Cold-Rolling Strain Hardening Effect on the Microstructure, Serration-Flow Behaviour and Dislocation Density of Friction Stir Welded AA5083

Abstract

5083 aluminium (Al) alloy materials have extensive structural applications in transportation industries because of their high strength-to-weight ratio and corrosion resistance. However, under conventional fusion weldings, these materials are limited by their porosity, hot cracking, and distortion. Herein, friction stir welding (FSW) was performed to join a similar AA5083 alloy. A post-weld cold-rolling (PWCR) process was applied on joint samples at different thickness-reduction percentages (i.e., 10%, 20%, and 40%) to identify the effect of strain hardening on the microstructure and mechanical properties of the friction-stir-welded joint of AA5083 while considering the serration-flow behaviour at stress–strain curves and dislocation density of the post-weld cold-rolled (PWCRed) samples. FSW induced a 20% reduction in the tensile strength of the joint samples relative to the base metal. PWCR also reduced the average grain size at the nugget zone and base metal because of the increase in plastic deformation imposed on the samples. Furthermore, PWCR increased the dislocation density because of the interaction among dislocation stress fields. Consequently, the tensile strength of the friction-stir-welded joint increased with the increased cold-rolling percentage and peaked at 403 MPa for PWCRed–40%, which significantly improved the serration-flow behaviour of stress–strain and welding efficiency up to 123%.

1. Introduction

Aluminium (Al) alloys are exceptionally adaptable materials that are particularly appropriate for the development of structural designs associated with the transportation industry, given their high corrosion resistance, high toughness, and light weight [1,2]. AA5083 is a frequently utilised Al alloy in the 5xxx series. The versatility of this strain-hardened alloy renders it ideal for applications related to the automotive industry [3] and for the fabrication of marine structures, pipelines, and aircraft components [4]. Given the problems encountered in the fusion welding of Al alloys, such as porosity [5], cracking [6], and dissolution of precipitation phases [7,8], a solid joining technique known as friction stir welding (FSW) was carried out for the joining of Al alloys [9,10,11,12,13]. In 1991, this welding process was introduced by The Welding Institute [14]. FSW is a joining process involving the use of a rotating tool that plunges and traverses along the interface of two or more plates [15]. FSW is used in the aerospace, automotive, railway, and shipbuilding industries. It is typically applied to butt joints on large Al parts [16]. Several advantages, including high-quality joints, low-energy process, precise external control, low pollution, and lack of special sample preparation, are provided via FSW [17,18,19]. In FSW, the deformation of the material under melting-point temperatures reduces the negative effects caused by distortion and residual stress [20,21].

Although FSW is a solid-state welding process, it is susceptible to the effects of residual stress and distortion [22], which diminishes the joint’s mechanical strength. The effect of residual stress in FSW has been extensively investigated. For example, Liu et al. [23] employed the neutron diffraction method to analyze residual stress in single- and double-sided friction-stir-welded (FSWed) Al 6005A-T6. The single-sided FSW of 6005A-T6 delivers a maximum tensile strength of 74.4% of base metal (BM) with low residual stresses of approximately 37.5% yield strength of BM and a peak yield strength extent of approximately 37.5%. Furthermore, the double-sided FSW of 6005A-T6 shows a residual stress of 70% yield strength of BM with a tensile strength of 80.8% of BM, which is double that derived through single-sided FSW.

The FSW of AA5083 has been widely investigated to optimise welding parameters and achieve the maximum joint strength [24,25,26,27]. Reduced mechanical properties still occur in AA5083 FSWed joints because of high solute concentration; the associated strain hardening; and the serration-flow behaviour effect (Portevin-Le Chatelier effect [28]) occurring in the microstructure of FSWed joint, especially at the heat affected zone (HAZ). For example, Behnagh et al. [24] reported that the hardness value dropped from 88 HV in BM to 72 HV in friction stir processed AA5083 because of precipitate dissolution. Leal et al. [29] reported that the lowest hardness occurred at the HAZ of FSWed Al5083 because of coarse grain occurrence. Therefore, a post welding process is required to improve the mechanical properties of FSWed AA5083. Post-weld cold rolling (PWCR) may have the most significant potential as a low heat input process to induce the strain hardening effect on non-heat-treated Al alloys, which is the main mechanism to strengthen this alloy. Gabrielli et al. [30] studied the effect of the PWCR process on FSWed AA5754. They reported that the CR process less than 10% in the thickness reduction of FSWed joint can improve the ultimate tensile strength of FSWed joints from 224 MPa to 243 MPa.

