- Email: [email protected]
Microstructure and anisotropic tensile behavior of laser additive manufactured TC21 titanium alloy
Materials Science & Engineering A 673 (2016) 204–212
Contents lists available at ScienceDirect
Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea
Microstructure and anisotropic tensile behavior of laser additive manufactured TC21 titanium alloy Qiang Zhang, Jing Chen n, Zhuang Zhao, Hua Tan, Xin Lin, Weidong Huang State Key Laboratory of Solidification Processing, Northwestern Polytechnical University, Xi’an, Shaanxi 710072, China
art ic l e i nf o
a b s t r a c t
Article history: Received 9 March 2016 Accepted 9 July 2016 Available online 11 July 2016
In the present work, two types of part with different geometries were fabricated by laser additive manufacturing (LAM) process under the same processing parameters. Tensile samples perpendicular to (horizontal) and parallel to (vertical) the build direction were produced after heat treated under different conditions. The results showed that the samples exhibit similar columnar β grains, but very different α phase characterizations. The tensile properties show significant anisotropic. The sample containing finest α laths shows highest ultimate tensile strength and yield strength. And the strength decreased with the increase of α laths width. The total elongation of the samples does not show regularity relationship with the strength. The horizontal samples show inferior ductility and brittle fracture due to the columnar β grain morphology and the presence of continuous grain boundary α layers. The coarse grain boundary α layers after double heat treatment further reduced the ductility. The vertical samples exhibit better ductility due to lack of continuous grain boundary α layers which perpendicular to the tensile load. The Vickers hardness test shows that the vertical samples were strengthened while the horizontal samples were little strengthened after tensile test. & 2016 Elsevier B.V. All rights reserved.
Keywords: Laser additive manufacture Microstructure Anisotropic tensile behavior Titanium alloy
1. Introduction Titanium alloys are widely used in the aeronautical industry for their high specific strength, excellent mechanical properties and outstanding corrosion resistance. However, the end products of titanium alloys are quite expensive due to the difficulties in refining, casting, forming and machining process [1]. Laser additive manufacturing is a kind of advanced processing technology that can be used to fabricate near fully-dense complex metal parts [2]. The LAM process has been recognized as an affordable and efficient solution that can meet the stringent requirements for aerospace titanium alloys components [3–5]. In order to achieve mechanical properties comparable or even superior to the traditional manufacturing, it is crucial to get a better and deeper understanding of the relationships between the processing parameters, microstructure evolution and the resulting mechanical properties. During the LAM process, the deposited layers undergo melting, rapid solidification, partial remelting and reheating. Therefore, the LAM process resulted in a specific microstructure. It has been well documented that during solidification of the molten pool, the solid substrate acts as a heat sink and the as-deposited microstructure generally presents a directional solidified morphology grown from n
Corresponding author. E-mail address: [email protected] (J. Chen).
http://dx.doi.org/10.1016/j.msea.2016.07.040 0921-5093/& 2016 Elsevier B.V. All rights reserved.
the substrate epitaxially [6]. As a result, typical prior β grain morphology of LAMed α þ β titanium alloys usually comprised of coarsen columnar grains which exhibit a strong 〈100〉 fiber texture [7–9]. The characterization of α phase precipitated in the β phase shows more complex features because the deposited layers experienced consecutive thermal cycles with a duration and amplitude which depend on the processing parameters and geometry of the part being fabricated [10]. Fine lamellar structure containing martensitic α′ [11], layer bands [12] and coarsen lamellar structure [13] have been reported. This specific microstructure characterization strongly effects the mechanical properties of LAMed titanium alloys. Generally, the LAMed titanium alloy components usually present anisotropic mechanical properties. The strongly textured prior β grains which growth along the build direction is regarded as the main reason for anisotropic mechanical properties [14–16]. Vilaro [17] found that the pores shape and orientation in the deposited Ti6Al4V components strongly influence the ductility. Qiu [18] found that the presence of planar pores due to incomplete remelting of previous layers in LAMed Ti6Al4V alloy also caused pronounced anisotropy in ductility. Carroll [4] studied the anisotropy tensile behavior in LAMed Ti6Al4V alloy cruciform component. They reported better ductility along the build direction in comparison to the direction perpendicular to the build direction. They believed that the columnar β grains and grain boundary α phase are the main reason for different ductility. The TC21 titanium alloy (Ti6Al2Sn2Zr3Mo1.5Cr2Nb, β transus
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
temperature is about 960 °C [19]) investigated in the present study is a α þ β titanium alloy. Due to the good balance of strength, ductility and fracture toughness, the TC21 titanium alloy has shown a good promising for applications in aerospace industry [20]. The previous study showed that the microstructure of LAMed TC21 titanium alloy is more inhomogeneous than the Ti6Al4V alloy due to the higher alloying elements [21]. Similar inhomogeneous microstructure could also been observed in other LAMed titanium alloys which contains more alloy elements [22,23]. However, there has been little effort to understand the connection between the mechanical properties and the more complex microstructure characterization. The aim of this study is to correlate the microstructure of LAMed TC21 titanium alloy to the corresponding tensile properties. In the present work, two types of part with different geometries were deposited under the same LAM processing condition. And then the tensile properties parallel to and perpendicular to the build direction after two different heat treatments were tested. Based on the experiment results, the anisotropic tensile behavior and the relationship with the microstructure characterization were discussed.
2. Experimental procedures The samples were fabricated by a LAM equipment that consists of a 4 kW continuous wave CO2 laser, a 5-axis numerical control working table, a coaxial powder feeder nozzle and an inert gas chamber filled with pure argon with oxygen content below 50 ppm. TC21 spherical powders produced by the plasma rotating electrode process (PREP) were used as the deposited material. The powder size ranges from 80 mm to 150 mm. The powders were dried in a vacuum oven for 2 h at 12075 °C to eliminate the moisture absorption and ensure the powders have good flow ability. The forged TC21 plates were used as substrate for the LAM process. The depositing surface was sanded with SiC paper and then degreased with acetone and ethanol. A cross-hatching scanning strategy was used in this study. Two types of part with different geometries were deposited in order to evaluate the tensile properties that parallel to and perpendicular to the build direction. The scanning strategy and part geometries are illustrated in Fig. 1. All the samples were deposited under the same processing condition (Laser power: 2000 W, scanning velocity: 10 mm/s, powder feed rate: 8 g/min, norminal increment of Z axis: 0.4 mm, overlap rate: 40%). The deposited samples were divided into two batches: the first one was subjected to ageing treatment (600 °C, 2 h/AC). The second one was subjected to double treatment (solution and ageing, 870 °C, 1 h/FCþ 600 °C,
Fig. 1. (a) cross-hatching scanning strategy, (b) horizontal sample, (c) vertical sample.
205
Table 1 The nomenclature and detail information of the tensile samples. Sample
Tensile direction
Heat treatment
H-AT V-AT H-DT V-DT
Horizontal Vertical Horizontal Vertical
600 °C, 2 h/AC 870 °C, 1 h/FC þ 600 °C, 2 h/AC
2 h/AC). After heat treatment, the tensile samples (6.0 mm in gauge width, 2 mm in gauge thickness and 30 mm in gauge length) were machined by electrical discharge machining and then sanded with SiC paper. The nomenclature and detail information of the tensile samples was listed in Table 1. All the samples were subjected to room temperature tensile testing at a constant crosshead displacement rate of 1 mm/min on an Instron-3382 testing machine. Strain evolution of the sample was investigated by using an optical system incorporated with digital image correlation system (DIC). All the samples were prepared with spray-painted black speckles on a white paint layer providing a random speckle pattern for the DIC system to follow. During the tensile test, the deformation was recorded by the DIC system and then the taken images were analyzed with the help of an in-house software. Microstructure of the gauge section was investigated by optical microscopy (OM, Keyence VH-Z50L) and scanning electron microscopy (SEM, Tescan VEGAⅡLMH). Kroll reagent (1 ml HF, 3 ml HNO3 and 50 ml H2O) was used to reveal the microstructure of all the samples after mounting, grinding and polishing. Vickers hardness before and after tensile test were measured at 1 mm intervals at the gauge section using a micro Vickers hardness meter (Struers Duramin-A300). The Vickers hardness tests were performed by applying a load of 9.8 N for 15 s. Vickers hardness and microstructure investigation before tensile test was performed on the samples taken from the same position where the corresponding tensile sample was cut out.
