2.1 WAAM Manufacturing of Test Samples

This work investigates the influence of the cooling process by comparing the microstructure and mechanical properties of specimens realized with an active interlayer cooling system, consisting of a flow of compressed dry air, and with uncontrolled cooling.

Specimens for the experimental tests were cut from plates realized by WAAM using grade ER 308LSi austenitic stainless-steel wire. Details of the printing process were already described by Laghi et al.,^[15] where a preliminary characterization of plates produced with the active cooling system was reported. All plates were manufactured with the same set of process parameters. In detail, 1 mm wire diameter was used, and the used welding speed and wire feed rate were within the ranges of 15–20 mm s⁻¹ and 4–8 m min⁻¹, respectively. The used shield gases were 98% Ar and 2% CO₂ at a flow rate between 10 and 20 L min⁻¹, and the current and arc voltage of the deposition process were within the range of 100–140 A and 18–21 V, respectively, while the deposition rate was between 0.5 and 2.0 kg h⁻¹. The welding source used was gas metal arc welding (GMAW), with pulse welding arc transfer. No arc correction was adopted during the printing process. The substrate was a printing plate of 1000 × 1000 × 30 mm³, with H-type beams welded as support.

A total of ten plates with nominal dimensions of 380 × 380 × 4 mm³ were realized at MX3D facilities,^[17] five manufactured with the active cooling system and five under uncontrolled cooling conditions, thus allowing natural cooling of the printed layer. For the sake of conciseness, hereafter the specimens cut from plates produced with active cooling are indicated with “A,” whereas those cut from plates produced under uncontrolled cooling conditions are indicated with “U.” In case of active cooling, immediately after the deposition a nozzle blew dry compressed air (8 bar) on the layer. The nozzle was located at ≈30 mm from the layer with a 30° inclination and followed the same trajectory as the torch, moving with a speed of 10 mm s⁻¹. In case of relatively small parts, as in this study, this procedure is able to reduce idle time between subsequent depositions up to 80%.

To evaluate a possible anisotropic mechanical behavior induced by the WAAM process, tensile specimens were cut from the plates following three main orientations to the deposition layer (Figure 1): 1) longitudinal (L), i.e., parallel to the deposition layer; 2) transverse (T), i.e., perpendicular to the deposition layer; and 3) diagonal (D), i.e., at 45° to the deposition layer.

Tensile specimens were cut from plates having dimensions according to ISO 6892-1.^[18] Given the inherent geometrical irregularities proper of WAAM printing process,^{[14, 19, 20]} the surface of the specimens was machined before testing to reduce the variations in thickness and cross-section provided by the layer-by-layer surface roughness. From a nominal thickness of the plates of 4 mm, the resulting machined specimens were 2.5–3 mm thick.

2.2 Chemical and Microstructural Characterization

Glow discharge optical emission spectroscopy (GDOES: GDA-650, Spectruma Analytik GmbH, Hof, Germany, with a sputtered burnt spot of 4 mm diameter) was used to check the chemical composition of WAAM plates.

Samples for microstructural analysis were sectioned from grip regions of longitudinal (L), transverse (T), and diagonal (D) tensile specimens. The samples were mounted in a phenolic resin, grounded with silicon carbide papers up to 1200 grits, and finally polished, according to ASTM E3 standard^[21] up to 1 μm polycrystalline diamond suspension for a mirror finish. Microstructural features were revealed by chemical etching with 20 s immersion in Vilella's reagent (1 g picric acid, 5 mL hydrochloric acid, and 100 mL ethanol).^[22] In addition, to highlight the possible presence of the δ-ferrite phase, a specific color etching was applied following the procedure described by Hetzner.^[23] Microstructural analysis was carried out by optical (OM, Zeiss Axio Imager A1 and Reichert MEF3) and field-emission gun scanning electron microscopy (FEG–SEM, TESCAN MIRA 3). The phase composition of representative samples was assessed by X-ray diffraction (XRD: X’Pert PRO diffractometer, PANAlytical, Almelo, NL), using Ni-filtered Cu Kα radiation generated from a conventional tube operating at 40 mA filament current and 40 kV acceleration voltage. Patterns were acquired over 40° < 2θ < 100° range, with a step size of 0.02° and 120 s step⁻¹ counting time, with a 1D array of solid-state detector (X’Celerator PANAlytical).

2.3 Mechanical Characterization

The mechanical behavior of the WAAM plates was investigated by means of tensile tests, hardness, and nanoindentation measurements.

