Investigation of 300M ultra-high-strength steel deposited by wire-based gas metal arc additive manufacturing

300 M UHSS wire with a diameter of 1.2 mm was used as the feedstock, and S355 steel plates with 300 mm × 200 mm × 12 mm were used as the substrates. The chemical compositions of the two materials are listed in Table 1. Figure 1a schematically shows the experimental setup for the GMA-WAAM process. The entire setup was placed in an enclosure filled with pure argon gas with an oxygen level controlled below 800 ppm. An EWM power source integrated with a wire feeder was used to provide the electric arc and feed the wire. An arc monitor was connected to the power source to monitor the arc current and voltage. A CMOS camera was placed on the side of the melt pool to monitor the metal transfer and melt pool behavior. The substrate was fixed to a workbench by clamps, and the motion of the torch was controlled by a 6-axis ABB robot. To investigate the influence of torch shielding atmosphere compositions on the process, deposit morphology, microstructure, and mechanical performance of the GMA-WAAM 300 M UHSS structures, single-layer beads and multi-layer single-pass wall structures were deposited using five different commercial shielding gases, including pure Ar, 97.5% Ar + 2.5% CO₂, 92% Ar + 8% CO₂, 80% Ar + 20% CO₂, and 60% Ar + 38% He + 2% CO₂, with key process parameters listed in Table 2 and Table 3. The deposition experiments were performed in a flexible transparent enclosure purged with pure argon (oxygen content < 800 ppm). For single-pass wall deposition, the travel direction rotated 180° between each deposition pass and the interlayer temperature was monitored and controlled at just below 100 °C. A total of 16 layers were deposited for each single-pass wall structure.

Figure 2 shows the sample extraction positions, dimensions, and reference frame. Full-height samples with a thickness of 5 mm were extracted from the normal-building (ND-BD) plane for metallographic observation. After being ground and polished, they were etched in 2% Nital (nitric acid in ethanol) to reveal the microstructure. Large-scale stitched macrographs were obtained using a microscope with an automated stage scanning function. The microstructure was then analyzed using a scanning electron microscope (SEM) equipped with energy-dispersive spectrometry (EDS) detector with an accelerating voltage of 20 kV. Small crystallographic orientation maps were obtained using a field emission gun SEM equipped with an electron backscattered diffraction (EBSD) detector with an accelerating voltage of 15 kV, a beam current of 6.5 nA, and a step size of 0.3 µm. The big-area EBSD map was obtained by a Thermo Scientific Apreo SEM with an Oxford Instrument Symmetry-2 detector, using 20-kV accelerating voltage, 51-nA beam current, and 1.2-µm step size. EBSD data was then post-processed using Oxford Instruments’ AZtecCrystal software. Prior austenite grains (PAGs) were reconstructed using the Nishiyama-Wasserman orientation relationship after filling in non-indexed pixels. The microhardness transition along the central line of the metallographic samples was examined from the bottom (the area near the substrate) to the top of the wall structure. It was measured using a hardness tester with a load of 300 g and a spacing of 1 mm. Round tensile specimens with a gauge length of 24 mm in the spirit of ASTM E8 standard were extracted from the top and bulk regions in TD to evaluate the mechanical properties on an Instron 5500R machine at room temperature with a constant strain rate of 1 mm/min. Three sets of tensile coupons were taken from each position of each wall structure to ensure data reliability. The resulting fracture surfaces were checked by SEM.

Figure 3 shows the typical arc characteristics of deposition processes under different local shielding atmospheres. A pure Ar shielding gas is shown as a reference in Fig. 3a, in which undesirable arc instability occurred, which is consistent with previous research [23]. Clearly, the arc length (L) decreased with the increase of CO₂ addition in the shielding gas, as shown in Fig. 3b–d. When the CO₂ content was 2.5%, the arc length was long enough to maintain a spray metal transfer mode. When the CO₂ content was increased to 8%, the arc length significantly decreased, and a globular metal transfer took place. This is because CO₂ is an electronegative gas with a higher ionization potential compared to inert gases like Ar. More CO₂ introduced into the shielding gas mixture will increase the electrical conductivity of the arc plasma, leading to a more stable arc that is easier to maintain at a shorter length [24]. In addition, alongside adding 2–2.5% of CO₂ in the shielding gas mixture, replacing 38% of Ar with He led to a broader arc core, a slightly shorter arc length and maintained a spray transfer, as shown in Fig. 3e. This is because He has a higher thermal conductivity than Ar, which increases the distance over which heat can flow.

