Review of laser powder bed fusion (LPBF) fabricated Ti-6Al-4V: process, post-process treatment, microstructure, and property

Ti-6Al-4V, also termed as TC4, titanium grade 5 and grade 23 (extra low interstitials), is the most commonly used Ti alloy^1,2. It combines excellent properties including good mechanical performance, outstanding corrosion resistance, and superior biocompatibility. Therefore, Ti-6Al-4V is widely applied in aerospace, automotive, marine and chemical industries. Furthermore, it has been widely used in biomechanical applications such as implants and prostheses^1–3. The details of the Ti-6Al-4V powders are available in the ASTM and AMS specifications^4,5.

Ti-6Al-4V is an α+β titanium alloy with a hexagonal close packed (HCP) α phase stable at low temperatures, and a body centered cubic (BCC) β phase stable at high temperatures¹. The β/α allotropic transformation temperature (β transus) is approximately 995 °C for Ti-6Al-4V², and this temperature can be increased by adding α stabilizers (Al, O, N, and C) and decreased by alloying with β stabilizers (V, Mo, Fe, etc.)^1,2. The crystal structures of α and β phases are shown in Fig. 1. Similar to another HCP/BCC metal Zr⁶, the crystallographic orientation relationship between HCP Ti (α) and BCC Ti (β) obeys the Burgers orientation relationship for which $ \left\{0001\right\} $_α//$ \left\{110\right\} $_β and <$ 11\overline {2}0 $>_α//<$ 111 $>_β¹.

Fig. 1  Crystal structures of HCP α and BCC β phases in titanium¹.

In Al-containing Ti alloys, the precipitation of ordered and coherent Ti₃Al (α₂) phase can occur in α grains after ageing at low temperatures of approximately 500 °C when the Al content is at 6 wt.%^7–9. Specifically, α₂ has a hexagonal D0₁₉ superlattice structure¹, in which the lattice parameter on the basal plane is twice that of α phase, whereas the lattice parameter remains the same on the c-axis. The α₂ solvus temperature is approximately 550 °C for Ti-6Al-4V¹. Similar to the influence on β transus, Al and O are α₂ stabilizers, which extend α+α₂ phase field to higher temperatures and promote the ordering transformation. Ordered α₂ phase introduces precipitation hardening, and it is an extremely brittle intermetallic phase^1,2. The presence of α₂ results in non-homogeneous planar slip, and increases the effective slip length and strain localization^10,11, which has detrimental effects on microcrack propagation, low cycle fatigue (LCF) resistance, and stress corrosion cracking resistance^1,12.

Rapid cooling from temperatures above the martensitic start temperature (M_s) leads to diffusionless transformation. The martensite start temperature (M_s) in Ti-6Al-4V has been reported in a wide temperature range from 575 °C to 800 °C^13–15, and a lower M_s has been observed at a higher cooling rate¹⁵. Similar to β transus and α₂ solvus temperatures, the martensitic start temperature is affected by the alloying elements. In general, the metastable martensite is hexagonal α’ in Ti-6Al-4V, and the Burgers orientation relationship is still valid between β and α’¹. Given that the interstitial O atoms only introduce a small elastic distortion in α’ martensite, the martensitic hardening leads to a slight increase in strength. Orthorhombic α’’ martensite is rarely observed in the low V containing Ti-6Al-4V. The martensite type can be determined by the temperatures from which Ti-6Al-4V is quenched and the concentration of β stabilizers in the β phase at that temperature³. In the α+β phase field, the V concentration in β is dependent on the temperature. The formation of α’’ is observed under conditions with a V concentration exceeding 9 wt.%^1,3. For example, the β phase is enriched with V concentration of up to 10 wt.% at a temperature range of 800−850 °C, and water quenching (WQ) from this temperature range leads to α’’ martensite transformation¹⁶. In contrast to the hardening effect of HCP α’ martensite, the orthorhombic α’’ martensite is soft and can provide deformability^2,3. Both metastable α’ and α’’ martensite phases can decompose into equilibrium α and β phases at suitable heat treatment temperatures.

The high cost of conventionally processed Ti alloys limits their applications, and crucial contributors to the expense are the significant material wastage, high energy input, and long lead times in conventional processing techniques¹⁷. Laser powder bed fusion (LPBF), as an advanced additive manufacturing (AM) technology, has attracted significant attention and has been implemented in research and industry over the last two decades. It has a wide range of advantages such as producing nearly fully dense and complex shaped components at a high resolution (which cannot be realized via conventional methods), high material utilization with limited machining, and reduced lead times^18,19.

Specifically, LPBF is a complicated metallurgical process, which involves powder melting and extremely fast directional solidification, re-melting and re-solidification during the subsequent scan tracks and layers, and post-solidification thermal cycling. The understanding of microstructure evolution during the LPBF process and post-process treatment is essential to obtain Ti-6Al-4V components with high quality and satisfied properties. In this review, the LPBF technique and its processing parameters are briefly introduced. Then, the possible types of defects, residual stress, and LPBF built surface are reviewed. Subsequently, the unique as-fabricated microstructure and microstructure evolution during the post-process treatment are summarized, and its effects on tensile, fatigue, fracture toughness, and creep properties are discussed.

LPBF, also known as selective laser melting (SLM) or direct metal laser melting (DMLM), was developed in the Fraunhofer Institute ILT in 1995²⁰. It employs a focused laser beam to melt metallic powders within a selected area based on the cross-sectional slice of a 3-dimensional (3D) CAD model, and the melt pool is then solidified at a high cooling rate. After a layer has been scanned, the build platform is lowered by a depth, which is defined as the layer thickness. A new layer of powder is spread by the re-coater and melted by the focused laser beam based on the next cross-sectional slice. The layer-by-layer deposition is repeated in a protective atmosphere until all the slices of the CAD profiles are complete. Fig. 2 illustrates a typical LPBF system, which includes a processing laser, an automatic powder feed and overflow system, and an inert gas protection system^21,22. After fabrication, unused powders can be collected from the overflow and powder bed. These collected powders can then be recycled and reused^23,24.

Fig. 2  Schematic diagram of a typical LPBF system²⁴.

