Application of materials with site-specific properties has been an field of active research due to the increasing demand of engineering structures that depend on a monolithic component with discrete properties [1]. When an engineering structure requires specialized properties at one specific location, the ability to manufacture the component with site-specific properties, in particular for dissimilar metal structures, could eliminate the need for the entire component to be made from advanced and costly material [2].

The joining of dissimilar metal components, including steels, aluminium alloys, magnesium alloys, titanium alloys, nickel-based superalloys and copper alloys, is frequently required in industrial sectors, such as aviation, aerospace, shipbuilding and automobile [[3], [4], [5]]. High quality dissimilar metal combinations by conventional welding or joining methods is a challenging task due to the potential formation of high levels of brittle intermetallic compounds (IMCs) during solidification [6]. To overcome this major impediment for obtaining a sound joint between dissimilar materials, the level of fusion heat input must be restricted by using solid-state bonding techniques. In recent years, various low temperature manufacturing process, such as friction welding, diffusion bonding as well as ultrasonic joining, have been developed to join dissimilar metals [7]. Nonetheless, significant deficiencies in these techniques, such as friction welding and ultrasonic welding lack versatility and diffusion bonding requires a long joining cycle time, then limit their application.

Additive manufacturing (AM), where materials are introduced to parts sequentially, has attracted much attention over the past ten years due to its ability to capture unrivalled design freedom and short lead times. Thanks to their uniqueness, AM techniques have offered exciting opportunities for production of structural components that fulfills various specific requirements or functions. AM offers the ability to tailor the composition at the interface to avoid the formation of detrimental intermetallics in the joining of dissimilar metals [8]. Furthermore, the joining of two materials with very different mechanical properties, can lead to poor interfacial bonding or high thermal residual stress due to different expansion, can lead to failure of the component [9]. Functionally graded materials (FGM) have been developed to overcome such problems by providing a gradual change in mechanical properties across the interface [10]. The benefits for improving interface bonding between dissimilar materials using FGM can be readily transferred to AM components, as a gradient in microstructure and mechanical properties can be achieved by manipulating the ratio of the feed stock. Hence, the use of additive manufacturing to produce materials is a promising technique to improve the bonding strength in the dissimilar metals.

Steel-nickel alloys are prospective alloys for nuclear energy structural materials, due to their good radiation damage resistance, high-temperature creep strength, outstanding mechanical properties and satisfactory corrosion resistance performance [11,12]. However, it is difficult to obtain high quality steel-nickel joints due to their different physical and chemical properties [13]. Abe and Sasahara [14] reported that a good combination of mechanical properties at the interface could be achieved by using wire arc additive manufacturing (WAAM) to join stainless steel and a Ni-based alloy.

In this work, a crack-free steel-nickel structural part with outstanding mechanical strength was successfully produced via WAAM using a build strategy by alternative deposition of overlapped weld beads. The microstructures, and related strengthening mechanisms, were explored and discussed. The present study provides a new route to produce steel-nickel bimetallic components with excellent quality performance for their structural application.

A WAAM system utilized in this study is based on gas tungsten arc welding (GTAW), as shown in Fig. 1. It consists of a 200 A-rate GTAW power supply (Master Tig MLS-Kemppi Co.), “cold” wire feeder (CK, Wordwild Co.), water cooling unit and travel mechanism (machine tool). The commercial ER70S-6 steel and ERNi-1 (Ni-3.5 wt%Ti) nickel welding wires with a diameter of 1.2 mm were employed as feedstock materials. A straight wall structure, with one half steel and another half nickel, was built using an alternative layered, overlapped bead strategy from twin wire feeding, as the build path shown in Fig. 1(c). The process parameters for deposition are given in Table 1. The angle between wire feeder and substrate was set to 60°, which was the optimum angle to produce stable melt pool during double wire feeding in this study. Pure argon (99.995% purity) was used as the shielding gas for GTAW welding torch. The final steel-nickel structural wall was fifteen layers, approximately 150 mm in length, 18 mm in width and 20 mm in height.

