as shown in Figures 4 and 5. Such decline is directly related to the microstructure resulting in each case and the thermal cycles exerted on the joint. With shorter friction time, the material had just enough time to initiate plastic deformation, start mixing with the material from the other side, and form new grain nucleation sites within the original grains. Such effect was demonstrated as tiny white dots, and large grains within the same structure for the sample welded at 4s. As the welding time increases, more material is mixed, and the new nucleating grains had enough time to grow, become fully developed, and have uniform grain size distribution. On average, the 5s welded sample had less plastic deformation than the 4 and 6s welded samples. Such a phenomenon can be attributed to just a weak welding sample that provided this result. Nevertheless, the samples welded at 4s without a doubt were the best in terms of tensile properties. As the welding time increases beyond 6s, the properties become unstable. The samples welded at 7 and 8s showed mixed indications. One of the 7s samples failed in a ductile manner, while the other two failed in a brittle way. Two of the 8s samples failed in a brittle way with evidence of some material joining while one failed in a simple brittle manner with little or no material joining. Such effects become even more distinctive for the 9s and 10s where all samples failed within the welding surface in a brittle way with almost no material joining. The thermal profiles can be related to such phenomena.
The maximum welding temperature occurs around 3-4s in all cases and reaches more than 180°C in most. The thermal profile during CDFW was investigated in [15,[21][22][23]. Most of the research efforts related to thermal profiles are about finite element analysis modeling and about connecting the thermal profile during the process in both experiments and models. The research related to thermal effects on the microstructure is almost none, especially for the considered material. Also, most of the thermal modeling research observed show only one peak of temperature and never two peaks. Any increase of welding time beyond 4s results in a gradual rise in the heat generated within the welding area, causing no additional material joining but acting like a thermal cycle which develops the grain structure. Grain structure seems to become fully developed at higher welding times resulting in weak tensile properties, as shown Figures 4 and 5. The microstructure illustrations support this claim, indicating more uniform grain structure at longer welding times and the development of the dark areas surrounding the welded area, indicating higher thermal cycles. The best welding conditions are when the process parameters are such that the material has just enough time for mixing and initiating new grains. Any process parameter combination that leads to fully developed grains and extended thermal cycles is not favorable and can flaw welding properties. The nonuniform grain size distribution obtained with lower welding times seems to affect the plasticity by introducing more grain boundaries and dislocations hence preventing the advancement of any cracks or defects within the welded area. Such a result appears to be logical since all samples welded at 4-6s (except for one-sample welded at 5s) failed within the base metal away from the welded areas. Also, less welding time indicates less material consumption, less energy required for welding, and, therefore, less operational cost. The effects of the process parameters of CDFW, especially the time of friction on the mechanical, thermal, and structural properties of AA6061 were investigated. The tensile tests indicated that no forging is required when joining such alloys with CDFW and that a short time of friction would result in strong joints. On the other hand, the thermal profiles of the welding times indicated that the peak temperature is reached within 3 to 4s of the welding process regardless of the duration of the welding procedure. Any further increase in the time of friction does not increase the temperature beyond the peak temperature obtained earlier but helps annealing the material. Macro and micro structure investigations showed that the process parameters should be handled in a way to produce a mixture of grain sizes and not allow the grains to fully develop and grow in size. The more uniform the grain size of the welding area, the weaker the welding joint.