FSW between Al alloy and Mg Alloy: the comparative study

3.1 Experimental optimization procedure (EOP)

The hardness of work pieces determine what type of the tool material [35] ought to be. First of all proper tool material (here HDS) was chosen for a particular workpiece materials (here AA 2024-T3 and AZ31B-O) combination. Also the tool dimensions are considered by the thickness of the workpiece only. The tool was chosen according to the self optimized tool geometry suggested by Prado and Murr [36].

Pin of length 4.7 mm, top pin diameter of 6 mm and bottom pin diameter of 4 mm, resulting in mean diameter at 5 mm and at bottom pin has rounded end; diameter of shoulder=20 mm, θ=2°. Plunge depth (PD) of tool=4.9 mm has set, leaving 0.1 mm (or a maximum of 0.15 mm since a low rpm of 300 rpm has employed during FSW) distance between tool pin tip and bottom surface of the workpiece during FSW. By this there will be no, not jointed interface between the two plates, below the tip of pin during and after welding. 200 rpm is too low rpm and 800 rpm is too high, and also V=120 mm/min is quite high for Mg alloy FSW, so, for a bead on plate welding of AZ31B-O plate the following parameters chosen by educated guess: rpm=300 to 800; θ=2°; PD=4.9; traverse speed (V)=100 mm/min, but this did not yield the good surface morphology or surface look of the weld. Keeping everything same and using a tool with shoulder diameter (D)=15 mm equal to 3 times the thickness of workpiece, 5 mm, the weld was performed yielding a good surface morphology. D will be>15 mm for still harder workpiece materials. Malarvizhi and Balasubramanian [38] claims that, in FSW of Al alloy to Mg alloy the ratio between shoulder diameter and thickness of workpiece should be 3.5 to get good weld, but here the author used the ratio, 3 to get good weld between Al alloy and Mg alloy. For all further FSW experiments reported here, the ratio 3 was used. For this weld (D=15 mm) a through hole inside the weld volume and along the length of the weld was obtained, which was ascribed to high V, and this was owing to lack of consolidation or forging action on the workpiece material in the weld volume. PD is already at its maximum and it no longer being increased further to eliminate through hole.

The welding speed contributes to the heat generation lesser than the heat generation by the chosen rotational speed. But in contributing to the forging action on weld, the welding speed comes next to the axial load. Interestingly, higher the welding speed decreases the forging action, and lower the welding speed increases the forging action. As the welding speed decreases, both the heat input to the material as well as the forging action being exercised on the material within weld volume increases. Because the welding speed is less, the tool spends more time at the region over the plate surface where it is rotating resulting in well consolidated weld. Forging action on the weld volume for lesser welding speed occurs for more time per mm of forward motion of FSW tool than the forging action for higher welding speed. This concept confutes the idea, i.e. welding speed does not have much effect on the quality of the weld.

Keeping all parameters and other things same as in previous experiment, welding was done for V=80 mm/min, again a smaller through hole inside the weld volume and along the length of the weld was obtained. Next another weld performed for V=60 mm/min keeping other parameters same and taking the same previous tool; a good weld was obtained at 340 rpm (Figure 1).

As the rotational speed increased, the flow stress of the material decreases and also the forging action increases owing to reduced welding speed of 60 mm/min. Consequently for this experiment the defects altogether disappeared at 340 rpm resulting in sound weld. Pressure on weld volume depends on plunge depth (PD) of tool and traverse speed, since both are constant and rpm increase from a lower value to higher value, plasticization of materials in weld volume increases from low at low rpm to high at higher rpm. At a certain rpm (340 rpm in above paragraph) and at corresponding plasticization of material, material fills each and every corner in the weld volume resulting in well consolidated and sound weld. The parameters at this state can be called optimized parameters and pressure in the weld volume is optimum pressure. After optimum rpm material plasticizes to more extent and flash results owing to more softened material escaping out of the weld volume resulting in defective weld.

Sound weld is directly dependent on flow stress of the WP material, which in turn depends on rpm of FSW tool. Pressure created by FSW tool shoulder on the material in the weld volume, which (pressure) in turn in turn depends on traverse speed of the tool. Low traverse speed (40 to 60 mm/min) gives highest pressure on the weld volume or highest forging action on the material in the weld volume thus consolidating the material in the weld volume yielding sound weld. Flow stress of material depends on rpm (and so temperature) and traverse speed (V); higher the rpm of the tool, higher the plastic deformation of the material in the weld volume and so higher is the temperature in the weld volume. Lower the V higher the rpm, per mm motion of the tool so higher the work done and higher the heat generation owing to longer duration of plastic deformation of the material below the shoulder of the tool and material adjacent to FSW tool pin. Duration in min (=T_pr ) in which pressure is acting on the material in the weld volume per mm motion of the tool=1/V in min/mm. Optimum pressure giving sound weld occurs when V is low for particular WP material and tool material combination.

