Tao Wang， Lei Zhu， Changhong Wang， Mingming Liu， Ning Wang，Lingchao Qin， Hao Wang， Jianbo Lei， Jie Tang and Jun Wu

（1. College of Aeronautical Engineering, Civil Aviation University of China, Tianjin 300300,China；2. Qingdao International Airport Group Co. Ltd., Qingdao 266000, China；3. Zhejiang Loong Airlines Co. Ltd., Hangzhou 311207, China；4. Center of Engineering and Technology,Civil Aviation University of China, Tianjin 300300, China；5. Research Center of Laser Technology, Tianjin Polytechnic University, Tianjin 300387, China）

Abstract: The dynamic mechanical properties of the Ti?6Al?4V (TC4) alloy prepared by laser ad?ditive manufacturing (LAM?TC4) under the high strain rate (HSR) are proposed. The dynamic compression experiments of LAM?TC4 are conducted with the split Hopkinson pressure bar(SHPB) equipment. The results show that as the strain rate increases, the widths of the adiabatic shear band (ASB), the micro?hardness, the degree of grain refinement near the ASB, and the dislo?cation density of grains grow gradually. Moreover, the increase of dislocation density of grains is the root factor in enhancing the yield strength of LAM?TC4. Meanwhile, the heat produced from the distortion and dislocations of grains promotes the heat softening effect favorable for the recrystalliz?ation of grains, resulting in the grain refinement of ASB. Furthermore, the contrastive analysis between LAM?TC4 and TC4 prepared by forging (F?TC4) indicates that under the HSR, the yield strength of LAM?TC4 is higher than that of F?TC4.

Key words: laser additive manufacturing；Ti?6Al?4V；dynamic mechanics properties；microstruc?ture

Because Ti?6Al?4V (hereinafter referred to as TC4) has excellent strength, plasticity, mater?ial toughness, and corrosion resistance[1?4], it is widely used for fabricating the fan blades and the compressor blades of aero?engines. During engine operation, foreign objects (such as gravel, birds,or tools) can be sucked in at a high speed, result?ing in deformation or fracture of the blades and seriously deteriorating aviation safety[5?8]. It is more economical to repair a damaged blade in?stead of replacing it with a new one.

In recent years, laser additive manufactur?ing (LAM), as an advanced repairing technology,has been gradually applied in aviation mainten?ance repair operation (MRO). Compared with traditional repairing techniques such as argon arc welding or thermal spraying, LAM has a smaller heat affected zone and results in better micro?structure of the repaired zone[9?13]. But there is another possibility: the blade repaired by LAM may be stuck again by foreign objects when it is reused in the aero?engine (as shown in Fig. 1),which may result in the deformation or fracture of the repaired region. Therefore, the study of the microstructure evolution and the dynamic mech?anical properties of LAM?TC4 under the high strain rate (HSR) is significant to evaluate LAM’s application in manufacturing or repairing aero?engine blades.