2.3.2 Nanoindentation and Hardness Tests

Standard microhardness measurements (HV_0.1) were also conducted in the interlayer regions of metallographic samples according to the Vickers test method described in the ISO 6507-1 standard,^[24] using a 100 g load and 15 s duration time.

For nanoindentations, a depth-sensing Berkovich nanohardness tester (NHT indenter, Anton Paar TriTec, Peseux, Switzerland) was used. Nanoindentation experiments were performed on polished samples under load-controlled mode, with maximum load of 250 mN and load rate 200 mN min⁻¹, according to ASTM E2546.^[25] Nanoindentation was performed on 5 × 5 indentation matrixes with 30 μm interspacing between the indents, which is large enough to avoid any overlap of deformation volume. The indentation depth (around 1.6 μm for all the samples) was ensured to be deep enough to ignore any indentation size effect.^[26] Three repetitions on each sample allowed obtaining statistically representative values from such a heterogeneous material. More details on the nanoindentation test are discussed in previous works,^[27-30] where the authors explain how it is possible to calculate the indentation hardness (HI, GPa) and indentation elastic modulus (E_IT, GPa). A representative optical image of the indentation and corresponding load–displacement curves are reported in Figure 2.

3.1 Compositional Analysis

Results of chemical composition analyses by GDOES are reported in Table 1 where the composition of samples with uncontrolled and active cooling is compared to the feedstock wire. Starting from an ER 308LSi wire, the composition of WAAM samples corresponded to the AISI 304 L chromium–nickel austenitic stainless-steel requirements, also reported in Table 1. This outcome was already evidenced in a previous work on the WAAM A plates^[15] and it is reasonable since ER 308LSi wire is a common filler wire for the welding of AISI 304 L.

Cr_eq and Ni_eq values were calculated according to the Schaeffler diagram^[34] from data reported in Table 1, and it was found Cr_eq= 20.5% and Ni_eq= 10.7% for WAAM A and Cr_eq= 20.7% and Ni_eq= 11.8% for WAAM U. Therefore, a microstructure with γ-austenite and δ-ferrite (≈10 vol%^[34]) is expected for both A and U conditions.

3.2 Microstructural Characterization

Microstructural analyses at low magnification, as well as fractographic analyses discussed in the following, highlighted the general absence of macrodefects (cracks, porosities) between any two adjacent layers, resulting in a dense microstructure. These results were also confirmed by density measurements of the printed plates. Samples processed with and without active cooling were characterized, in fact, by similar average density values equal to 7930 ± 240 kg cm⁻³. The density was evaluated according to the Archimedes principle, described in the ASTM B311-17 standard,^[31] by using an analytical balance and by measuring the specimen first in air and then immersed in distilled water, as also adopted in previous works.^{[14, 20]} Low-magnification micrographs highlighted also the typical hierarchical microstructure resulting from WAAM, consisting of subsequent deposited layers in which liquid metal solidifies following an epitaxial growth and, within long epitaxial grains, a fine columnar–dendritic solidification structure is formed.^{[3, 12]} Representative OM micrographs on metallographic samples extracted from the grip regions of L (a,d), T (b,e), and D (c,f) tensile specimens (according to the scheme in Figure 1) are reported in Figure 3. Deposition layers, epitaxial grain growth, perpendicular to deposition layers, and cellular–dendritic substructure inside long epitaxial grains, typical in WAAM-processed alloys, are visible. In the micrographs, layer boundaries are underlined with a yellow dashed line, and white dashed lines define borders of epitaxial grains that cross-over layers.

