Intermetallic Phase Layers in Cold Metal Transfer Aluminium-Steel Welds with an Al–Si–Mn Filler Alloy

Abstract

In welding of aluminium (Al) alloys to steels, a major challenge is excessive growth of brittle intermetallic phases along the bonded Al-steel interfaces. The formation and growth of these phases are influenced by the heat input and the alloying elements present. This work focuses on the phases formed between a low alloyed steel and an Al–Si–Mn alloy that was used as the filler wire in a cold metal transfer joint. During lap shear testing of the joint, fracture ran through the melted Al, and the joint reached a strength of 174 ± 21 MPa. Scanning and transmission electron microscopy showed that the formed ∼2.5 µm thick intermetallic phase layer consisted of polyhedral α_c-Al–(Fe,Mn)–Si, elongated or rounded θ-Fe₄Al₁₃ and near equiaxed η-Fe₂Al₅ grains. Electron backscatter diffraction was used to study the crystal orientations of the formed phases. Altogether this work aims to contribute to better understanding of the formation and growth of intermetallic phases in Al-steel welds where the Al alloy contains Si and Mn.

1. Introduction

Cold metal transfer (CMT) is a welding method in which a filler material (FM) wire is heated by an arc, and molten FM droplets are deposited in the seam between the materials to be joined, after which the weld is allowed to consolidate. The method is based on traditional metal inert gas welding but has been modified to include integrated FM wire feeding and back-drawing.^1,2) This allows welding with lower heat input and less spatter.^3–5) Low heat input is crucial for welding of aluminium (Al) alloys to steels, since Al alloys are sensitive to thermo-mechanical treatment and develop heat affected zones (HAZs) during welding. In addition, brittle intermetallic phases (IMPs) form along the Al-steel interfaces due to the limited solid solubility of iron (Fe) in Al,⁶⁾ and the growth of these phases is typically diffusion-limited and accelerates at elevated temperatures.⁷⁾ Mainly due to the relatively low heat input enabled by CMT, CMT can be used to fabricate sound Al-steel joints.^3,8,9)

The IMP layer that forms along the Al-steel interface during welding influences the mechanical properties of the joint. In some cases, the IMP layer has led to brittle fracture occurring along the interface during tensile or shear testing, and this is more likely to occur as the IMP layer thickness increases.¹⁰⁾ To understand the influence of the IMP layer on the mechanical properties of the joint, both the morphology, chemical composition and crystal structure of the formed IMPs must be assessed. Further, understanding of the crystal orientations of the IMPs provide insight into their nucleation and growth characteristics. Previously, orientation mapping has been performed for joints between low alloyed Al and steel, where the phases θ-Fe₄Al₁₃ and η-Fe₂Al₅ form along the interface.¹¹⁾ Several studies have focused on the growth of the η phase, and it has been found that η grains nucleate as fine equiaxed grains with random texture¹²⁾ that develop into elongated grains preferentially oriented with the c-direction normal to the Al-steel interface.^12–16) The texture has been explained based on the low occupation sites forming channels along the c-direction in the η crystal lattice.¹³⁾

The composition of the parent alloys has a significant effect on the IMP layer. Several studies have concluded that addition of Si to the Al FM leads to significant reduction in the IMP growth rate,^9,17,18) which is highly beneficial. Some studies have reported that the combined action of Mn and Si additions to the Al FM gave the most promising results.^9,18) With an Al alloy containing Si and Mn, ternary and tertiary IMPs consisting of Al–Fe(–Si)(–Mn) have been reported to form, together with the commonly reported θ and η phases. For instance, in some cases the cubic α_c-Al–(Fe,Mn)–Si phase has been reported to grow into Al,^9,19,20) while the hexagonal α_h-Al–Fe–Si phase has been reported in other cases.^17,20)

This work presents characterization of an Al-steel weld fabricated by CMT using an Al–Si–Mn alloy as FM. The mechanical properties of the joint were assessed by lap shear testing and hardness measurements, while the interface microstructure was investigated by optical microscopy and scanning and transmission electron microscopy (SEM and TEM). In particular, TEM was used to assess the morphology, chemical composition, and crystal structure of the formed interfacial IMP layer, while electron backscatter diffraction (EBSD) was used to study the crystal orientations of grains in the IMP layer. In total this work aims to contribute to better understanding of the IMP layer characteristics in CMT Al-steel joints fabricated with an Al FM containing Si and Mn.