Pressure on the weld volume is more for less traverse speed than higher traverse speed, because tool stays for more time per mm of traverse of the tool on the weld volume at less traverse speed. Let us say we have got a sound weld at high rpm and high traverse speed. The pressure in this case is lesser compared to pressure at low traverse speed and low rpm as said in previous paragraph, and also this pressure action takes place for a lesser duration owing to higher traverse speed than the pressure action in the case of low rpm and low traverse speed. In the case of low rpm and low traverse speed, owing to high pressure action for longer duration and lesser temperature rise in weld volume owing to low rpm, there occurs optimum diffusion between the plasticized materials in weld volume compared to high rpm, high V, resulting in higher strength weld. For lower traverse speed, pressure owing to PD of tool, acts more time per mm of movement of tool, on weld volume than that for higher traverse speed.

During the dissimilar FSW, 2024-T3 Al alloy and AZ31B-O Mg alloy plates were kept in AS and RS of FSW tool respectively. FSW Tool material was HDS and tool had threaded pin with top diameter 6 mm, bottom diameter 4 mm, tool shoulder diameter 15 mm and pin length=4.67 mm (this tool was also used for all the following welds). Rotational speed varied from 300 to 1000 rpm, welding speed=50 mm/min, tool plunge depth=4.88 mm. As a significant step, the author devised to set zero interface offset (IO). For zero IO, both the interface of plates and tool center line lie on same plane or line. During these parameters welding, welding was intercepted in between and weld tool was withdrawn since through crack had occurred after 400 rpm (Figure 2), owing to quick formation of brittle Al-Mg intermetallics (IMs) at higher rpm and so at higher temperatures. Then keeping all other parameters constant, a weld was carried out by varying rotational speed from 300 to 400 rpm (Figure 3). After welding, metallographic samples were prepared from the weld coupon (Figure 3). Sound weld was obtained at 305 rpm (Figure 4).

For the given set of parameters, there exists a window of interface position, in relation to the tool (0.5–1.5 mm away from the tool center) in the AS which gives the optimal strength and ductility without a joint line remnant (JLR), for similar workpieces thickness of 5 mm and bottom tool pin diameter of 4 mm. Similar study can be used to find out optimum location to position the initial interface in dissimilar welds; however the safe range is expected to change based on the tool, base metal properties and the processing parameters [39].

Varying interface weld performed for AA 2024 and AZ31B-O dissimilar material combination for the following parameters; rotational speed=300 rpm, starting interface offset is 2 mm in RS and ending interface offset is 2 mm in AS, welding speed=50 mm/min, plunge depth=4.88 mm, weld length=200 mm and tool tilt angle 2°. After welding, metallographic samples and tensile samples were extracted alternatively from weld coupon (Figure 5). In metallographic samples, Al alloy side was etched by using Kellers Reagent (1 ml HF+1.5 ml HCl+2.5 ml HNO₃+95 ml distilled water) for 30–50 s and Mg alloy side was etched by solution (14 ml out of “2g picric acid+20 ml ethanol”+2 ml acetic acid+2 ml water) for 5 s [13].

3.2 Macrostructure of joints

Exact position of the interface required for good weld for particular combination of materials can be determined by performing varying interface offset position FSW experiment for optimized parameters (tool centerline positioned, ±R_b mm on either side of interface line; R_b being the bottom radius of tool pin, here R_b=2 mm) and then finding out the position range for best weld [39]. Figure 8 shows outline for dissimilar material varying IO welding. Above (Figure 9) are the macro images of samples cut along the weld line shown in Figure 5 and Table 1, for interface offset varying weld between 2024-T3 Al alloy and AZ31B-O Mg alloy plates performed at a rotation speed of 300 rev min⁻¹ and 50 mm min⁻¹.

Figure 8:

Outline for dissimilar material (between 2024-T3 Al alloy and AZ31B-O Mg alloy) varying interface offset welding (image shown in Figure 5).

Figure 9:

Cross-sectional macrostructures of the joint samples S2 or (a) to S9 or (h), cut from interface varied weld coupon shown in Figure 5.

(C) Cross-sectional macrostructure of the joint sample S4. (D) Cross-sectional macrostructure of the joint sample S5. (E) Cross-sectional macrostructure of the joint sample S6. (G) Cross-sectional macrostructure of the joint sample S8. Similar defective macrostructures are obtained for samples S2, S3, S7, and S9.

Figure 9 shows macroscopic images of the cross sections of the dissimilar joints for IO varying weld. Joint of samples S4 or (C), S5 or (D) and S6 or (E) are free from defects while joint line remnant (JLR) was observed in middle region for these samples. The reason for vertical but little curved JLR or IM layer at the central portion of weld volume is FSW was carried out at a lower rotational speed (300 rev min⁻¹). The type of WN region structure and the IM distribution is directly dependant on rotational speed employed. The quality of the weld is assessed through various tests. In further sections the characterization of welds, by analyzing many aspects of properties of welds, is made.

