Microstructure and Mechanical Properties of Medium Carbon Steel Deposits Obtained via Wire and Arc Additive Manufacturing Using Metal-Cored Wire

: Wire and arc additive manufacturing (WAAM) is a 3D metal printing technique based on the arc welding process. WAAM is considered to be suitable to produce large-scale metallic components by combining high deposition rate and low cost. WAAM uses conventional welding consumable wires as feedstock. In some applications of steel components, one-o ﬀ spare parts need to be made on demand from steel grades that do not exist as commercial welding wire. In this research, a speciﬁcally produced medium carbon steel (Grade XC-45), metal-cored wire, equivalent to a composition of XC-45 forged material, was deposited with WAAM to produce a thin wall. The speciﬁc composition was chosen because it is of particular interest for the on-demand production of heavily loaded aerospace components. The microstructure, hardness, and tensile strength of the deposited part were studied. Fractography studies were conducted on the tested specimens. Due to the multiple thermal cycles during the building process, local variations in microstructural features were evident. Nevertheless, the hardness of the part was relatively uniform from the top to the bottom of the construct. The mean yield / ultimate tensile strength was 620 MPa / 817 MPa in the horizontal (deposition) direction and 580 MPa / 615 MPa in the vertical (build) direction, respectively. The elongation in both directions showed a signiﬁcant di ﬀ erence, i.e., 6.4% in the horizontal direction and 11% in the vertical direction. Finally, from the dimple-like structures observed in the fractography study, a ductile fracture mode was determined. Furthermore, a comparison of mechanical properties between WAAM and traditionally processed XC-45, such as casting, forging, and cold rolling was conducted. The results show a more uniform hardness distribution and higher tensile strength of the WAAM deposit using the designed metal-cored wires.


Introduction
Additive manufacturing (AM) processes produce 3D components directly from dedicated 3D CAD models by depositing material layer-by-layer. AM offers the advantage of building parts with geometric and composition complexities, which are difficult to produce by conventional subtractive manufacturing processes [1,2]. Wire and arc additive manufacturing (WAAM) is an AM technique that combines an electric arc as a heat source and a consumable wire as feedstock. It can be considered as a modification of the classical gas metal arc welding (GMAW) process. Due to its high deposition rate, In the present study, a medium carbon steel metal-cored wire is investigated for GMAW-based WAAM. The specific composition was chosen because it is of particular interest for the production of heavily loaded aerospace components and this chemical composition does not exist as commercial welding wire. The appearance, microstructure, yield/tensile strengths, and fractography of the deposited single-bead wall were studied for potential WAAM applications.

Materials and Methods
In this study, S355 structural steel [19] base plate of 250 × 60 × 10 mm 3 was used. XC-45 (standard AFNOR, NF A37-502) metal-cored wire with a diameter of 1.2 mm was employed as filler material. The chemical compositions of the materials used in this study are listed in Table 1. For the XC-45, the chemical composition referring to the composition of the deposited metal was measured by optical emission spectroscopy (OES). Prior to deposition, the base surface was cleaned with an acetone ((CH3)2CO) solution. The deposition was carried out using a Panasonic robotic arm ( Figure 2a) and a power source integrated by Valk Welding. The software used for designing the deposition tool path was Autodesk PowerMill (Autodesk B.V, Hoofddorp, The Netherlands). The experimental arrangement in this study is shown in Figure 2b, indicating the deposition direction, the build direction, and the positioning of the clamps. The torch was positioned perpendicular to the workpiece (PA position). The deposition strategy, i.e., reversing the deposition direction for each layer as shown in Figure 2c, was applied to avoid a height difference between the start and stop zones. A parametric experiment was carried out by ramping up the wire feed rate [20] (4-8 m/min) to select optimal deposition conditions based on good bead appearance and an appropriate width to height ratio [21]. The reference deposition parameters used in this research are summarized in Table 2.