The main in-process parameters include laser power, laser spot size, scanning speed, hatch distance, layer thickness, and build platform pre-heating temperature. Among them, the most widely manipulated parameters are laser power and scan speed. Fig. 3a presents a processing window between these two parameters. A high laser powder and a low scan speed can introduce an excessive energy input, which can lead to liquid melt overheating, evaporation within the melt pool, and keyhole defects²⁵. Conversely, a low laser power and a high scan speed can lead to insufficient energy density, which causes incomplete melting, inadequate penetration and lack-of-fusion defects^24,26,27. In addition to laser power and scan speed, inappropriate hatch distance and layer thickness can also result in lack-of-fusion defects. Furthermore, balling can occur at high laser power and scan speeds even for a suitable energy density as shown in the top right corner of Fig. 3a^28,29. Given that the increased laser power leads to a higher boosting effect on the maximum temperature achieved in the melt pool than the reverse effect from the increased scan speed^30,31, a higher peak melt pool temperature induces a more intense Marangoni flow, which can cause the Plateau-Rayleigh instability and balling^30,32,36. An example of different types of defects in LPBF fabricated Ti-6Al-4V is illustrated in Fig. 3b−d. For an intermediate laser power of 175 W, gradually decreasing scan speeds are located in the three different regions of incomplete melting, processing window, and overheating as shown in Fig. 3a. Additionally, another type of defect is gas porosity, which is typically small in size³³. The formation of gas pores can be related to two main causes: (a) the inert gas pores trapped within the powder feedstock during the powder atomization process are transferred to the part³⁴, and (b) after solidification, the protective gas can be trapped in the melt pool as bubbles due to the rapid cooling in LPBF fabrication³⁵.

Fig. 3  A process window and various types of defects in LPBF fabricated Ti-6Al-4V.

a A processing window for LPBF fabricated Ti-6Al-4V²⁸. b−d The influence of scan speed on porosity types in LPBF fabricated Ti-6Al-4V with a fixed laser power of 175 W and a hatch distance of 100 µm³⁶. b Lack-of-fusion due to insufficient penetration at a high scan speed. c A nearly fully dense microstructure density at an intermediate scan speed. d Keyhole type defects due to overheating at a low scan speed.

LPBF exhibits a unique thermal history including directional melting-solidifying against the heat transfer direction, subsequent re-melting/re-solidifying due to the following laser scan and deposited layer, and post-solidification cyclic heating/cooling^37–39. Fig. 4 is a simulated thermal history of a given location in a LPBF fabricated Ti-6Al-4V⁴⁰, and there are thermal cycles from different laser scan vectors within a layer and thermal cycles in the following layers. After solidification from the peak temperature in layer P1, the material undergoes a series of thermal cycles in the following laser scan tracks and layers. When the next track is scanned or the next layer is deposited, the previously fabricated material can be partially re-melted and re-solidified. The combined effect of high laser power, fast laser scan speed, and small laser spot size leads to an extremely high heating and cooling rate^41–44. According to an in-situ synchrotron experiment on LPBF fabricated Ti-6Al-4V, the maximum heating and cooling rates can reach the orders of 10⁶ °C/s and 10⁵ °C/s, respectively⁴⁵.

Fig. 4  Simulated thermal cycling history of a specific location in LPBF fabricated Ti-6Al-4V.

This includes intralayer and interlayer reheating and cooling. P1 to P5 are different layers⁴⁰.

During LPBF fabrication, the high cooling rate and large thermal gradients lead to high levels of residual stress⁴⁶. This can cause distortion and delamination, which results in parts failure and unacceptable dimensional change^47,48. In general, residual stress varies with the build height. Fig. 5 shows the residual stress distribution in Ti-6Al-4V cuboids fabricated on large and small substrates measured by the contour method. Specifically, the tensile residual stress presents at the bottom of the substrate and top surface of the built sample with a large compressive stress region located in between according to simulations⁴⁹ and experimental contour method⁵⁰. In the horizontal cross-section, high tensile residual stress/strain is generally located at the part edge, while compressive residual stress is at the part center⁵¹. More specifically, the tensile residual stress can exceed 900 MPa at the part edge⁵¹, which is close to the yield strength of Ti-6Al-4V.

Fig. 5  Residual stress of LPBF fabricated Ti-6Al-4V measured via contour-method in the rolling direction of the substrate.

Specimens are built on a large substrate with an overhang (left) and a small substrate without an overhang (right)⁵⁰.

Tensile residual stress is deleterious to fatigue properties, because it tends to facilitate fatigue crack propagation³. In addition to the influence on fatigue, the parts removal from substrate without stress-relieving heat treatment would introduce deformation through strain/stress relaxation^46,49. Therefore, efforts have been made to reduce the residual stress, including substrate heating during the LPBF process and post-process stress-relieving heat treatment. Furthermore, residual stress can be influenced by the in-process laser scan strategy and scan rotation between neighboring layers. A lower residuals stress was observed in the bi-directional scan than that resulting from a chessboard scan. Additionally, a scan rotation of 90° led to a reduced residual stress when compared with that in a scan rotation of 67°⁵².

The surface roughness can be controlled by the powder particle morphology and size, layer thickness, processing parameters, scan strategies, and the orientation relationship between the inclined surface and laser incident direction^53–58. In addition, the frequently used support structures adversely affect the surface finish of LPBF fabricated parts even after support removal⁵⁹.

The surface of LPBF fabricated parts is covered by partially melted powder particles⁵⁷. Hence, the powder characteristics affect the surface roughness. Given that LPBF is a layer-wise manufacturing process, a larger layer thickness leads to a rougher inclined surface owing to the staircase effect^53,54,58. More importantly, laser processing parameters can influence the melt pool morphology³⁰, which in turn affects the surface quality. A few studies employed contour scans to improve the quality of the as-fabricated surface^56,58. Other factors, including the location of the part on the substrate, the orientation relationship between the inclined surface and the laser incident direction, have also been found to influence the surface roughness⁵⁶.