3.3 Microstructure of joints

Equiaxed grains in Mg side of WN with much smaller size compared to those in Mg base metal and small lamellar like grains in Al side of WN compared to large lamellar like grains in Al base metal (Figure 10) are on left and right sides of the central IM layer respectively (Figure9C–E). In Figure 9C–E there is a continuous and very thin IM layer as visible in Figure 11A–C. Less thickness IM layer formed since diffusion is very less due to low strain rate owing to low rotation speed and so low temperature at the WN region; The IM layer mainly contains IM compound Al₁₂Mg₁₇ along with Al₃Mg₂; since the Al₁₂Mg₁₇ is formed at low temperature [11]. Yashan et al. [40] suggests in a study on FSW of 1100 Al and type 316 stainless steel that the diffusion is enhanced during plastic deformation with a high strain rate. Similar to micrographs of Figure 11, the author has taken other SEM images in each sample at the other three regions in the same sample WN, having almost same IM thickness.

Figure 10:

(1) Photomicrograph of Al 2024-T3 base metal. Average grain size=160×32 μm (2) Photomicrograph of AZ31B-O base metal. Average grain size=50×12 μm (3) SEM image of Al 2024-T3 side of WN. Average grain size=60×22 μm (4) SEM image of AZ31B-O side of WN. Average grain size=2.2 μm. Photomicrographs 1 and 2 are obtained from samples, which were used to obtain SEM images 3 and 4 respectively.

Figure 11:

SEM images taken at same magnification showing IM thickness “t” at central WN region of samples S4, S5 and S6.

(A) For sample S4 with IO=0.76 mm in RS, average t=1.2 μm. (B) For sample S5 with IO=0.30 mm in RS, average t=3 μm. (C) For sample S6 with IO=0.04 mm in AS, average t=2.5 μm.

3.4 X-ray diffraction (XRD) analysis

Figure 12 contains some peaks obtained from XRD analysis on sample S6 (Figure 9E). Large peaks of IM compound Al₁₂Mg₁₇ are detected. XRD patterns confirm that the presence of IM compounds Al₁₂Mg₁₇ and Al₃Mg₂ in the IM layer in WN. The IM formation of both types (Al₁₂Mg₁₇ and Al₃Mg₂) of compounds was reported in the Al-Mg laser welds [41] and FSW lap [42] and butt welds [43].

Figure 12:

X-ray diffraction pattern of the sample S6 (Figure 9E).

3.5 Tensile strength analysis

The tensile samples conform to ASTM standards. Dimensions of, one of the tensile samples is shown in Figure 13A. A nano UTM (BiSS Bangalore) with maximum capacity of 15 kN was used for tensile testing of the samples at a strain rate of 10⁻³ s⁻¹. Tensile strength of each sample recorded and plotted as shown in Figure 13B. In Figure 13B, positive values of IO positions are towards RS from the position of zero offset, negative values of IO positions are towards AS from the position of zero offset. Zero IO coincides with leading edge.

Figure 13:

(A) Dimension of tensile sample. (B) Tensile strength of tensile samples at various interface offset positions. (C) Plot of tensile strength versus IM thickness.

The reason for maximum tensile strength at D (Figure 13B) is, at that IO of D, IM layer of least thickness forms along with good bonding between IM layer and the recrystallized alloy materials on either side of IM layer. Sample S4 (Figure 11A) has IO=0.76 mm in RS and has least IM thickness, sample D (Figure 13B) has IO=0.66 mm in RS, which is near to S4, and so D also has almost same least IM thickness. On the same line one can conclude that sample C has IM layer of thickness as that of S5 and sample B has IM layer of thickness as that of S6. When the interface of the two alloy materials lies in the range of IO from B to D (Figure 13B), the interface experiences higher compressive strain and strain rates as well as medium (about π/2 rad) shear strain and shear strain rates owing to forward moving and rotating tool, because the extruded material by the forward moving tool is in crescent shaped cross section with wider top and thicker at top middle (i.e. at zero IO) and narrow bottom [44], [45]. So plastic deformation of the alloy materials is higher and also diffusion between alloy materials in weld volume is higher if the interface lies in the region between B to D, especially these are higher at region C. So sample C has larger IM thickness (Figure 11B) and also, it has comparatively lesser tensile strength than B and D. The reason for highest tensile strength at D, having least IM thickness is that strain energy increases as the volume of IM layer decreases [46]; volume of IM layer is least for tensile sample D having least IM thickness. As the strain energy is more, corresponding tensile strength is highest for tensile sample at D. Tensile strength decreases as the IM layer thickness increases in WN (Figure 13C). Highest tensile strength of 106.86 MPa was obtained for the tensile specimen with IMs thickness 1.2 μm and tensile strength of 93.5 MPa was obtained for tensile specimen with IMs thickness 3 μm.

Similar results were obtained by Venkateswaran and Reynolds [13]. Venkateswaran and Reynolds [13] have plotted the variation of the tensile strength as a function of the maximum IM thickness by taking many points. Their plot indicates that the tensile strength of the weld joint increases as the maximum thickness of the IM layer at the interface is decreased. They relate this higher bond strength for the lower IM thickness, to lower defect content in a smaller volume of IM. The reasoning by author, for highest tensile strength of D equal to 106.86 MPa (Figure 13B), with minimum IM layer thickness is already explained in previous paragraph.

Apart from the macro pictures (Figure 9) showing sound weld at 0.36 mm, 0.76 mm offsets in RS and at 0.04 mm offset in AS, there is good tensile strength, in turn sound weld (Figure 13B) between 0.3 mm (in AS) and 0.9 mm (in RS); this range of length=0.9+0.3=1.2 mm. During FSW, keep the interface at 0.3 mm in RS of the tool so that the tool can be constrained to move along this 0.3 mm interface offset line and can sway around this line by about 0.6 mm on either side.

