GTA Weldability of Rolled High-Entropy Alloys Using Various Filler Metals

: Gas tungsten arc (GTA) weldability of rolled CoCrFeMnNi high-entropy alloys (HEAs) was conducted using stainless steel (STS) 308L and HEA ﬁllers. Microstructure and mechanical properties of the welds were examined to determine GTA weldability of the rolled HEA. The welds had no macro-defects, and component behaviour between base metal (BM) and weld metal (WM) showed signiﬁcant di ﬀ erences in the weld using the STS 308L ﬁller. Macro-segregation of Fe components was conﬁrmed in the central region in the WM using the STS 308L ﬁller. Because the columnar grain sizes of all the WMs were larger than those of the rolled HEA BM irrespective of the ﬁller metals, the tensile properties of the GTA welds were lower than those of the rolled HEA BM, and the tensile fracture occurred in the centreline of each weld. In particular, the tensile properties of the weld using the STS 308L ﬁller deteriorated more than those of the HEA weld. This was induced by the formation of macro-segregation and severe martensite transformation in the centreline of WM. To enhance the weldability of the rolled HEA, the formation of macro-segregation and coarse grains in the WM of GTA welds must be prevented.

To apply HEAs as structural applications, the development of BMs and weldability evaluations should be prioritised [15,16]. HEA BMs have been actively developed; however, studies related to weldability evaluation are limited to low heat input welding: electron beam welding [17], laser beam welding [18][19][20][21], friction-stir welding [21][22][23][24][25]. Recently, the evaluation of gas tungsten arc (GTA) weldability on cast HEA using the developed HEA filler has been reported [26], and there have been studies on HEA weldability using high heat input welding [27,28]. The rolled HEA must be welded  Figure 2 shows the compositional behaviour of the obtained welds. Macro mapping of EPMA showed the compositional segregation behaviour of Fe, Ni, and Cr, the main components of the STS 308L filler (Figure 2a). However, the weld obtained using the HEA filler had a uniform composition with the BM, as shown by the little colour difference (Figure 2b) in the main components between BM and weld metal (WM). Significant variations were observed in the component behaviour of Fe and Ni in each WMs. In the centreline of the weld obtained using STS 380L filler, macro-segregation region of the Fe component and deficient region of the Ni component were formed because of the difference in the alloy compositions of the rolled HEA BM and that of each filler (HEA and STS 308L). The melting temperature of the STS 308 filler (~1723 K) was higher than that of the HEA BM (~1550 K), which is one of the causes of the inhomogeneous mixing during GTA welding [31,32].  Figure 3 shows the component behaviour quantitatively in the welds using HEA and STS 308L fillers. As depicted in Figure 1, EPMA quantitative analysis was performed along the red dotted lines from the rolled HEA BM to the centreline of the welds produced by the HEA and STS 308L fillers. For the WM produced by STS 308L filler, the composition of Fe in the WM (46.4 ± 0.5 wt.%) was significantly larger than that of the HEA BM (21.4 ± 0.2 wt.%), and slightly lower than that of STS 308L filler (67.9 ± 0.1 wt.%) (Figure 3a). Moreover, the compositions of Co, Ni, and Mn in the WM were lower than those in the HEA BM. The Cr component of the WM exhibited behaviour that was homogeneous with that of the HEA BM. Furthermore, small amounts of Si and C were detected in the WM, as shown in Figure 3a. The WM using STS 308L filler contained higher amounts of Ni, Mn, and Co than in case of the STS 380L filler, which stabilised the FCC because the STS 380L filler was diluted from the HEA BM to the WM. However, the compositions of all components in the WM using the HEA filler indicated behaviour that was homogeneous with that of the BM (Figure 3b).  Figure 2 shows the compositional behaviour of the obtained welds. Macro mapping of EPMA showed the compositional segregation behaviour of Fe, Ni, and Cr, the main components of the STS 308L filler (Figure 2a). However, the weld obtained using the HEA filler had a uniform composition with the BM, as shown by the little colour difference (Figure 2b) in the main components between BM and weld metal (WM). Significant variations were observed in the component behaviour of Fe and Ni in each WMs. In the centreline of the weld obtained using STS 380L filler, macro-segregation region of the Fe component and deficient region of the Ni component were formed because of the difference in the alloy compositions of the rolled HEA BM and that of each filler (HEA and STS 308L). The melting temperature of the STS 308 filler (~1723 K) was higher than that of the HEA BM (~1550 K), which is one of the causes of the inhomogeneous mixing during GTA welding [31,32].