Furthermore, thickness reduction in butt joints is a limitation in FSW process because of the tool shoulder’s forging effect [31]. As a result, the mechanical properties and formability of the FSWed blanks are reduced, and this condition deteriorates with a decrease in the thickness of the sheet. This decrease in thickness worsens the aesthetic appearance of the welded blank assembly. This condition reduces the applicability of FSW. Therefore, cold rolling can be applied as a post-weld technique to have a uniform sheet thickness after the welding process. The cold-rolling increases the strain hardening coefficient and enhances the mechanical strength and formability of the material while maintaining a uniform sheet thickness [32].

To date, research and information about the influence of cold rolling after the FSW of AA5083 are lacking. Therefore, the effect of the cold-rolling strain hardening process on the mechanical properties and deformability of the FSWed samples must be thoroughly investigated. Therefore, this study aimed to investigate the effect of the post-weld cold-rolling (PWCR) process on the microstructure, mechanical properties, and serration-flow behaviour of the stress–strain curves of FSWed samples at a different rolling percentages.

2. Materials and Methods

AA5083 sheets were used as the BM in this study.

Table 1. Chemical composition of AA5083-base metal (BM) (wt%).

Al	Mg	Mn	Cr	Si	Fe
Balance	4.46	1.0	0.14	0.25	0.45

shows the chemical composition of the AA5083 sheets. The sheets were cut and machined to approximately 150 mm × 100 mm × 5 mm. For the FSW process, the specimens were fixed and clamped in a butt joint configuration, and a 4.7 mm long H13 tool pin was used for the joining process, as shown in Figure 1. The pin and shoulder diameters were 5 and 20 mm, respectively. Figure 1b shows that the pin tool was set at a three-degree tilted angle to the tool axis during the FSW process. After several preliminary tests, the travel and rotational speeds were set at 100 mm/min and 1000 rpm, respectively.

The welded specimens were then subjected to cold-rolling parallel to the welding direction. The cold-rolling reduction percentage was set at 10%, 20%, and 40% from the original thickness of the AA5083 sheets. For simplicity, post-weld cold-rolled (PWCRed)–10%, PWCRed–20%, and PWCRed–30% were used to name the FSWed specimens at different reduction percentages, that is, 10%, 20%, and 40%, respectively.

Microstructures were examined at the cross-section of the joint perpendicular to the welding direction. Prior to microstructure observation, the FSWed specimens were cold-mounted and manually ground using sand papers ranging from #600 to #2000. The specimens were then polished, which involved the use of a diamond polishing suspension with a particle size of 1 µm. The microstructure of the AA5083 was revealed by electrochemical etching using Barker’s reagent (5 mL of HBF₄ and 100 mL of H₂O) at 20 VDC. Weld zone microstructures were analysed via optical microscopy. The measurement of the average grain size was in accordance with the ASTM Standard E112-10. X-Ray Diffraction (XRD) analysis was carried out to distinguish the dislocation density of the samples via a Bruker AXS Germany diffractometer with a monochromatic Cu–Kα radiation of λ = 0.1541 nm at 40 kV and 4 mA. The 100 kN Zwick universal testing machine with a crosshead speed of 1.8 mm/min was employed for tensile tests at room temperature. The three tensiles were tested for each condition. The specimens were machined normal to welding direction with the gauge length of 40 mm and a width of 6 mm. This condition complies with the ASTM Standard E8M-04. Zwick Vickers hardness machine was employed to evaluate micro-hardness along the cross-section of the FSWed samples using a load of 0.1 kgf and dwell time of 15 s. The fractography of the samples was performed after tensile test to study the fracture mode of samples via the ZEISS MERLIN field emission scanning electron microscope (FE-SEM) (Jena, Thuringia, Germany).

3. Results and Discussion

3.1. Microstructure

A cross section of the FSWed and PWCRed samples was observed to investigate microstructure and welding defects after welding and rolling. The as-weld sample showed no internal defects at the cross sections of the welding joints, as shown in Figure 2. The SZ was basin-shaped, as shown in Figure 2a. This shape has been reported for the first time by Sato et al. [33]. A clear borderline that indicates the nugget zone (NZ) and thermomechanically affected zone (TMAZ) at AS was located below the shoulder area.

