An Overview of Densification, Microstructure and Mechanical Property of Additively Manufactured Ti-6Al-4V — Comparison among Selective Laser Melting, Electron Beam Melting, Laser Metal Deposition and Selective Laser Sintering, and with Conventional Powder Metallurgy

This chapter begins with an introduction of the fundamental properties of Ti-6Al-4V, and its densification mechanism, typical microstructure and mechanical property achievable by conventional PM routes. This functions as a point of reference for the following discussion of the AM Ti-6Al-4V in terms of densification, microstructure, and mechanical property. The mostly popular laser-based AM techniques, namely selective laser melting (SLM), electron beam melting (EBM), laser metal deposition (LMD) and selective laser sintering (SLS), for the fabrication of Ti-6Al-4V have been overviewed based on an analysis of over 100 individual studies. Heat treatment is essential to most of the AM Ti-6Al-4V. Principles for selecting appropriate heat treatment for the AM Ti-6Al-4V are proposed based on martensite phase transformation and optimisation of mechanical properties. Oxygen impurity is an issue to most Ti materials and it is addressed in this chapter as well; counter measurements to mitigate oxygen have been suggested which involves the use of rare earth based materials.


Introduction
Additive manufacturing (AM) is the newest powder metallurgy (PM) technique [1][2][3] while Ti-6Al-4V is the single most important Ti alloy [4][5][6][7][8][9]. AM Ti-6Al-4V is being widely pursed by research community and industry for its capability to produce complicated, net-shape engineering parts and/or customised, biomedical implants. This chapter begins with an introduction of the fundamental properties of Ti-6Al-4V, and its densification mechanism, typical microstructure and mechanical property achievable by conventional PM routes. This functions as a point of reference for the following discussion of the AM Ti-6Al-4V in terms of densification, microstructure, and mechanical property. The mostly popular laser-based AM techniques, namely selective laser melting (SLM), electron beam melting (EBM), laser metal deposition (LMD) and selective laser sintering (SLS), for the fabrication of Ti-6Al-4V have been overviewed based on an analysis of over 100 individual studies. Heat treatment is essential to most of the AM Ti-6Al-4V. Principles for selecting appropriate heat treatment for the AM Ti-6Al-4V are proposed based on martensite phase transformation and optimisation of mechanical properties. Oxygen impurity is an issue to most Ti materials and it is addressed in this chapter as well; counter measurements to mitigate oxygen have been suggested which involves the use of rare earth based materials.  Table 1 and Table 2 list the fundamental physical and mechanical properties of the Ti-6Al-4V, respectively [14,15,27,28]. They serve as a point of reference and provide benchmark values for the following discussion on the AM Ti-6Al-4V.

Densification of SLS Ti-6Al-4V: Solid state sintering and liquid phase sintering
The fact that the as-built density of the SLS Ti-6Al-4V is similar to that of the conventional PM Ti-6Al-4V (see Figure 6) implies the two processing approaches should have analogous densification mechanism. This is understandable by referring to Figure 5 which shows that the SLS processing uses laser as the heating source while for the conventional PM Ti-6Al-4V, heating is mainly through conductive and/or radiation heat. Other than this the two processing pathways are essentially same. This further suggests that the driving force and activation energy for densification of the conventional PM Ti-6Al-4V should be applicable to the SLS Ti-6Al-4V. During SLS of Ti-6Al-4V, however, there is possibility that liquid phase sintering (LPS) occurs due to local overheating to temperatures higher than the liquidus temperature of Ti-6Al-4V (~1660°C). LPS provides extra driving force for sintering for it has extra surface energies, i.e. resulted from liquid surface and from liquid-solid interface [89,90]. It involves different sintering mechanisms, e.g. pore-filling, from the solid state sintering. These normally can contribute to an as-sintered density of up to 99.5% high [89,90]. Since the as-sintered density of the SLS Ti-6Al-4V is lower than this value, the overwhelming densification mechanism for the SLS Ti-6Al-4V should still remain as that of the solid state sintering.