3. Results 3.1. Microstructure Macrostructure and microstructure of the horizontal samples are presented in Fig. 2. Columnar prior β grains which extend over multiple layers can be observed in Fig. 2a and c. According to the previous studies, the columnar prior β grains present a strong 〈100〉 fiber texture [21]. As distinct from the prior β grain morphology, the α phase characterization is quite different. The sample H-AT exhibits a dense distribution of very fine α laths within the β grains (Fig. 2b). The width of α laths is about 0.14 mm. Very thin grain boundary α layer (αGB) can also be observed. Besides, the molten pool boundaries can be observed clearly in sample H-AT. The molten pool boundaries exhibit a regular structure. Coarse α laths with smaller aspect ratio can be observed in sample H-DT (Fig. 2d). The width of α laths in sample H-DT is about 1 mm. Both continuous and discontinuous αGB layers can be observed in the horizontal samples. The width of αGB layer is about 4 mm. As indicated by the black arrows in Fig. 2c, much thicker αGB layers can also be observed at β grain boundaries. In general, the morphology and scale of the precipitated α laths are relative uniform in the two horizontal samples. As shown in Fig. 3a and c, similar columnar prior β grains can be observed in the vertical samples. Layer bands structure can be observed clearly in the pictures. As shown in Fig. 3b, layer bands in sample V-AT consist of fine α laths. The width of α laths in the
206
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
Fig. 2. (a, b) Macrostructure and microstructure of sample H-AT, (c, d) macrostructure and microstructure of sample H-DT.
Fig. 3. (a, b) Macrostructure and microstructure of sample V-AT, (c) macrostructure of sample V-DT, (d) layer bands in the upper part, (e) layer bands in the lower part.
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
207
Fig. 4. Initial defects in the LAMed samples (a) lack of fusion defect in sample V-AT, (b) pore in sample V-DT.
layer bands and between the bands is about 0.54 mm and 0.81 mm, respectively. Layer bands of sample V-DT which presented as α colony alternately with α laths can be observed in Fig. 3c. Besides, the precipitated α laths are also non-uniform along the build direction. It can be seen from Fig. 3d and e, the scale of α laths at the upper part of the sample is much bigger than the α laths at the bottom part of the sample. The width of α laths in sample V-DT is maximum among the four samples. As shown in Fig. 4, a few initial defects, including lack of fusion defects and pores can be observed in the LAMed samples. The lack of fusion defect is rare in this experiment and only one example was observed in the sample V-AT (Fig. 4a). No lack of fusion defects was observed in other samples. A few circular pores with the diameter about 20 mm in the deposited samples is shown in Fig. 4b. The pores were formed when the gas did not escape from the molten pool timely during the solidification process [24]. The sharp angles of the lack of fusion defect could result in local stress concentrations and lead to early fracture, particularly in the vertical samples. In this study, more initial defects present in the vertical samples compared to the horizontal samples. This can be attributed to the fact that the build surface of the horizontal samples is flatter than the vertical samples. Some researchers [18] have used hot isostatic pressing to eliminate the pores as it is difficult to fabricate pore-free component only by adjusting the LAM process parameters. 3.2. Tensile properties The Young's modulus and tensile properties of the four different samples are listed in Table 2. The engineering stress-strain curves, as well as the curves of work hardening rate (θ ¼dst/dεt) vs true strain (εt), are shown in Fig. 5. The Young's modulus of the horizontal samples is slightly higher than that of the vertical samples. There is significant difference both in ductility and strength between the horizontal and vertical samples in two heat treatment conditions, indicating the tensile anisotropy of the LAMed parts. The Young's modulus of the ageing treated samples is slightly higher than the double treated samples. And the ageing Table 2 Young's modulus and tensile properties of the LAMed samples. Sample No. Young's modulus (GPa)
Yield stress (0.2% Offset) (MPa)
Ultimate tensile Total elongastrength (MPa) tion (%)
H-AT V-AT H-DT V-DT
995 969 868 845
1268 1060 997 933
112.4 109.7 107.8 102.7
3 11 2.8 16
treated samples are much stronger than the corresponding double treated samples. The ultimate tensile strength (UTS) and yield strength (YS) of sample H-AT is highest, followed by V-AT, H-DT and V-DT in a descending order. Yield strength was measured with the 0.2% offset method from the corresponding engineering stressstrain curve. Significant difference can also be noted in the curves of the alloy in four different conditions. The black pentagrams in Fig. 5a indicate the UTS in each curves. When the stress exceeds the UTS, the sample begins to necking. It can be seen that the sample H-AT fractured without necking after slightly uniform elongation. The sample H-DT fractured when the necking just getting start. The vertical samples show more obvious uniform elongation and necking when compared with the horizontal samples. The different stages of work hardening rate could be clearly observed from Fig. 5b. It can be seen that the sample H-AT exhibits high values of work hardening rate and displays a complex shape. Work hardening rate of the sample H-AT reduced dramatically after yielding and then increased again and maintains a higher value with increasing strain, and reduced again before fracture occurred. Work hardening rate of sample H-DT reduced dramatically after yielding and then increased slightly and then decreased again at a lower rate. The θ-εt curves of vertical samples exhibit very similar appearance. The two stages are characterized as follows: (1) the work hardening rate decreased sharply at the beginning of plastic deformation; (2) the work hardening rate decreased gradually. Small oscillations could be observed in the second stage of these curves, especially in sample V-DT. The second stage contributes significantly more toward plastic deformation compared to the first stage. As Mantri proposed [25], the differences in the work hardening behaviors indicate that dislocations tend to interact differently with the α laths with different morphologies. Fracture surfaces of the four different samples are shown in Fig. 6. The horizontal samples heat treated in two different conditions are characterized by a low strain to failure. Sample H-AT shows a brittle fracture way (Fig. 6a and e) and the evidence of grain boundary can also be observed. Analysis of the fracture profile reveals that the fracture occurred preferentially along the prior β grain boundaries in sample H-DT (Fig. 6g). This indicated that the voids nucleation and propagation at the interface between the grain boundary α and the β matrix. With respective strains to failure of 11% and 16%, the vertical samples exhibit a rather ductile behavior. As shown in Fig. 6f, the sample V-AT exhibit a transgranular and typical ductile fracture surface, with small dimples. The sample V-DT revealed a mixture feature of dimples and staircase (Fig. 6h). No grain boundaries fracture was observed on the fracture surface of the vertical samples. Given the detrimental effect of the lack of fusion defects on tensile properties, there is no
208
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
Fig. 5. (a) Engineering stress-strain curves, (b) work hardening rate vs true strain curves, of the samples in four different conditions.
evidence to suggest that the fracture surface is caused by lack of fusion defects, which indicated that no lack of fusion defects exist in the sample used for tensile test. The side surface topography of the four samples after tensile test are shown in Fig. 7. Significant different appearances can be observed in these pictures. Consistent with the engineering stressstrain curves, the horizontal samples showed no necking. Almost no signs of deformation can be observed in sample H-AT (Fig. 7a). Fig. 7b and c showed that the deformations in sample V-AT and H-DT is inhomogeneous. And the deformation feature in the two samples seemed to be related to the columnar prior β grains. This may be caused by the incompatible deformation between the adjacent prior β grains due to the different crystal orientations. Some deformation features which parallel to each other were indicated by the white arrows. Severe plastic deformation can be observed on the side surface of sample V-DT (Fig. 7d). The deformation near the fracture location is almost homogeneous. As indicated by the black rectangles, the deformation features which parallel to each other can also been observed. 3.3. Vickers hardness Fig. 8 shows the Vickers hardness at the gauge section of the four samples before and after tensile test. It can be seen from the pictures that the ageing treated samples (blue) are harder than the double treated samples (red). Vickers hardness of the horizontal samples changed little after tensile test (Fig. 8a). As distinct from the horizontal samples, Vickers hardness of the vertical samples increased after tensile test (Fig. 8b). The results show that more obvious work hardening happened in the vertical samples compared to the horizontal samples.
