3.1 Microstructure Characterization
Figure 3 shows the XRD patterns of the W powder, hot-rolled W and SLM W. The four observed diffraction peaks could be assigned to (110), (200), (211) and (220), as shown in Figure 3a. These results conformed to that of standard W diffraction with a body-centred cubic (bcc) lattice structure. Furthermore, the diffraction peak of SLM W was wider than those of hot-rolled W and W powder (Figure 3b), and this finding was consistent with that of previous research [23]. According to the Debye-Scherrer formula, the peak broadening was related to the grain refinement, which was benefited by the rapid solidification during the SLM process. In addition, the main peaks shifted to the left also indicated the presence of high residual stress in the SLM W, resulting in lattice distortion [23, 26].
Furthermore, the (110) diffraction angle of SLM W gradually shifted to a lower angle compared to those of W powder and hot-rolled W, as shown in Figure 3b, and according to the Bragg equation as follows:
$$2d \, \sin \theta = \lambda ,$$
(1)
where d is the interplanar spacing, θ is the diffraction angle and λ is the wavelength. According to Eq. (1), a smaller diffraction angle θ represents a larger interplanar spacing d, indicating the higher residual stress and lattice expansion of SLM W.
The optical micrograph surfaces of different directions of SLM W and hot-rolled W are shown in Figure 4. The individual scan tracks could be notably distinguished, and strip-shape grains were observed in the XY surface of SLM W, as shown in Figure 4a. Longitudinal and transverse cracks were present, and they extended along the grain boundaries. Columnar grains were observed in the YZ surface, which were different from the strip-shape grains noted in the XY surface, as shown in Figure 4b. The columnar grains along the building direction are typical microstructure in the SLM of pure metals due to the maximum heat flow direction is parallel to building direction, leading to the epitaxial growth of grains [23]. As shown in Figure 4c and d, the hot-rolled sample demonstrated a notable fibre structure combined with the small recrystallized grains, which could be attributed to the dynamic recrystallization occurring during the hot rolling process [4]. Few micropores and micro-cracks could be observed in the hot-rolled W. The average grain size of the hot-rolled W was significantly larger than that of SLM W.
Figure 5a and b show the EBSD orientation maps of the hot-rolled W and SLM W. It can be found that the deformed grains extended along the RD, as shown in Figure 5a. The average length and width of the grains were approximately 74.6 μm and 34.8 μm, respectively. The grain shape aspect ratio was 2.14. A typical microstructure morphology of SLM W with striped-shape grains was observed, as shown in Figure 5b. Similar grain morphologies were also reported in previous studies [22, 23, 27]. The average grain size of SLM W was approximately 14.2 μm, which was considerably smaller than that of hot-rolled W. The finer grains of SLM W could be a result of the occurrence of rapid solidification with an extremely high cooling rate during the SLM process [8, 28].
To further characterize the crystallographic textures of hot-rolled W and SLM W, the pole figures and inverse pole figures are shown in Figure 5c to g. The texture of the hot-rolled W had the following characteristics: <110>//RD, <110>//TD, <111>//ND and <001>//ND. Zhang et al. [4] compared the macro-texture and mechanical properties of W prepared using different rolling methods and found that the main texture <001> and <111> was parallel to the ND, which is in agreement with our result. The main texture orientation of SLM W was the <111> direction, parallel to the building direction (Z axis) with a higher texture intensity of 4.9 compared to the corresponding value of approximately 3.9 for hot-rolled W. The X and Y directions of the inverse pole figure exhibited a similar texture orientation due to the continuous rotation laser scan [29], as shown in Figure 5g.
Figure 6a and b show the distribution of grain boundaries of hot-rolled W and SLM W, respectively. The differently coloured lines represented different grain boundary angles: sub-grain boundary (SGBs, 2°–5°, red), low angle grain boundary (LAGBs, 5°–15°, green), high-angle grain boundary (HAGBs, > 15°, blue). The distribution of the grain boundary misorientation angles of the hot-rolled W and SLM W are shown in Figure 6c. The hot-rolled W had a large number of LAGBs and SGBs, distributed inside the elongated grains. The similar sub-grain structures were also observed in other hot-rolled W, which were caused by plastic deformation [30]. During plastic deformation, low deformation temperature and high strain rate can introduce more dislocations and inhibit the dynamic recovery of dislocations, which is beneficial to the formation of dislocation walls and sub-grain. Compared with HAGBs, the misorientation of LAGBs is so small that dislocations are easier to transfer across the boundary of adjacent grains, reducing the dislocations pile-up [31, 32]. As a result, the SLM W with higher fraction of HAGBs exhibits better strain compatibility than hot-rolled W. HAGBs are the preferred crack paths because that grain boundaries with large misorientation can promote crack propagation and decrease fracture energy [33]. Therefore, hot-rolled W shows better ductility than SLM W according to the grain boundary analysis.
3.2 Mechanical and Thermal Properties
Figure 7 shows the tensile stress–strain curves of the hot-rolled W and SLM W at different temperatures. The tensile direction of the hot-rolled W was parallel to the RD. The hot-rolled W exhibited typical brittle fracture at RT, 100 °C and 200 °C. When the temperature increased to 250 °C, the ultimate tensile strength (UTS) of the hot-rolled W was 496 MPa and the total elongation (TE) was 6.0%. It is well known that the DBTT is defined as the lowest temperature at which a sample undergoes a minimum elongation beyond 5% without failure [34]. According to the high temperature tensile test, it could be estimated that DBTT of the considered hot-rolled W was approximately 250 °C, which was similar to the reported results [35]. With the temperature increased from 250 °C to 500 °C, the UTS decreased from 496 MPa to 393 MPa. Furthermore, the samples at 400 °C and 500 °C demonstrated considerably ductility with a TE of ~15%.