Concurrent with the change from spray transfer to globular transfer, increasing the CO₂ portion increased spatter, as shown by the adherence of metal particles to the walls in Fig. 4. More metal particles adhered to the side surface can be found with increasing CO₂ proportion, meaning the surface roughness became coarser due to the severer spatter, leading to worse surface quality. Hence, surface finish quality was degraded with increasing CO₂ portion. A comparatively improved surface quality was found when replacing 38% of Ar with He under ⁓2% CO₂ (Fig. 5a, d), benefiting from a steady deposition process with less spatter generation. This can be attributed to generating a more constricted arc plasma with higher energy density resulting from the increasing CO₂ content, thereby leading to more heat transferred [24].

The bead morphologies deposited under different local shielding are displayed in Fig. 5, where bead dimensions are indicated. By comparing Fig. 5a and b, it can be seen that adding CO₂ in pure Ar increased both the bead width and height. CO₂ possesses a higher thermal conductivity than Ar, especially at the comparatively low temperatures that normally occur nearby the workpiece, so the electron work function will be reduced by the reaction between CO₂ and the melt pool, resulting in more heat being transferred to the workpiece and more material being melted [24]. In this case, more energy will be transferred to the boundary of the melt pool, and the fluid in the melt pool flows more strongly downwards, giving rise to both the increase in bead width and height [24]. The bead width and HAZ width became broader with the increase of CO₂ content from 2.5% to 20%, as shown in Fig. 5b–d. Also, replacing 38% of Ar with He led to a larger bead width and height (Fig. 5b and e). This is because He has a lower density but higher thermal conductivity than Ar, leading to more heat being transferred from the arc core to the arc periphery, which increases the arc potential gradient for energy loss compensation [24]. The bead dimensions remained similar when replacing 38% of Ar with He alongside ⁓2% CO₂, indicating He has less effect on heat transfer than CO₂. Following checking of the bead morphologies processed under various local shielding gas mixtures, the pure Ar build was excluded from further investigations owing to poor process stability and unsatisfying resulting morphology.

The changes in bead dimensions resulted in changes in wall height, as shown in Fig. 6. A decreasing trend in wall height can be found when CO₂ content in the shielding gas mixture increased from 2.5 to 8%, in line with the bead height results. However, with a further increase to 20%, the wall height does not decrease as expected. This is likely to be related to the misalignment deviation between the heat source and the centerline. Regarding wall width, the narrowest wall width could be found in W3 due to the material loss caused by the severest spattering phenomenon during the deposition process.

The etched cross-sections shown in Fig. 6 reveal a number of meso-scale features that need to be explained. First, there is a distinct light-etching region at the top of the build (roughly five layers high) followed by a darker-etching region that comprises the bulk of the build. Second, there is a “transition” region between the top and bulk regions of each build, where light- and dark-etching regions interpenetrate. Third, there are dark-etching bands and light-etching bands present periodically throughout the builds—the dark-etching bands, slightly inclined to the horizontal plane, only appear in the bulk material, whereas the curved light-etching bands appear throughout the bulk and top of the builds. Each of these features will be discussed in turn below.

The microhardness transitions from the top of the base plate to the top surface of the wall structures manufactured under different local shielding mixtures are presented in Fig. 12. It shows that the microhardness value fluctuated at around 450 HV in the bulk and then significantly increased to 620–650 HV when entering the top region. Such a hardness transition is consistent with the formation of tempered martensite and fresh untempered martensite in the bulk and top regions, respectively. During the in situ tempering process caused by the layer-by-layer thermal cycles, carbides would have gradually precipitated out from the martensitic matrix, leading to a decreasing supersaturation degree of carbon, and the dislocation density also would have likely decreased, leading to low hardness levels [37]. It might have been expected that the average microhardness values both in the bulk and top regions would decrease gradually with the increased CO₂ content in the local shielding atmosphere, since more heat would have been generated when introducing more CO₂ into the local shielding mixture, leading to a greater tempering effect and slower cooling rates. However, no trend could be distinguished above the scatter in results. Similarly, no statistically significant change in hardness was detected when replacing 38% of Ar with He.

