Active Slip Mode Analysis of an Additively Manufactured Ti-6Al-4V Alloy via In-Grain Misorientation Axis Distribution

Abstract

Selective laser-melted (SLM) Ti-6Al-4V alloy was quasi-statically compressed in the transverse and longitudinal directions at a strain rate of 1 × 10⁻³ s⁻¹ at room temperature. The twinning, in-grain misorientation axis (IGMA) distribution and active slip modes of individual grains in the deformed SLM Ti-6Al-4V alloy were studied in detail via transmission Kikuchi diffraction (TKD) and transmission electron microscopy (TEM). The α’/α phase was textured with the c-axis oriented either at ~45° or perpendicular to the building direction (BD). A combined analysis of the IGMA distribution and Schmid factor revealed that the prismatic <a> slip or pyramidal slip was easily activated in the soft grains with their c-axes perpendicular to the BD (or the loading direction) in the longitudinal compressed sample, while slip was hardly activated in the transverse compressed sample due to the lack of soft grains. Prismatic <a> slip with IGMA around <0001> Taylor axis also occurred in {10–11} twins. The observations revealed that the prismatic <a> slip played a key role in accommodating the external strain and, thus, well explained the anisotropy of mechanical properties in the SLM Ti-6Al-4V alloy.

1. Introduction

Additive manufacturing (AM) or 3D-printing is a promising material processing method that has gained significant momentum in the aerospace, medical and other fields [1,2], because it has the advantage of manufacturing parts with precision or special purposes, such as engine blades and shape memory alloys [3]. It is also flexibly used for depositing carbide coatings to improve the wear resistance of alloys [4,5]. The ever-expanding application of AM attracts more emerging technologies such as machine learning (ML) [6] and algorithms including topology optimization [7,8]. As a result, the designs have been significantly optimized for lightweight components and structures [9] and manufacturability [10,11].

AM Ti-6Al-4V alloy has recently attracted extensive interest due to its excellent properties [2,12,13,14,15,16,17,18,19,20]. The significant anisotropy of tensile strength, ductility and fatigue performance is frequently observed in the SLM Ti-6Al-4V alloy, which restricts its applications. It is widely accepted that the manufacturing process leads to special crystallographic textures, which have a direct link to the mechanical anisotropy controlled by the slip mode. Therefore, it is necessary to identify the slip mode of SLM Ti-6Al-4V samples deposited in different directions.

Several analytical techniques have been used to explore the slip mode in α-Ti and other hcp metals, but with different degrees of limitations. The Burgers vector analysis based on TEM requires a long time for testing and processing data, and it is not suitable for large strains with complicated dislocations. The trace analysis is mainly used for single crystals or coarse grains in polycrystalline materials during deformation under uniaxial loading. The common limitation is that only the deformation characteristics of extremely limited grains with obvious active slip or twinning could be shown, which does not directly reflect the whole material. The X-ray diffraction (XRD) or neutron diffraction technique of polycrystalline materials provides statistically averaged deformation features of a large area, but it is unable to explain how the slip of individual grains affects the overall deformation behavior. Therefore, IGMA analysis based on the electron backscatter diffraction (EBSD) technique was proposed by Chun [21,22] to build relationships between the overall deformation and every grain with chaotic orientation for pure titanium and magnesium alloy.

IGMA analysis has been more widely used in alloys with a hcp structure with the development of EBSD technique. The IGMA was mainly used to identify the slip mode of individual grains in different conditions, such as with specific orientations [23,24,25,26,27] by thermomechanical processing, in specific zones of welded joint [28], and with different concentrations of rare earth elements [29]. Furthermore, Yamasaki et al. [30,31] applied IGMA to quantitatively evaluate “kink deformation” and classify kink bands, and explained the role of kinking bands in improving the deformability by means of stress relaxation and crystal reorientation in grains with limited active slip systems [32,33]. Qiang et al. [34] extended IGMA not only to the limited slip systems at room temperature, but also linked it to the continuous dynamic recrystallization (CDRX) of high-temperature α phase in TiAl alloys.

However, the fine microstructures formed due to rapid cooling rate in SLM Ti-6Al-4V are difficult to be identified using the traditional EBSD technique; thus, there is currently no study on applying IGMA analysis to SLM Ti-6Al-4V. In the present study, the TKD technique is used to obtain crystallographic orientations of individual fine grains. The relationship between the slip mode and the special crystallographic texture is also discussed to explain the anisotropy of mechanical properties in the SLM Ti-6Al-4V.