The maximum tensile strength and the corresponding maximum elongation of the dissimilar welds are 106.86 MPa and 1.33% respectively. Owing to the formation of brittle IMs layer and concomitant decrease in the strength of weld, the FSW joint efficiency is 44.52% (=106.86/240); the baseline tensile strength being the tensile strength of AZ31B-O Mg alloy, equal to 240 MPa.

3.6 Fractographs of tensile fracture surfaces

Diagnosis made for the fractures in tensile sample and it follows. The tensile failure of the weld joints occurred only by fracture along the central continuous vertical but slightly curved IM layer. It can be categorically stated that all the weld joints failed along the IM layer during tensile testing owing to the minimum cross section of the joint along weld center line and brittle nature of IMs, which are present in the minimum cross section. Backward tilted tool gives rise to minimum cross section of the joint along the weld center line. Figure 14 shows the cross section of an Al alloy and Mg alloy weld joint’s, tensile fractured sample at B in Figure 13b. Figure 15 shows the SEM fractographs of the Al alloy side of the fractured sample at B in Figure 13b. The failure of the welded joint samples occurred through the very narrow IM layer leaving the IM compounds on both the Al and Mg sides of the fracture surfaces.

Figure 14:

Tensile Fractured Sample after the tensile test.

Figure 15:

SEM fractographs images of tensile fracture surface (Al alloy side).

Fractograph (B) indicating the wavy line (red color) through which initiation of fracture occurred, see also Figure 14. (A) Al alloy side fractograph. (B) Enlarged (A).

The transgranular cracking occurs in the brittle IM phases and the step like features appearing similar to cleavage facets (Figure 15). Figure 15 shows the SEM fracture morphology observed from the normal direction to the fracture surface on AS i.e. 2024-T3 Al alloy side. Cleavage like feature can be observed in the fracture surface, revealing that the dissimilar weld failed altogether through brittle fracture mode.

3.7 Hardness

The micro hardness profile of one of the samples (S5) is shown in Figure 16. Future-Tech micro hardness tester FM-800 was used for indentations. The micro hardness measurements were recorded across the weld nugget at middle depth along 3 dashed black lines shown in sample 4 (Figure 9C), on the transverse cross section of the welded plate normal to the welding direction. For sample S5 Vickers hardness (Hv) values versus distance along the cross section of the sample have been plotted (Figure 16). Each plotted point on the plot is the average of 3 readings (but on IM layer 4 readings) taken along on 3 parallel lines separated by 0.25 mm. The average Vickers hardness on the IM layer in S5 is 116.18 Hv, which is the maximum of all hardness values taken on IM layers only, on each sample. Similar measurements were recorded for samples S4 and S6 and are not shown in Figure 16; but the readings were almost same with a little variation similar to variations of readings in sample S5, except the readings on IM layer. For S4 the average Hv value on IM layer is 86.20 Hv and the same for S6 is 98.67 Hv. As shown in Figure 16, the hardness value fluctuates in the WN owing to dynamic recrystallization and formation of brittle IM compounds. The tensile fracture was brittle type, and its position was located at the mid position of IM layer.

Figure 16:

Vickers hardness plots of sample S5 obtained by micro indentation.

Since the thickness of IM layer is high in S5 than S4 and S6 the hardness is highest (116.18 Hv) on the IM layer of S5 and next highest (98.67 Hv) on IM layer of S6 and next highest (86.20 Hv) on IM layer of S4. Hardness is greater for S5 (with high IM thickness 3 μm) compared to S6 and S4 because there is more volume fraction of IM material in S5 with IM layer thickness of 3 μm, which participates in resisting the indentation. This observation is contrary to that of tensile strengths of samples shown in Figure 13b. It has known that (t)_S5>(t)_S6>(t)_S4, where ’t’ is the IM layer thickness; let “h” be the Vickers hardness on the IM layer for samples S4, S5, S6 then (h)_S5>(h)_S6>(h)_S4 so as the IM thickness increased the resistance to indentation increased and so hardness increased. The author has the images of hardness indentations; owing to shortage of space they have not been put.

For solid–solid transformations, there may be volume changes attendant to the formation of new phases. These changes may lead to the introduction of microscopic strains [47]. Crystal structures other than fcc or hcp have more volume per unit mass of substance, or crystal structures other than fcc or hcp have packing factor less than that of fcc; the IM layer has more volume than the parent materials which were in the place of IM layer, thus introducing microscopic stresses and strains in the material system (Al+IM+Mg). Higher the thickness of IM layers higher the compressive stresses in turn higher hardness value (here 116.18 Hv for highest IM thickness of 3 μm in S5). Higher compressive stresses owing to higher thickness of IM layer develop high shearing or sliding resistance between layers of materials Al, IM, Mg, for slip between layers (similar to frictional force proportional to normal force) during indentation. If a single atomic layer in case of edge dislocation develops compressive stresses, so definitely IM layer develops considerable compressive stresses and strains in it as well as in adjacent Al, Mg materials.