Fig. 1 Blades stroke by the foreign objects

Recently, most studies on the mechanical properties of LAM?TC4 focus on the quasic?stat?ic mechanical properties. For example, Gao et al.[14]and Li et al.[15]have reported the quasic?static mechanical properties of LAM?TC4 as well as the microstructure and fracture morphology. In their research, it has been found that the mechanical properties of LAM?TC4 are better than that of casting one that is cast and equivalent to one that is wrought annealed. In addition, some re?lated studies have concluded that the dynamic mechanical properties of metal material under HSR are much different from the quasi?static[16?18]. Currently, research on the microstruc?ture evolution and dynamic mechanical proper?ties of TC4 are mainly concentrated on TC4 pre?pared by forging (F?TC4). For example, Luo et al.[19]have discussed the anisotropy of mechan?ical properties of TC4 by hot rolling. Their re?search results have indicated that the yield strength is the lowest along the rolling direction,whereas it is the highest along the direction ver?tical to the rolling direction. Liu et al.[20]have proposed the effect of the content of alloying ele?ments on the dynamic mechanical properties of F?TC4, showing that both the yield strength and the strain of F?TC4 increase with the content in?crease of Al and V under HSR. Luo et al.[21]have found that the thickness of elongated α grains within ASB increases with the rise of deforma?tion temperature and that the dynamic recrystal?lization of α phase occurs in the warm compres?sion of the Ti?6Al?4V alloy. In the study of Zhou et al.[22], size effects have been studied by improv?ing the specimen structure for the split Hopkin?son pressure bar (SHPB) shear test to obtain dif?ferent shear band widths. Based on the shear testing results, the J?C constitutive model of ti?tanium alloy Ti?6Al?4V at an ultrahigh strain rate is modified. In the field of LAM, other works have not been retrieved yet except the research of Li et al[23]. They have pointed out that ASB can easily develop at all selected temperatures for LAM?TC4 samples under the high strain rate. In addition, the values of the ultimate strength and yield strength of LAM?TC4 are slightly lower than those of F?TC4 over the entire strain?rate range.

In a summary, current research on the dy?namic mechanical properties of LAM?TC4 and the microstructure evolution under HSR is rare.In the rest of this article, firstly, the specimens are prepared by LAM and impacted by the SHPB device under the different high strain rates. Secondly, the microstructure evolution in?cluding the adiabatic shear band (ASB), hard?ness, and grain size are analyzed. Afterwards, the dynamic mechanical properties under the differ?ent strain rates are discussed and compared with other research. Lastly, the conclusions are given.

1 Experiments

1.1 Preparing the specimens

Commercial TC4 titanium alloy is chosen as the substrate in the experiments and TC4 spher?ical powder, with chemical composition (wt %)0.3 Fe, 0.1 C, 0.05 N, 0.015 H, 0.2 O, 5.5 Al, 3.5 V and balance Ti, provided by BaoTi Group Co.,Ltd., as the material for LAM. The microscopic morphology of the spherical powder is shown in Fig. 2a, and the size of the powder ranges between 120 μm and 250 μm.

IPG fiber laser equipment is adopted in the experiment, as shown in Fig. 2b. The process parameters of the experiments after optimizing are as follows: laser power 2 000 W, laser scan?ning speed 10 mm·s–1, laser spot diameter 4 mm,overlapping rate 50%, and space between two neighboring layers 0.6 mm. The speed of synchro?nous powder feeding is set to 0.8 g/min, and the coaxial powder feeder device is shown in Fig. 2c.The LAM process is performed under the protec?tion of argon gas to ensure that the oxygen con?tent is less than 0.5%. The LAM test process is shown in Fig. 2d, and the prepared material block is shown in Fig. 2e.

After laser deposition with the coaxial powder feeder system, the LAM?TC4 is heated to 600 ℃ in the vacuum furnace. Heating to 600 ℃and cooling to room temperature can improve the internal stress and the anti?crack of the coating.This temperature is kept for 4 hours before the LAM?TC4 is naturally cooled to room temperat?ure in the vacuum furnace (shown in Fig. 2f). Then the cylindrical specimens for the SHPB experi?ments are sampled along the direction of laser deposition by the wire EDM machine, as shown in Fig. 2g. The length?to?diameter ratio of speci?mens should be from 0.5 to 1.0 so that the one?dimensional stress?strain status of specimens can?not be disturbed by the friction effect of the end surfaces and for the sake of strain and stress uni?formity of the specimen. The length and diamet?er of the specimens in the experiments used for the dynamic compression experiments are 6 mm and 6.5 mm, respectively, as shown in Fig. 2h.

