Selective Laser Melting of Near-α Titanium Alloy Ti-6Al-2Zr-1Mo-1V Parameter Optimization, Heat Treatment and Mechanical Performance

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Journal of Materials Science & Technology 57 (2020) 51–64

Contents lists available at ScienceDirect

Journal of Materials Science & Technology


journal homepage: www.jmst.org

Research Article

Selective laser melting of near-␣ titanium alloy Ti-6Al-2Zr-1Mo-1V:


Parameter optimization, heat treatment and mechanical performance
Chao Cai a , Xu Wu a , Wan Liu b , Wei Zhu a , Hui Chen c , Jasper Chua Dong Qiu d ,
Chen-Nan Sun d , Jie Liu a,∗ , Qingsong Wei a,∗ , Yusheng Shi a
a
State Key Laboratory of Materials Processing and Die & Mould Technology, Huazhong University of Science and Technology, Wuhan, 430074, China
b
Hubei Rehabilitation Assistive Technology Centre, Wuhan, 430074, China
c
Department of Mechanical Engineering, National University of Singapore, Singapore, 117575, Singapore
d
Singapore Institute of Manufacturing Technology, A*STAR, 73 Nanyang Drive, Singapore, 637662, Singapore

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents a comprehensive study conducted to optimize the selective laser melting (SLM)
Received 25 November 2019 parameters and subsequent heat-treatment temperatures for near-␣ high-temperature titanium alloy
Received in revised form 8 January 2020 Ti-6Al-2Zr-1Mo-1 V (TA15), which is widely used in the aerospace industry. Based on the surface mor-
Accepted 3 March 2020
phology and relative density analysis, the optimized process parameters were: laser power from 230 W
Available online 12 May 2020
to 380 W, scan speed from 675 mm/s to 800 mm/s, scan spacing of 0.12 mm, and layer thickness of
0.03 mm. The effects of the laser power and the layer thickness on the phase constitutions, microstruc-
Keywords:
ture features, as well as room-temperature and high-temperature (500 ◦ C) tensile properties, were then
Selective laser melting
Near-␣ titanium
studied to obtain an in-depth understanding of SLM-built TA15. Six typical temperatures (650, 750,
Heat treatment 850, 950, 1000 and 1100 ◦ C) covering three representative temperature ranges, i.e., martensite partial
High-temperature tensile properties decomposition temperature range, martensite complete decomposition temperature range and above ␤
transus temperature, were subsequently selected as heat-treatment temperatures. The heat treatment-
microstructure-mechanical property relationships of SLM-built TA15 were elucidated in detail. These
results provide valuable information on the development of SLM-built TA15 alloy for industrial appli-
cations, and these findings are also beneficial to additive manufacturing of other near-␣ Ti alloys with
desirable high-temperature properties.
© 2020 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science &
Technology.

1. Introduction porosity and composition segregation [3]. Machining of titanium


alloys is challenging owing to their poor thermal conductivity,
Titanium and its alloys are extensively applied in the aerospace, the propensity to harden under strain, a large amount of material
marine, defense industries, and biomedical devices as they have waste, and high manufacturing costs [4].
an excellent combination of outstanding mechanical strength, Selective laser melting (SLM) can be used to custom-fabricate
low density, extraordinary corrosion resistance, and remark- nearly fully-dense components directly from computer-aided-
able bio-compatibility. As a result, many titanium alloy parts design (CAD) models using a highly focused laser beam to melt
with good mechanical properties and complex geometries have metallic powders in a layer-by-layer manner selectively. SLM,
been designed and manufactured to meet the specific service allowing for a high degree of flexibility to achieve complex designs,
environments, especially those involving aerospace and defense shorter processing time, and high forming accuracy, can overcome
applications [1,2]. Currently, the manufacture of these titanium the limitations of conventional methods as above-mentioned [5].
alloy products mainly relies on casting and machining. Unfortu- Ever since the invention of the SLM process in the late 1990s, this
nately, cast titanium alloy parts generally have relatively weak promising technique has continuously attracted the attention of
mechanical properties resulting from metallurgical defects, e.g., academic and industrial researchers regarding the fabrication of
titanium alloys. Qiu et al. applied a square-island scanning strat-
egy to realize better homogeneous thermal distribution within
(␣ + ␤) type Ti-6Al-4 V parts, and then adopted hot isostatic press-
∗ Corresponding authors.
ing as a post-treatment to relieve thermal stress and close pores
E-mail addresses: hustliuj@hust.edu.cn (J. Liu), wqs xn@hust.edu.cn (Q. Wei).

https://doi.org/10.1016/j.jmst.2020.05.004
1005-0302/© 2020 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science & Technology.
52 C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64

Fig. 1. (a, b) Morphology and size distribution of the as-received TA15 powder, (c) an HRPM-II type SLM machine and the SLM scanning strategy, and (d) printed metallographic
and tensile specimens.

