Effect of welding process parameters on surface topography and mechanical properties of friction-stir-welded AA2024/AA2099 alloys

The joining of two materials with different chemical composition was a major setback for conventional methods of metal joining. The results of this welding were showing considerably great improvement in the aspects of quality of weld, amount of heat generated, uniform distribution of material, refined microstructure, enhanced tribology, materials flowing pattern, good strength with reduced internal stresses. Optimized parameters were estimated by using Desirabilty approach and Response surface methodology. Optimum parameter combination for dissimilar material welding was observed to be 913.74 rpm, 45 mm min−1 and 8kN. The desirability values for dissimilar welding process were 0.912 respectively. The grain enhancements were decreased in the range of WC > WT > HAZ > Parent material. The wear rate of dissimilar AA2024 and AA2099 were superior as the wear value increases from 0 to 50 μm in the nugget zone. The coefficient of friction value remains constant throughout the wear experiment ranges from 0.3 to 0.55. A steady state friction value of 3N to 5N is observed when sliding distance increases. The wear loss was measured by finding the difference between initial weight and final weight and found as in the range from 0.2283 g to 0.4866 g.


Introduction
The FSW process joins metals without addition of heat energy by external sources. The base material is held on the fixtures provided on special set up developed for FSW. Workpiece will be held together, and suitable tool will be rotated over the joining layer. The tool composed of shoulder and pin with support over the shoulder to hold in jig. The tool rotational speed (RS) will be controlled by an electrical motor and stepped gear arrangement. Based on tool rotation direction, two different sides are being distinguished as advancing side and retarding side. Owing to the frictional effect on advancing side, metal will be pulled and sent to the other end. Whereas the effect of heat generated during the friction would tend to melt the top layer of metal at retarding side. However, it will not be sufficient for melting. This will lead to the formation joint at interface of two metal plates. The effect of friction is adjusted by the load applied on the tool [1][2][3][4][5][6][7].
The modern technique of FSW was established by industries to overcome these problems by involving interlayer joining technique through frictional heat generated from tool and workpiece. It is creating more expectation amongst the industrial world owing to the special qualities such as energy efficient, green technology concept, free of harmful emissions, no external heat supply, reduced post weld treatment etc. The joining of different metal alloys using FSW is being researched in detail because of their immense application in various fields of marine technology, aerospace fuselage components, automobile structures, shipment industry and bullet trains development [8]. In addition, the introduction of interlayer material resulted in reducing the residual stress with confinement to intermetallic compound formation. Although FSW is superior welding in Al alloys low welding speed is the drawback. In order to overcome this drawback recently high speed FSW (HSFSW) has been established. Patel et al [9,10] employed HSFSW to increase manufacturing volume for lightweight crash-resistant battery trays in the EV market. Another variant of conventional FSW is the Stationary shoulder friction stir welding (SSFSW). Sejani et al [11,12] highlights the research development on SSFSW thoroughly reviewing microstructural evolution, mechanical properties, and variants to deal with particular issues. Using FSW, the persistence of this research is to examine the effect of tool profile on the mechanical and metallurgical properties of AA2024 and AA2099 alloys. The Desirability technique and Response surface methods were used to estimate the optimized values.