1.2 Dynamic compressing experiment

The dynamic compression experiments of LAM?TC4 are performed at a range of strain rates: 1 000 s–1, 1 500 s–1, 2000 s–1, and 3 500 s–1,by the SHPB device at room temperature. The fundamental theory of the experiment is shown in Fig. 3a. The SHPB device and methods are in?troduced in detail in Ref. [24]. The striker bar and the incident bar share the same material and manufacturing processes as the transmission bar,in which the diameter of bar is 16 mm, the elasti?city modulus of bar is 200 GPa, and the density of the bar is 7 830 kg/m3. Based on the one?di?mensional stress wave theory, the strain rate, the strain, and the stress of specimens can be ob?tained through the following equations as

Fig. 3 Schematic illustration of SHPB and specimens after impacting

The size of the sample before and after the SHPB test is shown in Tab. 1. Some samples are cracked after the SHPB tests, and the cracked specimens are fractured in the direction of 45°from the loading direction, as shown in Fig.3b. In order to analyze the microstructure evolution of the LAM?TC4 under HSR, the unbroken speci?mens are sectioned along the cylinder axis by the wire EDM machine (illustrated in Fig.3c). Then the surfaces of samples are grinded, polished, and etched for 25 seconds in the corrosive reagent with a HF∶HNO3∶H2O volume ratio of 1∶2∶50. Scanning electron microscopy (SEM) is em?ployed to observe the microstructure of the sec?tioned specimens; Transmission electron micro?graphs (TEM) is used to observe the disloca?tions; X?ray diffraction (XRD) is used to test the phase, and a Vickers hardness tester is used to measure the micro?hardness.

Tab. 1 SHPB test results

2 Results and discussions

2.1 Microstructure evolution

The area observed by SEM is shown in Fig.4a. As shown in Fig.4b, the microstructure of the original LAM?TC4 is reticular with a large needle?like α phase and intergranular β phase.The widths of the α phases are about 1 μm and the lengths are about 10 μm, except that the lengths of a few α phases reach more than 30 μm.The widths of the β phases are about 0.2 μm and the lengths are from 5 μm to 13 μm. In addition,there are no obvious initial voids and lack?of?fusion pores. The microstructure of LAM?TC4 prepared in this paper is more homogenous than that of LAM?TC4 prepared by Li et al[23]. The microstructure of LAM?TC4 specimens prepared by Li et al.[23]was mainly Widmanstatten struc?ture with larger α phases and β phases together with obvious initial voids and lack?of?fusion pores.

The microstructures of LAM?TC4 under the different strain rates are shown in Figs. 4c?4f.It is observed that the adiabatic shear band(ASB) and the loading direction are at an angle of about 45 degrees, similar to the results of Li et al[23]. The local microstructure transformation of the material has occurred near both sides of the ASB and the grain orientation near the ASB be?comes more uniform. On the left of the ASB, the grains closer to the ASB have smaller angles to the ASB, indicating significant plastic flow of the grains. However, on the right of the ASB, the grains near the ASB are almost perpendicular to the ASB.

Furthermore, the local microstructure trans?formation of the material is different under the different strain rates. From region 1 in Fig.4c(1 000 s–1) to region 4 in Fig.4f (3 500 s–1), the degree of the local microstructure transformation increases continuously, and the transformation area becomes larger and larger. Illustrated in Figs. 4c?4f, corresponding to the local microstruc?ture transformation beside the ASB, the widths of the ASB at 1 000 s–1, 1 500 s–1, 2 000 s–1, and 3 500 s–1are about 0.8 μm, 1.7 μm, 3.2 μm, and 5.8 μm respectively, indicating that the width of the ASB increases gradually as the strain rate in?creases.

Fig. 4 Microstructure revolutions of LAM?TC4 under different strain rates

The grain sizes in the transformation re?gions under different strain rates are illustrated in Fig.5a. Obviously, the closer the grain to the boundary of the ASB, the smaller the size of the grain. In addition, the higher the strain rate, the smaller the size of the grain. The microstructure within the ASB (the strain rate is 3 500 s–1) is as shown in Fig. 5c. From Fig. 5d, it is observed that the size of grains is about 200 nm in the ASB and the grain morphology in the ASB is equiaxed and uniform, and manifesting in the center of ASB are the ultrafine grains[21].