Materials and Methods
In this study, S355 structural steel [19] base plate of 250 × 60 × 10 mm 3 was used. XC-45 (standard AFNOR, NF A37-502) metal-cored wire with a diameter of 1.2 mm was employed as filler material. The chemical compositions of the materials used in this study are listed in Table 1. For the XC-45, the chemical composition referring to the composition of the deposited metal was measured by optical emission spectroscopy (OES). Prior to deposition, the base surface was cleaned with an acetone ((CH 3 ) 2 CO) solution. The deposition was carried out using a Panasonic robotic arm ( Figure 2a) and a power source integrated by Valk Welding. The software used for designing the deposition tool path was Autodesk PowerMill (Autodesk B.V, Hoofddorp, The Netherlands). The experimental arrangement in this study is shown in Figure 2b, indicating the deposition direction, the build direction, and the positioning of the clamps. The torch was positioned perpendicular to the workpiece (PA position). The deposition strategy, i.e., reversing the deposition direction for each layer as shown in Figure 2c, was applied to avoid a height difference between the start and stop zones. A parametric experiment was carried out by ramping up the wire feed rate [20] (4-8 m/min) to select optimal deposition conditions based on good bead appearance and an appropriate width to height ratio [21]. The reference deposition parameters used in this research are summarized in Table 2.
Ninety layers were deposited in total and each layer consisted of one bead. After deposition of each layer, the wall was air-cooled until the top surface was measured to be at room temperature, and the contact tip-to-work distance (CTWD) was kept 1 mm higher.
The samples of the as-deposited wall were prepared for metallurgical and mechanical investigations. The prepared cross section was etched by 2% Nital (98% ethanol and 2% HNO 3 ) for optical microscopy (Keyence VHX-5000, Osaka, Japan). The tensile samples were prepared in the vertical and horizontal directions, as shown in Figure 3, according to ASTM E8M-09 standard and tested at room temperature according to the DIN EN 6892-1 standard by means of an Instron-5550 tensile testing machine (Norwood, MA, USA). In addition, fractography was performed using a JEOL JSM-IT100 scanning electron microscope (SEM, Tokyo, Japan). Phase analysis was supported by energy dispersive spectroscopy (EDS). High-resolution microstructural characterization was carried out using a JEOL FEG-SEM JSM 5600F scanning electron microscope (SEM). Vickers hardness (HV2, with 2 kgf) was measured on a cross section of the wall, in the direction from the highest point towards the base, using a Struers DuraScan-70 machine (Struers Inc., Cleveland, OH, USA). The indentation path is shown in Figure 4a.
Ninety layers were deposited in total and each layer consisted of one bead. After deposition of each layer, the wall was air-cooled until the top surface was measured to be at room temperature, and the contact tip-to-work distance (CTWD) was kept 1 mm higher.
The samples of the as-deposited wall were prepared for metallurgical and mechanical investigations. The prepared cross section was etched by 2% Nital (98% ethanol and 2% HNO3) for optical microscopy (Keyence VHX-5000, Osaka, Japan). The tensile samples were prepared in the vertical and horizontal directions, as shown in Figure 3, according to ASTM E8M-09 standard and tested at room temperature according to the DIN EN 6892-1 standard by means of an Instron-5550 tensile testing machine (Norwood, MA, USA). In addition, fractography was performed using a JEOL JSM-IT100 scanning electron microscope (SEM, Tokyo, Japan). Phase analysis was supported by energy dispersive spectroscopy (EDS). High-resolution microstructural characterization was carried out using a JEOL FEG-SEM JSM 5600F scanning electron microscope (SEM). Vickers hardness (HV2, with 2 kgf) was measured on a cross section of the wall, in the direction from the highest point towards the base, using a Struers DuraScan-70 machine (Struers Inc., Cleveland, OH, USA). The indentation path is shown in Figure 4a.    Layer height 0.00177 (m) Figure 3. Schematic of the tensile test specimens extracted from the WAAM sample.

Macroscopic Inspection
The single bead wall composed of ninety layers is shown in Figure 4. The final dimensions were measured to be around 190 mm in length, 160 mm in height, and 7 mm in width. As seen in Figure  5, the peak-to-valley variation (the distance between the highest and lowest point of the surface) on the top part of the wall was measured to be ~600 μm. The peak-to-valley variation could be caused by bead shape variation due to molten pool instability. It has been reported that higher heat input contributes to decreased surface waviness [22]. Therefore, with the wall building up, the heat input fluctuation in the molten pool caused by the change of heat dissipation condition results in molten pool instability, such as oscillation. In addition, because of the adequate cooling after each layer, the heat accumulation was controlled, so that no significant distortion was noticed for both the base and the wall.   Table 2).  Table 2).

Macroscopic Inspection
The single bead wall composed of ninety layers is shown in Figure 4. The final dimensions were measured to be around 190 mm in length, 160 mm in height, and 7 mm in width. As seen in Figure 5, the peak-to-valley variation (the distance between the highest and lowest point of the surface) on the top part of the wall was measured to be~600 µm. The peak-to-valley variation could be caused by bead shape variation due to molten pool instability. It has been reported that higher heat input contributes to decreased surface waviness [22]. Therefore, with the wall building up, the heat input fluctuation in the molten pool caused by the change of heat dissipation condition results in molten pool instability, such as oscillation. In addition, because of the adequate cooling after each layer, the heat accumulation was controlled, so that no significant distortion was noticed for both the base and the wall.  Table 2).  Table 2).