Compositional and Microstructural Behaviour of GTA Welds Using Various Filler Metals
Metals 2020, 10, x FOR PEER REVIEW 3 of 10 of the fillers. These results show that when each filler was melted, a part of the BM was properly melted, resulting in sound welds irrespective of the fillers.  The melting temperature of the STS 308 filler (~1723 K) was higher than that of the HEA BM (~1550 K), which is one of the causes of the inhomogeneous mixing during GTA welding [31,32].  Figure 3 shows the component behaviour quantitatively in the welds using HEA and STS 308L fillers. As depicted in Figure 1, EPMA quantitative analysis was performed along the red dotted lines from the rolled HEA BM to the centreline of the welds produced by the HEA and STS 308L fillers. For the WM produced by STS 308L filler, the composition of Fe in the WM (46.4 ± 0.5 wt.%) was significantly larger than that of the HEA BM (21.4 ± 0.2 wt.%), and slightly lower than that of STS 308L filler (67.9 ± 0.1 wt.%) (Figure 3a). Moreover, the compositions of Co, Ni, and Mn in the WM were lower than those in the HEA BM. The Cr component of the WM exhibited behaviour that was homogeneous with that of the HEA BM. Furthermore, small amounts of Si and C were detected in the WM, as shown in Figure 3a. The WM using STS 308L filler contained higher amounts of Ni, Mn, and Co than in case of the STS 380L filler, which stabilised the FCC because the STS 380L filler was diluted from the HEA BM to the WM. However, the compositions of all components in the WM using the HEA filler indicated behaviour that was homogeneous with that of the BM (Figure 3b).  Figure 3 shows the component behaviour quantitatively in the welds using HEA and STS 308L fillers. As depicted in Figure 1, EPMA quantitative analysis was performed along the red dotted lines from the rolled HEA BM to the centreline of the welds produced by the HEA and STS 308L fillers. For the WM produced by STS 308L filler, the composition of Fe in the WM (46.4 ± 0.5 wt.%) was significantly larger than that of the HEA BM (21.4 ± 0.2 wt.%), and slightly lower than that of STS 308L filler (67.9 ± 0.1 wt.%) (Figure 3a). Moreover, the compositions of Co, Ni, and Mn in the WM were lower than those in the HEA BM. The Cr component of the WM exhibited behaviour that was homogeneous with that of the HEA BM. Furthermore, small amounts of Si and C were detected in the WM, as shown in Figure 3a. The WM using STS 308L filler contained higher amounts of Ni, Mn, and Co than in case of the STS 380L filler, which stabilised the FCC because the STS 380L filler was diluted from the HEA BM to the WM. However, the compositions of all components in the WM using the HEA filler indicated behaviour that was homogeneous with that of the BM (Figure 3b). To link compositional behaviour with the weld microstructure, we performed a phase analysis using XRD. Figure 4 shows the crystal structure obtained by XRD in the rolled HEA BM and welds produced using HEA and STS 308L fillers. Irrespective of the filler (HEA and STS 308L), the observed FCC single phase with diffraction peaks (2θ = 43.4°, 51.6°, and 74.7°) in the welds was the same as that in the rolled HEA BM. Typically, the STS 304 weld using STS 308L filler contains a small amount of δ-ferrite, which prevents solidification cracking during welding [33][34][35]. However, it was confirmed that the HEA weld using STS 308L filler comprised only the FCC single phase. The bodycentred cubic (BCC) phase was not formed in the WM using STS 308L filler, because the stabilising elements of FCC were diluted from the HEA BM [26]. XRD results of the rolled HEA BM and each weld were consistent with those of EPMA quantitative analysis. The WM produced by STS 308L filler contained a higher amount of FCC-stabilising components (Ni, Mn, and Co) than in the STS 308L filler (Figure 3a).  Figure 5 shows the microstructural behaviour of each region, such as BM, heat-affected zone (HAZ), fusion line, and WM, in the welds using the HEA and STS 308L fillers. The rolled HEA BM had fine equiaxed grains with an approximate size of 6 ± 0.3 μm, and twins generated by recrystallisation of the annealing treatment were observed ( Figure 5a). For all welds, the grain morphology of the HAZ was similar to that of the BM. However, the grain size of the HAZ was approximately 9 ± 0.5 μm, which was coarser than that of the BM (Figure 5b). However, when approaching the fusion line, finer grains were observed. These areas have been reported as partially melted zones (PMZs) in the case of nonferrous metals, such as aluminium and magnesium alloys [36,37].