As shown in Fig. 5b, the hardness of the ori?ginal LAM?TC4 material is about 410 HV; how?ever, the hardness of the specimen after being impacted at high speed increases obviously, indic?ating that the high strain rate leads to work hardening (WH)[25?27]. Moreover, the higher the strain rate, the harder the specimen after impact?ing. For one specimen, the hardness increases gradually from the outside to the center of the ASB and reaches the highest hardness within the ASB, which is in line with the research result of Zhan et al[28]. It indicates that the grains are finer in the ASB than outside.

Figs. 6a?6c show the TEM images of LAM?TC4 before and after the high?speed impacting.The low?density dislocations exist in the needle?like crystals of the original LAM?TC4 (Fig.6a),which are formed during the cooling process of the LAM. The high?density dislocations are formed in the specimen grains tightly close to the ASB at the strain rates of 1 000 s–1and 3 500 s–1(shown in Fig.6b) (1 000 s–1) and Fig.6c (3 500 s–1)).Obviously, the dislocations at the strain rate 3 500 s–1are more serious than at 1 000 s–1, indic?ating that the dislocation density of grains in?creases with the increase of strain rate. This may be because of the interaction between the hindered dislocation slipping and proliferation.The accumulation of high density dislocations could also greatly enhance the stress field in the grain boundaries. In addition, the dislocation slipping is greatly affected by the time effect at the high strain rate deformation of the material,resulting in the dislocations not having enough time to reach equilibrium.

Fig. 5 Micro?hardness and microstructure beside ASB

Fig. 6 TEM micrograph and dislocations of the LAM?TC4 before and after impacting

Fig. 6d shows the high density dislocation walls (DDW) and the slipping bands between the slipping bands. The width of the slipping band is from 100 nm to 200 nm. In Fig.6e, the orthogonal DDWs have cut a big grain into sev?eral varisized cell blocks (CBs). Moreover, there are still high density dislocations within the CBs,indicating that these CBs tend to be continually cut into smaller?size CBs. In addition, there ex?ist some nano?grains formed in the boundaries of the slipping bands and their sizes range from 10 nm to 50 nm (illustrated in Fig.6f). The dif?fraction patterns (DPs) shown in Fig.6g indicate that these nano?grains are α?Ti, the same as the matrix grain. The recrystallization nucleation mechanism is more similar to continuous dynam?ic recrystallization (CDRX), which is distinctive from rotational dynamic recrystallization(RDRX). Wang et al.[29]have pointed out that the RDRX mechanism is not possible in kinetics when the temperature of matrix is less than 573 K.The CDRX mechanism is a thermal activation process during the severe plastic deformation of the high stacking fault energy (HSFE) materials.For the HSFE materials, the dynamic recovery time after high?speed compressing is very short.Hence, the transition from the sub?grain bound?ary to the high?angle grain boundary does not need more energy.

The results of the specimen XRD test are il?lustrated in Figs. 7a?7d. The content (vol%) of the α phase of the original material is 83% and the content of the β phase is 17%. After the high?speed impacting, the content of the α phase at 1 000 s–1, 1 500 s–1, and 3 500 s–1are 98%, 97%,and 93%, respectively. Correspondingly, the con?tent of the β phase at 1 000 s–1, 1 500 s–1, and 3 500 s–1are 2%, 3%, and 7%, respectively. The content of the phase is mainly obtained by the Scherrer formula of the JADE software. The measurement error of the content of the phase is about 0.5%.