Figure 2b–d show the cross-section of PWCRed–10%, PWCRed–20%, and PWCRed–40%, respectively. Increasing the rolling percentage has compressed the thickness of the welding plate. The weld zone became wider by increasing of CR percentages compared with the welding zone of the as-welded sample without any cracks at the middle of cold-rolled samples. The NZ area of PWCRed–40% was the widest compared with other samples because of the highest strain imposed on the sample. Furthermore, Figure 3 shows the microstructures of the as-weld and PWCRed samples in advancing and retreating sides. As shown in Figure 3, the borderline between NZ and TMAZ was observed at AS, in contrast with RS. The matter on RS is much less transported around the tool before being deposited in the welding line when compared with the matter on AS [34,35].

Figure 4a shows the typical Al structure of as-received AA5083 with the average grain size of 21 µm. Dynamic recrystallisation and equiaxed grain microstructure with an average grain size of 14 µm were observed in the NZ of the as-welded sample because of the mechanical strain and frictional heat of the rotating tool, as presented in Figure 4b. As shown in Figure 4c–h, rolling was performed to compress and elongate the microstructure parallel to the rolling direction in weld zones of the rolled samples. The rolling process decreased the average grain size of FSWed samples, especially in SZ, in which they recrystallised and formed very fine microstructures. The average grain size of BMs was 19, 16, and 9 µm for PWCRed–10%, PWCRed–20%, and PWCRed–40%, respectively, as presented in Figure 4c,e,g.

The average grain size in the NZ also decreased because of the increase in the percentage of rolling, whereas the NZ grain sizes of PWCRed–10%, PWCRed–20%, and PWCRed–40% were 13, 9, and 6 µm, respectively. The plastic deformation and, consequently, the average grain size reduction in the welding zone could increase the dislocation density and improve the mechanical strength of materials [36,37]. The alteration of dislocation density after PWCR will be discussed further.

3.2. Tensile Properties

The tensile properties and stress–strain curves of AA5083-BM, as-welded, and PWCRed samples are shown in

Table 2. Tensile properties of AA5083-BM, as-welded, and post-weld cold-rolling (PWCR)ed samples.

Sample	Tensile Strength (MPa)	Elongation (%)	Welding Efficiency (%)
AA5083-BM	328	22	-
As-welded	262	7.3	80
PWCRed–10%	321	8.6	98
PWCRed–20%	350	8.4	107
PWCRed–40%	403	7.9	123

and Figure 5, respectively. FSW reduced the tensile strength of the AA5083-BM, whereas rolling improved the mechanical properties of the welded samples. The tensile strength of the BM decreased to 262 MPa after FSW because of frictional heat induced by the softening behaviour and increased grain size [38].

The cold-rolled strain hardening increased the yield and ultimate strength of joints as a result of the decreasing average grain size based on the Hall–Petch equation [39]. Furthermore, the elongation of the rolled samples (approximately 8%) was not significantly different compared with that of the as-weld sample. The tensile strength was increased to 321, 350, and 403 MPa by PWCRed–10%, PWCRed–20%, and PWCRed–40%, respectively. Correspondingly, the elongation of the rolled samples slightly decreased by increasing of reduction percentage as a result of average grain size reduction. Grain refinement is well known as a method that can increase the strength of metallic materials without decreasing their ductility. Moreover, welding efficiency increased from 98% to 123% by increasing the rolling percentages from 10% to 40%. Welding efficiency is the ratio of the tensile strength of PWCRed sample and the BM.

The serration-flow behaviour effect observed in the stress–strain diagrams of the as-welded and PWCRed samples is typical behaviour of Al–Mg alloys [28] that is directly attributed to the dislocation density in the structure of the sample, as presented in Figure 6. The dislocation movement in the AA5083 is hindered by the Mg atom in substitutional solid solution of Al–Mg alloy [40]. Furthermore, it influences the properties of AA5083 by the creation of serration-flow behaviour in the stress–strain curve during the tensile loading [41]. This serration-flow behaviour is called the Portevin-Le Chatelier (PLC) effect during the plastic deformation of the material [28]. The substitute and interstitial alloys, such as Al alloys [42], carbon steel [43], and copper alloys [44], tend to demonstrate PLC behaviour during tensile loading [45]. The PLC effect is related to pinning and unpinning processes, where the pinned process has occurred when the dislocation is stuck by obstacles formed from the forest dislocations and grain boundaries. Otherwise, the unpinned process occurs when the dislocation can be overcome from the solute cloud, leading to a stress drop during the tensile curve [46]. The problems caused by PLC are the decreasing of formability of materials and the formation of undesired band mark on the specimen surface [47].