Compositional and Microstructural Behaviour of GTA Welds Using Various Filler Metals
Irrespective of the fine grains generated in the PMZs using STS 308L and HEA fillers, coarse To link compositional behaviour with the weld microstructure, we performed a phase analysis using XRD. Figure 4 shows the crystal structure obtained by XRD in the rolled HEA BM and welds produced using HEA and STS 308L fillers. Irrespective of the filler (HEA and STS 308L), the observed FCC single phase with diffraction peaks (2θ = 43.4 • , 51.6 • , and 74.7 • ) in the welds was the same as that in the rolled HEA BM. Typically, the STS 304 weld using STS 308L filler contains a small amount of δ-ferrite, which prevents solidification cracking during welding [33][34][35]. However, it was confirmed that the HEA weld using STS 308L filler comprised only the FCC single phase. The body-centred cubic (BCC) phase was not formed in the WM using STS 308L filler, because the stabilising elements of FCC were diluted from the HEA BM [26]. XRD results of the rolled HEA BM and each weld were consistent with those of EPMA quantitative analysis. The WM produced by STS 308L filler contained a higher amount of FCC-stabilising components (Ni, Mn, and Co) than in the STS 308L filler (Figure 3a). To link compositional behaviour with the weld microstructure, we performed a phase analysis using XRD. Figure 4 shows the crystal structure obtained by XRD in the rolled HEA BM and welds produced using HEA and STS 308L fillers. Irrespective of the filler (HEA and STS 308L), the observed FCC single phase with diffraction peaks (2θ = 43.4°, 51.6°, and 74.7°) in the welds was the same as that in the rolled HEA BM. Typically, the STS 304 weld using STS 308L filler contains a small amount of δ-ferrite, which prevents solidification cracking during welding [33][34][35]. However, it was confirmed that the HEA weld using STS 308L filler comprised only the FCC single phase. The bodycentred cubic (BCC) phase was not formed in the WM using STS 308L filler, because the stabilising elements of FCC were diluted from the HEA BM [26]. XRD results of the rolled HEA BM and each weld were consistent with those of EPMA quantitative analysis. The WM produced by STS 308L filler contained a higher amount of FCC-stabilising components (Ni, Mn, and Co) than in the STS 308L filler (Figure 3a).  Figure 5 shows the microstructural behaviour of each region, such as BM, heat-affected zone (HAZ), fusion line, and WM, in the welds using the HEA and STS 308L fillers. The rolled HEA BM had fine equiaxed grains with an approximate size of 6 ± 0.3 μm, and twins generated by recrystallisation of the annealing treatment were observed (Figure 5a). For all welds, the grain morphology of the HAZ was similar to that of the BM. However, the grain size of the HAZ was approximately 9 ± 0.5 μm, which was coarser than that of the BM (Figure 5b). However, when approaching the fusion line, finer grains were observed. These areas have been reported as partially melted zones (PMZs) in the case of nonferrous metals, such as aluminium and magnesium alloys [36,37].