2. Materials and Methods

The starting material was spherical Ti-6Al-4V powders (in wt%: 6.54 Al, 4.10 V, 0.16 Fe, 0.087 O, 0.016 N, 0.007 C, 0.006 Si, 0.003 H, balance Ti), with the D10, D50 and D80 of the particle size distribution parameters being 20.2, 30.7 and 46.8 μm, respectively. In this study, the NRD-SLM-300A SLM equipment (Techgine Laser Technology Co., Ltd, Shanghai, China) and IPG 500W laser (IPG Photonics Corporation, Oxford, MA, USA) were used. A laser beam of 350 W, scanning velocity of 0.8 m/s, hatch spacing of 0.08 mm and layer thickness of 0.04 mm were applied during SLM (Figure 1a). The laser beam scanning was set to rotate at an angle of 105° for each of the following layers.

The cylindrical samples with a dimension of Φ8 mm were directly deposited on the substrate in an argon environment with the oxygen content below 0.1% in wt.%. Then, the samples were removed from the substrate by wire cutting and divided into Φ8 × 12 mm, which was suitable for compression, as shown in Figure 1b.

To identify the anisotropic mechanical properties in the as-built condition, two types of samples with the long axis along the transverse (X-direction in Figure 1b) and longitudinal (Z-direction in Figure 1b) directions were compressed in a Gleeble-3800 thermal simulator (DSI company, St. Paul, MN, USA) at a strain rate of 0.001 s⁻¹, at room temperature, to a strain of 0.15.

The microstructure was examined via optical microscopy (OM, Carl Zeiss Axios 5.0, Carl Zeiss Company, Oberkochen, Germany), TKD (Gemini 300 microscope equipped with an Oxford Symmetry S2 EBSD detector, Carl Zeiss Company, Oberkochen, Germany) and TEM (Tecnai G2 S-Twin F20, FEI Company, Hillsboro, OR, USA). XRD (Rigaku D/Max-2550, Rigaku Corporation, Tokyo, Japan) measurements at 2°/min were performed to determine the phases and orientation. The specimens for microstructural examinations and XRD measurements were taken from the mid-plane of the cylindrical samples, as shown in Figure 1b.

It is convenient to study the crystallographic orientation of individual grains via EBSD. However, conventional EBSD cannot be used to study nanoscale grains in SLM Ti-6Al-4V alloy without heat treatment because of its limited spatial resolution. In this study, the TKD technique of thin film materials was applied to study deformed SLM Ti-6Al-4V. The thin film specimens for TKD were prepared in the same way as the TEM specimens: mechanically grinding to a thickness of ~80 μm followed by twin-jet electropolishing at 50 V and 238 K in a solution of 6% perchloric acid, 34% N-butanol and 60% methanol. A fine step size of 10 nm for the TKD scan was used to obtain more details, and the data were processed using HKL Channel 5 software (Oxford Instruments, Oxford, UK). The microstructures characterized by TKD were presented as orientation maps (or inverse pole figure (IPF) maps) where the color corresponded to the crystallographic directions relative to the BD and Schmid factor distribution maps. To eliminate spurious boundaries caused by in-grain deformation and noise, a boundary misorientation of less than 2° was set as a cutoff. Grain boundaries with a misorientation of larger than 15° were defined as high-angle grain boundaries (HAGBs) and those with a misorientation between 2° and 15° were defined as low-angle grain boundaries (LAGBs), which were depicted as black and white lines, respectively.

Furthermore, the data on the IGMA distributions due to the deformation-induced crystal misorientation in individual grains were extracted and presented by projections with three-pivot orientations. Generally, grains with maximum intensities higher than 2 multiples of a random distribution (MRD) were considered to have preferential IGMA, while those with maximum intensities of less than 2 MRD were considered to have a uniform IGMA [21,22].

3. Results and Discussion

The microstructure of the as-built SLM Ti-6Al-4V was inhomogeneous with columnar crystals about 60–100 μm wide along BD (Z-direction), as shown in Figure 1c. A large amount of acicular α/α′ phase was distributed inside the columnar crystal, which originated from the rapid cooling of SLM. Figure 1d–g shows that the twin martensite structure was widely present in the α phase, corresponding to the {10–11}<10–12> twins.