4. Discussions 4.1. Microstructure of the samples in four different conditions The heat treated samples exhibit obvious anisotropic. As well documented that the mechanical properties of titanium alloy components are strongly affected by the characterizations of β grains and α phase [1]. The microstructure characterizations of LAMed titanium alloys are mainly determined by the thermal history they experienced both in the LAM and post heat treatment process. In this study, all the four samples exhibit coarse columnar β grains. The columnar β grains is resulted from the solidification conditions of the molten pool and the metallurgical characteristics of titanium alloy [26]. The columnar morphology of the prior β grains is essentially not changed when the LAMed samples were heat treated below β transus temperature [7]. However, significant differences in α phase characterizations can be observed, even in the samples which were fabricated by the same LAM processing parameters and experienced the same heat treatment. As shown in Section 3.1, microstructure of the horizontal samples is more uniform than that of the vertical samples. Microstructure differences between the horizontal and vertical samples after heat treatment mainly resulted from the as-deposited conditions. The differences in the as-deposited microstructure is expected as the thermal history during the LAM process is strongly affected by the part geometry. For the horizontal samples, the microstructure of the as-deposited sample is martensite α’ phase as a result of rapid cooling due to the longer scanning path and bigger heat dissipation area [27]. After ageing at 600 °C for 2 h, the martensite α’ phase transformed to fine acicular α laths (Fig. 2b). When solution treated at 870 °C for 1 h followed by furnace cooling, the
Fig. 6. Fracture surfaces of (a, e) sample H-AT, (b, f) sample V-AT, (c, g) sample H-DT, (d, h) sample V-DT.
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
209
Fig. 7. Photographs taken from the side surface of the tensile samples (a) sample H-AT, (b) sample V-AT, (C) sample H-DT, (b) sample V-DT.
precipitated α laths and the grain boundary α layers are more coarse (Fig. 2d). For the vertical samples, the thermal history they experienced is more complex, so the α laths with complex morphology and different scales can be obtained along the build direction [21]. Ageing treatment at 600 °C has little effect on the characterization of α phase. The characterization of layer bands was more apparent after double treatment mainly due to the higher solution temperature and slower cooling rate. 4.2. Strength and ductility As shown in Table 2, the highest UTS and YS are obtained in sample H-AT. This is due to the presence of fine α laths in β matrix resulting in a large number of α-β interfaces that can act as obstacles to dislocations motion [28]. The sample V-DT which presents the coarsest α laths and α colonies shows the lowest strength. Both the UTS and YS decreased with the increase of α laths width due to the reduction in volume fraction of α-β interfaces. This results showed that the α lath width plays a key role in strengthening the LAMed TC21 titanium alloy. However, the elongation of the samples do not show regularity relationship with the strength or α laths width. In general, the ductility of the horizontal samples is much inferior to that of the vertical samples. The inferior ductility of the horizontal samples can be attributed to the columnar β grain morphology and the presence of grain boundary α layers. Because these boundaries can serve as a path
along which the voids can preferentially nucleation and coalescence [29]. In the horizontal samples, the tensile load perpendicular to the αGB layers act to separate the adjacent prior β grains. The columnar prior β grains result in different amounts of αGB layers which perpendicular to the tensile load [4]. Most of the αGB layers in the horizontal samples is perpendicular to the tensile load while a little amount of αGB layers is perpendicular to the tensile load in the vertical samples. For the sample H-DT, both the UTS and YS is substantially lower than that of the sample H-AT. However, the reduction in strength does not increase ductility. In contrast, the sample H-DT exhibits even poorer elongation and more feature of intergranular fracture. The presence of coarse αGB layers after double treatment may be the main factor for the poor performance because the coarser αGB layers can further reduce the cohesion between the adjacent β grains. As a result, brittle intergranular fracture occurred at sample H-DT when the tensile load is low. As shown in Fig. 8, Vickers hardness of the horizontal samples is nearly unchanged after tensile test while Vickers hardness of the vertical samples increased after tensile test. As work hardening is mainly achieved by the accumulation of dislocations, for the horizontal samples, the fine α laths have a limited capacity to store dislocations and the premature fracture further limits the strain hardening. Fig. 9 shows the SEM photographs close to the fracture surface of the four samples. As shown in Fig. 9a, the presence of molten
Fig. 8. Vickers hardness of the gauge section before and after tensile test (a) the horizontal samples, (b) the vertical samples. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)
210
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
Fig. 9. The side view close to the fracture surface (a) sample H-AT, (b) sample V-AT, (C) sample H-DT, (b) sample V-DT.
pool boundaries is a noticeable feature of sample H-AT. A previous study investigated the effects of molten pool boundaries on the mechanical properties of the selective laser melted 316L stainless steel and indicated that the molten pool boundaries are the main factor that causes the fracture [30]. In the present study, the fracture does not along the molten pool boundaries. As shown in Fig. 9b–d, some voids could be observed in the α laths indicating the intrinsic fracture of the α laths. More voids are observed in the α-β interfaces. The amounts of voids in the four samples are different. Little voids were observed in the horizontal samples which fractured in a brittle way. More voids can be observed in the vertical samples. This is because the vertical samples have low working-hardening rate and this promotes easier voids nucleation [31]. As shown in Fig. 9b, voids coalescence along the αGB layer can be observed. However, the cracks in the vertical samples cannot propagate along the grain boundary because they are mainly parallel to the tensile load direction. This explains why intergranular fracture is rare in the vertical samples. Fig. 10 shows the SEM photograph away from the fracture surface of sample V-DT. More voids can be observed in “α colony” region indicating more serious deformation. It can be seen that the voids usually nucleate at the interface of coarse α laths and β matrix due to the continued plastic deformation and a severe strain gradient across α-β interface [28]. As the circle indicated, voids coalescence along the coarse α laths can be observed. This kind of coarse α laths which does not parallel to the tensile load is similar to the αGB layers in the horizontal samples. As distinct from the continuous αGB layer, the cracks along the coarse α laths do not lead to fracture. Almost no voids can be observed at the α-β interfaces which almost parallel to the tensile load.
Fig. 10. Inhomogeneous deformation in the layer band structure of sample V-DT.
4.3. Inhomogeneous deformation The DIC contour maps of the sample V-AT are shown in Fig. 11. Different colors in gauge section indicate different deformation degree. As indicated in Fig. 11a, the deformation of prior β grains starts at the elastic deformation stage. With the increase of tensile load, inhomogeneous deformation between layer bands can be observed in Fig. 11b. Then, the strain concentrations take place in the weaker region of the sample. The larger plastic strain (red region) in Fig. 11c indicates two possible locations where necking may occur. Finally, fracture happened in the necking location. The inhomogeneous deformation can be attributed to the coarse columnar prior β grains and non-uniform α laths. The small oscillations in the second stage of θ vs εt curves (Fig. 5b) may also
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
211
Acknowledgements This work was supported by the Research Fund of the State Key Laboratory of Solidification Processing (NWPU), China (Grant No.151-ZH-2016) and National Key Technologies R&D Program (2016YFB11000100) and National Natural Science Foundation of China (Grant No. 51475380, 51323008).
References
Fig. 11. The DIC contour maps of sample V-AT. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)
reflect the inhomogeneous deformation of the vertical samples.