Irrespective of the fine grains generated in the PMZs using STS 308L and HEA fillers, coarse columnar grains occurred from the fusion line and grew in the direction of the weld centreline ( Figure   Figure 4. XRD patterns of rolled HEA base metal (BM) and GTA welds using HEA and STS 308L fillers. Figure 5 shows the microstructural behaviour of each region, such as BM, heat-affected zone (HAZ), fusion line, and WM, in the welds using the HEA and STS 308L fillers. The rolled HEA BM had fine equiaxed grains with an approximate size of 6 ± 0.3 µm, and twins generated by recrystallisation of the annealing treatment were observed (Figure 5a). For all welds, the grain morphology of the HAZ was similar to that of the BM. However, the grain size of the HAZ was approximately 9 ± 0.5 µm, which was coarser than that of the BM (Figure 5b). However, when approaching the fusion line, finer grains were observed. These areas have been reported as partially melted zones (PMZs) in the case of nonferrous metals, such as aluminium and magnesium alloys [36,37]. associated with the significant compositional differences between the rolled HEA BM and WM using STS 308L filler (Figures 1a, 2a and 5c).
Finally, towards the centreline of the WM, coarse columnar grains near the fusion line were transformed into equiaxed grains of the top and middle parts in all welds. However, in the bottom part of the welds, coarse columnar grains were formed in the heat flow direction of the root gap of the V-groove (Figure 5e,f). Equiaxed grains at the centreline of the WM are typical microstructures formed by high heat input welding such as GTA welding [27,31]. Irrespective of the STS 308L and HEA fillers, the total grain size of columnar grains and equiaxed grains in the WM was similar (approximately 400 μm for WM using STS 308L filler and 410 μm for WM using HEA filler).  Figure 6 shows the hardness distribution in the transverse welds produced using the HEA and STS 308L fillers. The average hardness of the rolled HEA BM and WMs was 177 ± 1 Hv and 152 ± 2 Hv0.3, respectively. All WMs showed lower average hardness than that of the rolled HEA BM. As shown in Figures 4 and 5, no phase transformation and secondary phase occurred, owing to the welding heat input. The main reason for the significant difference in the average hardness between the HEA BM and WM in all welds was that the grain size of the BM (~6 μm) was approximately 70 times smaller than that of the WM (400-410 μm) in all welds. The largest hardness in the HAZ near the fusion line was associated with the fine grains locally generated in the PMZ, as shown in Figure  5c,d. Therefore, the hardness distribution in each weld was closely related to the grain size of each region (BM/HAZ/WM) in the welds, and the hardness distribution behaviour of each weld was mostly the same, irrespective of the filler. Irrespective of the fine grains generated in the PMZs using STS 308L and HEA fillers, coarse columnar grains occurred from the fusion line and grew in the direction of the weld centreline (Figure 5c,d). The columnar grains in the WM using the HEA filler showed the same colour as the grains in the PMZ, indicating epitaxial growth (Figure 5d). However, columnar grains in the WM produced by STS 308L filler were depicted in different colours than the grains in the PMZ (Figure 5c), which is associated with the significant compositional differences between the rolled HEA BM and WM using STS 308L filler (Figures 1a, 2a and 5c).