2. Experimental Procedure

2.1 Materials

A 5754 and a 4020 (Al–Si3–Mn1) Al alloy were used as base material (BM) and FM respectively, and their nominal compositions are listed in Table 1. The ultimate tensile strength (UTS) of the Al BM plate was 254 MPa. The steel BM was a plate of hot dip galvanized steel DX51D Z275 covered by a ∼19 µm thick Zn coating. The composition of the steel plate was determined by optical emission spectroscopy, and the result is listed in Table 2. The Al and steel BM plates had dimensions 2 × 70 × 350 mm and 2 × 160 × 350 mm, respectively, while the Al FM wire had a diameter of 1.2 mm.

Table 1 Nominal composition of the Al BM plate and Al FM in mass%.

Table 2 Composition (in mass%) of the DX51D steel BM plate determined by optical emission spectroscopy.

2.2 Cold metal transfer

Prior to welding, the BM plates were degreased in acetone, and the Al BM plate was ground with a steel brush. The Al BM was placed on top of the steel BM in a lap configuration and fastened by clamps or tacks, to minimize the vertical gap. Five CMT lap joints were produced, and selected welding parameters are presented in Table 3. For all joints produced, the arc was positioned at the edge of the Al BM plate so that the offset was zero, the welding speed was 9.7 mm/s, and argon shielding gas was supplied at a constant flow rate of 20 l/min. The CMT welding system was of type Fronius TransPuls Synergic 3200 Pipe HE.

Table 3 Selected welding parameters for joints A–E, including current, voltage, wire feed rate, calculated total heat input, and the vertical gap between the Al and the steel plate measured after welding.

2.3 Optical microscopy

Optical microscopy was performed on polished cross-sectional samples that had been etched in a 2% nital solution to reveal the microstructure of the steel.

2.4 Mechanical testing

Joint A was subjected to lap shear testing and hardness measurements. Three specimens were machined to obtain dog-bone shaped specimens with the same dimensions as reported in Ref. 21), with a transverse length of 12 mm and a parallel length of 62 mm. The testing was performed using a Zwick Roell Z030 testing machine with the cross-head speed set to 1 mm/min. The lap shear strength was calculated from the maximum force divided by the measured cross-sectional area of the Al of 24.96 mm² (12.48 × 2.00 mm).

Hardness measurements were performed on a polished cross-section of joint A using a Zwick Roell ZHV 30A hardness tester with an applied load of 0.2 kg.

2.5 Transmission electron microscopy

TEM lamellae were fabricated by focused ion beam (FIB) lift-out of sample A using a FEI Helios G2 and a FEI Helios G4 FIB instrument. The thinning procedure comprised coarse milling at 30 kV and fine polishing at 5 kV.

TEM characterization was performed using a JEOL-JEM2100F and a JEOL-ARM200F, both operated at 200 kV. The latter was equipped with spherical aberration corrects in both the probe and the image forming optics. Selected area electron diffraction (SAED) patterns were acquired, and X-ray energy dispersive spectroscopy (EDS) mapping was performed. The EDS data were visualized and analyzed using the python package hyperspy.²²⁾ The analysis included spectrum-by-spectrum model fitting using a model consisting of one Gaussian per identified X-ray line and a sixth order polynomial for the background. To create maps showing the relative composition, quantification was performed based on the Cliff-Lorimer method by considering the model fit for the Kα-lines of each element and by using calculated k-factors. Sum spectra from single phase regions were quantified using the same approach, to yield estimated compositions of individual phase layers.

2.6 Scanning electron microscopy

SEM characterization was performed on polished cross-sections with a Hitachi SU6600 using a beam energy of 20 keV. Three EBSD datasets (I–III) were acquired using a NORDIF UF-1100 EBSD detector with the sample tilted 70° from the horizontal. The step size was 0.1 µm, and the pattern size was 240 × 240 pixels with 8-bit pixel depth. The pattern centres per dataset were determined using Hough indexing in the Python package PyEBSDIndex.²³⁾

The EBSD patterns were pre-processed using the Python package kikuchipy v0.5.²⁴⁾ The signal-to-noise ratio of the raw experimental patterns was increased prior to indexing by subtracting a static background, followed by subtraction of a dynamic (per pattern) background. Finally, each pattern was averaged with its eight nearest neighbours using a Gaussian kernel with a standard deviation of 1.