in the printed parts [6]. The resultant parts showed tensile proper- ponents with irregular geometries for the aerospace industry. The
ties comparable to those of thermo-mechanically processed and moderate strength of this alloy at both room and high tempera-
annealed counterparts. Xu et al. designed innovative SLM pro- tures makes it optimal for the fabrication of bulkheads and nacelle
cessing routes to decompose unflavored ␣ martensite into a center beam frames [18]. The development of advanced SLM tech-
lath-like (␣ + ␤) microstructure in-situ with tunable length scales nique provides new opportunities to manufacture such key TA15
ranging from 0.15 ␮m to 0.18 ␮m by controlling the tempera- components with geometrical complexity. Therefore, in this work,
ture field in the previously solidified layers within the 600−850 ◦ C a systematic investigation on optimizing the SLM process parame-
temperature range [7]. The SLM-built parts with such lath-like ters and post-heat treatments for TA15 was carried out to evaluate
(␣ + ␤) microstructures exhibited superior tensile strength, which the feasibility of the SLM technique as an alternative method to
markedly exceeded ASTM standards as well as the majority of manufacture this near-␣ Ti alloy. The study focused on investigat-
Ti-6Al-4 V samples fabricated by other additive manufacturing pro- ing and establishing an in-depth understanding of the correlations
cesses. Sabban et al. explored a novel heat treatment procedure, between microstructural characteristics and tensile performance
involving repeated thermal cycling close to but below the ␤ transus (at room temperature and 500 ◦ C) for SLM-built TA15. This research
(␤tr ) temperature, to eliminate acicular martensite and gener- should be meaningful for the additive manufacturing of specific
ate globular ␣ grains to improve room-temperature strength and near-␣ Ti alloys applied in high-temperature environments, and
toughness [8]. Apart from (␣ + ␤) type Ti-6Al-4 V alloy, the process- should further promote the SLM process as a potential technique
microstructure-property relationships of many SLM-built ␣-type for the fabrication of difficult-to-process components demanding
titanium alloys (e.g. pure titanium [9] and Ti-5Al-2.5Sn [10]) and high mechanical properties in the aerospace field.
␤-type titanium alloys (e.g. Ti-24Nb-4Zr-8Sn [11], Ti-13Nb-13Zr
[12], and Ti-5Al-5V-5Mo-3Cr [13]) have also been investigated. 2. Experimental procedures
The studies mentioned above mainly focus on how to improve
room-temperature mechanical performance by optimizing the 2.1. Powder material
processing parameters and post-heat treatments. However, cer-
tain near-␣ Ti alloys are specially designed for harsh and Gas-atomized TA15 powder particles, showing near-spherical
high-temperature environments encountered in the aerospace shape with a mean size (D50 ) of 34.9 ␮m (Fig. 1(a) and (b)), were
applications. Evaluating their mechanical properties at the cor- purchased from AMC Powders Company, China.
responding usage temperatures is more valuable than evaluating
their room-temperature properties for technical guidance and
2.2. SLM experiment and heat treatment
industrial applications [14–17]. However, studies that report on
the high-temperature mechanical performance of SLM-fabricated
In this work, laser power (180 W to 380 W), scan speed
near-␣ Ti alloys are few.
(300 mm/s to 800 mm/s), layer thickness (0.03 and 0.06 mm)
Ti-6Al-2Zr-1Mo-1 V (TA15), is a type of near-␣ titanium alloy
and constant scan spacing (0.12 mm) were adapted on our self-
that is widely used in the large-scale production of integral com-
developed machine for sample preparation using a baseplate
C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64 53

Fig. 2. Chart of SLM processing parameters for TA15 alloy. Three typical optical metallographic (OM) images demonstrate over-melted, well-melted, and porous top surfaces
of printed samples, which are represented by red, green, and blue, respectively.

Fig. 4. X-ray diffraction patterns of the as-received powder and the SLM-processed
samples with different laser power.
Fig. 3. Relative density histogram of the printed samples.

tron backscatter diffraction (EBSD) analysis, the samples were first


preheating temperature of 200 ◦ C and the 67◦ scanning direction re-mechanically polished, and then electrolytically polished in 8
rotation between successive layers [19]. Cubes with a dimension of % perchoric acid and 92 % ethanol solution at 27 V under 25 ◦ C
5 mm × 5 mm × 5 mm were printed using these processing param- for 25 s. A scan speed of 5◦ /min was conducted over a wide range
eters for density testing. Then, metallographic specimens, and of 2 = 20◦ –80◦ on all samples. Transmission electron microscopy
room-temperature and high-temperature tensile samples, were (TEM) was conducted using an FEI Tecnai G20 machine.
fabricated using the preliminary optimized processing parameters, Room-temperature and high-temperatures (500 ◦ C) tensile
as shown in Fig. 1(d). tests were carried out on SLM-built samples with and without heat
Heat treatments were applied to the samples manufactured treatment using a Zwick/Roell Z010 machine with a cross-head
using the optimal SLM processing parameters. Six representative speed of 1 mm/min.
temperatures of 650, 750, 850, 950, 1000, and 1100 ◦ C were selected
as heat-treatment temperatures under a heating rate of 10 ◦ C/min, 3. Results
a holding time of 2 h and furnace cooling.
3.1. Effects of SLM processing parameter
2.3. Microstructural and mechanical testing
3.1.1. Preliminary processing parameter study
The densities of the samples were determined using Archimedes The metallurgical quality, densification level, and associated
method. The SLM-built samples were tested by an x-ray diffrac- mechanical performance of SLM-printed metal parts are dependent
tion (XRD) machine equipped with a Cu anticathode (␭ = 1.5406 Å) on two main parameters: laser power and scan speed [20,21]. Thus,
operated at 35 kV and 40 mA under a continuous scan mode for these two crucial parameters (laser power from 180 W to 380 W
phase identification. The metallographic samples were ground, pol- and scan speed from 300 mm/s to 800 mm/s) were chosen to print
ished and etched with Kroll reagent for 20 s. The microstructural 5 mm × 5 mm × 5 mm block samples for preliminary processing
observations were characterized by Quanta650 FEG, and the crys- parameter optimization. Based on the top-surface morphologies of
tallographic orientation of the samples was examined using Sirion the printed samples, the processing window was divided into three
200 instrument with a TSL/EDAX system (Holland). