Effect of Solution Treatment and Cooling Rate on the Microstructure and Hardness of Ti-6Al-4V Alloy Manufactured by Selective Laser Melting Before and After Hot Isostatic Pressing Treatment

SLM EOS M290 machine was used to create fatigue specimens. Argon gas was pumped into the EOS M290 machine's building chamber to prevent the titanium powder from oxidizing during the fabrication process. The scanning parameters were as follows: 80 μm spot diameter, 0.02 mm overlap stripe and 30 μm layer thicknesses. Three different laser power levels were employed to melt the powder at different locations to build the same sample, 370 W infill, 340 W down skin and 360 W up skin. The scanning process was carried out following a shell and core strategy. Firstly, the raster laser motion melted the inner layer to increase productivity, after which the contour of the layer was melted to gain surface quality and accuracy (Ref 24).

Subsequently, for each successive layer, the scanning laser was rotated by 67°. The samples employed in the experiment were printed horizontally and vertically in 3D form. The build platform was then covered with the first layer of the metal alloy powder, after which a powerful CO₂ laser was employed to trace out the layer. This process was carried out for each layer where laser ultimately melted the materials and fused them until complete fatigue specimens were formed horizontally and vertically with an overall length of 94 mm, a gauge section width of 7 mm, a grip section width of 19 mm and a thickness of 2.57 mm. Table 1 shows the chemical composition of the constructed specimens.

The specimens were placed in a vacuum furnace to undergo a stress-relief process without oxidation and avoiding deformation (Ref 25-28). Samples were then treated at 704 ± 14 °C for 2 ± 0.25 h according to ASTM F3301. Next, the samples were left to cool down in the furnace to room temperature (25 °C) to ensure that oxidation did not occur and minimize new residual stresses. Additionally, some samples were subjected to hot isostatic pressing to reduce pores and avoid defects (Ref 29). The detailed results of only stress-relieved samples were reported previously (Ref 8). Next, the samples were placed in an inert atmosphere at 920 °C for 120 ± 30 min and 100 ± 5 MPa, after which they were cooled down under an inert atmosphere to below 425 °C. The fatigue testing was then performed for both SR and SR+HIP samples. The samples were then cut from the grip section of the fatigue specimens. To evaluate the influence of HIP on porosity, the porosity of the specimens was tested and measured before and after HIP processing without the etching process to avoid any potential effect on the size and depth of the pores. In order to generate different microstructures, the samples underwent heat treatment at 930, 45 °C below the β-transus temperature. This process lasted for 15 min, after which they were cooled at different rates, including furnace cooling (FC), air cooling (AC) and water quenching (WQ). This is illustrated in Fig. 1. Finally, OM, SEM, XRD and EDS analyses were performed on the microstructures.

Before proceeding with any further heat treatment, the grip section of fatigue specimens was cut into eight pieces with a dimension of approximately 10×10 mm using wire electrical discharge machining (EDM) with a 5 mm/min cutting speed. Samples were then mechanically ground, polished down to a mirror finish and etched with Kroll's reagent. Samples were then cleaned with alcohol for 7-10 s, followed by distilled water. A Hitachi FlexSEM 1000 scanning electron microscope was used to examine the microstructure. The chemical composition of the samples was determined using EDS down to a few microns. Phase analyses were performed using the X’Pert HighScore Plus, and the fractions of phases were determined from SEM images using MIPAR image analysis software. Microhardness measurements were made using a Buehler micro-indentation hardness tester. A load of 0.30 kg was applied for ten seconds, with a minimum of seven indentations being performed per sample. The hardness measurement was conducted at random sites with indent size sufficiently large relative to the microstructural features concerned, as demonstrated in Fig. 2, and then, average values were calculated.

According to our previous study (Ref 8), the SR specimens which did not experience any additional treatment were characterized by the microstructure of martensite containing fine needle-shaped features with lath width up to 1.2 μm and microhardness value of 408 ± 4 HV0.3. In contrast, the SR + HIP specimens, before applying any further solution treatment, demonstrated coarser grains composed of a mixture of α with a typical lath width size of about 2-5 μm and ~ 6% β lamellar microstructure, with a reduction of volume fraction of porosity from 0.38 ± 0.24% to 0.20 ± 0.09 and a microhardness value of 346 ± 8 HV0.3. This is owing to the larger grain size, including the transformation of needle-shaped α′ to lamellar α + β. The driving factor is the simultaneous pressure and heat applied to the material during the HIP process, where the appearance of the β phase correlated positively with lower hardness (Ref 30, 31). The applied temperature is to raise the diffusivity while reducing the yield strength of the material allowing the pores to close properly. If the temperature is too high, however, undesirable amounts of grain development may occur, leading to changes in microstructure and material characteristics. In addition, the pressure works to ensure the solubility of any gas released due to collapsing pores. On the other hand, if the pressure was too low, some gases would escape to another pore vice the surface (Ref 32). Typical SEM images of the SR and SR + HIP samples are shown in Fig. 3.