Furthermore, the stress–strain curves of PWCRed specimens showed different serration-flow frequencies under a different rolling percentage, as shown in Figure 6. Increasing the cold-rolling percentage results in an increase in dislocation that piles up at the grain boundaries, as well as an increase in the strain hardening, leading to the higher tensile strength values [48]. Furthermore, the finer grains result in denser and uniform dislocation structure at higher deformation levels. In addition, the changes of wrinkle behaviour in the PWCRed–40% sample are consistent with the ideas of a homogeneously uniform distribution of the dislocations. The uniform plastic flow is suggested to occur in those subjected to a regular form and finer grain of Al [49,50,51].

Cold working leads to two remarkable changes in the microstructure of materials, namely, grain size and dislocation density, which in turn affect the performance enhancement of the materials. In general, materials with smaller grain size have higher mechanical properties than those with bigger grain size under the same cold-rolling conditions. This results because of the increase in dislocation density associated with high reduction rolling effect, leading to high plastic deformation and grain size reduction. Therefore, the influence of the dislocation density in FSWed and PWCRed samples must be considered. Figure 7 shows the XRD patterns of the as-welded and PWCRed samples.

The value of the dislocation density ρ (m⁻²) can be obtained from the average values of the crystallite size (d) and microstrain <η²>^1/2 using the following equation [52,53]:

$$\rho =\frac{2\sqrt{3}{\langle {\eta }^2\rangle }^{\frac{1}{2}}}{d\times b},$$

(1)

where d and η are calculated from the Williamson–Hall method, and Burgers vector (b) is calculated via b = $\frac{a}{\sqrt{2}}$ [54] (a is the lattice parameters of the as-welded and cold-rolled samples obtained from the Nelson–Riley method). Using Equation (2), the average crystallite size and microstrain can be calculated from the y-intercept and slope of the curves, respectively, which are obtained from the linear fitting of the Williamson–Hall plot, as shown in Figure 8:

$$FWHM\times \cos \left(\theta \right)=\frac{0.9\lambda }{d}+2\eta \sin \left(\theta \right),$$

(2)

where FWHM is the full width at the half-maximum intensity of the selected peak, θ is the Bragg’s diffraction angle, and λ is 0.1541 nm for the wavelength of Cu–Kα radiation. Figure 8 is described further in

Table 3. Mean crystallite size and strain amount of the rolled sample based on the Williamson–Hall equation.

Sample	Y = αX + β		d (n.m)	ⴄ (%)
	α	β
As-welded	0.0021	0.0011	126.08	0.21
PWCRed–10%	0.0024	0.0013	106.68	0.24
PWCRed–20%	0.0026	0.0014	99.06	0.26
PWCRed–40%	0.0030	0.0020	69.35	0.30

, which indicates a decrease in the samples’ crystallite size and an increase in strain with increased cold-rolling percentage. At 40% reduction, the crystallite size was decreased to a minimum value of 69.35 nm, whereas the lattice strain increased to a maximum value of 0.30. These findings demonstrate that the thickness reduction by cold-rolling could change the crystallite size (d) and strain (η).

Notably, the Burgers vector (B) can be changed during rolling because of microstrain-induced disturbances, which are caused by the lattice parameter of Al deviating from the database of FCC (a = 0.4094 nm [55]). The Nelson–Riley function (Equation (3)) was utilised to obtain the lattice parameter of Al, as follows [53,56]:

$$F\left(\theta \right)=\frac{1}{2}\left(\frac{{\cos }^2\theta }{\sin \theta }+\frac{{\cos }^2\theta }{\theta }\right),$$

(3)

where F(θ) is the Nelson–Riley function and θ is the Bragg’s diffraction angle. The lattice parameter of Al as a function of F(θ) is shown in Figure 9. The values of lattice parameter, Burgers vector, and dislocation density of the as-welded and PWCRed samples are summarised in

Table 4. The values of lattice parameter, Burgers vector, and dislocation density of the as-welded and PWCRed samples.

Sample	a (Nelson–Riley Method) (Å)	b (10−10 m)	ρ × 1014 (m−2)
As-welded	4.0729	2.87998	2.00340
PWCRed–10%	4.0721	2.87941	2.70643
PWCRed–20%	4.0715	2.87899	3.15797
PWCRed–40%	4.0705	2.87828	5.20672

.