The change of phase content verifies that there exists temperature variation during the high?speed impacting. For a frictionless deforma?tion process, the maximum increase in temperat?ure can be described as

Yang et al.[31]have found that the adiabatic temperature within α?Ti ASBs can reach 776K.Because the temperature of the specimen rises very fast during compression and then declines slowly to room temperature, the cooling temper?ature gradient of the compression test process is smaller than the solidification of the laser depos?ition process, promoting more β phases to be transferred to α phases. Hence, in the study of Luo et al.[21], the specimens were immediately put into the cold water to increase the temperature gradient and retain the β phases formed at the high temperatures. In addition, the higher the strain rate, the higher the ASB temperature and the greater the cooling temperature gradient,causing the content of β phases of the 3 500 s–1specimen to be slightly higher than that of the specimens at 1 000 s–1and 1 500 s–1.

Fig. 7 XRD results of the LAM?TC4 before and after impacting

2.2 Mechanical properties

Fig. 8 compares the flow stress curves of LAM?TC4 under different HSRs in the experi?ments. It is found that flow stress is generally larger at the high strain rate than that at the low strain rate. In Fig. 8, the flow stress linearly in?creases quickly to the yield strength in the elast?ic deformation stage and then almost keeps steady or slightly increases before the localized necking in the plastic deformation stage. With the beginning of plastic deformation, the disloca?tions accumulate rapidly, resulting in obvious WH behavior, but the increasing rate of flow stress significantly decreases. Because of the rel?atively high stacking fault energy of LAM?TC4,dynamic recovery (DRV) induced by the disloca?tion cross slip and climb easily occurs. With fur?ther straining, the plastic deformation energy gradually accumulates and DRV behavior is en?hanced. Finally, the competition between the WH and DRV behaviors reaches a dynamic bal?ance. Then, the steady flow stress appears. In ad?dition, it is found that the strain rate has a great impact on the flow behaviors of LAM?TC4. The flow stress decreases with decreasing strain rate.The decrease of strain rate could increase the time for energy accumulation, which enhances dynamic recovery and decreases flow stress.

Fig. 8 Flow stress curves under the different strain rates

During the compressing of specimens by SHPB, WH and DRV are the main deformation mechanisms in plastic deformation stage. There?fore, based on the classic stress?dislocation rela?tion, the flow stress ( σ) can be described as[32]

where M is Taylor factor, α is the dislocation in?teraction constant, μ is the temperature?depend?ent shear modulus, b is the Burgers vector, and ρiis the average dislocation density. Obviously,σiincreases with increasing ρi, which results in an increase of flow stress σ.

In Fig. 8, the plastic deformation stage in?creases accordingly with the increase of the strain rate. In other words, the stress?strain curves show that the fracture strain increases with in?creasing strain rate. Figs. 9a–9c present SEM fractures of three LAM specimens deformed at the strain rates of 1 500 s–1, 2 000 s–1, and 3 500 s–1,respectively. The dimple?like features on the three fracture surfaces indicate that the domin?ant fracture mechanism is transgranular ductile failure[34]. Furthermore, by comparing the three fracture surfaces, it is observed that the dimple density increases with the strain rate increase.

Fig. 9 Morphology of fracture under different strain rates and the comparative analysis of the yield strength stress between the LAM?TC4 and F?TC4

Comparing the yield strength of LAM?TC4 fabricated in this paper with that of F?TC4[19]under HSR (illustrated in Fig. 9d), the yield stre?ngth of LAM?TC4 is significantly higher than that of F?TC4. This indicates that LAM?TC4 has better dynamic mechanical properties than F?TC4. By analyzing the microstructures of LAM?TC4, it is found that the average size of grains is about 10 μm, but the average size of grains of F?TC4 is about 16 μm. Because the microstructure is not given in Ref. [19], the microstructure of a block of commercial F?TC4 from BaoTi Group in this laboratory is observed by SEM, illustrated in Fig. 9e. It can be seen that the crystal grains of this F?TC4 are coarse and elongated compared with Fig. 4b. The average size of the grains in Fig. 9d is between 18 μm and 22 μm.