Microstructure Evolution during WAAM Deposition
The microstructure was investigated at different locations within the WAAM deposited wall (as indicated in Figure 4). Representative etched cross sections at those locations are presented in Figure  6. The cross section revealed the ferrite (white), pearlite (black island), and bainite (black sheaf) phases.  Table 2).

Microstructure Evolution during WAAM Deposition
The microstructure was investigated at different locations within the WAAM deposited wall (as indicated in Figure 4). Representative etched cross sections at those locations are presented in Figure 6. The cross section revealed the ferrite (white), pearlite (black island), and bainite (black sheaf) phases.
Columnar grains were found at the top of the wall (region A). These columnar grains were directed perpendicular to the fusion line due to the preferential grain growth in the maximum thermal gradient. The prior austenite grains were decorated by grain boundary ferrite. The temperature was measured with a thermocouple at the middle of the deposited track after each layer was deposited, from which the cooling rate was measured by thermocouple to be approximately 85.7 • C/s from 1100 to 800 • C and 28.5 • C/s from 800 to 500 • C.
The microstructures observed by optical microscopy in regions B and C are shown in Figure 6b,c, respectively. From the entire of the two micrographs, it can be seen that the microstructure included ferrite (white) and pearlite (black). The microstructure, in general, became finer from the top to the bottom of the wall.
To have a clearer characterization of the microstructure, a higher resolution micrograph was taken by SEM at region C, which is shown in Figure 7. Some carbide precipitates and oxide particles (black spots in Figure 7a) were found in this region. The pearlite lamellae are shown in Figure 7b and the carbide precipitates are dispersed within the ferrite, as shown in Figure 7c. As seen in Figure 7b, the pearlite lamellae appeared to have thickened and contained edges. This was an indication that during reheating, the pearlite transformed into high carbon austenite, which then transformed to martensite during cooling. This was possible because of the high local hardenability of the carbon enriched austenite. The finest grains were observed in region D, as shown in Figure 6d. The grain size reduction was a result of the multiple thermal cycles experienced by the material. After the multiple thermal cycles, the ferrite and pearlite became finer. This could be proven by the measured maximum width and length of the prior austenite grains in region A and region D. The maximum width/length of the grain decreased from 37.5 µm/71.5 µm in region A to 5 µm/15 µm in region D. It can also be seen that the nucleation of ferrite occurred preferentially at prior columnar austenite grain boundaries.
Columnar grains were found at the top of the wall (region A). These columnar grains were directed perpendicular to the fusion line due to the preferential grain growth in the maximum thermal gradient. The prior austenite grains were decorated by grain boundary ferrite. The temperature was measured with a thermocouple at the middle of the deposited track after each layer was deposited, from which the cooling rate was measured by thermocouple to be approximately 85.7 °C/s from 1100 to 800 °C and 28.5 °C/s from 800 to 500 °C. The microstructures observed by optical microscopy in regions B and C are shown in Figure  6b,c, respectively. From the entire of the two micrographs, it can be seen that the microstructure included ferrite (white) and pearlite (black). The microstructure, in general, became finer from the top to the bottom of the wall.
To have a clearer characterization of the microstructure, a higher resolution micrograph was taken by SEM at region C, which is shown in Figure 7. Some carbide precipitates and oxide particles (black spots in Figure 7a) were found in this region. The pearlite lamellae are shown in Figure 7b and the carbide precipitates are dispersed within the ferrite, as shown in Figure 7c. As seen in Figure 7b, the pearlite lamellae appeared to have thickened and contained edges. This was an indication that during reheating, the pearlite transformed into high carbon austenite, which then transformed to martensite during cooling. This was possible because of the high local hardenability of the carbon enriched austenite. The finest grains were observed in region D, as shown in Figure 6d. The grain size reduction was a result of the multiple thermal cycles experienced by the material. After the multiple thermal cycles, the ferrite and pearlite became finer. This could be proven by the measured maximum width and length of the prior austenite grains in region A and region D. The maximum width/length of the grain decreased from 37.5 μm/71.5 μm in region A to 5 μm/15 μm in region D. It can also be seen that the nucleation of ferrite occurred preferentially at prior columnar austenite grain boundaries.