If one takes group of plates consisting of Al and Mg materials alternatively placed on each other and if one subjects the surface of the group of alternate plates to controlled heating or heat treatment then there occurs formation of IM layers in between Al and Mg materials at the interfaces of Al and Mg materials. As reported here hardness of Al, IM, Mg layers together directly proportional to thickness of IM layer, i.e. higher the thickness higher the hardness of system of materials. This type of material system exhibiting high hardness gives rise to new type of laminated composites. One has to do experimentation regarding this to get optimized parameters such as thickness of Al, Mg plates, heat treatment temperature and duration to which surface of plates have to be subjected to get the optimum thickness of IM layer etc. to get highest hardness of the material system.

3.8 Analysis of welds

K. Kumar in his PhD thesis, synopsis, writes “In order to obtain FSW welds with maximum joint efficiency, the welding temperature should not exceed the “softening temperature” of the base metal. If the weld formation temperature is less than the base metal softening temperature, the weld can be made with 100% joint efficiency. In order to optimize the FSW parameters, which gives defect free weld with lowest possible temperature an instrumented programmable FSW machine is to be designed and developed”. Here in this work (this paper) by the author, welds were obtained at lowest rpm, resulting in less temperature rise in weld volume, which is less than the base metals softening temperature.

As the thickness of plates decrease and so cross sectional area decreases, the volume of the IMs material in weld volume decreases; so tensile strength of lesser thickness FSW plate is higher than tensile strength of higher thickness FSW plate. This can be seen in the Table 2, where strength of FSW increases as the thickness of plates decreased. On the similar grounds welds obtained at low rpm having lesser IMs thickness capable to withstand higher stresses since they have lower volume of IMs material than that of the welds obtained at higher rpm having higher IMs thickness layers (owing to increased diffusion in weld volume) and so having higher volume of IMs.

Author should have got lesser tensile strength of the dissimilar weld for 5 mm thickness plates than tensile strength of weld for 4 mm thickness (Table 2), because as the thickness of plates increases the corresponding tensile strengths should reduce, as said in previous paragraph. Contrary to this the author have got tensile strength of 106.86 MPa for 5 mm thickness dissimilar plates, which is greater than 95 MPa for 4 mm thickness dissimilar plates as listed in Table 2. So one can intuitively conclude that for plates of thicknesses less than 5 mm or other thicknesses, this optimization procedure definitely yields higher values of tensile strength; i.e. higher than tensile strength of welds listed in Table 2, i.e. tensile strengths of 126 MPa and 132 MPa for 3.25 mm and 2 mm thickness plates respectively, because this EOP uses possible lowest rpm and medium V thus resulting in lowest thickness of IMs layers compared to higher rpm welds (as in Table 2); so one can conclude that, this experimental optimization procedure in FSW is especially suited for FSW of Al alloy to Mg alloy materials.

Every welding of plates of particular thickness can have its own highest strength, as said previously. One must not compare strengths of welds of different thicknesses plates. If one obtains weld for particular thickness of plate materials by following this optimization procedure he will get highest strength for the particular thickness of plates. In general as the thicknesses of plates decreases tensile strength of sound weld increases, irrespective of type of procedure or method used to obtain the sound weld, because as the thickness of plates reduces the volume of the IMs layer in weld volume decreases, resulting in increase in strength.

The unique characteristics of the optimization procedure are it yields highest strength of weld owing to use of lowest rpm generating lowest temperature level and use of medium welding speed, leading to a weld with least distortion and least residual stresses and least IMs thickness layer in weld volume compared to those of welds of high rpm. Welds with least distortion, least residual stress and least IMs thickness have high tensile strengths as reported here.

In the paper [38] the author nowhere reported about the IMs formed in the weld volume. But according to [13] formation of intermetallics (IMs) compounds are inevitable in the dissimilar Al–Mg weld joints under all conditions of welding. Strength of a chain is the strength of the weakest link in the chain. Since IMs have inevitably formed in the weld volume under all conditions of dissimilar Al to Mg alloy welding, IMs being brittle in nature and having very less strength, constitute the weakest part in the weld volume. So it can be concluded that the reported strength of 192 MPa by [38] is impossible to get in dissimilar Al to Mg alloy weld of plates thickness of 6 mm, and so it must be a fake result. Initially before FSW, Al and Mg alloy materials are in contact and also after FSW (under all conditions of welding) Al and Mg materials definitely will come into contact each other, so there is inevitable formation of IMs layers in the weld volume. In the paper [38] the authors, nowhere indicated or dealt about presence of IMs and thus leads to complete doubt about the data presented in the paper. Or another doubt is that the authors must have altered the data of results obtained by experiments. If you observe the above table, as the thickness of plates increases tensile strength decreases but, even though the authors used 6 mm plate, even then they have got 192 MPa high tensile strength, which is no way possible. Here the author has obtained a highest of 106.86 MPa, tensile strength for FSW between Al alloy and Mg alloy 5 mm thickness plates at possible lowest 300 rpm. So in the inevitable presence of IMs in weld volume the weld strength of dissimilar Al alloy to Mg alloy materials weld cannot be higher than 106.86 MPa (or around this value) for plates of thickness≥5 mm as reported here. The contribution of lesser IMs layer thickness, to the strength of weld is more than the contribution of the tortuous nature of weld to the strength of the weld.