Figure 3 shows the typical hierarchical microstructure with regular deposition layers, perpendicular, parallel, and 45° oriented to the loading direction in the samples T, L, and D, respectively. Epitaxial grains extended across several deposition layers, as clearly visible in Figure 3. Such grains had a preferential orientation along the building direction (Z) and were perpendicular to the bottom substrate (XY plane) up to the top sections. As WAAM technology essentially consists of an additive multipass welding process, during each new layer deposition the upper part of the previous layer partially melts and, when solidification occurs, grains grow epitaxially from the already solidified material following the direction of the maximum temperature gradient, perpendicular to the solid/liquid interphase. Furthermore, because of heating induced by the deposition of each new layer, coarsening of grains in the upper region of the bottom layer, near the layer boundary, can occur. Grain coarsening is indeed clearly visible in Figure 4, which reports a representative interlayer region of the A and U samples and, according to all microstructural analyses performed, was unaffected by the application of the active cooling system. The existence of different solidification morphologies inside each deposited layer was reported also by other researchers,^{[32, 33]} who evidenced, from the bottom to the top of each layer, a fine cellular structure in the proximity of the lower boundary, a fine columnar structure in the layer core, and a coarse columnar structure with secondary dendrite arms at the top of the layer, close to the upper boundary. This outcome is due to the variation of the GR⁻¹ ratio (where G is the cooling rate and R represents the solid–liquid interface growth) along the layer and thermal cycles resulting from the additive process.^[2] Moreover, despite the different cooling conditions, on both the U and A samples within the large epitaxial grains, the same fine columnar–dendritic submicrostructure is observed at high-magnification analyses (Figure 5), consisting of γ-austenite dendrites and vermicular δ-ferrite in the interdendritic spaces. Based on the chemical composition in Table 1 and according to the ternary Fe–Cr–Ni phase diagram, a ferrite–austenite (FA) solidification mode should be verified. The amount of δ-ferrite was evaluated by image analyses and was found to be in the range of 8–12 vol%, a value consistent with the one predicted by Schaeffler diagram (10 vol%).^[34] The coexistence of γ- and δ-phases in WAAM austenitic stainless steels was also observed in the literature^[12] and, besides chemical composition, it is related to both thermal cycles and nonequilibrium rapid solidification conditions. The presence of δ-ferrite in the interdendritic regions was confirmed by FEG–SEM analyses. The microstructure in the XZ plane of a representative diagonal sample is reported in Figure 6, showing the distribution of the δ-phase. Specifically, SEM image by secondary electrons (Figure 6a) shows that δ-ferrite (light gray phase in the micrograph) follows the large columnar grains, oriented at 45°, and it is homogenously distributed along the entire XZ section. Further details are given by the high-magnification SEM image by backscattered electrons (Figure 6b), which show the concentration of δ-ferrite in correspondence of cells boundaries. In addition, by comparison of metallographic cross-sections on XZ (Figure 5a,b) and XY (Figure 5c,d) planes, it can be seen that the morphology of grains and phases changes, being stretched along the building direction (Z-axis) on the XZ plane and approximately equiaxed in the XY one.

The phase composition of the printed materials was investigated by XRD and the diffraction patterns are reported in Figure 7. The analysis confirmed that samples processed with and without active cooling presented the same phases, and in fact γ-austenite and δ-ferrite main peaks were identified. Moreover, the peak distribution and intensity of γ-austenite evidenced an anisotropic crystallographic texture: in the D sample the main γ peak is the (111), whereas it is the (200) is the T and L samples. Such a difference in crystallographic texture can significantly influence the final mechanical properties. Experimental observations for polycrystalline austenitic stainless steels^[35-37] have shown that the mechanical properties, such as Young's modulus, depend on the crystallographic orientation because E_hkl increases in the order E₁₀₀ < E₁₁₀ < E₁₁₁.

3.3 Tensile, Hardness, and Toughness Data

Mean values and standard deviations of tensile properties are reported in Table 2. Overall, from the results of tensile tests it seems that the cooling strategy did not highly influence the tensile response of WAAM-produced AISI 304 L stainless steel specimens. The main difference between the two cooling strategies was related to the elongation at rupture, whose values were higher on average for specimens with uncontrolled cooling (U), especially for longitudinal (L) specimens. On the other hand, as expected, the mechanical parameters are more affected by the specimen orientation with regard to the deposition layers. In fact, diagonal (D) specimens registered the highest values of Young's modulus, 0.2% proof stress and UTS. This trend confirms the anisotropic nature of WAAM-produced AISI304L plates, as previously studied.^[14-16]

Figure 8 reports the bar charts comparing the mean values and standard deviations of the tensile properties for the three directions investigated (T, L, and D) with specimens extracted from plates printed with active cooling (A) or with uncontrolled cooling (U). Reference standard values of the conventionally manufactured 304 L stainless steel, according to European building code, is also provided (dotted red line).^[36]

In general, specimens with active cooling showed slightly reduced values (reductions between 3% and 25%) for most of the mechanical properties for the T specimens, with some exceptions for the L specimens (UTS) and the D specimens (0.2% proof stress and UTS). The largest reductions were observed for the elongation at rupture. For the latter, T specimens reported the highest dispersion of results (evidenced by the high standard deviation). Reasoning for this was found in the presence of a defect on the fracture surface of one of the specimens tested with active cooling, as also reported in a previous work.^[15]

As far as Young's modulus values are concerned (Figure 8a), the active cooling strategy led to slightly reduced values. As expected, high anisotropy was observed from the bar charts, as widely investigated in a previous work^[15] and also reported by Kyvelou et al.^[14]

When considering the 0.2% proof stress values (Figure 8b), both L and D specimens realized with active cooling resulted in slightly larger values. As far as the UTS is concerned (Figure 8c), the active cooling led to a slightly reduced values for both the T and L specimens, whereas no significant variation is registered for D specimens. As already mentioned, the highest discrepancy between specimens with different cooling strategies was found when considering the elongation at rupture (Figure 8d). Indeed, for all three orientations considered the specimens with active cooling led to significantly reduced values.