Surface treatments after LPBF fabrication are generally applied to improve surface quality. The most commonly used treatments include blasting⁵⁷, shot peening⁶⁰, vibratory grinding/tribo-finishing^57,60, machining/micro-machining^18,57, and electropolishing⁶⁰. The loosely attached powder particles on the surface of LPBF fabricated parts can be addressed by blasting and shot peening; however, the blasting and/or peening media can introduce deformations on the surface⁵⁷; hence, they can only slightly reduce the surface roughness. In addition, shot peening results in work hardening and compressive residual stress in the surface layer, which prevent the fatigue crack initiation from the surface⁶¹. Vibratory grinding leads to a smoother surface, and it provides a better surface finish (R_a) of a few μm when compared with blasting and shot peening. Machining and electropolishing are effective methods to remove the surface layer, which contributes to the lowest surface finish (R_a) below 1 μm^57,60.

The directional solidification, rapid cooling after melting, and subsequent thermal cycling result in a non-equilibrium hierarchical as-fabricated microstructure of epitaxial grown prior-β grains aligned in the opposite direction of the cooling direction, and these columnar β grains consist of a few sets of fine acicular α’ martensite variants^{16,31,38,62–67}. The processing parameters affect the as-fabricated microstructure. A recent study showed that a higher laser energy density leads to a larger prior-β grain width and a smaller α’ martensite size⁶⁸. In addition to the columnar grains, the partial re-melting of the previously solidified layers leads to a strong β solidification texture. As shown in Fig. 6, a preferred {$ 100 $}_β texture along the grain growth direction has been found on the BCC orientation map reconstructed from electron back-scatter diffraction (EBSD) data of HCP Ti (α and/or α’) based on the Burgers orientation relationship^62,63. It is important to note that the β grain growth direction can be influenced by the thermal flux direction, which is affected by the laser scan speed and strategy^38,62. As a result, the growth direction of β grain can be either parallel or inclined at a certain angle to the macroscopic build direction (Z)^31,62. The pole figure in Fig. 6c is an example where the {$ 100 $}_β texture in the grain growth direction is inclined approximately 20° away from the build direction⁶².

Fig. 6  Grain orientations and textures of LPBF fabricated Ti-6Al-4V.

a Orientation maps of HCP Ti and reconstructed BCC Ti taken from a vertical plane⁶². b Orientation maps of HCP Ti and reconstructed BCC Ti taken from a horizontal plane⁶³. c The corresponding pole figures of HCP Ti and reconstructed BCC Ti of the vertical plane in a⁶².

A few studies have shown that a small amount of β can be detected by synchrotron XRD⁶⁹, EBSD⁷⁰, and TEM⁷¹ in specimens fabricated at either high laser powers or short hatch distances. Fig. 7a shows an α+β microstructure at the middle height of a sample manufactured with a short hatch distance⁷¹. Both high laser power and short hatch distance can increase the in-process temperatures achieved and the time at high temperatures during the intrinsic thermal cycling⁴⁵. These factors facilitate the deposition of metastable martensite into α+β phases. In addition to β, a few studies have reported α₂ precipitates in the as-fabricated sample based on TEM and synchrotron XRD results^38,71. Fig. 7b shows the superlattice reflection of α₂ formed in α lamella, and the formation of α₂ is believed to be related to the thermal cycling at temperatures just below the α₂ solvus. Carefully controlled Al and O compositions in the Ti-6Al-4V powders and the low O concentration in the printing atmosphere can suppress the α₂ precipitation.

Fig. 7  β and α₂ phases detected in the as-fabricated LPBF Ti-6Al-4V.

a TEM bright field image showing an α+β microstructure in the middle height (Z~5 mm) of a Ti-6Al-4V sample⁷¹. b Selected Area Diffraction (SAD) pattern of ordered α₂ within an α plate⁷¹.

In general, the as-fabricated Ti-6Al-4V has a fully α’ martensitic microstructure. As a result of thermal cycles, α’ martensite has a hierarchical structure with different width and length, which can be identified as primary, secondary, tertiary and quartic α’, and their formation mechanism has been described in detail in a recent study⁷². Within the α’ martensite, lattice defects, including dislocations, stacking faults and twins, are present^73,74. A high density of dislocations has been identified in α’ plate via TEM⁶⁹ and diffraction peak broadening phenomenon in high energy XRD⁷⁰. The TEM dislocation analysis (g·b test) revealed the presence of both <α> and <c+α> dislocations within martensite, which indicates that significant plastic deformation occurs during the LPBF process⁶⁹.

In principle, {$ 10\overline {1}1 $}<$ 10\overline {1}\overline {2} $> and {$ 11\overline {2}2 $}<$ 11\overline {2}\overline {3} $> twins can be activated when compression is applied along the c-axis, while {$ 10\overline {1}2 $}<$ 10\overline {1}\overline {1} $> and {$ 11\overline {2}1 $}<$ 11\overline {2}\overline {6} $> twins can occur when tension is applied along the c-axis for HCP titanium⁷⁵. In LPBF fabricated Ti-6Al-4V, compression {$ 10\overline {1}1 $}^76,77 and {$ 11\overline {2}2 $} twins⁶⁹, and tension {$ 10\overline {1}2 $}^69,72,78,79 twins have been reported in hexagonal α’ martensite as shown in Fig. 8. A previous work has suggested that tension {$ 10\overline {1}2 $} twins can accommodate the tensile residual stress at the top region of LPBF fabricated parts⁷⁹, but this does not explain the co-existing compression and tension twins in the same TEM sample. Some studies have shown that twinning is related to the martensitic transformation of β→α’^76,80. However, this can only explain the activation of $ \left\{10\overline {1}1\right\} $ twins. A recent study has proposed that thermal cycles of fast heating to and rapid cooling from different peak temperatures and associated solid state phase transformation can lead to the formation of both compression and tension twins in LPBF fabricated Ti-6Al-4V⁸¹.

Fig. 8  Lattice defects of twins and dislocations in LPBF fabricated Ti-6Al-4V.

a $ \{10\overline {1}1 $}<$ 10\overline {1}\overline {2} $> type twin⁷⁶. b {$ 11\overline {2}2 $}<$ 11\overline {2}\overline {3} $> type twin⁶⁹. c {$ 10\overline {1}2 $}<$ 10\overline {1}\overline {1} $> type twin⁶⁹.