To index the EBSD patterns, dictionary indexing^25,26) was performed as implemented in kikuchipy. Dynamical EBSD master patterns were simulated using EMsoft v5.0²⁷⁾ for Al, ferrite (α-Fe), and the IMPs identified based on the TEM characterization. The parameters used in the simulations are listed in the Supplementary Information (SI) found on Zenodo,³⁴⁾ Section S1. The simulated patterns were projected from master patterns created with an energy of 20 keV. The respective orientation spaces of the phases were uniformly sampled using cubochoric sampling,²⁸⁾ as implemented in the Python package orix v0.9,²⁹⁾ with an average misorientation angle of 1.4°. Every pre-processed experimental pattern was compared to the dictionary via the normalised cross-correlation (NCC) coefficient.³⁰⁾ The orientation of the simulated pattern with the highest NCC score was chosen as the initial solution per pattern. Each initial orientation was then refined by varying the orientation slightly to optimise the NCC score while keeping the pattern centre fixed, as implemented in kikuchipy. For this, the Nelder-Mead optimisation algorithm was used, as implemented in the Python package SciPy v1.7.³¹⁾ Refined orientations corresponding to NCC scores below 0.1 were removed prior to further analysis. Also, orientations were removed prior to further analysis if they corresponded to EBSD patterns with low total intensity due to significant sample topography, as these were observed to be incorrectly indexed. Figures showing the total intensities and histograms showing the NCC scores are shown in the SI Section S4 for each dataset. The final orientations were analyzed using the MATLAB toolbox MTEX v5.8.³²⁾

Whether misorientations across phase boundaries between Al and adjacent IMPs could be described by previously reported orientation relationships (ORs) was investigated. The misorientations were clustered using the hierarchical clustering method implemented in MTEX, which resulted in misorientation clusters corresponding to each of the major IMP grains adjacent to Al. The average misorientation of each cluster was then compared to the previously reported ORs listed in SI Section S2. Except for the specific clustering method used, the approach is similar to that detailed in a previous work.³³⁾

The SI and the EBSD datasets have been made available open access on Zenodo,³⁴⁾ while the code used for data analysis has been made available open access on GitHub.³⁵⁾

3. Results and Discussions

3.1 Optical microscopy

Micrographs showing the cross-sections of joints A–E are displayed in Figs. 1(a)–(e), respectively. Figures 1(f)–(j) show the interface microstructure in the middle of the bond line of joints A–E, respectively. All specimens showed a continuous IMP layer at the bonded interface. With increasing heat input, the bond line length, the IMP layer thickness, and the fraction of IMPs located within Al, all increased. Note that the specimens from joint D and E broke during specimen preparation and during welding, respectively. The fracture ran through the IMP layer, and it can be assumed that the reason for the interfacial fracture was the large thickness ($ \gtrsim 20$ µm) of the brittle IMP layer. In general, the best mechanical properties can be expected for the sample with the thinnest IMP layer,¹⁰⁾ and therefore, only joint A was investigated further.

Fig. 1

Optical microscopy images of the joint cross-sections, showing overviews over the weld areas (left column) and the microstructure in the middle of the bond line (right column). The samples are arranged from top to bottom after increasing heat input: (a) and (f) A, (b) and (g) B, (c) and (h) C, (d) and (i) D and (e) and (j) E.

3.2 Mechanical properties

Joint A was subjected to mechanical testing in the form of lap shear tests and hardness measurements, as shown in Fig. 2. The measured hardness profiles are shown Fig. 2(a), which shows that the HAZ extended 14 mm into the Al BM plate from the fusion line. The hardness in the Al FM was measured at four locations, giving values in the range of 59 to 76 HV_0.2, where the notation HV_0.2 indicates an applied load of 0.2 kg. This indicated that the Al BM-FM interface region was the softest zone with hardness values <60 HV_0.2. The force-displacement curves resulting from lap shear testing of three specimens from joint A are shown in Fig. 2(b). All specimens showed brittle fracture characteristics with no necking. The lap shear strengths were 158 MPa, 161 MPa and 203 MPa, which gave an average and standard deviation of 174 ± 21 MPa. Considering the UTS of the Al BM, the corresponding joint efficiencies were 62%, 63% and 80%, and in average 69 ± 8%. Images of the broken specimens are shown in Fig. 2(c), together with a schematic indicating the fracture modes. The fracture ran close to the interface between the HAZ in the Al and the fusion zone in the specimen reaching 203 MPa (green in Figs. 2(b) and (c)). In the two other specimens (red and blue in Figs. 2(b) and (c)), the fracture propagated further into the fusion zone toward the top. The spread in the joint efficiencies and the variation in fracture path could possibly be explained by differences in the distribution and sizes of pores. The formation of porosity was likely related to entrapment of Zn vapour following evaporation of the Zn layer on the steel BM.^21,36)

Fig. 2

Mechanical testing of joint A. (a) Hardness profiles measured from the fusion line and into the Al BM. The inset shows an optical microscopy image (same as in Fig. 1(a)), where the locations of the indents are shown schematically. (b) Force-displacement curves resulting from lap shear testing of three specimens. (c) Images of the tested specimens after fracture, where the frame colours correspond to the colours in (b). The bottom inset framed in black shows a schematic joint cross-section where the fracture path for the specimen marked in green is illustrated with a green line, and the fracture paths for the specimens marked with blue and red is illustrated in purple.