Hardness Distribution Behaviour of the Welds
Finally, towards the centreline of the WM, coarse columnar grains near the fusion line were transformed into equiaxed grains of the top and middle parts in all welds. However, in the bottom part of the welds, coarse columnar grains were formed in the heat flow direction of the root gap of the V-groove (Figure 5e,f). Equiaxed grains at the centreline of the WM are typical microstructures formed by high heat input welding such as GTA welding [27,31]. Irrespective of the STS 308L and HEA fillers, the total grain size of columnar grains and equiaxed grains in the WM was similar (approximately 400 µm for WM using STS 308L filler and 410 µm for WM using HEA filler). Figure 6 shows the hardness distribution in the transverse welds produced using the HEA and STS 308L fillers. The average hardness of the rolled HEA BM and WMs was 177 ± 1 Hv and 152 ± 2 Hv 0.3 , respectively. All WMs showed lower average hardness than that of the rolled HEA BM. As shown in Figures 4 and 5, no phase transformation and secondary phase occurred, owing to the welding heat input. The main reason for the significant difference in the average hardness between the HEA BM and WM in all welds was that the grain size of the BM (~6 µm) was approximately 70 times smaller than that of the WM (400-410 µm) in all welds. The largest hardness in the HAZ near the fusion line was associated with the fine grains locally generated in the PMZ, as shown in Figure 5c,d. Therefore, the hardness distribution in each weld was closely related to the grain size of each region (BM/HAZ/WM) in the welds, and the hardness distribution behaviour of each weld was mostly the same, irrespective of the filler. Metals 2020, 10, x FOR PEER REVIEW 6 of 10 Figure 6. Hardness distribution behaviour of the welds using HEA and STS 308L fillers. Figure 7a,b show the tensile stress-strain curves and fracture positions of rolled HEA BM and welds using HEA and STS 308L fillers, respectively, at 298 K. The yield strength (YS), tensile strength (TS), and elongation-to-fracture (Ef) of the rolled HEA BM were measured to be approximately 377 MPa, 672 MPa, and 53%, respectively. For welds produced using various fillers (HEA and STS 308L), the tensile properties of all welds were worse than those of the rolled HEA BM. The YS of each weld was approximately 374 ± 2 MPa, which was almost the same as that of the BM, and the TS and Ef of the welds using the HEA filler were 58 MPa and 5% higher than those of the welds using the STS 308L filler. Nevertheless, the tensile fracture of all welds occurred near the centreline in the WMs (Figure 7b). To understand the mechanism of the tensile properties in the welds, microstructural analysis was performed near the tensile fracture of welds produced using the HEA and STS 308L fillers. As shown in Figure 7b, all welding specimens were fractured near the centreline of the WM. Figure 8 shows the microstructure near the fracture position in the welds using HEA and STS 308L fillers. We analysed the microstructure of each weld via the IPF, IQ, and phase maps of the EBSD. Deformation twins were observed by the CSL boundary of ∑3, which is indicated by the red line in the IQ maps. The fraction of deformation twins (CSL boundary) of WM using STS 308L filler was 0.08 and that of WM using HEA filler was 0.22. The STS 308L filler produced fewer CSL boundaries than in the case of the HEA filler. The phase map distinguished austenite (yellow), α′-martensite (red), and ε-martensite (green). The STS Figure 6. Hardness distribution behaviour of the welds using HEA and STS 308L fillers. Figure 7a,b show the tensile stress-strain curves and fracture positions of rolled HEA BM and welds using HEA and STS 308L fillers, respectively, at 298 K. The yield strength (YS), tensile strength (TS), and elongation-to-fracture (E f ) of the rolled HEA BM were measured to be approximately 377 MPa, 672 MPa, and 53%, respectively. For welds produced using various fillers (HEA and STS 308L), the tensile properties of all welds were worse than those of the rolled HEA BM. The YS of each weld was approximately 374 ± 2 MPa, which was almost the same as that of the BM, and the TS and E f of the welds using the HEA filler were 58 MPa and 5% higher than those of the welds using the STS 308L filler. Nevertheless, the tensile fracture of all welds occurred near the centreline in the WMs (Figure 7b).