So the optimization procedure reported here yields possible highest strength welds for all thicknesses of plates since possible lowest rpm, and so lowest IMs thickness, as well as moderate tortuous welds, can be achieved. If this optimization procedure is assimilated for FSW of similar or dissimilar Al alloys, definitely one can get higher strength FSW joints due to evolution of high strength HAZ, than strength reported in literature e.g. [39].

One can find in FSW of similar materials literature (Mg alloys), FSwelds have been obtained at higher rpm (>600 rpm) i.e. at higher temperatures exhibiting higher tensile strengths. Tensile strengths were increased as the rpm increased. As the rpm increased temperature of the weld volume also increased and diffusion between the two similar material plates also increased which ensures effective bondage of the two materials at and around the interface of the two materials plates resulting in higher strength, owing to complete elimination of interface two surfaces and formation of continuous material body in place of interface. It is difficult to find higher rpm at which sound weld occurs giving highest tensile strength for similar materials (same or different Mg alloys). By EOP (experimental optimization procedure) one can determine lowest rpm at which sound weld occurs. Then one has to subject the sound welds to a constant high temperature (350–550°C for Mg alloys) for justified duration, and obtain the weld with highest tensile strength.

There are innumerable alloy combinations with innumerable hardness values; higher the hardness higher the rpm at which sound weld occurs for the given material combination. So, one has to repeat the trial and error methods, to fix the optimum parameters for different alloys, of the same thickness. So it becomes laborious procedure. But by EOP one can quickly arrive at optimum parameters irrespective of alloy material combination and thicknesses of materials to be FSWed.

Here in this dissimilar FSW highest strength of weld occurred for the sample having IO of 0.7 mm in RS. So one can conclude as follows; In AS, forward motion of the tool is more than that of in RS. So for similar materials FSW IO should be set in or towards AS. Material in AS is subjected to more deformation owing to more forward deformation plus less length of frictional or plastic deformation rubbing by tool; material in RS, less forward deformation plus long length of rubbing by tool. The net effect is almost equal plastic deformation for material in AS and for material in RS. In case of complete (in AS harder and in RS softer material) dissimilar materials welding, harder material has to be deformed for more length and more time, this is possible if and only if interface has kept in RS only.

This EOP yields sound weld at medium traverse speed and at possible lowest rpm. It is not possible to obtain sound similar or dissimilar welds at rpm>500 rpm from this EOP. One can notice that, in order to obtain sound weld at higher rpm (>600 rpm), one has to adapt trial and error method (till now no one has obtained sound weld by following the EOP), so it is not an easy task to arrive to FSW parameters yielding sound weld at rpm>600 rpm, but it is quite an easy task to obtain sound weld at lower rpm (<500 rpm) by following the EOP. This sound weld may not be having strength > that of sound weld obtained at higher rpms since there is no proper mixing of similar materials at the interfaces of the materials and so there will be no effective bondage between the two materials at the materials interface, owing to diffusion between the materials across both surfaces of interface, at low rpm, compared to that of weld at higher rpm. As explained previously, one can increase the strength of weld obtained by EOP by subjecting the weld to heat treatment i.e. one can determine practically the temperature to which the weld should be subjected and also its duration, so that resulting weld will have the highest strength.

Heat treatment above a particular temperature (Topt) will not give rise to higher values of tensile strengths, since at and around Topt, diffusion among the similar materials at interface completely eliminates interface or joint line remnant and produces almost homogeneous material, so any further increase in temperature above Topt will not give additional rise in strength. The author thinks that for FSW/FSP of similar precipitatable Al alloys, there is no necessity of heat treatment after FSW/FSP according to EOP developed here, because other than FSW heating, material may overage in HAZ region resulting in lower strength weld, but for FSW/FSP of Mg alloys heat treatment is a must since FSW of Mg alloy at higher rpm have greater strength compared to lower rpm welds.

3.9 Effect of intermediate material

Considering Exp. 2 or 3, intermediate metal Zn (or Cu) cannot be dissolved in Al alloy and Mg alloy materials owing to very less diffusion among the materials, owing to less rpm and so less temperature rise (around 350°C) in the weld volume; and also diffusion is very low because weld volume experiences less temperature level for only a few minutes of duration. Even at higher rpm welds (>1000 rpm) the temperature reached in the weld volume is around 475°C which is well below melting points of base metals or solidus temperature and also duration to which the materials will experience this temperature is of the order of a few minutes only. For diffusion to occur between metals not only the temperature is to be higher (>500°C) but also the metals must be maintained at the temperature for sufficiently long duration (hours to days) [47], [48]. But in FSW the temperature reached will be around 350 to 475°C [35] and also this will not last long for more than a few minutes. So owing to less temperature rise and insufficient diffusion, solid solutions could not be formed between and along the interfaces (Al, Zn and Mg, Zn). But IMs (if formed) will be formed at low temperatures or at low rpms (around 300 rpm); e.g. Al₁₂Mg₁₇ and Al₃Mg₂ IM compounds formed between Al alloy and Mg alloy at 300 rpm (Exp. 1).