The relationship among the yield stress, the resistance of grain boundaries, and the grain sizes accords with the Hall?Petch law[35]as shown in

In the study by Li et al.[23], the yield streng?ths of LAM?TC4 prepared were 1 280 MPa at the strain rate 1 000 s–1and 1 480 MPa at 5 000 s–1.In this paper, the yield strengths are 1 320 MPa at 1 000 s–1and 1 520 MPa at 3 500 s–1. Obvi?ously, LAM?TC4 in this paper has better dynam?ic mechanical properties than that prepared by Li et al. As mentioned before, the microstructure of the LAM?TC4 prepared by Li et al. is mainly Widmanstatten structure where α phases are about 1?2 μm in width and dozens of microns in length and the length of the β grains can reach several millimeters. In addition, there are many α?laths structures with widths of 2.5?3.5 μm and lengths of 10?50 μm woven into basket?weave mi?crostructures. Particularly, there exist a few de?fects, including 1?10 μm initial voids in diameter with cycloidal or elliptic shape and lack?of?fusion pores without regular geometrical morphology or size, which could substantially degrade the mech?anical properties of the material. The voids are mainly formed when gas wrapped by hollow powders does not escape in time during the solid?ification process in the molten pool, or when there are impurities in the powders. The lack?of?fusion defects occur at the interface between two depositing layers or between the substrate and one depositing layer. It is resulted from that the proceeding layer cannot become completely re?molten during the subsequent depositing process when there is insufficient temperature in the mol?ten pool, or an excessively high powder?feeding rate causes the inadequate melting of the powders.

However, there are finer grains, more homo?geneous reticular microstructure, and no obvious voids or lack?of?fusion defects in the specimens of this research, maybe because the process of LAM is better than that of Li et al. In summary, the inhomogeneous microstructure, bigger grains, ini?tial voids and lack?of?fusion pores result in the worse dynamic mechanical properties of LAM?TC4 prepared by Li et al.

3 Conclusions

In this work, to understand the dynamic mechanical properties of TC4 prepared by laser additive manufacturing, the compression tests of LAM?TC4 specimens are systematically per?formed by the SHPB device over a wide range of high strain rates: 1 000 s–1, 1 500 s–1, 2000 s–1, and 3 500 s–1. Special attentions are paid to the ef?fects of high strain rate on the microstructure evolution behaviors and dynamic mechanical properties of LAM?TC4. The results are con?cluded below.

① Under the high strain rates adopted in this research, ASBs are formed in all specimens.Moreover, the width of the ASB increases as the strain rate rises. The grains become finer near the ASB; particularly, there exist about 200 nm nano?grains within ASB. The continuous dynam?ic recrystallization (cDRX) mechanism may res?ult in the formation of these fine grains within and near the ASB.

② During high?speed compression, high density dislocations come into being in the grains. Moreover, the dislocation density in?creases with the rise of the strain rate, leading to the strain rate strengthening effect of the LAM?TC4. Meanwhile, during the severe plastic de?formation of the LAM?TC4, much heat is gener?ated from the dislocations, which activates the cDRX mechanism.

③ Compared with the forging TC4 in the re?search of Luo et al.[19], the yield strength of the LAM?TC4 in this research is significantly higher under the high strain rate. In addition, the res?ults are also contrasted with that of TC4 pre?pared by the similar LAM process in the re?search of Li et al.[22], indicating the dynamic mechanical properties of LAM?TC4 in this paper are better than that prepared by Li et al.

④ By comparison and analysis, it is found that the microstructure of LAM?TC4 material in?cluding the grain size, the microstructure uni?formity, voids, and lack?of?fusion defects determ?ine its dynamic mechanical properties. In this re?search, by adopting the process parameters after optimizing and designing the appropriate craft process, finer grains and more homogeneous retic?ular microstructure are obtained. Furthermore,obvious voids and lack?of?fusion defects are avoided, resulting in better dynamic mechanical properties of the LAM?TC4 than that prepared by the congeneric LAM.