In addition to hexagonal α’ martensite, orthorhombic martensite α’’ has been reported in as-fabricated Ti-6Al-4V in a few studies^66,77,78. For samples fabricated at the same LPBF parameters, α’’ was observed in one of the Ti-6Al-4V cuboids with a larger horizontal (X-Y) cross section area. The results suggested that faster cooling related to the larger heat conduction area on the horizontal cross section could lead to α’’ formation⁷⁷. However, the discussion is missing regarding the influence of horizontal cross-sectional area on the temperature achieved during the post-solidification thermal cycling. During LPBF, thermal cycling can heat some previously deposited layers to intermediate temperatures in the α+β phase field. At these temperatures, element segregation occurs and β phase is enriched in V. Thus, rapid cooling from such temperatures would result in the formation of α’’ martensite⁷⁸.

When compared to thermo-mechanically processed Ti-6Al-4V, as-fabricated LPBF Ti-6Al-4V exhibits higher strength but lower ductility due to the non-equilibrium α’ martensitic microstructure^64,65,72,82 and residual stress built up during the LPBF process^83,84. As a result, in-process measures and post-process treatments have been explored to promote the metastable martensite decomposition to equilibrium α+β phases and to minimize residual stress.

During the LPBF process, a preheated substrate generally reduces the temperature gradients within the parts and decreases residual stress⁸⁵. Substrates preheated to 370 °C and 570 °C can effectively reduce the residue stress by 71% and almost 100%, respectively, when compared to the sample built on a substrate heated to 100 °C⁸⁶. For LPBF fabricated α’ martensite, TEM and in-situ XRD investigations revealed that martensite decomposition commences at a low temperature at 400 °C⁸⁷. β phase has been observed in the as-deposited Ti-6Al-4V fabricated on a preheated substrate at 200 °C and with a tight hatch distance of 40 µm⁷¹. A substrate preheated at 500 °C results in a partially decomposed α’ martensite microstructure¹⁶, which improves elongation to approximately 10%⁸⁶. However, high temperature preheating is not energy efficient, and can account for 40% of the total energy consumed in LPBF fabrication⁸⁸. In addition, in a recent study, it was indicated that a preheated substrate at 550 °C leads to significant oxidation⁸⁹, because titanium suffers a high chemical affinity to oxygen at high temperatures. High oxygen content contributes two consequences, namely increased brittleness and deteriorated powder recyclability^90,91.

Other in-process measures, including manipulation of laser beam focal offset distance (FOD) and interlayer time, were used to increase the temperature achieved during the intrinsic thermal cycling and to promote the α’ martensitic decomposition⁹². Specifically, FOD can control the laser spot diameter and the local applied energy within the laser beam⁹³, and an increased FOD generally leads to a decreased temperature and a shortened time window for in-situ martensite decomposition during the intrinsic thermal cycling⁹⁴. Given that LPBF is a layer-by-layer deposition process, a short interlayer time of less than 12 s have been found to be effective in promoting the in-process martensite decomposition, which leads to a coarser α+β lamellar microstructure^95,96. With a suitable FOD of 2 mm, an ultrafine fine α+β lamellar microstructure has been achieved in the as-deposited Ti-6Al-4V fabricated on support structures with a substrate preheated at 200 °C⁶⁴. In addition to the FOD manipulation, all the aforementioned investigations included lattice support structures to decrease the amount of heat transferred to the substrate, thereby increasing the in-situ thermal cycling temperature. Support structures are known to introduce surface roughness even after support removal⁵⁹, which affect the as-built surface quality and its application in the as-built condition.

In general, stress-relieving treatments are carried out in the temperatures between 450 °C and 650 °C followed by air cool (AC) or furnace cool (FC) prior to the electrical discharge machining (EDM)^3,97. For instance, a 2−4 h stress-relieving treatment at 595 °C followed by AC is recommended for Ti-6Al-4V weldment³. Such a heat treatment is generally performed in a protective argon atmosphere, which reduces oxidation and the formation of α case. As shown in Fig. 9, the time required for stress relief is dependent on temperature. A higher stress-relieving temperature decreases the required heat treatment time. For example, 1 h at 595 °C is equivalent to 5 h at 540 °C in terms of stress relief effectiveness³. As the stress-relieving temperature is relatively low in the α+β field, the as-deposited microstructure will be slightly coarsened although the size and variants of primary and secondary α/α’ laths are maintained⁹⁸.

Fig. 9  Effect of stress-relieving temperatures on residual stress in Ti-6Al-4V³.

It should be noted that the α₂ solvus temperature is approximately 550 °C for Ti-6Al-4V¹, and it can be influenced by the Al and O content. A stress-relieving temperature below the α₂ solvus can lead to the precipitation hardening of α phase by α₂ particles⁷³, which is dependent on the Al and O concentration in the α phase. If the LPBF fabricated parts are used after stress relief, the stress-relieving temperature should be cautiously determined by whether α₂ formation is ideal for the application.

It is widely accepted that the α’ martensite formed in LPBF is brittle and responsible for the low elongation below 10% in the as-fabricated Ti-6Al-4V^{31,64,70,78,83,84,96}. Such a low ductility does not satisfy the elongation requirement of 10% as per the specification of AM fabricated Ti-6Al-4V in ASTM 2924–14⁴. Therefore, investigations have been focused on promoting the non-equilibrium α’ martensite decomposition in the as-fabricated microstructure to an equilibrium α+β microstructure through annealing. Previous studies have revealed that annealing in the range of 750-900 °C can effectively facilitate the α’ martensite decomposition and improve the ductility^76,99,100.

When compared with the stress-relieving heat treatments, annealing is generally performed at relatively higher temperatures in the α+β phase field and followed by AC or FC⁹⁷. A direct annealing can promote martensite decomposition into α and β phases, and the lamellar microstructure will be significantly coarser than the as-fabricated microstructure. A recent study has explicitly investigated the static coarsening behavior of lamellar structures in LPBF fabricated Ti-6Al-4V, which suggests that heat treatment temperatures at 900 °C or higher significantly accelerate the coarsening kinetics¹⁰¹.