3.3 Scanning electron microscopy

Figure 3(a) shows an overview secondary electron (SE) SEM image of the Al-steel interface region, where the interfacial IMP layer can be seen together with IMP particles located within the Al FM. Note that some of the IMP particles within the Al FM were damaged during specimen preparation, resulting in topography. Figure 3(b) shows a SEM image of the continuous and ∼2.5 µm thick IMP layer. The steel-IMP interface was relatively straight, while the Al-IMP interface was irregular, and elongated IMP particles extended into the Al FM.

Fig. 3

SE SEM images of the Al-steel interface. (a) Overview image and (b) close-up of the IMP layer. Steel appears light grey, Al appears dark grey, and IMPs appear medium grey.

3.4 Transmission electron microscopy

The IMP layer was further characterized by TEM. Overview bright field (BF)-TEM and high angle annular dark field (HAADF)-STEM images from one lamella are shown in Figs. 4(a) and (b), respectively. Three distinct IMP layers were seen, and SAED was used to assess the crystal structures, as shown in Fig. 4(c). The first layer towards Al was discontinuous, up to 2 µm thick and consisted of polyhedral ∼1–2 µm long grains. SAED patterns from this layer could be indexed with respect to the cubic α_c-Al–(Fe,Mn)–Si phase (Al₁₅(Fe,M)₃Si₂, α-AlFeSi, a = 12.6 Å, $Im\bar{3}$ (204)³⁷⁾). The second middle layer was continuous and consisted of grains with a variety of shapes and sizes, up to ∼1 µm thick. In some regions the second layer had a relatively high density of smaller rounded grains. In other regions it had mainly larger elongated grains that were protruding into the first layer in an irregular manner. Some smaller Al or α_c regions were found in between the grains of the second layer in some areas. Thus, the boundary between the first and the second layer was far from straight and highly irregular when viewed at higher magnification. By SAED the second layer was identified to consist of the phase θ-Fe₄Al₁₃ (FeAl₃, a = 15.5, b = 8.1, c = 12.5 Å, β = 108°, C2/m (12)³⁸⁾). The third layer closest to the steel was discontinuous, up to ∼1 µm thick and consisted of near equiaxed grains. The SAED patterns from this layer were consistent with the phase η-Fe₂Al₅ (a = 7.7, b = 6.4, c = 4.2 Å, Cmcm (63)³⁹⁾). The phase sequence and grain morphologies identified here - coarse polyhedral α_c-Al–(Fe,Mn)–Si, elongated θ and equiaxed η - have been reported also in studies of other CMT joints where similar Al alloys were used as FMs.^9,20)

Fig. 4

TEM of the interfacial IMP layer. (a) BF-TEM image and (b) HAADF-STEM image of the same region. In (b), parts of the border between the η and the θ phase layers are highlighted with dashed white lines. (c) SAED patterns of the three IMPs identified; α_c-Al–(Fe,Mn)–Si, θ-Fe₄Al₁₃, and η-Fe₂Al₅.

The chemical composition of the interfacial IMP layer was assessed using EDS. A HAADF-STEM image and corresponding element maps based on STEM-EDS are shown in Fig. 5. It can be seen that the three IMPs contained Al, Fe, and small amounts of Si, and that the α_c in addition contained Mn. Zn and Mg had segregated to the boundary between the two α_c grains seen in Fig. 5, which may suggest that the α_c phase has low solid solubility of Zn. While it has been reported that several transition elements such as Mn and Cr may substitute for Fe and promote formation of the α_c phase,^40,41) it has also been reported that the presence of Zn does not promote formation of α_c.⁴¹⁾ The Zn originated from the Zn coating on the steel BM, which in general evaporates during CMT and leaves Zn-rich areas in the Al FM near the toe of the weld.^21,42) It has been reported that for liquid/solid joining of low-alloyed Al and low carbon steel, a Zn coating on the steel promotes formation of an even IMP layer, without affecting the type of phases that forms.⁴³⁾

Fig. 5

HAADF-STEM image of interfacial IMPs, and element maps based on STEM EDS that show the relative compositions of the major constituents Al, Fe, Si, Mn, Zn and Mg in at%.

Table 4 lists the relative composition of the major constituents obtained by quantification based on the maps shown in Fig. 5. It must be emphasised that Table 4 was based only on the single map displayed, and that local composition variations might occur. Further, the numbers listed were subjected to systematic errors due to the use of calculated k-factors and absorption of X-rays, which presumably resulted in an underestimation of the relative contents of Al and Si. The cubic α_c phase has been found to have a body-centred crystal structure with space group $Im\bar{3}$ for high Fe/Mn-ratios, e.g., ∼1–5,^44,45) and a simple cubic structure with space group $Pm\bar{3}$ for low Fe/Mn-ratios, e.g., $ \lesssim 1$.^44,45) Here the Fe/Mn-ratio was 3, which supported the choice of $Im\bar{3}$ as determined by SAED (see Fig. 4(c)). When it comes to the η phase, the Si content of 5 at% was higher than previously reported values of 4 at%⁴⁶⁾ and 2 at%.⁴⁷⁾

Table 4 Estimated relative content of major elements of the three phases α_c, θ and η in at% based on the EDS data used to create the element maps shown in Fig. 5.