Hardness of the WAAM Deposited Wall
The variation of hardness along the vertical direction of the deposited wall is plotted in Figure  8a. In general, the deposited material undergoes several thermal cycles, which is expected to affect the hardness [23,24]. In the present case, the average measured hardness of bottom, middle, and top regions were 238 ± 8 HV, 243 ± 5 HV, and 250 ± 4 HV, respectively. The hardness of the top layers was slightly higher than the bottom layers because of the presence of non-equilibrium phases such

Hardness of the WAAM Deposited Wall
The variation of hardness along the vertical direction of the deposited wall is plotted in Figure 8a. In general, the deposited material undergoes several thermal cycles, which is expected to affect the hardness [23,24]. In the present case, the average measured hardness of bottom, middle, and top regions were 238 ± 8 HV, 243 ± 5 HV, and 250 ± 4 HV, respectively. The hardness of the top layers was slightly higher than the bottom layers because of the presence of non-equilibrium phases such as bainite and Widmanstätten ferrite, as seen in Figure 6a. The measured hardness (Figure 8a) showed that the fluctuation of hardness in the middle region was the smallest, while that of the top and bottom regions was higher. This indicated that the mechanical properties within the middle region were suspected to be more uniform. resulting in a relatively homogeneous hardness. The middle part of the wall also experienced re-heatings, therefore tempering effects also contributed to the mechanical property values. Tempering effects were pronounced in the middle region because there was no influence of the base plate during cooling. c) The top part of the wall, region A. This was the last deposited section of the wall, in which the cooling rate was as in the previous case b), but the number of re-heating cycles experienced to induce significant tempering effects was reduced. Therefore, this region was likely to exhibit higher hardness values.
The hardness of the WAAM deposited wall was compared with the hardness of XC-45 from different manufacturing processes, as shown in Figure 8b. It shows that the WAAM deposited XC-45 metal-cored wire had a comparable hardness with the other processes.

Tensile Strength Evaluation of the WAAM Deposited Wall
The measured yield and tensile strength together with the elongation of the samples prepared in different directions are summarized in Figure 9. The measurements showed that there was anisotropy in mechanical behavior within the WAAM deposited wall. The tests showed that both yield and ultimate strength decreased from the top to bottom layers. This was in agreement with the hardness profile shown in Figure 8a. A combination of microstructural factors affects hardness values. In the present case, the hardness depended on the local phase constituents, the grain size, and the fraction and size of the precipitates. The phases present could exhibit a hardness variation due to their exposure to high temperatures. High-temperature exposure induced tempering, recovery, and recrystallization effects, which decreased the hardness, while at the same time, precipitation of carbides and martensitic transformation contributed to a hardness increase. These effects were different along the build direction, as it was largely dependent on the cooling rate and the number of subsequent heat cycles that each layer was exposed to. Therefore, three different zones were distinguished: (a) The lower part of the wall, region D. In this region the cooling rate was influenced by the base.
The lower regions experienced the highest number of re-heating cycles due to the subsequent layers. Additionally, the composition of the first layers varied slightly, as the dilution of the consumable with the base took place. All these factors were expected to contribute to a lower hardness value and possibly higher elongation in the tensile test. (b) The middle part of the wall, regions C and B. In these regions, cooling, apart from convection and radiation, which was similar for all layers, was governed by conduction through the already deposited layers. Therefore, the cooling rate of this region was lower than that of the region D, resulting in a relatively homogeneous hardness. The middle part of the wall also experienced re-heatings, therefore tempering effects also contributed to the mechanical property values. Tempering effects were pronounced in the middle region because there was no influence of the base plate during cooling. (c) The top part of the wall, region A. This was the last deposited section of the wall, in which the cooling rate was as in the previous case (b), but the number of re-heating cycles experienced to induce significant tempering effects was reduced. Therefore, this region was likely to exhibit higher hardness values. The hardness of the WAAM deposited wall was compared with the hardness of XC-45 from different manufacturing processes, as shown in Figure 8b. It shows that the WAAM deposited XC-45 metal-cored wire had a comparable hardness with the other processes.