Densification of EBM, SLM, and LMD Ti-6Al-4V: Solidification from liquid
Full or nearly full denseness is achievable in EBM, SLM and LMD Ti-6Al-4V (see Figure 6), implying that the densification mechanism for these AM approaches is different from that of the SLS. Indeed, for these three laser-based processing approaches, the densification process is more of solidification from liquid rather than the normal sense of sintering. The so-called time-temperature-transformation (T-T-T) of Ti-6Al-4V regulates phase selection and phase constitution of the solidified microstructure. Figure 7 provides the T-T-T curve of the Ti-6Al-4V alloy containing different levels of oxygen [91]. Under equilibrium conditions (i.e. low cooling rates), the resultant microstructure will be a mixture of thermodynamically stable α and β phases, while high cooling rates can enable formation of martensite phases.

Solidification map
As aforementioned, regarding EBM, SLM and SMD, the AM processing is essentially a solidification process . In this regard, the solidification map (see Fig. 9) suggested by Kobryn and Semiatin [1] is useful for predicting the solidification microstructure and can even be served for microstructural design. The two parameters, R and G, that are crucial for composing the solidification map can be measured and/or calculated as follows: where R is the solidification velocity which can be calculated based on a distance (dz) moved by solidus isotherm over a certain amount of time (dt), and G is the thermal gradient which can be obtained using a temperature window (dT) between solidus and liquids isotherms over a certain amount of distance (dz).

Texture and heterogeneity issue
Texture and heterogeneity are widely observable microstructural features in EBM, SLM and SMD Ti-6Al-4V . They are formed mainly due to the following two reasons: (a) different temperature distribution in the as-built sample and (b) different cooling rates in the various parts of the sample. Figure 10 provides optical graphs for an EBM Ti-6Al-4V at the transverse direction (Fig. 10a) and longitudinal direction (Fig. 10b) [4]. The former shows a lamellar microstructure while the later presents directional growth along the build direction. Greater thermal gradient in the latter is suggested to be the main reason for the formation of this columnar type of microstructure [4]. Figure 11 shows other two examples to illustrate the microstructural heterogeneity between the surface (Fig. 11a) and the bulk material (Fig. 11b) [71]. In this case, the much faster cooling rate in the surface of the AM Ti-6Al-4V has contributed to the formation of the acicular, metastable martensite phases in Fig. 11(a) while the bulk material presents a bimodal, equilibrium microstructure due to a much lower cooling rate, see Fig.11(b).

Martensite phase transformation
Cooling rate of SLM, EBM and LMD can be 10 4 K/s -10 6 K/s high . This enables martensite phases to form in the microstructure. Figure 12 proposes the dependency of phase selection on the cooling rate for Ti-6Al-4V [92,93]. Acicular α̍ , massive martensite α m or equilibrium (α+β) is proposed for each representative cooling rate. A minimum cooling rate of 20°C/s is suggested to be necessary for the formation of the martensite phases while when cooling rate is higher than 525°C/s the entire microstructure can be featured as acicular α̍ martensite. One still needs to note that the real cooling rate during AM processing of Ti-6Al-4V, other materials as well, is still a research question to be further investigated. Figure 13 provides an example from an EBM Ti-6Al-4V where featherless α̍ can be found in the microstructure [94]. The stable α phase is also observable in the microstructure although it is suggested that the cooling rate during EBM can be much higher than the critical cooling rate for the martensite phase transformation and accordingly the microstructure should have been overwhelmingly martensite. This is not consistent with the microstructural observation.

Martensite phase transformation
Cooling rate of SLM, EBM and LMD can be 10 4 K/s-10 6 K/s high . This enables martensite phases to form in the microstructure. Figure 12 proposes the dependency of phase selection on the cooling rate for Ti-6Al-4V [92,93]. Acicular α, massive martensite α m or equilibrium (α +β) is proposed for each representative cooling rate. A minimum cooling rate of 20°C/s is suggested to be necessary for the formation of the martensite phases while when cooling rate is higher than 525°C/s the entire microstructure can be featured as acicular α martensite. One still needs to note that the real cooling rate during AM processing of Ti-6Al-4V, other materials as well, is still a research question to be further investigated. Figure 13 provides an example from an EBM Ti-6Al-4V where featherless α can be found in the microstructure [94]. The stable α phase is also observable in the microstructure although it is suggested that the cooling rate during EBM can be much higher than the critical cooling rate for the martensite phase transformation and accordingly the microstructure should have been overwhelmingly martensite. This is not consistent with the microstructural observation.