3.5 Electron backscatter diffraction

The interfacial IMP layer was further characterized by EBSD to study the phase fractions, the orientations of the phases and potential orientation relationships. Figures 6(a) and (b) show SEM images of an interface area from which EBSD dataset I was acquired. The EBSD patterns were pre-processed before each individual pattern was compared to a dictionary of simulated patterns for the phases identified by TEM: Al, α_c, θ, η and α-Fe. The patterns were compared using the NCC score as metric, as explained in Section 2.6. Figure 6(c) shows a plot of the final NCC scores, and as expected, the scores decreased at (sub-)grain and phase boundaries. One point of high NCC score is marked for each phase in Fig. 6(c), and the experimental patterns from these locations are shown in the SI Section S3 together with the best matching simulated patterns. The phase map for dataset I is shown in Fig. 6(d). The large polyhedral grains located towards Al were identified to be α_c grains, while the middle layer was found to consist of the θ phase, and the layer towards the steel was identified to be a layer of the η phase, in agreement with the TEM results. From datasets I–III, the relative phase fractions of the three identified IMPs were: 61% α_c, 12% θ and 26% η.

Fig. 6

SEM images and EBSD results from dataset I. (a) and (b) SE SEM images of the interfacial IMP layer. Both (c) and (d) show EBSD results corresponding to the area highlighted in (b). (c) Normalised cross-correlation (NCC) scores between each experimental pattern and its best matching simulated pattern. (d) Phase map where the phase boundaries are marked in black.

Orientation maps coloured according to lattice vectors pointing east (right) and south (down) are shown in Figs. 7(a) and (b), respectively. Similar maps for the other two EBSD datasets are shown in the SI Section S4. The corresponding orientation colour keys per phase are shown in Fig. 7(c). One main orientation was found in the Al region, although the orientation map indicates that several sub-grains had formed. The θ and η regions showed numerous grains, while the steel and the α_c areas showed few and well-defined grains. A previous study on Hough indexing of EBSD patterns from the α_c phase reported that such patterns have a tendency of being frequently misindexed unless caution is taken, due to pseudosymmetry following the similarity of the α_c phase to an icosahedral quasicrystal.⁴⁸⁾ In this study, no apparent misindexing was observed, which correspond well with a previous study reporting that indexing methods relying on comparisons to dynamical simulations are less prone to this challenge.⁴⁹⁾

Fig. 7

Orientation mapping results. (a) and (b) Orientation maps corresponding to the area shown in Fig. 6, where the phase boundaries are coloured black. The orientation maps are coloured according to lattice vectors (a) pointing east (right) and (b) south (down). (c) Inverse pole figures displaying the orientation colour keys used for each phase in (a) and (b).

The two large polyhedral α_c grains displayed to the middle left and middle right of Figs. 7(a) and (b) showed two distinct orientations. The dendritic α_c grains found embedded within Al further from the interface also showed similar orientations to the polyhedral α_c grains they were adjacent to, which indicated that they were connected in three dimensions. It has been reported that α_c crystals forming during solidification of Al alloys can develop various morphologies with specific habit planes and growth directions.^50,51) A previous study found that α_c crystals developed polyhedral morphology near the Al surface and script in the Al bulk,⁵²⁾ similar to the morphologies seen here. Both morphologies have been found previously in Al alloys, and it was explained that the polyhedral crystals first grew as primary particles during solidification, while their morphologies changed to a convoluted eutectic structure once the Al dendrites started to form.⁵³⁾ The polyhedral shape of the interfacial α_c phase grains towards Al suggest a possible habit plane or OR.

Various ORs between Al and α_c have been reported previously for α_c particles forming as constituent phases or dispersoids in Al alloys containing Mn, Fe and Si.^54–59) Here the misorientations at the phase boundaries between α_c and Al were clustered, and the average misorientation of each cluster was compared to the previously reported ORs listed in the SI Section S2, as described in Section 2.6. From dataset I, the largest cluster had an average misorientation that deviated 4.8° from the OR [111]_Al || $[100]_{\alpha_{\text{c}}}$, $(20\bar{2})_{\text{Al}}$ || $(001)_{\alpha_{\text{c}}}$.^58,59) This cluster corresponded to the phase boundary between Al and the large polyhedral α_c grain to the middle right and between Al and the smaller script α_c grains to the top right of Figs. 6 and 7. In total, seven of the average cluster misorientations extracted from datasets I–III were within 8.2° of six previously reported ORs. The SI Section S2 lists, for each of the seven clusters, the average cluster misorientations in axis-angle notation, the previously reported OR, the misorientation angle to the OR, and the spread of the misorientation angles within each cluster.