Tensile Strength Evaluation of the WAAM Deposited Wall
The measured yield and tensile strength together with the elongation of the samples prepared in different directions are summarized in Figure 9. The measurements showed that there was anisotropy in mechanical behavior within the WAAM deposited wall. The tests showed that both yield and ultimate strength decreased from the top to bottom layers. This was in agreement with the hardness profile shown in Figure 8a. It also can be seen that there was an apparent elongation difference between the horizontal and vertical direction. The existence of ferrite grains nucleated at columnar prior austenite grain boundaries contributed differently to the strength and elongation in the horizontal (deposition) and vertical (build) direction [25]. The anisotropic nature of additive manufactured material properties was also reported by other researchers [26,27].
These average values of the WAAM deposited with XC-45 metal-cored wires were compared with the results of traditional metal processing (based on AISI 1045 steel) methods, which is shown in Figure 10. The yield strength and ultimate strength produced by WAAM were higher than most of the conventional manufacturing methods, while the elongation was lower. This corresponded well with the hardness results presented in Figure 8, which also shows that the average hardness in the quenched condition was higher than for the other techniques. In order to improve the ductility of the XC-45 WAAM deposited material, we considered additional post-deposition heat treatments to acquire the desired ductility. The possible heat treatments for XC-45 WAAM is a topic that was not covered in the current study. It also can be seen that there was an apparent elongation difference between the horizontal and vertical direction. The existence of ferrite grains nucleated at columnar prior austenite grain boundaries contributed differently to the strength and elongation in the horizontal (deposition) and vertical (build) direction [25]. The anisotropic nature of additive manufactured material properties was also reported by other researchers [26,27].
These average values of the WAAM deposited with XC-45 metal-cored wires were compared with the results of traditional metal processing (based on AISI 1045 steel) methods, which is shown in Figure 10. The yield strength and ultimate strength produced by WAAM were higher than most of the conventional manufacturing methods, while the elongation was lower. This corresponded well with the hardness results presented in Figure 8, which also shows that the average hardness in the quenched condition was higher than for the other techniques. In order to improve the ductility of the XC-45 WAAM deposited material, we considered additional post-deposition heat treatments to acquire the desired ductility. The possible heat treatments for XC-45 WAAM is a topic that was not covered in the current study.

Fractography
According to the results of tensile test, the samples 3, 6, 8, and 11 (shown in Figure 3) were the samples from the top, middle, bottom, and vertical positions, respectively. Therefore, the fracture surfaces of these tensile samples (number 3, 6, 8, and 11) were observed by means of SEM, as shown in Figure 11. All examined tensile samples showed a non-porous ductile fracture, which was reflected by the dominance of dimples in all fracture surfaces, as seen in Figure 11a. Inclusions were also observed inside dimples. These sites had a higher possibility to nucleate cracks.

Fractography
According to the results of tensile test, the samples 3, 6, 8, and 11 (shown in Figure 3) were the samples from the top, middle, bottom, and vertical positions, respectively. Therefore, the fracture surfaces of these tensile samples (number 3, 6, 8, and 11) were observed by means of SEM, as shown in Figure 11. All examined tensile samples showed a non-porous ductile fracture, which was reflected by the dominance of dimples in all fracture surfaces, as seen in Figure 11a. Inclusions were also observed inside dimples. These sites had a higher possibility to nucleate cracks.
Comparing Figure 11b-e, the dimple size in the horizontal samples (numbers 3, 6, and 8) were found to be similar to that in the vertical sample (number 11). Due to the micro-void coalescence effect [28], some large dimples appeared locally.
The particle (A) could be found inside the dimples. The existence of the metallic phases or oxides could promote the formation of dimples. Smaller dimples were probably related to the voids initiating at grain boundaries or other microstructural features. Larger dimples tended to nucleate at oxide particles [29]. Energy-dispersive X-ray spectroscopy (EDX) analyses were performed at particle A ( Figure 11f) and the results are shown in Figure 11g. It can be seen that O, Fe, Mn, Cr, Si, and S were the main elements in particle A, which could be identified as either non-metallic inclusion, such as MnS [30] or oxide. Comparing Figure 11b-e, the dimple size in the horizontal samples (numbers 3, 6, and 8) were found to be similar to that in the vertical sample (number 11). Due to the micro-void coalescence effect [28], some large dimples appeared locally.

Conclusions
The microstructure and mechanical properties of WAAM material deposited with XC-45 metal-cored wires were investigated. Despite the challenging material composition, due to the high carbon content, the deposited wall showed good structural integrity and as-deposited mechanical properties attractive for industrial application. From the study, the following conclusions can be drawn: Based on the results found in this study, it has great potential to apply metal-cored wires in WAAM applications. Funding: This research was funded by the Dutch organization for scientific research (NWO-Nederlandse Organisatie voor Wetenschappelijk Onderzoek) in the framework of project GradWAAM, Project Number S16043. A special thanks to Mr. Vincent Wegener for providing additional financial support from RAMLAB.