Literature review
Aluminium alloys AA2024 and AA2099 were used as the workpiece materials for this project. The materials were chosen for their use in aircraft applications. It has unique properties like as fracture toughness, better tensile strength, and corrosion resistance [13]. The chemical make-up of AA2024 is listed in table 1. Feed rate and rotating speed were chosen as the study's variable parameters [14]. The microstructure of components welded on 7075-O alloy using a recursive FSW procedure has been investigated [15]. During the procedure, many constraints such as WS, TS, number of passes, weight applied, and so on were investigated. The influence of welding speed on various structural parameters of AA6082 was investigated using several metallurgical tests including as TEM, XRD, OM, Fractography, EBSD, and others [16]. TIG and MIG welding were used to join the materials, which were then aged. Heat created per unit length was found to be greater at lower welding speeds, which improved the welded joint's fatigue performance. Full and partial softening was predicted for low and high welding rates. RSM was used to study the impacts of process factors on ultimate tensile strength and micro hardness [17]. In order to build mathematical models for forecasting UTS and HV, several parameters were studied. Li et al [18] used the FSW technique to create 7A04-T6 aluminium joints. Fine grains were found at the nugget zone, with the grain border misaligned. Micro crack propagation was greatly reduced because of dynamic recrystallization. Laser local heat treatment was used to refine grain microstructure and increase mechanical properties (LLHT).
Mehta et al [19] examined Al-Mg junctions created using the FSW approach using a variety of mechanical and microstructural tests. Throughout the testing at the nugget zone, intermetallic complexes Al 3 Mg 2 and Al 12 Mg 17 were seen. By reducing intermetallic compounds in the weld, cooling assisted welding has better strength. With the help of cooling, the joining efficiency of the process increased by about 73 percent. Choi et al [20] investigated the fracture strength of pure Ti and Al metal joints created by FSW and attempted to optimize it. The optimum dispersion of titanium metal within the aluminium base metal can be observed at very minimum RS. The reduction of Ti fragment production during material flow in FSW was used to achieve superior tensile strength of the Ti/Al junction. Increased strength and correct intermetallic layer development in the nugget zone were achieved by reducing probe offsetting and optimizing tool rotation. Shen et al [21] investigated the effect of tool profile on the formation of voids in Al6022-T4/Al7075-T6 weld joints formed by refill spot FSW. To manufacture a defect-free joint, trials were carried out with same operating settings and only a tool modification. The proposed instrument aided in the betterment of bonding and material intermixing mentation. By decreasing fracture formation, a tool sleeve with a groove at the bottom enhanced joint strength.

Experimental setup
The experiments were conducted on a FSW step-up with custom work-holding and tool-holding configurations as shown in figures 1 and 2. The workpiece materials for this project were aluminium alloy AA2024 & AA2099 with a thickness of 3 mm and a cross section of 100 mm × 60 mm. It has unique properties like as fracture toughness, better tensile strength, and corrosion resistance [13]. Feed rate and rotating speed were chosen as the study's variable parameters [14]. For cylindrical tool geometry, three different levels in each parameter were chosen, as shown in table 1. With 20 trials, design expert software was utilized to generate the composite design for the current inquiry. The tilt angle and penetration depth were both adjusted to 1.50 and 0.3mm during the operation [22].   [1,2,9,19,20,26,27]. The traverse speed also depicts inverse relation while ranging from 45 mm min −1 to 70 mm min −1 . It is revealed that, for dissimilar welding both the speed parameters show inverse relation. The raise in axial load displayed an increasing average tensile strength.

Response surface methodology
The hardness value of dissimilar joints was analyzed for variance and the same is presented in table 3. The model F-value was estimated to be 38.49 which is significant with a possibility of error to be 0.01%. The 'Prob > F' lesser than 0.05 which is very much influencing the hardness of dissimilar weld joints made of AA2024 and AA2099 materials. In this analysis, the model contains a greater number of terms having 'Prob > F' values greater than 0.05. The terms may be either removed from the model or not which may result in improvisation of model with reduction in possibility of error. A large value of 21.01 for 'Lack of Fit F-value' was observed and it should be treated with importance. This improved the possibility of error to 0.23%. Hence more concentration must be involved while recording the results.
The tensile strength value of FSW joints developed from dissimilar materials of AA2024 and AA2099 were analyzed for variance and presented in the table 4. The F-value is 34.48 for the model developed which is noteworthy with only a 0.01% chance of error to occur. The error value will be very small in analyzing the design space with utmost care. Like the former analysis, the model terms with the 'Prob > F' lesser than 0.05 showed impact on tensile strength of welded joints. The important terms seen were A, B, C, AC, BC, B2. The other terms in the model with the 'Prob > F' values greater than 0.1 might be eliminated for improving the accuracy and model hierarchy. The process of removing these terms are called 'Model improvement' [6,11,17,28,29].
The influence of independent variables on hardness was analyzed statistically by ANOVA. On the same way, perturbation analysis was also used for identification of generalized solution from the original solution. It states about the dependency of input variables over output responses. The association between the input parameters and the output answers, on the other hand, was not fully explained. This could be easily achieved by the response surface models developed in figures 3 and 4. The figure 3 shows about the relation between the input parameters RS (A), TS (B), AL (C) and output response weld hardness of the joints developed from dissimilar base metals AA2024 & AA2099. Hardness of the welded joint is maximum of about 185 when the RS is minimum at 650 rpm and TS is maximum at 70 mm min −1 respectively as seen in figure 3(a). The RS produced an inverse relationship with hardness, as RS increases with decrement in hardness. Upon increasing the TS, the hardness value tends to increase.
The lower value of hardness could be seen for the combination of 1050 rpm and 45 mm min −1 at constant load friction stir welding process. Hardness decreases with increase in RS as seen in figure 3(b). Axial load does not show much variation when it comes to hardness of dissimilar metal joints. However, only a slight increase in hardness could be seen with the axial load shift from 5 kN to 7 kN. Maximum values of about 170 could be observed for 6 kN load and 650 rpm rotating speed. Similarly, a value of 110 for hardness may be seen at 8 kN load and 1050 rpm rotating speed combination. When value of rotating speed and axial load were increased