One can say, there is very small scale diffusion occurring between layers of materials (between Al and Zn, and between Zn and Mg) in the weld volume because of low temperature and so low diffusion leading to neither formation of solid solution nor the formation of IMs at the materials interfaces. So these materials interfaces will become loosely held boundaries without any bonds in between them; so fracture occurs along these interfaces at almost zero stress.

Considering Exp. 4 and 5, any metal goes on softening at a constant rate (or flow stress of metal goes on decreases) from solid state (at room temp) to liquid state (melting point). The extent of the softening of the metal depends on the temperature level to which the metal is subjected to. Since weld volume (and the Ni) is subjected to lower temp (around 300–400°C at 305 rpm) the Ni foil is not softened enough (since Ni melting point is 1455°C) so that atoms of Ni would diffuse to nearby Al and Mg material and form IMs between Al and Mg. Since there is insufficient diffusion between Ni, Al and Ni, Mg there is no proper bonding and no proper IMs formation between Ni, Al and Ni, Mg; the interfaces between these become weak or have less strength leading to failure of weld in weld volume, at low tensile stress=53 MPa. The lesser the material is softer (owing to material’s lower temperature) the less is the diffusion of atoms from the material to the nearby other materials. Tensile strength of weld with 0.1 mm Ni foil=50 MPa<tensile strength of weld with 0.25 mm Ni foil=53 MPa, owing to slightly more diffusion in higher thickness (0.25 mm) Ni foil.

In bare Al alloy to Mg alloy weld at 300 rpm there is sufficient diffusion between, enough softened materials Al alloy and Mg alloy (The melting points are 502–638°C and 605–630°C respectively compared to the melting point of Ni 1455°C) leading to formation of lowest thickness (enough thickness to withstand higher stresses) IMs Al₁₂ Mg₁₇, Al₃Mg₂ giving rise to maximum tensile strength of 106.86 MPa. The weld between Al alloy and Mg alloy at lowest rpm (300 rpm) does not require any intermediate material since bare weld of Al alloy to Mg alloy itself has possible highest strength owing to formation of least thickness IMs. But for Al alloy to Mg alloy weld at higher rpm (>600 rpm), intermediate material such as Ni gives better strength. No other metals have melting points less than melting points of Al alloy and Mg alloy. So with regard to mutual diffusion and formation of IMs no metal will form IMs with Al alloy and Mg alloy having proper least thicknesses than IMs formed between Al alloy to Mg alloy at low rpm (305 rpm).

Analysis of results obtained by Woong [24] (see section 1, paragraph 7) by the author is as follows: the increase in tensile strength (=115 MPa) was owing to formation of lesser thickness layer of IM Ni₃Al compared to high thickness of IM layer Al₁₂Mg₁₇ owing to higher diffusion between Al alloy (melting point=638°C) and Mg alloy (melting point=630°C) since less diffusion occurred between Ni and Al materials owing to high melting point of Ni material. At 800 rpm, 35 mm/min parameters, higher temperature develops, leading to softening of materials (Al, Mg), involved in the welding. So there is considerable diffusion between Al alloy and Mg alloy materials; owing to higher diffusion at higher temperature, slightly higher thickness IM layer (Al₁₂Mg₁₇) forms yielding tensile strength of 95 MPa. When the Ni foil introduced and welding is performed at 800 rpm 35mm/min, Ni having high melting point (1455°C) softens less compared to Al alloy or Mg alloy at 800 rpm or at corresponding temperature. Diffusion from Ni to Al or Mg is less resulting in lesser thickness IM layer (Ni₃Al) formation in the weld volume compared to higher thickness IMs layer formed in weld volume of bare Al alloy to Mg alloy weld, and so higher strength resulted for the weld of Al alloy to Mg alloy keeping Ni as the intermediate material. IMs cannot be ductile in nature, so above said is the proper reason for increase of strength of weld with Ni as the intermediate material.

Phases shown in phase diagrams (PDs) can be obtained for equilibrium mixture of liquids of the materials involved. For solid to solid contacts one may not get these phases (shown in PDs). However for solid to solid diffusion 50–50% compositions of the solids in contact can be assumed. Strictly speaking only 50–50% mixture of liquid materials, give rise to phases in PDs, at 50–50% composition value.

3.10 Friction

As the seizure occurs between FSW tool and work piece, there will be no more relative motion between tool surface which is in contact with work piece material and work piece material immediately adjacent to FSW tool surface. So there is no more friction but there will be shear deformation between layers of work piece materials throughout the FSW volume, which contributes to temperature rise in and adjacent to weld nugget. Since material gets softened or plasticized in weld volume, total shear resistance (or equivalent frictional resistance) among the layers of material will be lesser compared to frictional resistance which occurs during initial plunging of tool and beginning of transverse motion of tool before seizure takes place. For Mg alloy plate defect free weld was obtained at 340 rev min⁻¹ (Figure 1).