The metallographic specimens were extracted from the mid-point of the as-built wall hight, where properties were expected to be consistent. The cross-section samples were hot mounted, ground and polished according to standard preparation procedures. Etching bimetallic materials required a two-step etching technique; samples were firstly etched in 2% Nital reagent for steel region examination and then etched in a mixed solution of 5 mg oxalic acid and 95 mL HCl by using an electrolytic approach (with 6 V DC), for nickel region inspection. Microstructural observation was performed using a Leica DMR optical microscope (OM) and qualitative microanalysis of the interfacial area was conducted using a JEOL JSM-6490LA scanning electron microscope (SEM), with an energy dispersive spectroscopy (EDS), at 15 kV. In-situ observation of solidification process for the interface region was performed on a high temperature confocal microscope (HTCM). The microhardness was investigated using a Vickers automatic hardness tester, with a test load of 500 g and dwell time of 15 s. Tensile tests were conducted on an Instron universal testing machine, at a constant crosshead displacement rate of 0.25 mm/min, at room temperature. The fracture surfaces of the tensile specimens were analyzed using the same SEM as stated above.

A cross-section sample of the interface between the Fe and Ni weld layers was studied by OM and different examined regions are given in Fig. 2. For the nickel side, the primary γ matrix phase was observed, displaying a hypoeutectic dendritic structure. A significant amount of primary carbide formed along dendritic regions. It is also found that the slightly coarse grains occur in the top regions (Fig. 2(b)) compared to bottom regions, which may be caused by the less thermal cycles. The as-deposited steel side has a morphology dominated by coarse columnar grains with a fine cellular microstructure of α-Fe, especially in first few layers (Fig. 2(h)). Carbide particles were predominately found at the grain boundaries as well. It is apparent that the grain growth direction in both steel and nickel regions was at an angle of around 60° with the substrate, which coincided with the maximum temperature gradient at the solidification front. Due to the woven path pattern used during deposition, distinctive interface that displays an interlocking microstructure was achieved (Fig. 2(e)), implying good bonding between the two materials.

SEM and EDS maps for interface regions are presented in Fig. 3, showing an interlocking interface between the Fe and Ni due to the overlapping weld layer strategy. The line analysis revealed that intermixing between the Fe and Ni occurred across the interface over a relatively wide region through atomic diffusion and mixing during solidification. The Fe and Ni phase diagram (Fig. 3(b)) indicates that Fe and Ni are completely soluble in each other at high temperatures, so substantial intermixing can be expected, leading to solid solution strengthening. The Ti line analysis showed some enrichment on the Ni side of the interface and two large spikes indicate that Ti-rich particles or intermetallic had formed. These enriched Ti could lead to severe crystal lattice distortion of nickel and then produce large resistance to prevent dislocation, giving enhanced strength [15].

To better understand this solidification process, a detailed laser scanning confocal microscopy investigation was conducted. The related results are shown in Fig. 4. It is found that pectinate nickel regions melted first due to its comparatively lower melting point (Fig. 4(b)), then dissolved into Fe-riched areas (Fig. 4(c) and (d)) to form solid solutions. Referring to the equilibrium phases of Fe-Ni system (Fig. 3(b)), γFe-Ni solid solutions will form. Diffusion Ni into Fe rich areas may result in δ-Fe to precipitate at high temperatures and α-Fe form at temperatures below 912 °C, as a bcc structure.

EDS data were taken at specific locations of the interface, in the same region of the confocal experiments, as shown in Fig. 5. Substantial amounts of Fe dissolved into the Ni rich side (D, E, F and G). Point C indicated that close to 50/50 mixing can occur. The enhanced mobility of Fe atoms into Ni regions may came from the activation energy of arc source when steel wire deposited. Here, it needs to be mentioned that the FeNi₃ intermetallic compounds were not acceptable in transition regions as their formation requires a first-order order-disorder transformation below 517 °C, much lower than molten pool temperature during solidification [16].

Fig. 6 shows the hardness distribution from the bottom to the top of the samples for the Ni side, Fe side and interface. For nickel side, hardness was relatively stable along the buildup direction, until the very top, were hardness dropped suddenly. For the steel side, a large increase in hardness (around 30%) was recorded with the increase in height, which may be ascribed to the refined grain size and increased carbide precipitation. The hardness at the interface also showed an increase in hardness values at a similar height position, although hardness fluctuated significantly. It is noted that the microhardness dropped at very top of wall structures, which was mainly due to the microstructure evolution caused by less thermal cycle to last layer.

Fig. 7 displays the tensile properties of as-deposited material, extracted along the travel direction. At room temperature, the build volume exhibits an average ultimate tensile strength (UTS) of 634 MPa, which significantly exceeds that of ISO9001 standard of 537 MPa for ER70 steel and 455 MPa for ERNi-1 nickel feedstock materials. Therefore, the WAAM technique with woven deposition path used for this study led to higher strength at the interface than the individual strengths for each component. For better comparison, Table 2 provides a summary of the mechanical properties for AM-produced steel-nickel part in existing literature.