5. Conclusions The microstructure and anisotropic tensile behavior of LAMed TC21 titanium alloy under two different heat treatments were analyzed. The primary conclusions from the present study are as follows: (1) The samples in four different conditions exhibit similar columnar β grains and very different α laths. The horizontal samples exhibit uniform α laths. Layer bands can be observed in the vertical samples. The part geometry and heat treatment have significant effect on the characterization of α laths. (2) The tensile tests show significant anisotropic mechanical properties in two different heat treatment conditions. The horizontal samples exhibit stronger strength than the vertical samples. And the total elongation of the horizontal sample is much inferior to that of the vertical samples. (3) The ultimate tensile strength and yield strength decreased with the increase of α laths width. The elongation of the samples do not show regularity relationship with the strength. The inferior elongation of the horizontal samples is caused by a larger amounts of αGB layers which perpendicular to the tensile load. Coarsening of αGB layers can further reduce the ductility. (4) Vickers hardness test shows that the vertical samples were strengthened obviously after tensile test due to strain hardening. The horizontal samples were little changed after tensile test due to the premature fracture. (5) The coarse columnar prior β grains and the non-uniform α laths of the vertical samples resulted in inhomogeneous deformation.
[1] G. Lutjering, J.C. Williams, Titanium, Springer Science & Business Media, Berlin, Germany, 2007. [2] W.D. Huang, X. Lin, Research progress in laser solid forming of high-performance metallic components at the state key laboratory of solidification processing of China, 3D Print. Addit. Manuf. 1 (2014) 156–165. [3] A.R. Nassar, E.W. Reutzel, Additive manufacturing of Ti6Al4V using a pulsed laser beam, Metall. Mater. Trans. A 6 (2015) 2781–2789. [4] B.E. Carroll, T.A. Palmer, A.M. Beese, Anisotropic tensile behavior of Ti6Al4V components fabricated with directed energy deposition additive manufacturing, Acta Mater. 87 (2015) 309–320. [5] A.J. Sterling, B. Torries, N. Shamsaei, S.M. Thompson, D.W. Seely, Fatigue behavior and failure mechanisms of direct laser deposited Ti6Al4V, Mater. Sci. Eng. A 655 (2016) 100–112. [6] M. Gaumann, C. Bezencon, P. Canalis, W. Kurz, Single-crystal laser deposition of superalloys: processing-microstructure maps, Acta Mater. 49 (2001) 1051–1062. [7] C. de Formanoir, S. Michotte, O. Rigo, L. Germain, S. Godet, Electron beam melted Ti6Al4V: microstructure, texture and mechanical behavior of the asbuilt and heat-treated material, Mater. Sci. Eng. A. 652 (2016) 105–119. [8] B. Vrancken, L. Thijs, J.P. Kruth, J. Van Humbeeck, Microstructure and mechanical properties of a novel β titanium metallic composite by selective laser melting, Acta Mater. 68 (2014) 150–158. [9] S.S. Al-Bermani, M.L. Blackmore, W. Zhang, I. Todd, The origin of microstructural diversity, texture, and mechanical properties in electron beam melted Ti6Al4V, Metall. Mater. Trans. A. 41 (2010) 3422–3434. [10] A. Crespo, Modelling of Heat Transfer and Phase Transformations in the Rapid Manufacturing of Titanium Components. Convection and Conduction Heat Transfer, INTECH Open Access Publisher, 2011, pp. 315–340. [11] E. Brandl, F. Palm, V. Michailov, B. Viehweger, C. Leyens, Mechanical properties of additive manufactured titanium (Ti6Al4V) blocks deposited by a solid-state laser and wire, Mater. Des. 32 (2011) 4665–4675. [12] S.M. Kelly, S.L. Kampe, Microstructural evolution in laser-deposited multilayer Ti6Al4V builds: part I. Microstructural characterization, Metall. Mater. Trans. A 35A (2004) 1861–1867. [13] B. Baufeld, Effect of deposition parameters on mechanical properties of shaped metal deposition parts, Proc. Inst. Mech. Eng. B: J. Eng. 226 (2012) 126–136. [14] Y.Y. Zhu, X.J. Tian, J. Li, H.M. Wang, The anisotropy of laser melting deposition additive manufacturing Ti6.5Al3.5Mo1.5Zr0.3Si titanium alloy, Mater. Des. 67 (2015) 538–542. [15] F.D. Wang, S. Williams, P. Colegrove, A.A. Antonysamy, Microstructure and mechanical properties of wire and arc additive manufactured Ti6Al4V, Metall. Mater. Trans. A. 44 (2013) 968–977. [16] Z. Liu, Z.X. Qin, F. Liu, X. Lu, H.M. Wang, The microstructure and mechanical behaviors of the Ti6.5Al3.5Mo1.5Zr0.3Si alloy produced by laser melting deposition, Mater. Charact. 97 (2014) 132–139. [17] T. Vilaro, C. Colin, J.D. Bartout, As-fabricated and heat-treated microstructures of the Ti6Al4V alloy processed by selective laser melting, Metall. Mater. Trans. A. 42 (2011) 3190–3199. [18] C.L. Qiu, G.A. Ravi, C. Dance, A. Ranson, S. Dilworth, M.M. Attallah, Fabrication of large Ti6Al4V structures by direct laser deposition, J. Alloy. Compd. 629 (2015) 351–361. [19] Y.H. Wang, H.C. Kou, H. Chang, Z.S. Zhu, F.S. Zhang, J.S. Li, L. Zhou, Influence of solution temperature on phase transformation of TC21 alloy, Mater. Sci. Eng. A 508 (2009) 76–82. [20] Z.F. Shi, H.Z. Guo, J.Y. Han, Z.K. Yao, Microstructure and mechanical properties of TC21 titanium alloy after heat treatment, Trans. Nonferr. Met. Soc. China 23 (2013) 2882–2889. [21] Q. Zhang, J. Chen, P.F. Guo, H. Tan, X. Lin, W.D. Huang, Texture and microstructure characterization in laser additive manufactured Ti6Al2Zr2Sn3Mo1.5Cr2Nb titanium alloy, Mater. Des. 88 (2015) 550–557. [22] X. Wu, R. Sharman, J. Mei, W. Voice, Microstructure and properties of a laser fabricated burn-resistant Ti alloy, Mater. Des. 25 (2004) 103–109. [23] C.L. Qiu, G.A. Ravi, M.M. Attallah, Microstructural control during direct laser deposition of a β-titanium alloy, Mater. Des. 81 (2015) 21–30. [24] P.H. Li, W.G. Guo, W.D. Huang, Y. Su, X. Lin, K.B. Yuan, Thermomechanical response of 3D laser-deposited Ti6Al4V alloy over a wide range of strain rates and temperatures, Mater. Sci. Eng. A 647 (2015) 34–42. [25] S.A. Mantri, D. Choudhuri, A. Behera, J.D. Cotton, N. Kumar, R. Banerjee, Influence of fine-scale alpha precipitation on the mechanical properties of the beta titanium alloy beta-21S, Metall. Mater. Trans. A 46 (2015) 2803–2808. [26] J. Donoghue, A.A. Antonysamy, F. Martina, P.A. Colegrove, S.W. Williams, P.
212
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
B. Prangnell, The effectiveness of combining rolling deformation with wire-arc additive manufacture on β-Grain refinement and texture modification in Ti6Al4V, Mater. Charact. 114 (2016) 103–114. [27] Q. Zhang, J. Chen, H. Tan, X. Lin, W.D. Huang, Influence of solution treatment on microstructure evolution of TC21 titanium alloy with near equiaxed β grains fabricated by laser additive manufacture, J. Alloy. Compd. 666 (2016) 380–386. [28] R. Sarkar, A. Mukhopadhyay, P. Ghosal, T.K. Nandy, K.K. Ray, Effect of aged microstructure on the strength and work hardening behavior of
Ti15V3Cr3Sn3Al alloy, Metall. Mater. Trans. A. 46 (2015) 3516–3527. [29] C. Sauer, G. Luetjering, Thermo-mechanical processing of high strength β-titanium alloys and effects on microstructure and properties, J. Mater. Process. Technol. 117 (2001) 311–317. [30] S.F. Wen, S. Li, Q.S. Wei, C.Z. Yan, S. Zhang, Y.S. Shi, Effect of molten pool boundaries on the mechanical properties of selective laser melting parts, J. Mater. Process. Technol. 214 (2014) 2660–2667. [31] D. Banerjee, J.C. Williams, Perspectives on titanium science and technology, Acta Mater. 61 (2013) 844–879.