The η phase grains depicted in Figs. 7(a) and (b) showed a tendency of being oriented with the ⟨001⟩ plane normal perpendicular to the interface. This agrees with previous reports of η grains growing preferentially along the ⟨001⟩ direction.^12–16) From the phase map shown in Fig. 6(d), it can be seen that the η phase layer was nearly continuous. One study showed that in a discontinuous η phase layer $ \lesssim 2$ µm thick, the η grains had random orientations, and that texture developed as the layer grew and became continuous.¹⁶⁾ Based on this, it could be expected that if the layer had been allowed to grow thicker, the texture would have become stronger.

4. Conclusion

Al-steel cold metal transfer joints were produced with an Al–Si3–Mn1 filler wire, and the joint produced with the lowest heat input (0.12 kJ/mm) was characterized in terms of mechanical properties and interface microstructure. Fracture occurred in the fusion zone during lap shear testing, and the strength reached was 174 ± 21 MPa. Scanning and transmission electron microscopy revealed that the formed ∼2.5 µm thick interfacial intermetallic phase layer consisted of polyhedral α_c-Al–(Fe,Mn)–Si grains, elongated or rounded θ-Fe₄Al₁₃ grains and near equiaxed η-Fe₂Al₅ grains. Electron backscatter diffraction indicated that the polyhedral α_c-Al–(Fe,Mn)–Si grains were connected in three dimensions to α_c grains with script morphology located further into Al. Also, some of the misorientation angles between Al and α_c grains corresponded to previously reported orientation relationships. The η grains showed a tendency to be preferentially oriented with the ⟨001⟩-direction normal to the interface.

Acknowledgments

Elin Pettersen is acknowledged for performing optical microscopy of joint cross-sections. The reported work is based on activities within the centre for research-based innovation SFI Manufacturing that is partially funded by the Research Council of Norway (237900). The support by the Research Council of Norway to NORTEM (197405) and NorFab (245963/F50) is acknowledged. EBSD indexing was performed on resources provided by the NTNU IDUN/EPIC computing cluster.⁶⁰⁾

REFERENCES

• 1)   J.  Bruckner: The Fabricator, August, (2003).
• 2)   J.  Bruckner: Weld. J. 84(6) (2005) 38–40.
• 3)   K.  Furukawa: Weld. Int. 20 (2006) 440–445. doi:10.1533/wint.2006.3598
• 4)   H.T.  Zhang,  J.C.  Feng,  P.  He,  B.B.  Zhang,  J.M.  Chen and  L.  Wang: Mater. Sci. Eng. A 499 (2009) 111–113. doi:10.1016/j.msea.2007.11.124
• 5)   J.  Feng,  H.  Zhang and  P.  He: Mater. Des. 30 (2009) 1850–1852. doi:10.1016/j.matdes.2008.07.015
• 6)   X.  Li,  A.  Scherf,  M.  Heilmaier and  F.  Stein: J. Phase Equilibria Diffus. 37 (2016) 162–173.
• 7)   S.  Kobayashi and  T.  Yakou: Mater. Sci. Eng. A 338 (2002) 44–53. doi:10.1016/S0921-5093(02)00053-9
• 8)   H.T.  Zhang,  J.C.  Feng,  P.  He and  H.  Hackl: Mater. Charact. 58 (2007) 588–592. doi:10.1016/j.matchar.2006.07.008
• 9)   L.A.  Jácome,  S.  Weber,  A.  Leitner,  E.  Arenholz,  J.  Bruckner,  H.  Hackl and  A.R.  Pyzalla: Adv. Eng. Mater. 11 (2009) 350–358. doi:10.1002/adem.200800319
• 10)   T.  Tanaka,  T.  Morishige and  T.  Hirata: Scr. Mater. 61 (2009) 756–759. doi:10.1016/j.scriptamat.2009.06.022
• 11)   A.  Bahadur and  O.N.  Mohanty: Mater. Trans. 32 (1991) 1053–1061. doi:10.2320/matertrans1989.32.1053
• 12)   N.  Chen,  M.  Wang,  H.P.  Wang,  Z.  Wan and  B.E.  Carlson: J. Manuf. Process. 34 (2018) 424–434. doi:10.1016/j.jmapro.2018.06.024
• 13)   T.  Heumann and  S.  Dittrich: Z. Metallk. 50 (1959) 617–625. doi:10.1515/ijmr-1959-501010
• 14)   W.J.  Cheng and  C.J.  Wang: Appl. Surf. Sci. 257 (2011) 4663–4668. doi:10.1016/j.apsusc.2010.12.118
• 15)   N.  Takata,  M.  Nishimoto,  S.  Kobayashi and  M.  Takeyama: Intermetallics 67 (2015) 1–11. doi:10.1016/j.intermet.2015.07.011
• 16)   Z.  Ding,  Q.  Hu,  W.  Lu,  X.  Ge,  S.  Cao,  S.  Sun,  T.  Yang,  M.  Xia and  J.  Li: Mater. Charact. 136 (2018) 157–164. doi:10.1016/j.matchar.2017.12.024
• 17)   J.L.  Song,  S.B.  Lin,  C.L.  Yang and  C.L.  Fan: J. Alloy. Compd. 488 (2009) 217–222. doi:10.1016/j.jallcom.2009.08.084
• 18)   Z.  Silvayeh,  R.  Vallant,  C.  Sommitsch,  B.  Götzinger,  W.  Karner and  M.  Hartmann: Metall. Mater. Trans. A 48 (2017) 5376–5386. doi:10.1007/s11661-017-4277-5
• 19)   L.  Agudo,  N.  Jank,  J.  Wagner,  S.  Weber,  C.  Schmaranzer,  E.  Arenholz,  J.  Bruckner,  H.  Hackl and  A.  Pyzalla: Mater. Technol. 79 (2008) 530–535. doi:10.1002/srin.200806162
• 20)   J.  Chen,  B.  Shalchi Amirkhiz,  R.  Zhang and  B.  Rona: Metall. Mater. Trans. A 51 (2020) 5198–5212. doi:10.1007/s11661-020-05917-8
• 21)   S.  Yang,  J.  Zhang,  J.  Lian and  Y.  Lei: Mater. Des. 49 (2013) 602–612. doi:10.1016/j.matdes.2013.01.045
• 22)  F. de la Peña et al.: hyperspy/hyperspy: HyperSpy v1.5.2. Zenodo, (2019) doi:10.5281/ZENODO.3396791.
• 23)  D. Rowenhorst and H.W. Ånes: USNavalResearchLaboratory/PyEBSDIndex: PyEBSDIndex, version 0.1rc2, (2021) https://github.com/USNavalResearchLaboratory/PyEBSDIndex.
• 24)  H.W. Ånes, L. Lervik, O. Natlandsmyr, T. Bergh, Z. Xu and E. Prestat: pyxem/kikuchipy: kikuchipy, version 0.5.7, (2022) doi:10.5281/zenodo.5833053.
• 25)   Y.H.  Chen,  S.U.  Park,  D.  Wei,  G.  Newstadt,  M.A.  Jackson,  J.P.  Simmons,  M.  De Graef and  A.O.  Hero: Microsc. Microanal. 21 (2015) 739–752. doi:10.1017/S1431927615000756
• 26)   M.  Jackson,  E.  Pascal and  M.  De Graef: Integr. Mater. Manuf. Innov. 8 (2019) 226–246. doi:10.1007/s40192-019-00137-4
• 27)   P.G.  Callahan and  M.  De Graef: Microsc. Microanal. 19 (2013) 1255–1265. doi:10.1017/S1431927613001840
• 28)   S.  Singh and  M.  De Graef: Model. Simul. Mat. Sci. Eng. 24 (2016) 085013. doi:10.1088/0965-0393/24/8/085013
• 29)  H.W. Ånes, B. Martineau, P. Crout, P. Harrison, D. Johnstone, N. Cautaerts and S. Høgås: pyxem/orix: orix, version 0.9.0.post0, (2022) doi:10.5281/zenodo.6560385.