Regression equation modelling
The regression equation is an elegant way of representing the working of any physical phenomenon that is happening. In this process, the results are considered as dependent variable and inputs are treated as independent variables or regressor. These regressor have the capability to vary on their own without disturbing any other input variable involved in the formulation of system equations [30][31][32][33][34][35].
The regression equation for predicting tensile strength value is given in equation (2) with an R-squared value of 0.968, Pred R-squared value of 0.837. The standard deviation and mean of the reading were observed to be 6.94 & 277.45 respectively. The 'Pred R-squared' value was in reasonable agreement with that of 'Adj R-squared' value of 0.940. The 'Adeq Precision' value of equation is 21.399, which is greater than 4 (expected). Hence, this model can be used for analyzing the design space.

Validation of regression model
The validation of the regression model developed using the above analysis procedure is presented in this section. The equations (1) and (2) were used for predicting the mechanical property results by substituting input parameter values. Simple mathematics were used for estimating the resultant values.
The predicted values and experimental values were compared in figures 5 and 6. The hardness of weld joints (AA2024 & AA2099) could be predicted successfully by the equation (1). The R-squared value is 0.970 and predicted values are comparatively presented. The confidence level of this equation was 97% and only 3% possibility was allowed for error. The confidence level is higher than 85% and the same can be used for predicting the results within the operating range of experimentation. The tensile strength of weld joints (AA2024 & AA2099) could be predicted successfully by the equation (2). The R-squared value is 0.960 and predicted values are comparatively presented. The confidence level of this equation was 96% and only 4% possibility was allowed for error. The confidence level is higher than 85% and the same can be used for predicting the results within the operating range of experimentation.

Optimization using desirability
The desirability for results obtained vary with respect to the circumstances and criteria will be framed accordingly to the objective functions [10,[36][37][38][39][40]. The regression equations developed earlier were considered to be objective functions for optimization. In this analysis approach, the function for weld hardness was assigned with 'maximization' condition [22,[41][42][43]. Similarly, 'maximization' condition was assigned for the tensile strength function. The desirability values for achieving the expected response were estimated to be 0.916. Optimized process parameters were estimated to be 913.74 rpm RS, 45 mm min −1 TS and axial load of 8 kN. The optimized results were estimated as 116.462 for weld hardness and 325.647 N mm −2 for tensile strength of dissimilar weld joints in figure 7.

Microstructural analysis
The samples were 3 mm thick and made from aluminium alloys AA2024 and AA2099. A longitudinal butt weld was performed along the rolling direction with a vertical milling machine (VMC) [20,44]. Figure 7 depicts a schematic illustration of joint geometry.
AA2024 was available on advancing side of weld, whereas AA2099 was on the retreating side ( figure 8). The welding tool was made of H13 tool steel, which had been heat treated. The diameter of the pin, tool shoulder and pin length are 18, 3.8 and 6.4 mm respectively, as shown in figure 7. The clockwise rotation of the instrument was chosen. The tool was set at a 3 degree angle to the longitudinal axis. The tool rotated at 1140 revolutions per minute and advanced at 32 millimetres per second. Cromel-Alumel type thermocouple was used for measuring the weld zone temperature of the procedure. Two thermocouples were installed with 11 mm intermediate distance from the welded centre on both the advancing and retreating sides for this purpose. The after weld heat treatement was carried out fot 2 h at solution annealing temperature of 460°C . The process for post-weld heat treatment was carried out after the solution treatment. Mechanical properties of tensile and micro-hardness were used to determine the post-weld heat treated joints [16,33]. Samples were prepared as per the ASTM E8 standards for testing [25]. Optical and scanning electron microscopes were used to characterise the microstructure of different locations of welded connections (SEM). A pin on disc tribometer was used to determine the wear rate. Figures 9 & 10 shows the EDX and XRD pattern corresponds to the nugget zone of dissimilar weld joint specimen of AA2024 and AA2099. The above pattern confirms the polycrystalline nature of material and peaks corresponds to cubic Al of JCPDS file # 85-1324. It is observed that the peak at 65°corresponds to (220) reflection is in high intensity then followed by (311) at 80°, (200) at 46°and (111) at 40°. On observing above graph, the strengthening precipitates of MgZn 2 and Al 2 CuMg were dissolved in welded joint. It contains elements such as Fe as FeAl 3 compound in the base material structure. MgZn 2 and Al 2 CuMg particles nucleate and grow as a result of post weld heat treatment. Al 2 CuMg particles gets distinguish by applying aging treatment to the welded specimen. Since the weld region consists of particles corresponds to AA2024 and AA2099, the XRD pattern from 0°to 39°does not show any crystalline phase. The possibility of the formation of MgZn 2  could be observed in the graph during PWHT [44]. The XRD pattern does not show the peak corresponds to the formation of S(Al 2 CuMg) due to lack of sufficient temperature and time for the formation.