Figure 7b shows the tensile results of the SLM W at various temperatures. Even at the highest testing temperature of 500 °C, the SLM W exhibited almost no ductility and a rather low tensile strength. However, the compressive strength of the SLM pure W was approximately 1500 MPa, which was higher than coarse grain W (CG W) fabricated through powder metallurgy (PM) [5, 36]. Thus, it could be noted that SLM W exhibited an interesting performance involving a relatively high compressive strength and an extremely low tensile strength. According to our previous research [22], due to the high residual stress of the SLM process, some micro-cracks appear in the sample, resulting in the low plasticity of pure W at room temperature. Therefore, for the SLM W, the tensile properties were sensitive to the micro-cracks in the samples, while the compression properties were not sensitive. However, no effective method currently exists to eliminate the micro-cracks by adjusting the process parameters or adding toughening materials [37,38,39]. Compared with SLM W, hot-rolled W has been subjected to more uniform temperature, less thermal gradient and higher external pressure during hot rolling process, so hot-rolled W exhibits higher density and almost no pre-existing micro-cracks. According to the comparative study, suppressing micro-cracks and increasing strength will be a research hotspot of SLM W.
Figure 8 shows the fracture surfaces of the SLM W and hot-rolled W at different test temperatures. The pre-existing micro-cracks acted as the source of cracks propagation, as shown in Figure 8a to c. Under tensile loading, the micro-cracks grew along the grain boundaries. According to Figure 4a and b, a large number of micro-cracks were distributed along the grain boundaries, which significantly weakened the bonding strength between the neighbouring grains. Thus, the fracture surfaces of SLM W were dominated by intergranular fracture, which propagated along the strip-shaped grain boundaries. Figure 8d shows that the fracture surfaces of hot-rolled W at room temperature were dominated by transgranular fracture (green dotted rectangles). Some intergranular fracture was also present along the rolling direction (red dotted rectangles), in which a fine and equiaxed recrystallization microstructure was observed. As shown in Figure 8e, the sample demonstrated a smaller cleavage surface and exhibited a certain amount of ductility. Nevertheless, at 500 °C the hot-rolled samples exhibited extremely ductile in Figure 8f. Strong necking phenomenon and many dimples can be observed on the fracture surfaces.
To further study the influence of micro-cracks on the mechanical performance of the hot-rolled W and SLM W, nano-indentation test was carried out. As shown in Figure 9, the load–depth curves of SLM W and hot-rolled W were present. Three segments were observed in the curves: loading, holding at maximum load and unloading. A larger depth was observed for SLM W at the same load, demonstrating that SLM W exhibited a lower hardness than hot-rolled W. In this work, the Oliver-Pharr analysis [40] was applied to calculate the hardness and elastic modulus. The hardness (H) is defined as follows [12, 41]:
$$H=\frac{{P}_\text{max}}{{A}_{c}}$$
(2)
$${A}_{c}=f\left({{h}_{c}}^{2}\right)=24.5{{h}_{c}}^{2}$$
(3)
where Pmax is the maximum load, Ac is the area of the indentation, hc is the contact depth at the maximum load. It could be determined that the hardness of the hot-rolled W (5.85 ± 0.18 GPa) was slightly higher than that of SLM W (5.51 ± 0.24 GPa). The results were in accordance with those presented in previous reports, indicating the results were reliable [37, 42]. It could be concluded that the rolled W was harder than the SLM W. Furthermore, the curves of hot-rolled W and SLM W demonstrated a similar unloading segment, implying a similar elastic modulus.
Even though the grain size of SLM was smaller, hot-rolled W exhibited higher nano-hardness and tensile strength. It could be speculated that the SLM W samples had large numbers of micro-cracks, which could easily expand. This W was sensitive to cracking, because of its high DBTT (~ 400 °C) [36]. The cracking of W was difficult to suppress due to the high thermal stress, which was larger than the intrinsic ductility. Wang et al. [35] reported that the segregated nano-pores at the grain boundaries (GBs) acted as crack nucleation zones.
The thermal conductivities of the hot-rolled W, SLM W, SPS W and ITER grade W are shown in Figure 10. At room temperature, the thermal conductivity of hot-rolled W was 182 W/m·K, that was 23% higher than that of SLM W. As is well known, the thermal conductivity was dependent on many factors, for instance the relative density, grain size, grain shape, grain boundary, defect and porosity. The relative density of the hot-rolled W and SLM W was 97.4% and 98.7%, respectively. The grain size of the hot-rolled W along the ND was 34.8 μm, which was approximately 2.5 times that of the SLM W. The GB density of the hot-rolled W was significantly less compared with that of the SLM due to larger average grain size of hot-rolled W. When there are more grain boundaries, the electrons are more likely to scatter at the interface and thus reduce thermal conductivity. At the same time, the SLM W exhibited a higher thermal conductivity compared with SPS W, because the columnar grain of the SLM W along the building direction was conducive to the thermal conduction [25].