The microstructural examination of the stress-relieved sample followed by solution treatment at 930 °C indicated that the quenched samples were characterized by a mixture of α and α′ phases identified as needles with size 1.8-80 μm in length and 0.3-2.4 μm in width, with a microhardness value of 336 ± 6 HV0.3. This result was attributed to the fact that at 930 °C, the microstructure mainly consists of β with a small amount of α phase, and during quenching, that β phase will transform into α′ phase. Meanwhile, the air cooling produced a structure dominated by the α phase with a size of up to 27.6 μm in length and up to 2.5 μm in width, with ~ 7.5% of the β phase and microhardness value of about 330 ± 4 HV0.3. In contrast, furnace cooling from 930 °C led to a microstructure with a coarse plate-like mixture of α phase with a size of up to 45 μm in length and up to 4.3 μm in width with ~17% of the β phase and no α′ martensite phase being noted as illustrated in Fig. 4. The average hardness for this microstructure was ~ 327 ± 6 HV0.3. XRD analysis validated that the microstructure of the water-quenched specimens did contain martensite and that β emerged at lower cooling speeds (air/furnace). As can be seen in Fig. 5, the patterns indicate the presence of the β phase after air cooling and furnace cooling, whereas only α/α' phase is noted with no β phase after quenching. Compared with air cooling, the XRD peak of the β phase was sharper following furnace cooling, most likely because of the different internal stresses associated with each process. Figure 6 provides the EDS spectra with the chemical composition in wt.% in various regions for heat-treated samples. WQ samples show no significant difference in the chemical composition in both white and dark areas, compared with the noticeable difference in the aluminum and vanadium values after air and furnace cooling in the two regions. The low value of vanadium indicates that the α phase dominates the microstructure.

The microhardness values were compared with those quenched samples from 1015 °C studied previously (Ref 8). When the treatment temperature was decreased by 85 °C, the microhardness values decreased by 16.5% from 391 ± 3 HV0.3 to 336 ± 6 HV0.3, while the size of martensite needles grew from 1.1 to 55 μm lengths and 1.1-1.8 μm widths to 1.8-80 lengths and 0.8-2.3 widths. This conclusion is consistent with the Hall–Petch strengthening effect, in which an increase in needle size results in a reduction in total needle boundary area, resulting in high ductility and low hardness (Ref 33, 34). This increase inhardness at 1015 °C might also be due to the differences in microstructure before WQ. As 1015 °C is above β-trances, only β phase will exist; hence, nearly 100% of α′-martensite will form after WQ. In contrast, at 930 °C, which is below β-trances, a mixture of β + α phase will form. In this case, only the β-phase would transform after quenching into α'-martensite. Furthermore, the composition of α before WQ will be different. The aluminum content at a higher solution treatment temperature of 1015 °C is different than 930 °C. Al would be more dissolved at higher solution temperature than at lower temperature. (Ref 35). Table 2 shows the comparisons of the microhardness values to those treated from 1015 °C.

As stated above, the microstructure of the SR+HIP samples consists of a mixture of α and β phases. During post-heat treatment at 930 °C, most of the α phase will transform to β-grains, and thus, microstructure will be mainly a β phase. Following quenching, the microstructure of the samples mainly consisted of several microns of α' needles with 0.8-41 μm in length and up to 1 μm in width with a microhardness value of 377 ± 2 HV0.3. Prior-β/_GB was observed because of the thermal gradient along the build direction, Fig. 7(a). The transformation of the β phase into a fine α′ martensite after the laser melting process is due to this process' very high cooling rate. A small amount of β was likely retained between the martensite needles. However, it would be low in volume and hardly detected. Figure 7(b) shows the microstructure after air cooling. The β-phase transformed to the α-phase leading to a mixture of α-lamella with up to 76 μm in length and up to 5 μm in width and a small amount of retained β phase. However, there are some fine needles of the α′ phase left in the microstructure.

Meanwhile, in the case of the specimens after furnace cooling, a complete transformation of β to α + β mixture was observed, and the pattern of α lath was coarser up to 62.8 μm in length and 5.6 μm in width, with ~ 14% of β as illustrated in Fig. 7(c). Vickers microhardness values decreased as the cooling rate decreased, reaching a minimum of 322 ± 4 HV0.3 with a 17% decrease after furnace cooling and 12% after air cooling to 336 ± 3 HV0.3 to the water-quenched samples. XRD analysis validated that the microstructure of the water-quenched specimens did contain martensite and that β emerged at lower cooling speeds (air or furnace cooling). As can be seen in Fig. 8, the patterns indicate the presence of the β phase at around 2θ = 39.3° and 39° after air cooling and furnace cooling, respectively, whereas the only α/α′ phase is noted, with no β phase after quenching. Comparing the peaks intensity and its full width at half maximum (FWHM) value with those samples not exposed to HIP treatment illustrated in Fig. 5, the peaks of the WQ specimen are wider following HIP treatment while no significant difference after air and furnace cooling. This wider peak after WQ may indicate higher internal stress and a smaller grain size. Figure 9 shows the EDS spectra presenting the measured chemical composition in the different regions on the microstructure. The result shows a higher fraction of vanadium in the white area after air and furnace cooling indicating the formation of β phase.

By comparing the results of solution-treated microstructure and hardness with and without the HIP process, an exciting relationship was observed between the heat treatments and their resulting hardness values. The hardness value of the solution-treated samples at 930 °C after the HIP process decreased from 377 ± 2 HV0.3 after water quenching to 336 ± 3 HV0.3 after air cooling and then slightly decreased to 322 ± 4 HV0.3 after furnace cooling. The increase inhardness value after quenching was due to the formation of α' martensite with a much finer microstructure. In contrast, samples not exposed to HIP processing show a lower microhardness value; 336 ± 6 HV after water quenching, 330 ± 4 HV after air cooling and 327 ± 6 HV after furnace cooling. This is related to the smaller grain size obtained after HIP followed by solution treatment compared to those solution-treated samples not preceded by the HIP. A smaller grain size means more grain boundaries, so the hardness value is higher. It might also be due to the lower porosity level after HIP illustrated in our previous study (Ref 8).