• 30)  R.C. Gonzalez and R.E. Woods: Digital Image Processing, (4th. Pearson Education Limited, 2017).
• 31)   P.  Virtanen et al.: Nat. Methods 17 (2020) 261–272. doi:10.1038/s41592-019-0686-2
• 32)   F.  Bachmann,  R.  Hielscher and  H.  Schaeben: Ultramicroscopy 111 (2011) 1720–1733. doi:10.1016/j.ultramic.2011.08.002
• 33)   D.N.  Johnstone,  B.H.  Martineau,  P.  Crout,  P.A.  Midgley and  A.S.  Eggeman: J. Appl. Cryst. 53 (2020) 1293–1298. doi:10.1107/S1600576720011103
• 34)   T.  Bergh,  H.W.  Ånes and  S.  Wenner: Zenodo (2022). doi:10.5281/zenodo.6634354.
• 35)   H.W.  Ånes and  T.  Bergh: GitHub, (2022) https://github.com/hakonanes/ebsd-al-steel-joint-intermetallic-particles.
• 36)   M.R.  Ahsan,  Y.R.  Kim,  C.H.  Kim,  J.W.  Kim,  R.  Ashiri and  Y.D.  Park: Sci. Technol. Weld. Joining 21 (2016) 209–215. doi:10.1179/1362171815Y.0000000084
• 37)   M.  Cooper: Acta Crystallogr. 23 (1967) 1106–1107. doi:10.1107/S0365110X67004372
• 38)   J.  Grin,  U.  Burkhardt,  M.  Ellner and  K.  Peters: Z. Kristallogr. 209 (1994) 479–487. doi:10.1524/zkri.1994.209.6.479
• 39)   U.  Burkhardt,  Y.  Grin,  M.  Ellner and  K.  Peters: Acta Crystallogr. 50 (1994) 313–316. doi:10.1107/S0108768193013989
• 40)   D.  Munson: J. Inst. Met. 95 (1967) 217–219.
• 41)   C.Y.  Sun and  L.F.  Mondolfo: J. Inst. Met. 95 (1967) 384.
• 42)   H.  Zhang and  J.  Liu: Mater. Sci. Eng. A 528 (2011) 6179–6185. doi:10.1016/j.msea.2011.04.039
• 43)   H.  Springer,  A.  Szczepaniak and  D.  Raabe: Acta Mater. 96 (2015) 203–211. doi:10.1016/j.actamat.2015.06.028
• 44)   A.L.  Dons: Z. Metallk. 76 (1985) 151–153. doi:10.1515/ijmr-1985-760214
• 45)   P.  Donnadieu,  G.  Lapasset and  T.H.  Sanders: Philos. Mag. Lett. 70 (1994) 319–326. doi:10.1080/09500839408240993
• 46)   S.P.  Gupta: Mater. Charact. 49 (2002) 269–291. doi:10.1016/S1044-5803(03)00006-8
• 47)   M.C.J.  Marker,  B.  Skolyszewska-Kühberger,  H.S.  Effenberger,  C.  Schmetterer and  K.W.  Richter: Intermetallics 19 (2011) 1919–1929. doi:10.1016/j.intermet.2011.05.003
• 48)   H.  Becker and  A.  Leineweber: Mater. Charact. 141 (2018) 406–411. doi:10.1016/j.matchar.2018.05.013
• 49)   G.  Nolze,  A.  Winkelmann and  A.P.  Boyle: Ultramicroscopy 160 (2016) 146–154. doi:10.1016/j.ultramic.2015.10.010
• 50)   T.  Gao,  Y.  Wu,  C.  Li and  X.  Liu: Mater. Lett. 110 (2013) 191–194. doi:10.1016/j.matlet.2013.08.039
• 51)   A.  Bjurenstedt,  D.  Casari,  S.  Seifeddine,  R.H.  Mathiesen and  A.K.  Dahle: Acta Mater. 130 (2017) 1–9. doi:10.1016/j.actamat.2017.03.026
• 52)   H.  Becker,  T.  Bergh,  P.E.  Vullum,  A.  Leineweber and  Y.  Li: Materialia 5 (2019) 100198. doi:10.1016/j.mtla.2018.100198
• 53)   C.M.  Dinnis,  J.A.  Taylor and  A.K.  Dahle: Scr. Mater. 53 (2005) 955–958. doi:10.1016/j.scriptamat.2005.06.028
• 54)   P.  Liu and  G.L.  Dunlop: Acta Metall. 36 (1988) 1481–1489. doi:10.1016/0001-6160(88)90215-5
• 55)   Y.  Li and  L.  Arnberg: Acta Mater. 51 (2003) 3415–3428. doi:10.1016/S1359-6454(03)00160-5
• 56)   Y.  Li,  A.  Muggerud,  A.  Olsen and  T.  Furu: Acta Mater. 60 (2012) 1004–1014. doi:10.1016/j.actamat.2011.11.003
• 57)   A.M.F.  Muggerud,  Y.  Li and  R.  Holmestad: Philos. Mag. 94 (2014) 556–568. doi:10.1080/14786435.2013.857796
• 58)   A.M.F.  Muggerud,  Y.  Li,  R.  Holmestad and  S.J.  Andersen: Acta Crystallogr. B70 (2014) 888–896. doi:10.1107/S2052520614017880
• 59)   A.M.F.  Muggerud,  Y.J.  Li and  R.  Holmestad: Mater. Sci. Forum 794–796 (2014) 39–44. doi:10.4028/www.scientific.net/MSF.794-796.39
• 60)  M. Själander, M. Jahre, G. Tufte and N. Reissmann: arXiv:1912.05848.