3.10.1 Equivalent coefficient of friction in bead on plate FSW of AZ31B-O Mg alloy

There is full seizure [49], [50], [51], [52], [53] between the tool (comprising shoulder and pin) and the material at the 3D surface (comprising shoulder surface and pin surface) on the tool. 1-2-3-4-5-6 (Figure 17) is the cross section of this 3D surface, which is also profile of the tool (Figure 17). Shearing of material takes place everywhere in the weld volume, i.e. between 1-2-3-4-5-6 and 1-B-C-O₂-D-E-6, the 3D profiles or surfaces; owing to this, shearing resistance occurs all over weld volume, around the tool pin. But if one assumes that the shearing resistance or equivalent frictional force (nearly similar to radius of gyration in mechanics), is restricted to or acts only on the weld volume outer profile 1-B-C-O₂-D-E-6 or on the middle profile (red line in Figure 17) or on the tool profile 1-2-3-4-5-6, then the equivalent coefficient of friction (μ_s) on the corresponding profiles can be determined as follows.

Figure 17:

Tool profile and FSW volume cross section profile in friction stir welding of Mg alloy.

Consider the profile, 1-B-C-O₂-D-E-6. One can imagine a tool with shoulder diameter of the length “1-6” (from point 1 to point 6 in Figure 17) and pin of profile 1-B-C-O₂-D-E-6. One can consider resisting shear torque T_s with equivalent coefficient of friction μ_s on this surface or correctly, on 3D surface containing this profile.

The following are the formulae [54] for flat pivot and conical pivot under the assumption of uniform pressure.

(1)Ts=[2/3]μsZ R; R is radius; Z is the Axial load.

(2)Ts=[2μsZ(R13−R23)]/[3(Sinα)(R12–R22)]; R1,R2 are radii; α is semi cone angle.

Two conical surfaces are defined, first one which includes profiles 1-B and 6-E, projected area of this surface is A₁ and second, which includes BC, ED profiles, and projected area of this surface is A₂. Another surface which includes horizontal profile CD, having surface area A₃. The author measured the dimensions of lines (O6=6.5 mm; O₁E=3.5 mm; O₂D=2.5 mm; OO₁=2.5 mm=O₁O₂) by actual measurement on the physical sample shown in Figure 17, which was obtained for the following FSW parameters: rotational speed=340 rev min⁻¹, welding speed=60 mm min⁻¹.

So A₁=94.24 mm²; A₂=18.85 mm²; A₃=19.634 mm²; A₁+A₂+A₃=132.724 mm²; if the vertical load is Z (=12.326×10³ N, accessed from the archives of FSW machine computer system) on all surfaces together, load on each area is,

Z1=(94.24/132.724) Z=8.752×103 N;Z2=(18.85/132.724) Z=1.75×103 N;Z3=(19.634/132.724) Z=1.823×103 N.

Let resisting shear torque be T_s with equivalent coefficient of friction μ_s.

Then from equation (2),

Ts1=[2×μs×8.752×103(6.53−3.53)]/[3×(sin 50)(6.52−3.52)]=58.837×103 μsN mm

here (in the above equation), R₁=6.5 mm; R₂=3.5 mm; Z₁=8.752×10³ N; α=50;

Ts2=[2×μs×1.75×103(3.53−2.53)]/[3×(sin 21.80)(3.52−2.52)]  =14.267×103 μsN mm;

here R₁=3.5 mm; R₂=2.5 mm; Z₂=1.75×10³ N; α=21.80°; and from equation (1),

Ts3=2×μs×1.823×103×2.5/3=3.0383×103 μsN mm

here R=2.5 mm; Z₃=1.823×10³ N.

Not only axial load Z along tool axis acts but also transverse load (F_t) along welding direction acts on tool. Owing to this transverse load (F_t=1.838×10³ N, this data accessed from the archives of FSW machine computer system), acting on front ½ of the weld volume peripheral surface, additional resisting torque will be developed, this is T_s4.

Ts4=μs×1.838×103×4=7.352×103×μs N mm;

where 4 mm is the average radius from Figure 4.

But the indicated torque accessed from the archives of FSW machine computer system=19.66×10³ N mm; so

Ts=Ts1+Ts2+Ts3+Ts4=19.66×103.

Substituting values for T_s1, T_s2, T_s3, and T_s4 from above calculations, one gets the equation for μ_s in canonical form as

83.4943×103×μs=19.66×103

μ_s=0.234 or 0.23. This is for the profile 1-B-C-O₂-D-E-6.

Considering red colored profile (Figure 17), by following the procedure mentioned earlier one can get, μ_s=0.24. Similarly considering tool profile 1-2-3-4-5-6 (Figure 17), one can get, μ_s=0.25.

If one divides the region between 1-2-3-4-5-6 layer and 1-B-C-O₂-D-E-6 layer into 10 layers, former being the 0^th layer (where seizure occurs) and latter being the 10^th layer. Then assuming equal, equivalent frictional force on each layer (excluding 0^th layer but including 10^th layer) with equal, equivalent μ_s on each layer; one can take 1/10 of vertical load, 1/10 of transverse load and 1/10 of torque for each layer. If one performs calculations as mentioned earlier for each layer, one can approximately determine equivalent μ_s on each layer. These μ_s values can then be used effectively in simulations, approaching actual conditions.