Contents lists available at ScienceDirect
Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea
Microstructure and anisotropic tensile behavior of laser additive manufactured TC21 titanium alloy Qiang Zhang, Jing Chen n, Zhuang Zhao, Hua Tan, Xin Lin, Weidong Huang State Key Laboratory of Solidification Processing, Northwestern Polytechnical University, Xi’an, Shaanxi 710072, China
art ic l e i nf o
a b s t r a c t
Article history: Received 9 March 2016 Accepted 9 July 2016 Available online 11 July 2016
In the present work, two types of part with different geometries were fabricated by laser additive manufacturing (LAM) process under the same processing parameters. Tensile samples perpendicular to (horizontal) and parallel to (vertical) the build direction were produced after heat treated under different conditions. The results showed that the samples exhibit similar columnar β grains, but very different α phase characterizations. The tensile properties show significant anisotropic. The sample containing finest α laths shows highest ultimate tensile strength and yield strength. And the strength decreased with the increase of α laths width. The total elongation of the samples does not show regularity relationship with the strength. The horizontal samples show inferior ductility and brittle fracture due to the columnar β grain morphology and the presence of continuous grain boundary α layers. The coarse grain boundary α layers after double heat treatment further reduced the ductility. The vertical samples exhibit better ductility due to lack of continuous grain boundary α layers which perpendicular to the tensile load. The Vickers hardness test shows that the vertical samples were strengthened while the horizontal samples were little strengthened after tensile test. & 2016 Elsevier B.V. All rights reserved.
Keywords: Laser additive manufacture Microstructure Anisotropic tensile behavior Titanium alloy
1. Introduction Titanium alloys are widely used in the aeronautical industry for their high specific strength, excellent mechanical properties and outstanding corrosion resistance. However, the end products of titanium alloys are quite expensive due to the difficulties in refining, casting, forming and machining process [1]. Laser additive manufacturing is a kind of advanced processing technology that can be used to fabricate near fully-dense complex metal parts [2]. The LAM process has been recognized as an affordable and efficient solution that can meet the stringent requirements for aerospace titanium alloys components [3–5]. In order to achieve mechanical properties comparable or even superior to the traditional manufacturing, it is crucial to get a better and deeper understanding of the relationships between the processing parameters, microstructure evolution and the resulting mechanical properties. During the LAM process, the deposited layers undergo melting, rapid solidification, partial remelting and reheating. Therefore, the LAM process resulted in a specific microstructure. It has been well documented that during solidification of the molten pool, the solid substrate acts as a heat sink and the as-deposited microstructure generally presents a directional solidified morphology grown from n
Corresponding author. E-mail address: [email protected] (J. Chen).
http://dx.doi.org/10.1016/j.msea.2016.07.040 0921-5093/& 2016 Elsevier B.V. All rights reserved.
the substrate epitaxially [6]. As a result, typical prior β grain morphology of LAMed α þ β titanium alloys usually comprised of coarsen columnar grains which exhibit a strong 〈100〉 fiber texture [7–9]. The characterization of α phase precipitated in the β phase shows more complex features because the deposited layers experienced consecutive thermal cycles with a duration and amplitude which depend on the processing parameters and geometry of the part being fabricated [10]. Fine lamellar structure containing martensitic α′ [11], layer bands [12] and coarsen lamellar structure [13] have been reported. This specific microstructure characterization strongly effects the mechanical properties of LAMed titanium alloys. Generally, the LAMed titanium alloy components usually present anisotropic mechanical properties. The strongly textured prior β grains which growth along the build direction is regarded as the main reason for anisotropic mechanical properties [14–16]. Vilaro [17] found that the pores shape and orientation in the deposited Ti6Al4V components strongly influence the ductility. Qiu [18] found that the presence of planar pores due to incomplete remelting of previous layers in LAMed Ti6Al4V alloy also caused pronounced anisotropy in ductility. Carroll [4] studied the anisotropy tensile behavior in LAMed Ti6Al4V alloy cruciform component. They reported better ductility along the build direction in comparison to the direction perpendicular to the build direction. They believed that the columnar β grains and grain boundary α phase are the main reason for different ductility. The TC21 titanium alloy (Ti6Al2Sn2Zr3Mo1.5Cr2Nb, β transus
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
temperature is about 960 °C [19]) investigated in the present study is a α þ β titanium alloy. Due to the good balance of strength, ductility and fracture toughness, the TC21 titanium alloy has shown a good promising for applications in aerospace industry [20]. The previous study showed that the microstructure of LAMed TC21 titanium alloy is more inhomogeneous than the Ti6Al4V alloy due to the higher alloying elements [21]. Similar inhomogeneous microstructure could also been observed in other LAMed titanium alloys which contains more alloy elements [22,23]. However, there has been little effort to understand the connection between the mechanical properties and the more complex microstructure characterization. The aim of this study is to correlate the microstructure of LAMed TC21 titanium alloy to the corresponding tensile properties. In the present work, two types of part with different geometries were deposited under the same LAM processing condition. And then the tensile properties parallel to and perpendicular to the build direction after two different heat treatments were tested. Based on the experiment results, the anisotropic tensile behavior and the relationship with the microstructure characterization were discussed.
2. Experimental procedures The samples were fabricated by a LAM equipment that consists of a 4 kW continuous wave CO2 laser, a 5-axis numerical control working table, a coaxial powder feeder nozzle and an inert gas chamber filled with pure argon with oxygen content below 50 ppm. TC21 spherical powders produced by the plasma rotating electrode process (PREP) were used as the deposited material. The powder size ranges from 80 mm to 150 mm. The powders were dried in a vacuum oven for 2 h at 12075 °C to eliminate the moisture absorption and ensure the powders have good flow ability. The forged TC21 plates were used as substrate for the LAM process. The depositing surface was sanded with SiC paper and then degreased with acetone and ethanol. A cross-hatching scanning strategy was used in this study. Two types of part with different geometries were deposited in order to evaluate the tensile properties that parallel to and perpendicular to the build direction. The scanning strategy and part geometries are illustrated in Fig. 1. All the samples were deposited under the same processing condition (Laser power: 2000 W, scanning velocity: 10 mm/s, powder feed rate: 8 g/min, norminal increment of Z axis: 0.4 mm, overlap rate: 40%). The deposited samples were divided into two batches: the first one was subjected to ageing treatment (600 °C, 2 h/AC). The second one was subjected to double treatment (solution and ageing, 870 °C, 1 h/FCþ 600 °C,
Fig. 1. (a) cross-hatching scanning strategy, (b) horizontal sample, (c) vertical sample.
205
Table 1 The nomenclature and detail information of the tensile samples. Sample
Tensile direction
Heat treatment
H-AT V-AT H-DT V-DT
Horizontal Vertical Horizontal Vertical
600 °C, 2 h/AC 870 °C, 1 h/FC þ 600 °C, 2 h/AC
2 h/AC). After heat treatment, the tensile samples (6.0 mm in gauge width, 2 mm in gauge thickness and 30 mm in gauge length) were machined by electrical discharge machining and then sanded with SiC paper. The nomenclature and detail information of the tensile samples was listed in Table 1. All the samples were subjected to room temperature tensile testing at a constant crosshead displacement rate of 1 mm/min on an Instron-3382 testing machine. Strain evolution of the sample was investigated by using an optical system incorporated with digital image correlation system (DIC). All the samples were prepared with spray-painted black speckles on a white paint layer providing a random speckle pattern for the DIC system to follow. During the tensile test, the deformation was recorded by the DIC system and then the taken images were analyzed with the help of an in-house software. Microstructure of the gauge section was investigated by optical microscopy (OM, Keyence VH-Z50L) and scanning electron microscopy (SEM, Tescan VEGAⅡLMH). Kroll reagent (1 ml HF, 3 ml HNO3 and 50 ml H2O) was used to reveal the microstructure of all the samples after mounting, grinding and polishing. Vickers hardness before and after tensile test were measured at 1 mm intervals at the gauge section using a micro Vickers hardness meter (Struers Duramin-A300). The Vickers hardness tests were performed by applying a load of 9.8 N for 15 s. Vickers hardness and microstructure investigation before tensile test was performed on the samples taken from the same position where the corresponding tensile sample was cut out.
3. Results 3.1. Microstructure Macrostructure and microstructure of the horizontal samples are presented in Fig. 2. Columnar prior β grains which extend over multiple layers can be observed in Fig. 2a and c. According to the previous studies, the columnar prior β grains present a strong 〈100〉 fiber texture [21]. As distinct from the prior β grain morphology, the α phase characterization is quite different. The sample H-AT exhibits a dense distribution of very fine α laths within the β grains (Fig. 2b). The width of α laths is about 0.14 mm. Very thin grain boundary α layer (αGB) can also be observed. Besides, the molten pool boundaries can be observed clearly in sample H-AT. The molten pool boundaries exhibit a regular structure. Coarse α laths with smaller aspect ratio can be observed in sample H-DT (Fig. 2d). The width of α laths in sample H-DT is about 1 mm. Both continuous and discontinuous αGB layers can be observed in the horizontal samples. The width of αGB layer is about 4 mm. As indicated by the black arrows in Fig. 2c, much thicker αGB layers can also be observed at β grain boundaries. In general, the morphology and scale of the precipitated α laths are relative uniform in the two horizontal samples. As shown in Fig. 3a and c, similar columnar prior β grains can be observed in the vertical samples. Layer bands structure can be observed clearly in the pictures. As shown in Fig. 3b, layer bands in sample V-AT consist of fine α laths. The width of α laths in the
206
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
Fig. 2. (a, b) Macrostructure and microstructure of sample H-AT, (c, d) macrostructure and microstructure of sample H-DT.
Fig. 3. (a, b) Macrostructure and microstructure of sample V-AT, (c) macrostructure of sample V-DT, (d) layer bands in the upper part, (e) layer bands in the lower part.
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
207
Fig. 4. Initial defects in the LAMed samples (a) lack of fusion defect in sample V-AT, (b) pore in sample V-DT.
layer bands and between the bands is about 0.54 mm and 0.81 mm, respectively. Layer bands of sample V-DT which presented as α colony alternately with α laths can be observed in Fig. 3c. Besides, the precipitated α laths are also non-uniform along the build direction. It can be seen from Fig. 3d and e, the scale of α laths at the upper part of the sample is much bigger than the α laths at the bottom part of the sample. The width of α laths in sample V-DT is maximum among the four samples. As shown in Fig. 4, a few initial defects, including lack of fusion defects and pores can be observed in the LAMed samples. The lack of fusion defect is rare in this experiment and only one example was observed in the sample V-AT (Fig. 4a). No lack of fusion defects was observed in other samples. A few circular pores with the diameter about 20 mm in the deposited samples is shown in Fig. 4b. The pores were formed when the gas did not escape from the molten pool timely during the solidification process [24]. The sharp angles of the lack of fusion defect could result in local stress concentrations and lead to early fracture, particularly in the vertical samples. In this study, more initial defects present in the vertical samples compared to the horizontal samples. This can be attributed to the fact that the build surface of the horizontal samples is flatter than the vertical samples. Some researchers [18] have used hot isostatic pressing to eliminate the pores as it is difficult to fabricate pore-free component only by adjusting the LAM process parameters. 3.2. Tensile properties The Young's modulus and tensile properties of the four different samples are listed in Table 2. The engineering stress-strain curves, as well as the curves of work hardening rate (θ ¼dst/dεt) vs true strain (εt), are shown in Fig. 5. The Young's modulus of the horizontal samples is slightly higher than that of the vertical samples. There is significant difference both in ductility and strength between the horizontal and vertical samples in two heat treatment conditions, indicating the tensile anisotropy of the LAMed parts. The Young's modulus of the ageing treated samples is slightly higher than the double treated samples. And the ageing Table 2 Young's modulus and tensile properties of the LAMed samples. Sample No. Young's modulus (GPa)
Yield stress (0.2% Offset) (MPa)
Ultimate tensile Total elongastrength (MPa) tion (%)
H-AT V-AT H-DT V-DT
995 969 868 845
1268 1060 997 933
112.4 109.7 107.8 102.7
3 11 2.8 16
treated samples are much stronger than the corresponding double treated samples. The ultimate tensile strength (UTS) and yield strength (YS) of sample H-AT is highest, followed by V-AT, H-DT and V-DT in a descending order. Yield strength was measured with the 0.2% offset method from the corresponding engineering stressstrain curve. Significant difference can also be noted in the curves of the alloy in four different conditions. The black pentagrams in Fig. 5a indicate the UTS in each curves. When the stress exceeds the UTS, the sample begins to necking. It can be seen that the sample H-AT fractured without necking after slightly uniform elongation. The sample H-DT fractured when the necking just getting start. The vertical samples show more obvious uniform elongation and necking when compared with the horizontal samples. The different stages of work hardening rate could be clearly observed from Fig. 5b. It can be seen that the sample H-AT exhibits high values of work hardening rate and displays a complex shape. Work hardening rate of the sample H-AT reduced dramatically after yielding and then increased again and maintains a higher value with increasing strain, and reduced again before fracture occurred. Work hardening rate of sample H-DT reduced dramatically after yielding and then increased slightly and then decreased again at a lower rate. The θ-εt curves of vertical samples exhibit very similar appearance. The two stages are characterized as follows: (1) the work hardening rate decreased sharply at the beginning of plastic deformation; (2) the work hardening rate decreased gradually. Small oscillations could be observed in the second stage of these curves, especially in sample V-DT. The second stage contributes significantly more toward plastic deformation compared to the first stage. As Mantri proposed [25], the differences in the work hardening behaviors indicate that dislocations tend to interact differently with the α laths with different morphologies. Fracture surfaces of the four different samples are shown in Fig. 6. The horizontal samples heat treated in two different conditions are characterized by a low strain to failure. Sample H-AT shows a brittle fracture way (Fig. 6a and e) and the evidence of grain boundary can also be observed. Analysis of the fracture profile reveals that the fracture occurred preferentially along the prior β grain boundaries in sample H-DT (Fig. 6g). This indicated that the voids nucleation and propagation at the interface between the grain boundary α and the β matrix. With respective strains to failure of 11% and 16%, the vertical samples exhibit a rather ductile behavior. As shown in Fig. 6f, the sample V-AT exhibit a transgranular and typical ductile fracture surface, with small dimples. The sample V-DT revealed a mixture feature of dimples and staircase (Fig. 6h). No grain boundaries fracture was observed on the fracture surface of the vertical samples. Given the detrimental effect of the lack of fusion defects on tensile properties, there is no
208
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
Fig. 5. (a) Engineering stress-strain curves, (b) work hardening rate vs true strain curves, of the samples in four different conditions.
evidence to suggest that the fracture surface is caused by lack of fusion defects, which indicated that no lack of fusion defects exist in the sample used for tensile test. The side surface topography of the four samples after tensile test are shown in Fig. 7. Significant different appearances can be observed in these pictures. Consistent with the engineering stressstrain curves, the horizontal samples showed no necking. Almost no signs of deformation can be observed in sample H-AT (Fig. 7a). Fig. 7b and c showed that the deformations in sample V-AT and H-DT is inhomogeneous. And the deformation feature in the two samples seemed to be related to the columnar prior β grains. This may be caused by the incompatible deformation between the adjacent prior β grains due to the different crystal orientations. Some deformation features which parallel to each other were indicated by the white arrows. Severe plastic deformation can be observed on the side surface of sample V-DT (Fig. 7d). The deformation near the fracture location is almost homogeneous. As indicated by the black rectangles, the deformation features which parallel to each other can also been observed. 3.3. Vickers hardness Fig. 8 shows the Vickers hardness at the gauge section of the four samples before and after tensile test. It can be seen from the pictures that the ageing treated samples (blue) are harder than the double treated samples (red). Vickers hardness of the horizontal samples changed little after tensile test (Fig. 8a). As distinct from the horizontal samples, Vickers hardness of the vertical samples increased after tensile test (Fig. 8b). The results show that more obvious work hardening happened in the vertical samples compared to the horizontal samples.