Tensile and Microstructural Behaviour of the Welds
Metals 2020, 10, x FOR PEER REVIEW 6 of 10 Figure 6. Hardness distribution behaviour of the welds using HEA and STS 308L fillers. Figure 7a,b show the tensile stress-strain curves and fracture positions of rolled HEA BM and welds using HEA and STS 308L fillers, respectively, at 298 K. The yield strength (YS), tensile strength (TS), and elongation-to-fracture (Ef) of the rolled HEA BM were measured to be approximately 377 MPa, 672 MPa, and 53%, respectively. For welds produced using various fillers (HEA and STS 308L), the tensile properties of all welds were worse than those of the rolled HEA BM. The YS of each weld was approximately 374 ± 2 MPa, which was almost the same as that of the BM, and the TS and Ef of the welds using the HEA filler were 58 MPa and 5% higher than those of the welds using the STS 308L filler. Nevertheless, the tensile fracture of all welds occurred near the centreline in the WMs (Figure 7b). To understand the mechanism of the tensile properties in the welds, microstructural analysis was performed near the tensile fracture of welds produced using the HEA and STS 308L fillers. As shown in Figure 7b, all welding specimens were fractured near the centreline of the WM. Figure 8 shows the microstructure near the fracture position in the welds using HEA and STS 308L fillers. We analysed the microstructure of each weld via the IPF, IQ, and phase maps of the EBSD. Deformation twins were observed by the CSL boundary of ∑3, which is indicated by the red line in the IQ maps. The fraction of deformation twins (CSL boundary) of WM using STS 308L filler was 0.08 and that of WM using HEA filler was 0.22. The STS 308L filler produced fewer CSL boundaries than in the case of the HEA filler. The phase map distinguished austenite (yellow), α′-martensite (red), and ε-martensite (green). The STS To understand the mechanism of the tensile properties in the welds, microstructural analysis was performed near the tensile fracture of welds produced using the HEA and STS 308L fillers. As shown in Figure 7b, all welding specimens were fractured near the centreline of the WM. Figure 8 shows the microstructure near the fracture position in the welds using HEA and STS 308L fillers. We analysed the microstructure of each weld via the IPF, IQ, and phase maps of the EBSD. Deformation twins were observed by the CSL boundary of 3, which is indicated by the red line in the IQ maps. The fraction of deformation twins (CSL boundary) of WM using STS 308L filler was 0.08 and that of WM using HEA filler was 0.22. The STS 308L filler produced fewer CSL boundaries than in the case of the HEA filler. The phase map distinguished austenite (yellow), α -martensite (red), and ε-martensite (green). The STS 308L filler produced more α and ε martensites (Ms~0.24) than the HEA filler (Ms~0.15). The tendency of fraction of martensite transformation was opposite to that of the CSL boundary, that is, a smaller CSL boundary and more martensite for the weld produced using the STS 308L filler. These results were associated with the macro-segregation observed near the WM centreline (Figure 2a). The FCC stabilising elements of Ni, Mn, and Co were deficient in the WM centreline (Figure 3a), which facilitated the formation of martensite transformation in the WM using the STS 308L filler.

Tensile and Microstructural Behaviour of the Welds
Metals 2020, 10, x FOR PEER REVIEW 7 of 10 308L filler produced more α′ and ε martensites (Ms ~0.24) than the HEA filler (Ms ~0.15). The tendency of fraction of martensite transformation was opposite to that of the CSL boundary, that is, a smaller CSL boundary and more martensite for the weld produced using the STS 308L filler. These results were associated with the macro-segregation observed near the WM centreline (Figure 2a). The FCC stabilising elements of Ni, Mn, and Co were deficient in the WM centreline (Figure 3a), which facilitated the formation of martensite transformation in the WM using the STS 308L filler.  Figure 9 shows the fracture morphology of the welds obtained using the HEA and STS 308L fillers tested at 298 K. The fracture surface of all welds primarily comprised dimple morphology. However, the fracture morphology of quasi-cleavage (QC) was observed more frequently near the central part of the fracture surface of the weld using the STS 308L filler than that using the HEA filler. The large fraction of QC fracture and the deterioration of TS and Ef were associated with considerable macro-segregation and martensite transformation produced in the WM using STS 308L filler. Conclusively, the commercialised STS 308L filler was not sufficient to produce CoCrFeMnNi HEA welds that are stronger than the rolled HEA BM. The authors previously reported the successful use of the STS 308L filler for the cast HEA weld and produced a stronger WM than the BM [26]. Future studies are required to enhance grain refinement and dispersion strengthening to enable the use of the HEA for structural applications.   Figure 9 shows the fracture morphology of the welds obtained using the HEA and STS 308L fillers tested at 298 K. The fracture surface of all welds primarily comprised dimple morphology. However, the fracture morphology of quasi-cleavage (QC) was observed more frequently near the central part of the fracture surface of the weld using the STS 308L filler than that using the HEA filler. The large fraction of QC fracture and the deterioration of TS and E f were associated with considerable macro-segregation and martensite transformation produced in the WM using STS 308L filler. Conclusively, the commercialised STS 308L filler was not sufficient to produce CoCrFeMnNi HEA welds that are stronger than the rolled HEA BM. The authors previously reported the successful use of the STS 308L filler for the cast HEA weld and produced a stronger WM than the BM [26]. Future studies are required to enhance grain refinement and dispersion strengthening to enable the use of the HEA for structural applications.