From stress‒strain plots, two fracture locations are observed in the plastic region for each sample, indicating that fracture instabilities occurred before final fracture. Specifically, due to the comparative soft, the first fracture failure tended to occur at the nickel side, which then led the fracture mode change from stretching to in-plan shearing during tensile test. The second fracture point indicated the rupture of steel side. For steel-nickel component, this fracture behaviour often brings low toughness to the interface regions [17].

The fractographs for the tensile specimen at three selected regions are presented in Fig. 8. For the steel side, the fracture surface was cleavage, with ductile micro-voids at the fracture face, indicating a small degree of toughness. Parallel cracks were observable on this side, related to the dendritic orientation shown in the microstructure. A large number of enlarged ductility-dip-cracking cavities have also been found, identifying as initiation sites for these cracks [18]. The nickel region responded differently to rupture, intergranular laves phases were present across the fracture surface, due to the failure along the dendrite grain boundaries, where a stress concentration exists and a crack initiates by in-plan shearing stress [19]. This fracture mode belongs to brittle intercrystalline fracture, meaning a low ductility to the as-deposited part.

According to Fig. 2, the steel side showed fine, cellular ferrite grains with carbides surrounding the grain boundaries, while the nickel side, displayed dendritic grains with some carbides between grains. Fig. 2 also revealed that the pronounced columnar grain growth, in steel and nickel parts, showed preferred orientation in line with the maximum thermal gradient, which was determined by both the melt flow state of current layers and heat accumulation effects from the previous deposit. As the overlap of layers resulted in the variation of localized thermal distribution in the molten pool, the maximum temperature gradient was altered to along the feedstock delivery direction (60°). In addition, it is found that these dominant grains in deposit could grow without interruption, even a short dwell time between layers existed. This is mainly ascribing to the combined effects of thermal accumulation and minimal grain nucleation that can provide conditions for the development of columnar grains [20].

Regarding to the steel part, the grains appeared to have a smaller size in the upper regions compared to the lower regions, which would lead to lower strength in the lower region (Fig. 7), in accordance with the Hall-Petch relationship. A significant increase in microhardness can be found at top regions of steel side, mainly addressing the combined effects of refined grains and carbide precipitation. Due to slow cooling conditions in the built-up wall, away from substrate, a small temperature gradient and low solidification rate would be expected, leading to the formation of large carbides. Moreover, due to such thermal effects, heat accumulation allowed increased elements evolution in grain bondries, particularly carbides precipitation, which provides starting locations for nucleation events, contributing to a refined/cellular α-Fe morpoholgy at upper steel regions [21], as shown in Fig. 2(f). The microhardness maintained a relatively constant level from the bottom to top for the nickel side, probably due to the limited alloying additions in the Ni wire.

This paper proposes that it is possible to produce steel-nickel dissimilar metal parts with enhanced mechanical properties, when being fabricated by WAAM, using alternative layers of overlap bead strategy. Fig. 7 shows that the interface region had higher strength compared to the individual feed stock materials, and hence, a strengthening mechanism at the interface occurred. The good bonding strength at the interface occurred for two reasons.

Firstly, the interlocking interface between the Fe and Ni was generated by using the overlapping alternating twin wire deposition strategy, eliminating the formation of a weak fusion boundary. This strategy prevented premature failure along a weak interface layer, which can occur when joining dissimilar metals. Observed intertwining microstructure in as-received steel-nickel part is similar to the so-called ‘Kelvin-Helmholtz instability (KHI)’ phenomenon, in which two viscous materials move in parallel at different velocities to generate interfacial perturbation [22]. Uunder such reinforced microstructural configuration, the movement of two dissimilar materials was restrained when they are under tensile stress, producing synergetic strengthening and then improving tensile strength of monolithic part.

Secondly, solid solution strengthening (SSH) occurred at interface and this would play an important role in mechanical strengthening. During deposition, Fe and Ni atoms were diffused and dissolved between the two materials, leading to a wide range of compositions to form across the interface and significant solid solution strengthening. Due to the variation of atomic diameter, the mutual dissolution of Fe and Ni atoms (single FCC solid solution), was able to produce localized elastic strain fields that interacted with dislocations movement, hence, giving rise to SSH in the diffusion zones [23]. Electronic interactions between the Fe and Ni atoms may also contribute to the solid solution hardening due to electron drag effect [24].