In thermomechanically processed titanium alloys, solution treating and aging (STA) treatments are generally used to improve strength. Such a heat treatment has also been employed as a post-process treatment for LPBF fabricated Ti-6Al-4V, which can be categorized into β STA and α+β STA based on the solution temperature. As shown in Fig. 10, the β STA microstructure has significantly coarsened prior-β grains when heat treated at temperatures above the β transus, and β grains lose the columnar morphology on vertical plane^84,102. Because all the α’/α lamellae transform to β phase solid solution at such temperatures, and β grains begin to grow⁸⁴. For example, a β solution treatment at 1015 °C for 2 h led to an increase in β grain width with unchanged length⁸⁴, and a heat treatment of 2 h at 1050 °C resulted in equiaxed prior β grains in LPBF fabricated Ti-6Al-4V¹⁰². The width of α lamellae is related to the cooling rate after β solution treatment. After the β STA, there are fewer α variants left. This results in increased α colony size, which leads to a longer effective slip length for dislocations.

Fig. 10  Effect of heat treatment temperature on microstructure.

a EBSD orientation map of post-process heat treated at sub-transus 850°C/2h/FC⁸⁴. b EBSD orientation map of post-process heat treated at super-transus 1020°C/0.5h followed by sub-transus 730°C/2h/AC⁸⁴. The deposition direction in both images is in vertical direction.

When compared to β STA, α+β STA is versatile in terms of achieving different microstructures, in which the size, volume fraction, and morphology of various phases can be manipulated. A 2 h α+β solution treatment followed by a slow cooling method, such as FC, generally leads to an α+β lamellar microstructure, and a small amount of equiaxed α can form at solution temperatures close to β transus^102,103. If fast cooling methods such as AC or WQ are used, some β phases can transform to fine α_s (secondary α) or martensitic microstructure during cooling. Several studies have indicated that a bi-modal microstructure can be obtained in LPBF fabricated Ti-6Al-4V through post-process α+β STA treatments^102,104,105. In addition to LPBF, the bi-modal microstructure has been reported in similar post-process sub-transus STA treated Ti-6Al-4V fabricated by other AM techniques including electron beam melting (EBM)¹⁰⁶, direct laser deposition (DLD)¹⁰⁷, and wire arc additive manufacturing (WAAM)¹⁰⁸.

As show in Fig. 11a, the microstructure is still mainly α+β lamellae for low to intermediate solution temperatures of up to 920 °C for 2 h followed by AC. This is due to the low equilibrium volume fraction of β phase when the solution temperature is exceedingly below the β transus, and the relatively slow AC used after solution treatment. Fig. 11b−d show that increases in solution temperatures lead to an increased β_t (transformed β) region, in which fine α_s precipitates are formed. In the meantime, the volume fraction of α_p (primary α) decreases, and the aspect ratio of α_p decreases with the formation of some equiaxed α_p grains. Fig. 11 also presents the influence of solution time, and a longer solution time results in a reduced number density of α_p, a larger sized α_p and a more globular α_p morphology.

Fig. 11  Effect of sub-transus solution temperature and time on microstructure.

SEM images of LPBF fabricated Ti-6Al-4V after different α+β STA heat treatments. The results show the effect of solution temperature and time on the microstructure evolution; all the solution treatments are followed by AC and a subsequent aging treatment at 700 °C/2 h/FC¹⁰⁵.

In general, plastic deformation prior to heat treatment is essential for inducing globularization in thermomechanically processed Ti-6Al-4V. While in LPBF fabricated Ti-6Al-4V, plastic deformation is not required for the globularization to occur. Because the high dislocation density inherited from the LPBF process can initiate the globularization. The globularization mechanism is similar for LPBF fabricated and thermomechanically processed Ti-6Al-4V. Tangled dislocations in α’ martensite can rearrange into dislocation arrays and form sub-grain boundaries during the α+β solution treatment¹⁰⁷. In the initial stage of globularization, dislocation arrays and sub-grain boundaries are the feasible locations for the boundary splitting process, which includes the growth of β phase into the sub-grain boundary (thermal grooving) and edge spheroidization at lamellae ends¹⁰⁹. For a prolonged solution treatment, the globularization is driven by the second mechanism related to the α lamellae coarsening by termination migration consisting of mass transfer from curved surface to flat surfaces. This results in shorter and thicker α_p lamellae, which gradually change the lamellar morphology to a more equiaxed morphology¹⁰⁹.

In addition to α+β STA, a recent study has reported that a sub-transus cyclic heat treatment between 975 °C and 875 °C with slow heating and cooling rates for 24 h is able to provide a bi-modal microstructure with coarse globularized α_p and fine α_s within the β_t region¹¹⁰. The globularization mechanism during cyclic heat treatment differs from that mentioned in the α+β STA. The cyclic heat treatment leads to variations in the volume fraction of α and β, which promote α lamellae to β transformation in the heating cycle and facilitates β to a more equiaxed α transformation during the slow cooling cycle. During the heating cycle, the transformation from α to β also stimulates faster thermal grooving and boundary splitting processes¹¹⁰.

Typically, LPBF fabricated metals and alloys often contain defects such as gas pores, contour related pores, lack-of-fusion, keyhole pores, and partially melted powders^25,31,38,111. These defects are detrimental to the fatigue properties and can affect the quality control of the LPBF process. As shown in Fig. 12, hot isostatic pressing (HIP) is an effective post-process method for tackling defects, including lack-of-fusion, keyhole, and gas pores, but it does not address pores connected to the surface as indicated by a few defects retained after HIP in Fig. 12a. Similar evidence of pore closure in HIP have been reported in other studies^112,113. In addition, HIP can consolidate partially melted or unmelted powders¹¹¹. A few studies have revealed that post-HIP annealing, including both sub-transus and super-transus heat treatments, re-opens the HIP ‘closed’ pores^111,114,115. This indicates that these defects, containing Ar gas, are not eliminated by HIP and instead converted into small high-pressured pores, which are below the resolution of X-ray tomography.