Table 3 reports the values of the A/U ratio for all tensile properties distinguishing the different orientations. In general, the ratio are within +/-5% to +/-10%, in line with the inherent variability of the mechanical parameters for each type of specimen tested (as reported in Table 2). The main differences were registered for the elongation at rupture, whose values were higher for U specimens along all the three directions considered. The main increase in A_% is registered for the L direction, with a 25% increase.

Table 4 reports the relative ratios of the three orientations considered, with either active (A) or uncontrolled (U) cooling. Overall, the trends do not seem to be highly affected by the different cooling strategies. In fact, a similarly marked anisotropic behavior is registered for both types of specimens (A and U).

3.4 Fracture Surfaces

Representative SEM micrographs of fracture surfaces are reported in Figure 9. All samples, regardless of cooling condition and extraction direction, exhibited a ductile fracture mechanism. A ductile fracture is characterized by nucleation and growth of microvoids, called dimples, clearly visible at high magnification analyses (Figure 9a,b). Because of the very fine microstructure, dimples were still very fine, with an average dimension of the order of a few micrometers. As a confirmation of the microstructural and tensile results discussed previously, no major differences in dimples size and shape were evidenced among specimens obtained with or without the active cooling system or extracted along different directions, as also discussed in a previous study.^[15]

3.5 Microhardness

Vickers microhardness measurements (HV_0.1) have been performed on metallographic L, T, and D samples, and results are reported in Figure 10. With the aim of assessing the effect of the active cooling on the microstructure, microhardness was evaluated in the interlayer region of samples. As shown by the scheme in Figure 10, measurements were conducted on the coarse columnar–dendritic structure at the top of the bottom layer, the columnar–dendritic structure at the basis of the upper layer, and at the layer boundary. All samples exhibited a comparable value of hardness of ≈250 HV_0.1 and, by considering the standard deviation of the experimental results, no difference can be evidenced either according to different interlayer areas or cooling conditions, in agreement with previously discussed microstructural analyses.

3.6 Nanoindentation

One of the main features of metal additive processes is the location dependence of mechanical properties due to the layer-by-layer nature of the process, as well as heating and cooling rates during manufacturing. To assess the small-scale mechanical properties of the as-printed materials, depth-sensing nanoindentation experiments were therefore conducted. Instrumented indentation is a semidestructive, robust, and convenient approach to locally assess the mechanical properties of materials (e.g., location-dependent properties).^[37] The nanoindentation technique is a unique tool to track variations in local mechanical properties when dealing with fine and inhomogeneous microstructure, such as the one observed in metal additive manufacturing.^{[15, 38]} Past investigations showed the possibility to obtain hardness data from nanoindentation tests for a single laser track or within an individual melt pool, too small to be studied with conventional microhardness tests.^[39-41] The nanoindentation size is 10 μm and localized properties are measured rather than average ones like in Vickers hardness measurement.

The differences in properties between different zones could be related to the crystallographic texture^[42] and some studies confirmed that nanoindentation measurements can be suitable to analyze crystal orientation.^[43] Repetitions on the same component are possible and it is a convenient technique to asses mechanical properties, including hardness (H_IT) and Young's modulus (E_IT). Indentation hardness (H_IT) and indentation Young's modulus (E_IT) were evaluated from the force–displacement curves applying the Oliver and Pharr method. Figure 11 shows the indentation hardness and elastic modulus for the D, L, and T processed with or without active cooling. As shown, all the samples tested presented almost the same mean hardness values, close to 4.4 GPa, but the elastic modulus for D is considerably higher than for T and L specimens, perfectly in agreement with tensile test results. These results reflect the grain orientation evidenced by XRD. E is highest in the <111> direction^{[14, 37-46]} and this is also the main grain orientation in the D specimen.

The micromechanical behavior of WAAM 304 L austenitic stainless steel is very similar to the macromechanical one, confirming that instrumented nanoindentation as an easy, robust, and nondestructive technique to investigate mechanical properties, anisotropy, and texture useful in process design and optimization.