Mechanical property of AM Ti-6Al-4V (As built)
UTS and elongation of as-built AM Ti-6Al-4V are summarised into Figure 14 (a) and (b) . ASTM requires that as-prepared Ti-6Al-4V should be no lower than 860 MPa in UTS and no lower than 10% in elongation. These two benchmark values are marked in the two figures using dotted lines. These data suggest that, comparatively speaking, EBM is able to provide a combination of good facture strength and good ductility which satisfy the ASTM specifications. Ti-6Al-4V made by SLM tends to show the highest fracture strength among the four AM approaches yet the corresponding ductility is the lowest which is mostly below the corresponding benchmark value (10%). There are no adequate data for LMD but it is reckoned that it should show a similar tendency to that of the SLM due to the similarity between the two processing techniques. Because of poor density (see Figure 6), SLS Ti-6Al-4V normally relies on post treatment such as HIP to achieve good fracture strength as well as ductility.

Heat treatment temperature: A collection from AM Ti-6Al-4V and corresponding mechanical property
Results show that the residual stress in AM Ti-6Al-4V can be higher than 1000 MPa [33].
Considering that the yield strength of Ti-6Al-4V is merely around 880 MPa (see Table 2), residual stress can initiate premature failure especially during dynamic testing such as fatigue.
It is a major concern for the AM Ti-6Al-4V. Most as-built AM Ti-6Al-4V therefore need heat treatment to mitigate the impact of the residual stress on the mechanical property performance. A collection of heat treatment temperatures being adopted by current studies is shown in Fig.  15  . Temperatures of the martensite phase transformation (at ~610°C), α → β (at ~750°C) and β-transus (at ~980°C) are also marked in the figure using dotted lines. The figure indicates that most of the heat treatments are distributed at temperatures between 610°C and 980°C. This is very likely due to the necessity of eliminating the hard-but-brittle martensite phases via phase transformation from the α' and/or the α m phases to stable (α and/or β) phases for the LMD, EBM and SLM approaches. The microstructure of the SLS Ti-6Al-4V is similar to that of the equilibrium state, and can be free from any post heat treatment. Figure 15. Collection of heat treatment temperatures adopted by various studies of the AM Ti-6Al-4V . Dotted lines in the figure represent critical temperatures for Ti-6Al-4V (refer to Table 1).
UTS and elongation of the as-annealed AM Ti-6Al-4V are shown in Fig.16  . If comparing with the as-built data (see Fig. 14), the general trend is that the fracture strength will be reduced after heat treatment while ductility can be improved. This is mainly resulted from both the reduced residual stress due to the heat treatment and the partially or even fully transformed martensite phases in the as-built microstructure.

Stress relief annealing and corresponding microstructure of as-annealed AM Ti-6Al-4V
Heat treatment at temperatures below the temperature of the martensite phase transformation may be not able to change the microstructure. Figure 17 provides such an example via an EBM Ti-6Al-4V annealed at 600°C for four hours [69]. The martensite phases are still observable after the annealing. Such heat treatment may only offer stress relief to reduce the residual stress in the as-built AM Ti-6Al-4V.

Heat treatment at temperature above the martensite phase transformation
Contrary to the low temperature annealing, heat treatment at temperatures higher than the martensite phase transformation may assist in eliminating the brittle, metastable martensite phases via phase transformation . Increasing the heat treatment temperature to higher than the α → β (at ~750°C) or even the β-transus (at ~980°C) will facilitate the transformation from α to β. Figure 18 provides an example of this where heat treatment was conducted up to 1200°C to form equilibrium lamellar (α+β) microstructure [69]. Martensite phases were not observable after such high-temperature heat treatment. Figure 18. Microstructure of an LMD Ti-6Al-4V annealed at 1200°C for two hours. Colony α-phase is observable along with prior β grains and grain boundary α phases [69].

Dependence of α colony size on heat treatment and microstructural dependency of mechanical property
The model developed by Tiley et al. [95,96] suggests that the mechanical performance of the (α+β) dual phase Ti-6Al-4V can be estimated based on a variety of microstructural parameters including the size of the α colony. The schematic graph in Figure 19 suggests that increasing size of the α colony may lead to reduced ductility (ε) and yield strength (σ 0.2 ) but in the meantime it may also contribute to better resistance to macro-cracks [95,96]. It is noted that heat treatment can coarsen the α colony of the AM Ti-6Al-4V, see Fig. 20 [54]. From this perspective, heat treatment of AM Ti-6Al-4V will have to be selected according to specific requirement and serve the application purpose of the material. Figure 19. Dependency of mechanical property on the α colony size of the Ti-6Al-4V alloy [95,96].