Fig. 12  Effect of HIP treatment on defects closure in LPBF fabricated Ti-6Al-4V.

a Keyhole and gas pores closed by HIP. b Lack-of-fusion and gas pores closed by HIP. The voxel size is 5 µm for the X-ray tomography experiment¹¹¹.

The microstructure evolution during HIP is similar to that of heat-treated conditions. According to the ASTM standard for powder bed fusion fabricated Ti-6Al-4V⁴, the recommended HIP scheme corresponds to 100 MPa at 895 to 955 °C for 3 to 4 h. Several studies have performed HIP at 900−950 °C/100-150 MPa/2-4 h^{31,112,113,116}, the resultant microstructure is mainly coarse α+β lamellar microstructure containing a few equiaxed α grains. For super-transus HIP, the prior-β grain will grow and the lath thickness in the lamellar microstructure will be controlled by the cooling rate in the HIP cycle.

The most influential microstructural parameter that affects mechanical properties of LPBF fabricated lamellar structures is the α/α’ width^{35,64,76,96,117}. Under extremely fast cooled condition, the effective slip length for dislocation is limited to the individual martensite width rather than the α colony size in thermomechanically processed Ti-6Al-4V¹. The relationship between the strength and α/α’ width can be described by Hall-Petch relation as shown in Fig. 13a, in which a linear relationship can be established between the yield strength and the inverse square root of α/α’ width. Such a relationship is still valid after post-process annealing and HIP treatments in α+β phase field according to Fig. 13a. In general, the ductility exhibits an inverse relationship with the yield strength as shown in Fig. 13b. The as-fabricated LPBF Ti-6Al-4V has a fine α’ martensitic microstructure¹¹⁸, which results in a high strength but low ductility as shown in Fig. 13b and Table 1. Post-process heat treatments and HIP can promote the martensite decomposition to α+β lamellae, which reduce the strength and increase the ductility. A recent study found that heat treated LPBF Ti-6Al-4V with an equilibrium microstructure possesses a superior ductility (elongation at fracture (El.) > 18%) similar to that of wrought Ti-6Al-4V⁷⁶. Details on the martensite decomposition in post-process treatments have been summarized in the previous section of annealing. With respect to the lamellar microstructure, the overall trend is a higher heat/HIP treatment temperature (below the β transus) and/or a longer heat/HIP treatment time lead to a lower strength but a higher ductility as shown in Table 1.

Fig. 13  The effect of microstructure on tensile properties.

a A Hall-Petch relation between yield strength and the width of α’ or α laths in Ti-6Al-4V with a lamellar microstructure fabricated by LPBF^{31,54,63,64,94,112,119}, LPBF and post-process heat treatment^{16,31,120,121,60,63,76,83,84,94,103,119}, and LPBF and post-process HIP^{31,60,103,112,113,119,122}. b Yield strength vs. total elongation to failure for Ti-6Al-4V fabricated by LPBF with α’ lamellar microstructures^{[16,31,54,60,63,76,84,119,120}, α+β lamellar microstructures^{31,63,121,123,124,64,83,84,94,96,103,119,120}, and bi-modal microstructures^102,104,110.

In addition to the lamellar microstructure, another bi-lamellar/bi-modal microstructure can be achieved by post-process α+β STA and/or cyclic heat treatments which are described in the sections of solution treating and aging. As shown in Fig. 13b and Table 1, the bi-modal microstructure generally has an improved ductility without sacrificing strength when compared to the tensile properties of lamellar microstructure. The simultaneously enhanced strength and ductility are related to the fine α_s region and the coarse equiaxed α_p, respectively¹⁰⁷.

A further information that can be extracted from Fig. 13 is the scattered tensile property in LPBF fabricated Ti-6Al-4V, which can be attributed to the various specimen orientations, compositions, porosity levels, and microstructures resulting from different powder characteristics (size and composition), LPBF systems, exposure parameters/strategies, and post-LPBF treatments.

Despite the efforts on transforming a non-equilibrium α’ martensitic microstructure into the an equilibrium α+β lamellar microstructure, a few studies have reported similar ductility in a fully martensitic microstructure when compared to that in an α+β microstructure^80,126–128. In hot deformed Ti-6Al-4V, a rapid heating (> 50 °C/s) to the β phase field followed by water quenching lead to the formation of fine prior-β grains (~8 µm) containing only α’ martensite, which simultaneously improves strength and ductility¹²⁶. The increase in strength is attributed to the refinement in both martensite width and prior-β grain size, and the enhanced ductility is related to the reduced strain localization due to the shortened martensite, which is limited by the prior β grain size¹²⁶. Two similar observations have been reported in LPBF fabricated Ti-6Al-4V, in which a fully martensitic lamellar microstructure with a fine prior-β grain width between 50-100 µm results in high strength and ductility^127,128. For such an approach to achieve good tensile properties, the key is to have a small prior-β grain size¹²⁶. It is known that the prior-β grain width is related to the hatch distance in laser continuous scan strategy and the point distance in pulsed laser strategy. A small hatch distance can provide a narrow prior-β width, but it sacrifices the productivity. In addition, cautiously controlled laser processing parameters are required to avoid the β formation during the in-process thermal cycling, because the low volume of thin β film causes stress concentration, which decreases ductility^127,128.

The mechanical properties of LPBF fabricated materials have been reported as anisotropic due to the columnar prior-β grains inclined close to the build direction^113,129,130 and the manufacturing defects^16,31. In the former case, horizontally built samples generally have a higher strength but a lower ductility than those of vertically built samples, because the width of columnar β grains is aligned in the tensile direction of horizontal samples^113,129,130. The situation is changed when lack-of-fusion defects present between deposited layers and scan vectors. Vilaro et al. have reported higher elongation in horizontally built samples than that in vertically built samples in as-fabricated and post-process heat treated LPBF Ti-6Al-4V¹⁶. The reason is that tension in the vertical direction tends to open the lack-of-fusion defects and results in a low ductility. Recently, a plasticity model has been developed to predict the stress state dependent anisotropic plasticity behavior of LPBF fabricated Ti-6Al-4V¹³¹.