Microstructure analysis
The figure 11 represents the microstructural images of various zones in the welded region of dissimilar aluminium alloys of AA2024 and AA2099. The reagent used to study the microstructural feature is Keller's reagent.
The intersect image of a, b and c from figure 10, shows the microstructure of parent metal in solution treated and rolled condition. The grain flow is along the direction of rolling. There is a formation of micro cracks, twinning defects, dimples, fractured area in the nugget zone, solution treated and rolled condition of the parent metal when compared to microstructures of same material weld joint developed by FSW. The microstructure shows elongated grains of primary aluminium with precipitated eutectic particles in the solid solution of aluminium grains. The intersect image of d and e shows the heat affected zone of the process at p-1 region with the nugget zone. The microstructure shows partially recrystallized grains through the influence of FSW process. The magnifications are 100 and 200. The matrix shows more precipitated particles. The intersect image of f and g shows the interface zone of the nugget and the heat affected zone of the process. The centre line is the fusion line. Left side shows the parent metal heat affected zone and the right side shows the nugget zone with fragmented grains of primary and eutectic particles precipitated. Some grain direction change is observed along the direction of revolution of the tool. The effect of re-crystallization is more pronounced, and the grain size has changed at the heat affected zone. The field also shows formation of TMT zone [45].

Macro structure analysis
The stir zone was wider, whereas the nugget zone shrank as it approached the vortex region. Because of the lower heat input during friction welding. The HAZ area on the progressing side is greater than on the withdrawing side, according to the findings. Because of the tool's rotation orientation, this event is typical in the FSW process   [1,7,25,43,46]. We see that material near to the tool is subjected to a lot of shear, deformation (compressive), and heat, leading the material towards the condition of large plastic deformation, dynamic recrystallization, and an increase in the width of the weld stir zone. The reduction of tool rotating speed and raise of travelling speed resulted in the smaller joint area on comparing with the macrostructure of other specimen (figure 12).

SEM analysis
In the as-weld junction of dissimilar aluminium alloys, there are two sorts of particles. Compounds obtained from precipitation-hardening like S(Al 2 CuMg) and (MgZn 2 ), as well as compounds combining Fe and Mn like (Cu, Fe, Mn) Al 6 , are examples. It is well known that during high temperature heat treatment, strengthening  The microstructure of the weld determines the mechanical properties of the joints, as shown in figure 13. The presence of the major alloying elements such as Cu, Mg and Li in the stir zone can be seen more clearly by SEM-EDS maps. Presence of the precipitates such as T1 (Al 2 CuLi), θ΄ (Al 2 Cu) and S΄(Al 2 CuMg) are found in the  FSWed region. T1 appears as very thin shaped platelets and is the most common strengthening phase in Al-Cu-Li series alloys and θ΄ precipitates are in shape of thin plates. The S΄(Al 2 CuMg) precipitates were indicated by thin needles where, they have a cohesive interaction with the matrix and so, rather than dispersions, are responsible for inhibiting dislocation movement [47].
There are four distinct microstructural zones discovered in all joints: the BM, the heat affected zone (HAZ), the thermomechanically affected zone (TMAZ), and the stir zone (SZ) [48][49][50]. Optical and SEM showed that all of the samples' microstructures were characterized by fine equiaxed grain in the stir zone. With high heat input, that is, at a low welding speed, intensive mixing of the material and movement of the material towards the upper surface occurs. This results in dynamic recrystallization and grain refinement [51,52].
The strength of welded aluminium alloy joints, for example, is determined by the grain size and precipitate. There are various methods available for selection of tool profile that can be used to control the strengthening factors and improve the mechanical properties [18,27,28,[53][54][55]. There is a formation of twinning defect, micro cracks, cleavages, fractured area as compared to similar welding of aluminium alloy but the particles formed here are fine in nature. The amount of heat energy developed in the welding process affects the microstructure in general. In this case, the lowest welding speed will appear to have the highest heat input, and vice versa. The welding regime correlates to fine microstructure, while the regime corresponds to coarse microstructure. On the subject of structure, there is a substantial quantity of published study in the open literature.