4. Discussions 4.1. Microstructure of the samples in four different conditions The heat treated samples exhibit obvious anisotropic. As well documented that the mechanical properties of titanium alloy components are strongly affected by the characterizations of β grains and α phase [1]. The microstructure characterizations of LAMed titanium alloys are mainly determined by the thermal history they experienced both in the LAM and post heat treatment process. In this study, all the four samples exhibit coarse columnar β grains. The columnar β grains is resulted from the solidification conditions of the molten pool and the metallurgical characteristics of titanium alloy [26]. The columnar morphology of the prior β grains is essentially not changed when the LAMed samples were heat treated below β transus temperature [7]. However, significant differences in α phase characterizations can be observed, even in the samples which were fabricated by the same LAM processing parameters and experienced the same heat treatment. As shown in Section 3.1, microstructure of the horizontal samples is more uniform than that of the vertical samples. Microstructure differences between the horizontal and vertical samples after heat treatment mainly resulted from the as-deposited conditions. The differences in the as-deposited microstructure is expected as the thermal history during the LAM process is strongly affected by the part geometry. For the horizontal samples, the microstructure of the as-deposited sample is martensite α’ phase as a result of rapid cooling due to the longer scanning path and bigger heat dissipation area [27]. After ageing at 600 °C for 2 h, the martensite α’ phase transformed to fine acicular α laths (Fig. 2b). When solution treated at 870 °C for 1 h followed by furnace cooling, the
Fig. 6. Fracture surfaces of (a, e) sample H-AT, (b, f) sample V-AT, (c, g) sample H-DT, (d, h) sample V-DT.
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
209
Fig. 7. Photographs taken from the side surface of the tensile samples (a) sample H-AT, (b) sample V-AT, (C) sample H-DT, (b) sample V-DT.
precipitated α laths and the grain boundary α layers are more coarse (Fig. 2d). For the vertical samples, the thermal history they experienced is more complex, so the α laths with complex morphology and different scales can be obtained along the build direction [21]. Ageing treatment at 600 °C has little effect on the characterization of α phase. The characterization of layer bands was more apparent after double treatment mainly due to the higher solution temperature and slower cooling rate. 4.2. Strength and ductility As shown in Table 2, the highest UTS and YS are obtained in sample H-AT. This is due to the presence of fine α laths in β matrix resulting in a large number of α-β interfaces that can act as obstacles to dislocations motion [28]. The sample V-DT which presents the coarsest α laths and α colonies shows the lowest strength. Both the UTS and YS decreased with the increase of α laths width due to the reduction in volume fraction of α-β interfaces. This results showed that the α lath width plays a key role in strengthening the LAMed TC21 titanium alloy. However, the elongation of the samples do not show regularity relationship with the strength or α laths width. In general, the ductility of the horizontal samples is much inferior to that of the vertical samples. The inferior ductility of the horizontal samples can be attributed to the columnar β grain morphology and the presence of grain boundary α layers. Because these boundaries can serve as a path
along which the voids can preferentially nucleation and coalescence [29]. In the horizontal samples, the tensile load perpendicular to the αGB layers act to separate the adjacent prior β grains. The columnar prior β grains result in different amounts of αGB layers which perpendicular to the tensile load [4]. Most of the αGB layers in the horizontal samples is perpendicular to the tensile load while a little amount of αGB layers is perpendicular to the tensile load in the vertical samples. For the sample H-DT, both the UTS and YS is substantially lower than that of the sample H-AT. However, the reduction in strength does not increase ductility. In contrast, the sample H-DT exhibits even poorer elongation and more feature of intergranular fracture. The presence of coarse αGB layers after double treatment may be the main factor for the poor performance because the coarser αGB layers can further reduce the cohesion between the adjacent β grains. As a result, brittle intergranular fracture occurred at sample H-DT when the tensile load is low. As shown in Fig. 8, Vickers hardness of the horizontal samples is nearly unchanged after tensile test while Vickers hardness of the vertical samples increased after tensile test. As work hardening is mainly achieved by the accumulation of dislocations, for the horizontal samples, the fine α laths have a limited capacity to store dislocations and the premature fracture further limits the strain hardening. Fig. 9 shows the SEM photographs close to the fracture surface of the four samples. As shown in Fig. 9a, the presence of molten
Fig. 8. Vickers hardness of the gauge section before and after tensile test (a) the horizontal samples, (b) the vertical samples. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)
210
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
Fig. 9. The side view close to the fracture surface (a) sample H-AT, (b) sample V-AT, (C) sample H-DT, (b) sample V-DT.
pool boundaries is a noticeable feature of sample H-AT. A previous study investigated the effects of molten pool boundaries on the mechanical properties of the selective laser melted 316L stainless steel and indicated that the molten pool boundaries are the main factor that causes the fracture [30]. In the present study, the fracture does not along the molten pool boundaries. As shown in Fig. 9b–d, some voids could be observed in the α laths indicating the intrinsic fracture of the α laths. More voids are observed in the α-β interfaces. The amounts of voids in the four samples are different. Little voids were observed in the horizontal samples which fractured in a brittle way. More voids can be observed in the vertical samples. This is because the vertical samples have low working-hardening rate and this promotes easier voids nucleation [31]. As shown in Fig. 9b, voids coalescence along the αGB layer can be observed. However, the cracks in the vertical samples cannot propagate along the grain boundary because they are mainly parallel to the tensile load direction. This explains why intergranular fracture is rare in the vertical samples. Fig. 10 shows the SEM photograph away from the fracture surface of sample V-DT. More voids can be observed in “α colony” region indicating more serious deformation. It can be seen that the voids usually nucleate at the interface of coarse α laths and β matrix due to the continued plastic deformation and a severe strain gradient across α-β interface [28]. As the circle indicated, voids coalescence along the coarse α laths can be observed. This kind of coarse α laths which does not parallel to the tensile load is similar to the αGB layers in the horizontal samples. As distinct from the continuous αGB layer, the cracks along the coarse α laths do not lead to fracture. Almost no voids can be observed at the α-β interfaces which almost parallel to the tensile load.
Fig. 10. Inhomogeneous deformation in the layer band structure of sample V-DT.
4.3. Inhomogeneous deformation The DIC contour maps of the sample V-AT are shown in Fig. 11. Different colors in gauge section indicate different deformation degree. As indicated in Fig. 11a, the deformation of prior β grains starts at the elastic deformation stage. With the increase of tensile load, inhomogeneous deformation between layer bands can be observed in Fig. 11b. Then, the strain concentrations take place in the weaker region of the sample. The larger plastic strain (red region) in Fig. 11c indicates two possible locations where necking may occur. Finally, fracture happened in the necking location. The inhomogeneous deformation can be attributed to the coarse columnar prior β grains and non-uniform α laths. The small oscillations in the second stage of θ vs εt curves (Fig. 5b) may also
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
211
Acknowledgements This work was supported by the Research Fund of the State Key Laboratory of Solidification Processing (NWPU), China (Grant No.151-ZH-2016) and National Key Technologies R&D Program (2016YFB11000100) and National Natural Science Foundation of China (Grant No. 51475380, 51323008).
References
Fig. 11. The DIC contour maps of sample V-AT. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)
reflect the inhomogeneous deformation of the vertical samples.
5. Conclusions The microstructure and anisotropic tensile behavior of LAMed TC21 titanium alloy under two different heat treatments were analyzed. The primary conclusions from the present study are as follows: (1) The samples in four different conditions exhibit similar columnar β grains and very different α laths. The horizontal samples exhibit uniform α laths. Layer bands can be observed in the vertical samples. The part geometry and heat treatment have significant effect on the characterization of α laths. (2) The tensile tests show significant anisotropic mechanical properties in two different heat treatment conditions. The horizontal samples exhibit stronger strength than the vertical samples. And the total elongation of the horizontal sample is much inferior to that of the vertical samples. (3) The ultimate tensile strength and yield strength decreased with the increase of α laths width. The elongation of the samples do not show regularity relationship with the strength. The inferior elongation of the horizontal samples is caused by a larger amounts of αGB layers which perpendicular to the tensile load. Coarsening of αGB layers can further reduce the ductility. (4) Vickers hardness test shows that the vertical samples were strengthened obviously after tensile test due to strain hardening. The horizontal samples were little changed after tensile test due to the premature fracture. (5) The coarse columnar prior β grains and the non-uniform α laths of the vertical samples resulted in inhomogeneous deformation.