As presented in Table 4, the values of lattice parameter and Burgers vector are not significantly different. However, the dislocation density increased with the increasing of rolling percentages. The dislocation density increased significantly at 40% rolling percentage. This phenomenon is believed to be attributed to the intrinsic capacity of dislocations that interact with one another during rolling by creating stress fields within the material. The interaction between the stress fields of dislocations can act as an obstacle against dislocation motion by attractive or repulsive interactions. At this time, the mechanical properties of the materials can be improved owing to the disrupted motion of dislocations. The frequencies of serration are different because of the stress fields of the grain size, in which the serration frequency is high for the as-welded sample. After rolling, the serration causes the stress to increase with increased rolling percentage because of the pinning effect [42]. Thus, the strain hardening increasing during rolling results in decreasing crystallite size and lattice strain for the sample.

3.3. Fractography

The fracture surface of the as-welded and cold rolled samples was examined after the tensile test. Figure 10 shows the fracture surfaces of as-welded, PWCRed–10%, and PWCRed–40% samples. The fracture occurred from the boundary between NZ and TMAZ in the as-welded sample to HAZ in PWCRed–10% and PWCRed–40%.

SEM observation showed dimple fracture of the as-welded sample (Figure 10a) compared with the PWCRed samples. This result indicates a ductile fracture behaviour of the as-welded joint showing the lowest tensile strength in the as-welded condition (Figure 5). The various sizes of dimples with shear effect were observed on the fracture surfaces of PWCRed–10% and PWCRed–40% samples. Dimple size decreased with increased rolling from 10% to 40%, as presented in Figure 10b and c. Imperfect feature, as well as the small size of dimples, has been an indication for the brittleness with low ductility of joints [57]. Shallow dimples and flat surfaces were obtained at PWCRed–10%, and PWCRed–40% can be related to brittle fracture behaviour compared with the as-welded sample after the tensile test. This result provides evidence of the increased tensile strength in PWCRed samples.

3.4. Vickers Hardness

Evolution of the hardness values along the welded joints of the specimens was measured using the Vickers micro-hardness tester. The hardness profiles measured starting from the base metal of the advancing side to the retreating side are shown in Figure 11. The values of hardness are closely related to the microstructural evolutions that occurred within the weldments at each process, that is, FSW, cold-rolling processes. Hardness profiles are different depending on the welding regions and cold-rolling percentage. The initial hardness of AW55 was 90 ± 2 HV. The hardness of weldments then started to decrease when approaching the HAZ. Minimum hardness was attained at the TMAZ (i.e., 82–85 HV). Cold rolling improved the hardness of the welded joint, and the hardness increased from 93 ± 2 HV to 106 ± 2 HV when the thickness-reduction percentage was increased from 10% to 40%. This phenomenon is attributable to the decreasing grain sizes that led to the improved hardness of the materials in association with the Hall–Petch equation [58]. Furthermore, good agreement was found between hardness profile results, tensile strength, and the fracture location of the samples after the tensile test. Thus, the fracture occurred in the region of low hardness values, that is, HAZ and TMAZ.

4. Conclusions

FSW was performed to join a similar AA5083 Al alloy and the welded specimens obtained without any defects. The FSWed specimens were then subjected to the PWCR process. The mechanical properties of the PWCRed specimens were improved owing to the induced strain hardening. PWCRed samples without any defects were successfully produced at different thickness reductions of 10%, 20%, and 40%. The average grain size in NZ and BM reduced more with the increase of cold-rolling percentage because of the increase in plastic deformation imposed on the samples. The average grain size of SZ in the as-welded sample decreased from 14 µm to a smaller grain size of 13, 9, and 6 µm by cold-rolling at 10%, 20%, and 40% thickness reduction, respectively. Furthermore, PLC behaviour was observed in the stress-strain curve of as-welded and PWCRed samples such that the serration-flow frequency decreased with increased thickness reduction because of the strain hardening induced during rolling and dislocation motion. The tensile strength of the BM was reduced by approximately 20% after FSW. However, the PWCR process reduced the serration effect in tensile behaviour and improved the tensile strength of the joint samples. The tensile strength increased to 403 MPa at 40% of PWCR and demonstrated a 123% improvement in welding efficiency owing to the disrupted motion of dislocations and increasing dislocation density from 2.00340 × 1014 m⁻² in the as-welded sample up to 5.20672 × 1014 m⁻² in the PWCRed–40% sample.