Metals 2020, 10, x FOR PEER REVIEW 7 of 10 308L filler produced more α′ and ε martensites (Ms ~0.24) than the HEA filler (Ms ~0.15). The tendency of fraction of martensite transformation was opposite to that of the CSL boundary, that is, a smaller CSL boundary and more martensite for the weld produced using the STS 308L filler. These results were associated with the macro-segregation observed near the WM centreline (Figure 2a). The FCC stabilising elements of Ni, Mn, and Co were deficient in the WM centreline (Figure 3a), which facilitated the formation of martensite transformation in the WM using the STS 308L filler.  Figure 9 shows the fracture morphology of the welds obtained using the HEA and STS 308L fillers tested at 298 K. The fracture surface of all welds primarily comprised dimple morphology. However, the fracture morphology of quasi-cleavage (QC) was observed more frequently near the central part of the fracture surface of the weld using the STS 308L filler than that using the HEA filler. The large fraction of QC fracture and the deterioration of TS and Ef were associated with considerable macro-segregation and martensite transformation produced in the WM using STS 308L filler. Conclusively, the commercialised STS 308L filler was not sufficient to produce CoCrFeMnNi HEA welds that are stronger than the rolled HEA BM. The authors previously reported the successful use of the STS 308L filler for the cast HEA weld and produced a stronger WM than the BM [26]. Future studies are required to enhance grain refinement and dispersion strengthening to enable the use of the HEA for structural applications.

Conclusions
GTA weldability of the rolled CoCrFeMnNi HEA using STS 308L and HEA fillers was investigated. Microstructures and mechanical properties of the GTA welds were examined at room temperature, and the conclusions of this study are as follows: (1) No macro-defects such as internal pores or cracks were observed for any of the GTA welds.
However, the macro-segregation region of the Fe component and deficient regions of Ni, Mn, and Co components were formed in the centreline of the weld using an STS 308L filler. (2) For WMs produced using different fillers (HEA and STS 308L), an FCC solid solution phase was observed. The weld using the STS 308L filler had no BCC phase because of the dilution of the stabilising element of FCC introduced from the HEA BM. Furthermore, the columnar grains exhibited unidirectional growth from the fusion line in the WM using the STS 308L filler than those using the HEA filler. (3) The main reason for the low hardness of the WM was that the grain size of WM was approximately 70 times larger than that of the rolled HEA BM regardless of the filler metals. (4) The tensile properties of all welds were worse than those of the rolled HEA BM, and the tensile fracture of all welds occurred near the centreline in the WMs. Furthermore, the tensile properties of the weld using the STS 308L filler deteriorated more than those of the weld using the HEA filler. This was associated with the macro-segregation and severe martensite transformation formed in the centreline of WM. Therefore, to enhance the weldability of the rolled HEA, it is necessary to prevent the formation of macro-segregation and enhance grain refinement in the WM of GTA welds.