A recent study has revealed that β solution treatment can effectively mitigate the anisotropy in mechanical property due to the fact that β grains lost the columnar morphology after such a treatment¹³². However, the strength and ductility significantly decrease after β solution treatment, because excessive β grain growth occurs at temperature above the β transus^84,132.

For LPBF fabricated Ti-6Al-4V, the fatigue performance is influenced by the surface finish, manufacturing defects, residual stress and microstructure^133,134. The surface quality has been found to be the most crucial factor. Fig. 14b shows a rough as-built surface covered by partially melted powders. Without surface treatments, the fatigue performance of as-fabricated LPBF specimen is worse than that of the traditional cast products¹³³. After surface treatment, the fatigue strength is significantly increased in LPBF Ti-6Al-4V with^57,58,60,103 and without post-process HIP treatments⁵⁸. Fig. 14a shows that various surface treatments can improve the fatigue performance, and a reduced surface roughness of R_a = 0.3 µm (Fig. 14c), achieved by milling, improves the tension-tension (R = 0.1) high cycle fatigue (HCF) strength to 775 MPa⁵⁷, which is superior than that of wrought Ti-6Al-4V at 450–650 MPa¹³⁵. Similarly, with respect to tension-compression fatigue (R = −1), Fig. 14d shows a similar observation that an improved surface quality increases the HCF strength, and a reduced surface roughness (R_a < 1 µm) provides a HCF strength comparable to that of wrought Ti-6Al-4V. This is consistent with the fractographs shown in Fig. 14e and f, which indicate that surface related defects are responsible for the inferior fatigue strength in the as-built sample, and fatigue crack initiation site is at the sample interior when the surface defects are removed by machining. The comparable or even higher HCF strength in improved surface conditions is because the LPBF fabricated Ti-6Al-4V has a much finer microstructure than wrought and cast Ti-6Al-4V, which means a short effective dislocation slip length and a high resistance to fatigue crack initiation.

Fig. 14  Effect of surface finish on HCF strength.

a Fatigue performance of post-process HIP treated LPBF Ti-6Al-4V after different surface treatments with an axial fatigue stress of R =0.1⁵⁷. b SEM micrograph of as-built surface (R_a = 17.9 µm)⁵⁷. c SEM image of milled surface (R_a = 0.3 µm)⁵⁷. d Fatigue performance of post-process heat treated or HIP treated LPBF Ti-6Al-4V with an axial fatigue stress of R =-1^1,58. e Fatigue fractograph of a sample with as-built surface (R_a > 7 µm)⁵⁸. f Fatigue fractograph of a sample with machined surface (R_a < 1 µm)⁵⁸.

If the defects are on the surface and/or at the sub-surface which are connected to the surface, then HIP cannot close such the surface related defects and does not improve the fatigue strength of samples with an as-built surface quality as shown in Fig. 14d. In contrast, internal defects can be addressed by post-process HIP, which has been summarised in the section of hot isostatic pressing. Fig. 14d shows that HIP improves the fatigue performance of machined samples (R_a < 1 µm), which is through eliminating the internal defects. In addition, HIP can effectively narrow down the fatigue data scatter⁶⁰. As described in the section of residual stress, tensile residual stress, which accelerates the fatigue crack propagation and is detrimental to the fatigue properties³, presents at the top and side surfaces of the part⁵⁰. Therefore, surface peening treatments have been used to tackle the tensile residual stress, which result in a compressive stress below the surface and improves the fatigue strength^60,136.

For the influence of microstructures, previous investigations are mainly focused on the lamellar microstructure. In general, a finer microstructure has a higher fatigue strength. A sub-transus HIP treatment at 920 °C shows a better fatigue performance than the super-transus HIP treatment at 1050 °C¹¹⁶. Similar observations has been reported in the heat treated conditions¹¹⁶. Whereas, the situation is different for fatigue crack growth (FCG). In the hypo-transitional region, FCG is generally influenced by the microstructure. A coarse microstructure has a higher fatigue crack threshold stress intensity factor (ΔK_th) and a slower FCG rate¹³⁷. At the threshold range, a coarser lamellar microstructure displays a stronger ability to deflect cracks along the lath boundaries^137,138. Even at a higher ΔK, crack deflection can be controlled by the colony/packet size¹³⁸. On the contrary, the cyclic plastic zone is sufficiently large in the hyper-transitional region, which exceeds the colony size, thus FCG is insensitive to the microstructure¹³⁸.

As described in tensile and fatigue properties, a fine microstructure generally provides high yield and HCF strengths. In contrast, a fine microstructure leads to a low fracture toughness¹. For LPBF Ti-6Al-4V, the fracture toughness has been reported at values below 30 MPa∙m^−1/2 in the as-fabricated condition with α’ martensitic microstructure¹³⁹. This is lower than that of the wrought Ti-6Al-4V with a lamellar microstructure at 45-75 MPa∙m^{−1/2 1,3}. This is due to the extremely fine (sub-micro) and brittle α’ in the as-fabricated condition. It has been found that post-process stress-relieving and annealing can improve the fracture toughness to approximately 50 MPa∙m^{−1/2 139,140}, which is comparable to that of wrought alloys. In addition, anisotropic fracture toughness has been observed in both as-fabricated and post-process stress-relieved/annealed conditions, which is related to porosity levels and columnar β grains aligned along the build direction. A recent study has showed that a two-stage heat treatment of α+β STA effectively improves the fracture toughness to approximately 100 MPa∙m^{−1/2 138}. A short solution treatment at 920 °C followed by aging results in a coarse α+β Widmanstätten microstructure, and this microstructure can provide two benefits in terms of improving the fracture toughness. One is the α+β microstructure is more ductile than the α’ martensitic microstructure, and the other is a coarser lamellar microstructure means a rougher crack front profile and an increased crack tortuosity^1,138.

In the temperature range between 450 °C and 650 °C, the as-fabricated Ti-6Al-4V has a comparable creep performance with that of the wrought Ti-6Al-4V¹⁴¹. After post-process heat treatments, a Widmanstätten microstructure showed a relatively lower creep strain and a lower steady-state creep rate than that in the as-fabricated condition. This indicates that the heat treated condition has a higher creep resistance¹⁴². This is due to the fact that the as-fabricated martensitic microstructure can be decomposed at creep testing temperatures, and such a microstructure evolution reduces the creep resistance¹⁴². The α/β interfaces in Widmanstätten microstructure act as obstacles to the dislocation glide during the creep deformation, which also improves the creep resistance^143,144.