Oxygen equivalent and oxygen level in AM Ti-6Al-4V
Oxygen is detrimental to ductility of Ti-6Al-4V especially when it exceeds certain level [97][98][99]. Miura et al. [26] suggests the critical oxygen level for PM Ti-6Al-4V is 0.33 wt.%, above which ductility drops rapidly and can be much lower than the corresponding ASTM specification (see Fig. 1b). It needs to be noted that C, N and Fe may affect the mechanical performance of Ti-6Al-4V in a similar way to that of the oxygen, as suggested by the following Eq. (3) [100] and Eq.(4) [101].
( ) Generally speaking, AM processing only mildly increases oxygen level and the oxygen level in most of the as-built AM Ti-6Al-4V remain low . This is mostly contributed to the extra-low interstitial (ELI) raw powders used, good vacuum condition and/or inert gas protection achievable during AM processing. Only occasional report can be found where oxygen level is higher than the 0.2 wt.%O benchmark value, see Figure 21 [29-88].   Figure 21 indicates that the oxygen issue may have not been serious to the AM Ti-6Al-4V so far. This is mainly thanks to the use of ELI powders that are prepared by gas-atomisation or rotating electrode. The high cost of the ELI raw powders, however, can be an issue to AM Ti-6Al-4V from cost perspective. Figure 22 compares the price of a variety of powders, showing that the cost on the AM powders is extraordinary [4][5][6][7][8][9][10][11][12][13]. They are mostly higher than $400 per kg while hydrogenation-dehydrogenation (HDH) Ti powders that are widely used in conventional PM Ti are merely around $30 per kg. The high cost can be the bottle-neck issue for AM Ti-6Al-4V to be fully embraced by industries. For this reason, low-cost, high-interstitial (i.e. oxygen), fine powders are more desired for developing cost-effective AM Ti alloys. Sun et al. [102] recently developed an approach to manipulate irregularly-shaped HDH powders to make round-shaped, fine particles that are suitable for AM. The cost of their powders is higher than the HDH Ti powder but still markedly lower than the current AM powders, making it be more reliable for developing AM Ti and Ti alloys.

Counter measurements
Counter measurements need to be consulted when the high impurity (oxygen) powders are to be used for developing AM Ti-6Al-4V. Research has shown that the RE elements are capable to scavenge oxygen and their oxygen scavenging capability follows this sequence: Y>Er>Dy>Tb>Gd (see Figure 23a) [103,104]. The high potency of yttrium in scavenging oxygen from AM titanium alloys is demonstrated in Figure 23 (b), where the uniformly distributed Y 2 O 3 dispersoids are resulted from an addition of 0.1 wt.%Y to an EBM titanium alloy [105].
RE hydrides such as YH 2 have been further demonstrated to be able to scavenge, aside from oxygen, Cl, another important impurity element to Ti materials, Fig. 24 [106,107]. Figure 23. The free Gipps energy (∆G) of formation of various oxide materials, showing their thermodynamic stability as well as their affinity for oxygen [104]. Rare earth elements such as Y and some alkaline earth elements such as Ca have higher affinity for oxygen than Ti. (b) Precipitation of fine Y 2 O 3 dispersoids (due to the addition of 0.1 wt.%Y) in the α-Ti matrix of Ti-6Al-2.7Sn-4Zr-0.4Mo-0.45Si-0.1Y (wt.%) which contained 0.07 wt.% oxygen. The alloy was additively manufactured by EBM [105].

Concluding remarks
3D printing has becoming a focusing topic not only to research community and industry but also to the general public, and AM Ti and Ti alloys is one of the most promising and interested areas to be further developed. For the time being, although a few issues persist such as the microstructural inhomogeneity in the as-built material, some of the AM Ti-6Al-4V have already been able to achieve mechanical properties no lower than the corresponding ASTM specifications. The cost of the ELI AM Ti powders is one of the most challenging issues to limit the scale-up of the AM products. Employing low-cost powders to replace existing expensive powders to reduce the overall cost is a valuable research direction for further developing AM Ti-6Al-4V. Counter measurements to deal with the impurity issue associated with the highinterstitial Ti powders can be one of the key research elements for such development.