Due to the location dependence of microhardness values, three tensile coupons were extracted from both the top and bulk regions of the wall structures under different shielding gas mixtures to further examine the tensile performances. The room-temperature tensile stress–strain curves and corresponding properties of the specimens are shown in Fig. 13. It can be seen that the strength in the top region is higher than that in the bulk region, while the ductility showed the contrary trend regardless of the shielding gas mixture conditions. Necking can be clearly observed on the fractured tensile specimens extracted from bulk regions (Fig. 13b) and is not so evident in the top region specimens, reflecting better plasticity in the bulk of the walls. There are several strengthening mechanisms in polycrystalline alloys, including solid solution, precipitation, dislocation, and grain boundary strengthening. Due to the multiple thermal cycles in the deposition process, which can be viewed as an in situ tempering treatment, carbide precipitates (e.g., cementite) are formed, which contribute to strengthening. Nevertheless, it is well known that their strengthening effect is outweighed by the loss of carbon presence in solution or pinning dislocations and also by the loss of dislocation density during the tempering thermal cycles (this is why tempering martensite decreases its strength in general). Since the average grain size of the top region (whether martensite or original austenite) is coarser than the bulk part, the effect of grain boundary strengthening here is clearly offset by martensite strengthening. In comparison, the shielding gas mixtures from the torch exert a much lesser influence on strength and ductility compared with deposition location—no trends can be confidently discerned with respect to shielding gas mixture.

The fracture morphologies in different regions of the wall components deposited under different local shielding gas mixtures are shown in Fig. 14. Distinct differences in fracture morphologies can be found between the top and bulk regions. The fracture cross-section in the top region samples is larger than in the bulk samples, which is consistent with the reduction of area and ductility in the top region being smaller than that in the middle. By magnifying, it can be found that the fracture morphology in the top region consists of quasi-cleavage facets, tearing ridges and shallow dimples, indicating a mixed mode of ductile and brittle fracture. The development of quasi-cleavage facets could be attributed to the difficulty inhibiting crack initiation and propagation by untempered martensite at the top region [4]. In comparison, the fractured bulk region specimens mainly consist of deep dimples with cementite particles inside and some fine-facet cracks, indicating a mixed fracture but more inclined to ductile fracture. These cracks are formed where cleavage occurs, indicating that a cleavage-detaching mechanism may have been associated with the boundaries of the martensite blocks [38]. Therefore, it can be seen that the ductile characteristics in tensile fracture morphology increase from the top region to the reheated bulk region, which is in accordance with the elongation data. Besides, the effects of local shielding gas mixtures on the fracture morphology appear to be negligible compared to the deposition location. Nevertheless, the largest concentration of dimples was measured in the bulk of W4 and the smallest concentration in the bulk of W3 in Fig. 14b, d, f, and h, consistent with the highest and poorest elongation being achieved in W4 and W3, respectively.

Table 4 summarizes the tensile properties of 300 M UHSS produced by various AM processes in previous studies. The bulk region of GMA-WAAM 300 M UHSS possesses comparatively lower strength but higher ductility than the CMT-WAAM products, which could be attributed to a higher thermal input increasing the level of tempering. However, GMA 300 M bulk possesses similar strength as the laser-powder-based AM product in the as-deposit condition, but the elongation is much higher, which could be attributed to different energy density, porosity level (this tends to be higher in powder-based AM), and tensile coupon design. Compared with the AMS6419K standard of 300 M UHSS forgings, the top region of GMA 300 M studied here possesses comparable strength but comparatively lower ductility. While the ductility of the GMA 300 M bulk region is much higher than the forged one, the strength is far away from the lower limit of the standard. These observations are consistent with the top region being under-tempered, and the bulk region being over-tempered in the as-built state. Therefore, it is clear that in the as-deposit condition, the tensile properties of all AM 300 M UHSS do not meet the standard and need to be heat treated. Therefore, in future research, comprehensive post-heat treatment investigations will be carried out to make the final mechanical properties meet the standard.