Wear analysis
The wear resistances in the weld joint region were measured using pin on disc tribometer. The parameters like wear rate, friction force and friction coefficient were measured in the nugget zone of welded joint and in the nugget zone of specimen (noted as 11 in the figure). The wear rate of dissimilar AA2024 and AA2099 were lower with the wear value increases from 0 to 50 μm in the nugget zone of the specimen ( figure 14). This happens due to the material transfer from Alumina ball to the wear track. During pin on disc tribometer measurement as the sample is harder compared than the ball (in the experiment), there is a possibility for wear happens from the ball. The applied load was 10 N, sliding velocity of 1.5 m s −1 , sliding distance of 800 m, sliding diameter of 30 mm and 955 rpm at a time period of 8.887 min. The wear debris formed will transfer to the wear track as the dissimilar AA2024 and AA2099 nugget zone shows positive value [8,17,44].
The CoF of dissimilar sample was shown in figure 15. The value remains constant throughout the wear experiment ranges from 0.3 to 0.55 as the alumina ball from the tribometer does not experiences much friction on the wear track of both samples. The wear loss was measured by finding the difference between initial weight and final weight and found as in the range from 0.2283 g to 0.4866 g as shown in the figure 15. Wear surface analysis reveals that delamination is the primary wear mechanism at low frictional heat, whereas metal flow and severe delamination are identified as wear mechanisms at high frictional heat [56,57]. In this instance, formation of debris and delamination was observed [58]. The dissimilar weld samples showed increased hardness thus the resistance to abrasive wear improved, therefore the susceptibility for ploughing and the creation of surface deformation decreased.

Conclusions
The first set of plates consists of identical aluminium alloys AA2024 and AA2099 plates that were welded to improve the process settings in order to investigate structural, microstructure, macrostructure, mechanical, and wear behaviour. For wear resistance qualities, the welded connections were examined. Experiments were carried out with various combinations of rotating speed, transverse speed, welding time, and axial stress. The welding parameters influence on the findings was explored with the help of ANOVA. Results were acquired because of the research.
• Hardness was (maximum) for lower amount of RS, average TS and AL. The tensile strength was 361.24 N mm −2 (maximum) for average RS and TS with maximum load conditions.
• The parameters influencing hardness can be arranged in descending manner as RS > AL > TS. The influencing parameters for tensile strength were TS and RS. • Mathematical models were developed for predicting the results with a confidence level above 96% for all the results.
• Optimum parameter combination for dissimilar material welding was observed to be 913.74 rpm RS, 45 mm min −1 T −1 S −1 and 8kN AL. The desirability values for dissimilar welding process were 0.912 respectively.
• The XRD pattern confirms the polycrystalline nature of material and peaks corresponds to cubic Al of JCPDS file # 85-1324.
• The effect of recrystallization is more pronounced, and the grain size has changed at the heat affected zone. The field also shows formation of TMT zone.
• Tunnel defects were observed in welded joints. It was minimum for the 1st specimen having maximum tensile strength where we can be able to see a narrow heat affected zone (HAZ) was formed on the side lines of SZ in PC side.
• The grain refinements were decreased in the range of WC > WT > HAZ > Parent material.
• The wear rate of dissimilar AA2024 and AA2099 were superior as the wear value increases from 0 to 50 μm in the nugget zone.
• The coefficient of friction value remains constant throughout the wear experiment ranges from 0.3 to 0.55. A steady state friction value of 3N to 5N is observed when sliding distance increases.
• The wear loss was measured by finding the difference between initial weight and final weight and found as in the range from 0.2283 g to 0.4866 g.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).