In titanium alloys, the corrosion resistance is attributed to the protective surface oxide film¹. Therefore, the surface quality affects the corrosion performance of LPBF fabricated Ti-6Al-4V, and surface treatments can effectively improve the corrosion resistance by surface machining¹⁴⁵ and electropolishing¹⁴⁶. For the influence of microstructure, metastable α’ martensite is regarded as less corrosion resistant than the α phase, and the V-containing β phase is more corrosion resistant than α’ martensite¹⁴⁷. Therefore, as-fabricated LPBF Ti-6Al-4V has an inferior corrosion resistance (in 3.5 wt.% NaCl aqueous solution at room temperature) when compared with conventionally processed Ti-6Al-4V with an α+β lamellar microstructure due to the non-equilibrium α’ martensitic microstructure^147,148. The passive film thickness has been found to be thinner in laser fabricated Ti-6Al-4V than that in wrought counterpart, which also indicates a faster corrosion rate¹⁴⁹. In terms of anisotropy, there is a negligible difference in corrosion properties between vertical and horizontal planes in 3.5 wt.% NaCl aqueous solution, while anisotropic corrosion behavior has been observed in a harsher environment of 1 M HCl aqueous solution, and the authors have related the anisotropy to different β volume fractions on different orientations¹⁵⁰.

After post-process heat treatments, previous studies showed two distinctly different corrosion phenomena. A recent study has reported that both sub-transus and super-transus heat treatments deteriorated the corrosion resistance in 3.5 wt.% NaCl aqueous solution at room temperature due to the increased grain size¹⁵¹. While, other studies have suggested that post-process heat treatments improve the corrosion resistance in Ringer’s solution at 37 °C¹⁵² and in 3.5 wt.% NaCl aqueous solution at room temperature¹⁴⁷. This is attributed to the limited element segregation in the μm-sized α+β lamellar microstructure after the heat treatment, which mitigates the galvanic corrosion between α and β phases¹⁴⁷. Nevertheless, further investigations on the corrosion performance of SLM-produced Ti-6Al-4V are required to understand the influence of post-process heat treatments.

In this paper, the LPBF process, post-process treatments, microstructures, and properties of LPBF fabricated Ti-6Al-4V have been briefly reviewed. At the beginning, the influences of processing parameters on the defect formation, residual stress, and surface roughness are presented. This is followed by the as-fabricated microstructure and its evolution during the in-process and post-process treatments. Lastly, the effects of microstructure, defects, and surface roughness on tensile, fatigue, fracture toughness, and creep properties are discussed. A summary of specific findings is provided below:

● The formation of defects and surface quality can be controlled by tuning the processing parameters and scan strategies.

● The residual stress and pores in the as-fabricated Ti-6Al-4V can be addressed via post-process HIP treatments.

● The as-fabricated Ti-6Al-4V generally has a hierarchical microstructure containing textured prior-β grains that are aligned close to the build direction and intragranular hexagonal α’ martensitic microstructure. In addition, a few studies have observed a small amount of β, α₂, and orthorhombic α’’ martensite in the as-fabricated condition.

● The microstructure evolution, including martensite decomposition and formation of lamellar and/or bi-modal α+β microstructures, can be realized via in-process thermal cycling and post-process treatments.

● For the lamellar microstructure, the yield strength follows a Hall-Petch relationship with the lamellar width in the as-fabricated and post-process treated samples. The typical trade-off between strength and ductility is observed within the sample with a lamellar microstructure. In contrast, the bi-modal microstructure can provide a simultaneous enhancement in strength and ductility.

● After surface treatments, the fine microstructure in as-LPBF and post-process treated conditions leads to a high HCF strength. In contrast, a coarse α+β Widmanstätten microstructure is preferred for obtaining good fracture toughness and creep properties.

● In terms of corrosion resistance, the as-fabricated Ti-6Al-4V has an inferior corrosion performance than that of wrought counterpart, which is related to the metastable α’ martensitic microstructure. A more detailed understanding of the influence of post-process heat treatments on the corrosion behavior is required.

● Most post-process treatments, including heat treatments, HIP, and surface treatments, lead to comparable or even superior tensile, fatigue, fracture toughness, and creep properties than those of cast and/or wrought Ti-6Al-4V.

In principle, LPBF fabricated Ti-6Al-4V has great advantages in preparing complex near-net shape parts and reducing lead times. However, there is still a gap between the current research and the aforementioned advantages. For example, surface roughness and internal defects require post-process surface treatments and HIP to address. Additionally, disadvantages of anisotropic microstructure and properties must be considered, and the mitigation of anisotropy requires further investigation.

In terms of microstructure and properties, LPBF fabricated Ti-6Al-4V possesses a fine microstructure, which offers great advantages, including superior tensile and HCF strengths. However, a fine microstructure is not favorable for fracture toughness and creep resistance. A careful microstructure design through post-process treatments is required for specific applications.

In contrast to the inverse relationship between tensile strength and ductility in the lamellar microstructure, the bi-modal microstructure shows a great potential in terms of achieving satisfied strength and ductility. In addition, the bi-modal microstructure shows a higher HCF strength, a slower fatigue crack propagation rate of microcracks, and a higher LCF strength in wrought and cast alloys. As a two-step α+β STA treatment is generally used to obtain a bi-modal microstructure, the alloy element partitioning effect inevitably occurs. Future studies are required to understand the effect of strength difference between the initially formed α_p and the subsequently formed α_s on properties at a fine scale in LPBF fabricated Ti-6Al-4V.

This work is funded by Australian Research Council IH130100008 “Industrial Transformation Research Hub for Transforming Australia's Manufacturing Industry through High Value Additive Manufacturing”. The authors gratefully acknowledge useful discussions with Kai Zhang and Jianwen Liu from the University of Shanghai for Science and Technology.