The alternating stacking sequence gave rise to a concentration gradient between the Ni and Fe layers. This occurred due to both the elemental mixing dissolution of the previous layer with the molten pool of the next layer and a diffusion process of the Fe and Ni atoms. The thermal gradient generated in the welding process provided the driving force for the Fe/Ni diffusion (Fig. 5), between the alternating layers, which is explained by the Fick’s first law [25]:

where J_x refers to the number of diffusing atoms per unit time, D is the diffusion coefficient, c is the concentration, x is the diffusion distance. According to this principle, the diffusion of foreign atoms within the previous layer will be maintained until there is no thermal gradient across the diffusion regions. Hence, diffusion of Fe into the Ni part and vice versa, would contribute to the concentration gradient and improvement in bonding strength at the interface. For this reason, to produce a structural component with multiple metal components using WAAM, the diffusion of alloying elements in combined zones should be carefully controlled, to obtain optimal performance of production. Moreover, the carbide precipitation at grain boundaries may also play a critical role in the improvement of mechanical strength. These carbides can naturally impede the movement of grain boundaries, thus leading to an enhanced strength.

From Fig. 7, a variation in mechanical performance occurred when the layers were built up. This may be related to the inhomogeneous microstructures between the base layers and the middle to top layers of the buildup see Fig. 2. For the base layers, heat extraction is high due to the base acting as a heat sink [26]. In the middle layers, heat transfer was controlled by radiation and convection to the surrounding environment, which is less effective. The variation in the heat dissipation resulted in changes to the microstructure during deposition, generating inhomogeneous material properties. In addition, uneven interface from alternate thermal input may also one of the reasons for producing ragged material properties.

The woven build strategy using WAAM introduced in this article can produce entire steel-nickel structural components with enhanced interfacial strength. Through tailoring alternate deposition of Fe and Ni, a diffusion gradient occurred in transition zone, generating solid solution strengthening, which allowed the bonding strength to improve. However, although enhanced mechanical strength was realized, low elongation still occurred in the produced part. Hence, ancillary interpass or post processing, such as interpass rolling or post-heat treatment, may be required to achieve improved ductility.

Strategic advancements in build process imparted by multiple wires with alternative deposition of overlap dissimilar layer could potentially be used to design and produce monolithic metal components with site-specific properties, where the local material features could be developed for the local conditions. This is particularly beneficial to structural applications. The design and manufacturing process presented, could be adapted for joining other dissimilar alloys, such as aluminium-copper, and magnesium-aluminium, to enhance mechanical properties. As WAAM involves a complex thermal history, the primary process designed in this study may provide the opportunity to overcome the material processing challenge in terms of diffusion control of dissimilar feed metals.

In summary, this work successfully achieved a metallurgically sound bond for a steel-nickel dissimilar metal component using WAAM with an interweaving build path. The tensile-strength at the interface was greater than the sum of the parts, which were an ER70S-6 steel wire and ERNi-1 nickel wire. The underlying reason for the observed mechanical response was due to the interweaving strategy that produced an interlocking dissimilar microstructure, which eliminated a potentially weak fusion boundary at the interface. Furthermore, the intermixing of the Fe and Ni produced a gradient of Fe-Ni solid solution strengthening, which increased the strength of the interface. A relative fine, cellular microstructure was observed for both the Fe and Ni side and dendrite orientation was aligned with the maximum hear flux, along the feedstock delivery direction, 60° to the substrate. Fracture orientation was aligned with the dendrite orientation and failure occurred first at the nickel edge of the sample, owing to its comparatively softer microstructure. The as-fabricated steel part was sensitive to thermal history, due to heat built-up as the wall height increased, the change in cooling rate allowed increased carbide precipitation, leading to improved hardness along the build direction. Conventional welding or joining methods are available to manufacture dissimilar metal structures, however the WAAM process provided in this study, offers a promising pathway to produce dissimilar alloy components, where greater control of site-specific properties and improved interface strength can be achieved.

The present research was carried out at the Welding Engineering Research Group, University of Wollongong, and it was supported by China Scholarship Council (No.201506680056) and the National Natural Science Foundation of China (No.51805085). The authors would express sincere acknowledgement to Australian Institute for Innovative Materials (AIIM) center for the use of their electrical microscopy facilities.