[1] G. Lutjering, J.C. Williams, Titanium, Springer Science & Business Media, Berlin, Germany, 2007. [2] W.D. Huang, X. Lin, Research progress in laser solid forming of high-performance metallic components at the state key laboratory of solidification processing of China, 3D Print. Addit. Manuf. 1 (2014) 156–165. [3] A.R. Nassar, E.W. Reutzel, Additive manufacturing of Ti6Al4V using a pulsed laser beam, Metall. Mater. Trans. A 6 (2015) 2781–2789. [4] B.E. Carroll, T.A. Palmer, A.M. Beese, Anisotropic tensile behavior of Ti6Al4V components fabricated with directed energy deposition additive manufacturing, Acta Mater. 87 (2015) 309–320. [5] A.J. Sterling, B. Torries, N. Shamsaei, S.M. Thompson, D.W. Seely, Fatigue behavior and failure mechanisms of direct laser deposited Ti6Al4V, Mater. Sci. Eng. A 655 (2016) 100–112. [6] M. Gaumann, C. Bezencon, P. Canalis, W. Kurz, Single-crystal laser deposition of superalloys: processing-microstructure maps, Acta Mater. 49 (2001) 1051–1062. [7] C. de Formanoir, S. Michotte, O. Rigo, L. Germain, S. Godet, Electron beam melted Ti6Al4V: microstructure, texture and mechanical behavior of the asbuilt and heat-treated material, Mater. Sci. Eng. A. 652 (2016) 105–119. [8] B. Vrancken, L. Thijs, J.P. Kruth, J. Van Humbeeck, Microstructure and mechanical properties of a novel β titanium metallic composite by selective laser melting, Acta Mater. 68 (2014) 150–158. [9] S.S. Al-Bermani, M.L. Blackmore, W. Zhang, I. Todd, The origin of microstructural diversity, texture, and mechanical properties in electron beam melted Ti6Al4V, Metall. Mater. Trans. A. 41 (2010) 3422–3434. [10] A. Crespo, Modelling of Heat Transfer and Phase Transformations in the Rapid Manufacturing of Titanium Components. Convection and Conduction Heat Transfer, INTECH Open Access Publisher, 2011, pp. 315–340. [11] E. Brandl, F. Palm, V. Michailov, B. Viehweger, C. Leyens, Mechanical properties of additive manufactured titanium (Ti6Al4V) blocks deposited by a solid-state laser and wire, Mater. Des. 32 (2011) 4665–4675. [12] S.M. Kelly, S.L. Kampe, Microstructural evolution in laser-deposited multilayer Ti6Al4V builds: part I. Microstructural characterization, Metall. Mater. Trans. A 35A (2004) 1861–1867. [13] B. Baufeld, Effect of deposition parameters on mechanical properties of shaped metal deposition parts, Proc. Inst. Mech. Eng. B: J. Eng. 226 (2012) 126–136. [14] Y.Y. Zhu, X.J. Tian, J. Li, H.M. Wang, The anisotropy of laser melting deposition additive manufacturing Ti6.5Al3.5Mo1.5Zr0.3Si titanium alloy, Mater. Des. 67 (2015) 538–542. [15] F.D. Wang, S. Williams, P. Colegrove, A.A. Antonysamy, Microstructure and mechanical properties of wire and arc additive manufactured Ti6Al4V, Metall. Mater. Trans. A. 44 (2013) 968–977. [16] Z. Liu, Z.X. Qin, F. Liu, X. Lu, H.M. Wang, The microstructure and mechanical behaviors of the Ti6.5Al3.5Mo1.5Zr0.3Si alloy produced by laser melting deposition, Mater. Charact. 97 (2014) 132–139. [17] T. Vilaro, C. Colin, J.D. Bartout, As-fabricated and heat-treated microstructures of the Ti6Al4V alloy processed by selective laser melting, Metall. Mater. Trans. A. 42 (2011) 3190–3199. [18] C.L. Qiu, G.A. Ravi, C. Dance, A. Ranson, S. Dilworth, M.M. Attallah, Fabrication of large Ti6Al4V structures by direct laser deposition, J. Alloy. Compd. 629 (2015) 351–361. [19] Y.H. Wang, H.C. Kou, H. Chang, Z.S. Zhu, F.S. Zhang, J.S. Li, L. Zhou, Influence of solution temperature on phase transformation of TC21 alloy, Mater. Sci. Eng. A 508 (2009) 76–82. [20] Z.F. Shi, H.Z. Guo, J.Y. Han, Z.K. Yao, Microstructure and mechanical properties of TC21 titanium alloy after heat treatment, Trans. Nonferr. Met. Soc. China 23 (2013) 2882–2889. [21] Q. Zhang, J. Chen, P.F. Guo, H. Tan, X. Lin, W.D. Huang, Texture and microstructure characterization in laser additive manufactured Ti6Al2Zr2Sn3Mo1.5Cr2Nb titanium alloy, Mater. Des. 88 (2015) 550–557. [22] X. Wu, R. Sharman, J. Mei, W. Voice, Microstructure and properties of a laser fabricated burn-resistant Ti alloy, Mater. Des. 25 (2004) 103–109. [23] C.L. Qiu, G.A. Ravi, M.M. Attallah, Microstructural control during direct laser deposition of a β-titanium alloy, Mater. Des. 81 (2015) 21–30. [24] P.H. Li, W.G. Guo, W.D. Huang, Y. Su, X. Lin, K.B. Yuan, Thermomechanical response of 3D laser-deposited Ti6Al4V alloy over a wide range of strain rates and temperatures, Mater. Sci. Eng. A 647 (2015) 34–42. [25] S.A. Mantri, D. Choudhuri, A. Behera, J.D. Cotton, N. Kumar, R. Banerjee, Influence of fine-scale alpha precipitation on the mechanical properties of the beta titanium alloy beta-21S, Metall. Mater. Trans. A 46 (2015) 2803–2808. [26] J. Donoghue, A.A. Antonysamy, F. Martina, P.A. Colegrove, S.W. Williams, P.
212
Q. Zhang et al. / Materials Science & Engineering A 673 (2016) 204–212
B. Prangnell, The effectiveness of combining rolling deformation with wire-arc additive manufacture on β-Grain refinement and texture modification in Ti6Al4V, Mater. Charact. 114 (2016) 103–114. [27] Q. Zhang, J. Chen, H. Tan, X. Lin, W.D. Huang, Influence of solution treatment on microstructure evolution of TC21 titanium alloy with near equiaxed β grains fabricated by laser additive manufacture, J. Alloy. Compd. 666 (2016) 380–386. [28] R. Sarkar, A. Mukhopadhyay, P. Ghosal, T.K. Nandy, K.K. Ray, Effect of aged microstructure on the strength and work hardening behavior of
Ti15V3Cr3Sn3Al alloy, Metall. Mater. Trans. A. 46 (2015) 3516–3527. [29] C. Sauer, G. Luetjering, Thermo-mechanical processing of high strength β-titanium alloys and effects on microstructure and properties, J. Mater. Process. Technol. 117 (2001) 311–317. [30] S.F. Wen, S. Li, Q.S. Wei, C.Z. Yan, S. Zhang, Y.S. Shi, Effect of molten pool boundaries on the mechanical properties of selective laser melting parts, J. Mater. Process. Technol. 214 (2014) 2660–2667. [31] D. Banerjee, J.C. Williams, Perspectives on titanium science and technology, Acta Mater. 61 (2013) 844–879.