The {10–11} twins in titanium alloys were a kind of high-temperature compression twin, which have been reported in quasi-static compression deformation above 400 °C [35] and equal channel angular pressing (ECAP) at 250 °C [36], and also appeared in the shear zone under dynamic compression at room temperature [37] due to the local high temperature. In addition to high temperature deformation, {10–11} twins were also produced during the quenching of Ti-6Al-4V alloy from 1150 °C [38], which was related to the rapid cooling rate being higher than 10⁴ K/s. The Ti-6Al-4V alloy with martensite α′ phase also formed twin martensite under cyclic tension and compression loads at room temperature [39], because the original α’/α’ interface and the internal stress introduced by cyclic loads reduced the activation energy of twin nucleation. The morphology of twin martensite under cyclic loads was consistent with the observation in this study. The initial martensite α/α′ phase, a large amount of internal stress generated in the rapid cooling process and the thermal cycles of layer-by-layer deposition facilitated the nucleation and growth of {10–11} twins.

Figure 2a shows the XRD patterns of the original and compressed samples. The peaks corresponded well to the characteristic peaks of α phase (hcp) and there was no obvious peak corresponding to the β phase (bcc). The Full Width at Half Maximum (FWHM) values of the strongest peaks (10–11)_α of the three samples were all close to 0.4° of 2θ, being much higher than the FWHM of α phase 0.2° of 2θ [40], and it could be inferred that martensite α’ phase should have been present, since the lattice parameters differed slightly from those of α phase along with obvious peak broadening [40].

Figure 2c shows the stress–strain curves of the transverse (blue) and longitudinal (red) samples. It is clear that SLM samples were anisotropic with the transverse sample exhibiting a higher strength. Figure 2b shows the cross-sectional microstructures of the longitudinal and transverse cylindrical compressed samples, respectively. The longitudinal sample retained the columnar grains along the BD, being consistent with the as-built sample (Figure 1c), but the columnar grains of the transverse sample were broken.

For hcp materials, the orientation changes of a single crystal or individual grains in a polycrystal were strongly associated with the active slip systems, in addition to twinning. During deformation, all crystallographic planes in an individual grain rotated about an axis (called the Taylor axis), and the Taylor axis was in a crystallographic direction lying in the slip plane and perpendicular to the slip direction [21,33,41]. By extracting the orientation information of all lattice points belonging to a special deformed grain from the TKD data, the density of all possible misorientation axes for neighboring points could be determined. Based on the method, the number of Taylor axes corresponding to various deformation modes can be found and listed in

Table 1. Typical slip modes and the corresponding Taylor axes in α-Ti [21,23,26].

Slip Mode	Number of Slip Systems	Taylor Axis	Number of Variants of Taylor Axis
{01–10}<−2110>	3	<0001>	1
{0002}<−2110>	3	<0−110>	3
{01–11}<0−112>	6	<2–1–10>	3
{01–11}<−2110>	6	<0–112>	6
{01–11}<−1–123>	12	<13–8–53>	12
{11–22}<−1–123>	6	<1–100>	3
{−12–11}<−1–123>	12	<6−1–53>	12

.

The distribution patterns of IGMA in the longitudinal compressed samples are presented with the microstructures and microtextures of individual grains in Figure 3. In addition to a step size of 10 nm, selecting a reasonable distribution range of misorientation angles also affected the accuracy of the IGMA analysis. The distribution of the fraction and the accumulated fraction of neighboring points of the entire TKD data was plotted against the misorientation angle, as shown in Figure 3b. It is seen that the 95% misorientation angles were both below 0.85°. These misorientations were considered to be related to the limited angular resolution of the diffraction techniques [23]. In view of the fact that the grains with different orientations caused different levels of influence on diffraction [21,23,27], the minimum misorientation angle included in the IGMA statistics was taken as 1.05°. The maximum misorientation angle was usually taken as 2°, because the larger misorientation may have been a result of shear bands or kink bands during deformation and was understood as LAGBs or sub-grain boundaries, whose corresponding Taylor axes did not display information about activated slip modes.

Figure 3a presents the IPF map with respect to the BD, and shows that the α/α′ phase exhibited the typical orientation of SLM titanium alloys: the c-axes (i.e., the <0001> orientation of the hcp structure) oriented either at ~45° or perpendicular to the BD [42,43,44,45,46,47]. This was also corroborated by the pole figures. Some typical grains with different orientations marked by letters in Figure 3a were selected, and their IGMA distribution is presented in Figure 3c. It was observed that each grain underwent lattice rotation to a small extent and their distributions of IGMA and intensity were dependent on grain orientation. For example, in some grains (grains F, G and H) with the c-axes perpendicular to the BD (green in Figure 3a), the distribution of IGMA was concentrated toward the <0001> axis and the maximum intensities were much higher than 2 MRD (8.38, 3.32 and 8.73 MRD, respectively). In other grains (grains D and I) with the c-axes perpendicular to the BD, the distribution was concentrated toward the <10–10> axis and the maximum intensities were slightly higher than 2 MRD. In addition, some grains (grains A, B, C and E) with the c-axes oriented at ~45° to the BD (pink in Figure 3a) exhibited relatively uniform distribution with their maximum intensities being lower than 2 MRD.

By matching the Taylor axis to the IGMA values listed in Table 1, the deformation modes of some grains (grains F, G and H) could be attributed to the predominant activation of prismatic <a> slip with <0001> as their only significant Taylor axis. At the same time, the slip modes of some other grains (grains D and I) could be attributed to the predominant activation of basal <a> slip or pyramidal<c + a> slip because the distribution of IGMA was a weak agglomeration near the <10–10> Taylor axis. For other selected grains, it was not possible to accurately predict their slip modes in the same way, even if their IGMA was distributed close to some Taylor value; for example, grain B showed agglomeration on the <10–12> axis but with an intensity of only 1.76 MRD.

Prismatic <a> slip was regarded as the main slip mode in hcp metals, such as α titanium and α zirconium [23,48,49,50], because it had the lowest critical resolved shear stress (CRSS) in the α phase. In general, the c/a ratio in the unit cell of hcp materials would reflect the slip modes. The c/a ratio of ~1.587 for near-α titanium alloys was less than the ideal c/a ratio of 1.633; thus, the prismatic <a> slip was more likely to occur than other slip modes [51,52,53]. The deformation generated by the prismatic <a> slip was on the (10–10) plane with a Burgers vector of 1/3 [11–20]; thus, two arbitrarily chosen neighboring points in this deformation in the TKD measurement should be misoriented about the <0001> direction, which is in the (10–10) plane and perpendicular to the [11–20] direction. This method was proven to be supported by an original dislocation model [21], which was used to explain the formation of tilt boundaries and extended for the deformation substructures in homogeneous grains.

Figure 4a shows the Schmid factor distribution of the prismatic <a> slip in the same area as Figure 3a, with respect to the loading direction. Prediction of the activation of slip systems based on the Schmid factor was fully in agreement with the IGMA analysis. The grains (grains F, G and H) with <0001> as their Taylor axis had the largest Schmid factor of prismatic <a> slip. These grains, with easy activation of prismatic <a> slip, were called soft grains, which often appeared in low-temperature severe plastic deformation (SPD) [54], contributing largely to the coordinated deformation.

Regarding the weak intensity of IGMA around the <10–10> axis, it might have been caused by basal <a> slip (i.e., {0002}<−2110> slip) or {11–22} <−1–123> slip, which did not easily occur at room temperature. The basal <a> slip could be easily excluded by the distribution of the Schmid factors (Figure 4b), which were extremely low for the corresponding grains, approaching pure blue. In fact, the CRSS value of basal <a> slip was reported to be obviously higher than that of prismatic <a> slip [55,56,57]. In contrast, the same grains showed higher Schmid factor values of {11–22} <−1–123> slip in comparison with the surrounding grains (Figure 4c). However, pyramidal <−1–123> slip was observed on the {10–11} plane more frequently than on the {11–22} plane [56,57,58], and the Taylor axis of the former slip mode was <13–8–53>. This was not inconsistent with the results of this study, because a weak intensity of IGMA distribution around the <13–8–53> axis was also found in grains D and I. This further showed that pyramidal slip plays a leading role in this type of grain.

If different slip modes appeared in the same grain, a uniform distribution of IGMA would have been possible. In addition, the presence of slip systems that belonged to the same slip mode but on different planes could have led to the dispersed distribution of IGMA, because the latter slip would tilt the original misorientation axes in neighboring areas. This means that the IGMA analysis was not suitable for body-centered-cubic or face-centered-cubic materials and pyramidal <c + a> slip in hcp materials, because the former had a large number of independent slip systems [21,27] and the latter had more slip variants than prismatic slip [21]. With this limitation, the IGMA approach was more suitable for the analysis of the deformation of hcp materials at low temperatures, because the pyramidal slip would basically activate at high temperatures.

Indeed, the XRD analysis depicted in Figure 2a also pointed to the more pronounced prismatic <a> slip and pyramidal slip during the deformation of longitudinal compressed sample, since the peak of the prismatic plane (10–10)_α and the pyramidal plane (10–11)_α exhibited a higher relative intensity than the original and transverse compressed samples [15], as summarized in

Table 2. Intensity values of main peaks of XRD analysis in Figure 2a.

Samples	Intensity Values of (10–10)α	Intensity Values of (0001)α	Intensity Values of (10–11)α	Relative Intensity of (10–10)α/(0001)α	Relative Intensity of (10–11)α/(0001)α
Original	580	2186	4684	0.265	2.143
Longitudinal	177	595	1761	0.297	2.960
Transverse	1364	5187	13255	0.263	2.555

.

Although some grains had a similar orientation with their c-axes perpendicular to the BD, they showed completely different slip modes, such as grains F, G and H with prismatic <a> slip and grains D and I with pyramidal slip. This could be inferred from the morphologies of the grains, where they were all lath-shaped but the long axis of the laths was inclined and perpendicular to the loading direction, respectively. Prismatic <a> slip dominated the deformation and caused the corresponding soft grains to produce a large amount of strain, but the strain of the grains perpendicular to the load direction was more restricted by the surrounding grains. The grains had to activate pyramidal slip (meaning more slip systems) or twinning, and even grain break occurred [59,60].

The appearance of soft grains in the longitudinal compressed sample was due to the c-axes of such grains being perpendicular to the loading direction, which was the BD, and the <11–20> axes parallel to the loading direction. However, for the grains with c-axes perpendicular to the BD in the transverse compressed sample, their c-axes were parallel to the loading direction; thus, they did not meet such a condition for soft grains. As a result, the SLM Ti-6Al-4V samples exhibited anisotropy to some extent, as shown in Figure 2c, where the transverse sample showed a higher strength while the longitudinal sample yielded earlier. This also corroborated the findings that the SLM titanium alloy has better plasticity in the BD, which were reported in several studies [61,62,63,64].

A large number of {10–11} twins were distributed in the as-built samples, as shown in Figure 1d–g, and the twins in the sample compressed to a strain of 0.15 were identified by TKD technology with a resolution of 10 nm. The IGMA analysis was also carried out to explore the influence of the existing twins during deformation, as shown in Figure 5. The minimum misorientation angles was taken as 1.05° according to the accumulated fraction of neighboring points, as shown in Figure 3b. Compared with the matrix (grains A, C, E, G and I in Figure 5a), the twins B, D, F and H showed a typical IGMA distribution of the <0001> Taylor axis, which corresponded to prismatic <a> slip, being similar to the soft grains in the longitudinal sample. Slip was activated in twins, which suggested that twinning could help coordinate local strains.

4. Conclusions

The microstructures of as-built and compressed SLM Ti-6Al-4V alloy were studied. The IGMA distribution, slip mode and twinning of individual grains were systematically analyzed to elucidate the deformation mechanisms. The following conclusions could be drawn:

(1)

Due to the rapid cooling and thermal cycling in the SLM process, a large number of {10–11} twins appeared in the as-built SLM Ti-6Al-4V alloy, which contributed to the accommodation of local strains.

(2)

In the longitudinal compressed sample, the IGMA distribution and slip modes of individual grains were directly related to the grain orientation and morphology. The grains with their c-axes perpendicular to the loading direction (i.e., the BD) were observed to facilitate the activation of dislocation slip, so they were called soft grains. The long axes of lath-shaped soft grains were inclined or perpendicular to the loading direction, which determined their tendency to activate prismatic <a> slip or pyramidal slip, respectively.

(3)

The presence of soft grains rationalized the mechanical property anisotropy of the SLM Ti-6Al-4V alloy: Soft grains in the longitudinal compressed sample played an important role in accommodating the external strain. In contrast, in the transverse compressed sample, the grains with their c-axes parallel to the loading direction (which was perpendicular to the BD) did not satisfy the conditions for soft grains. Therefore, the longitudinal compressed sample showed better ductility.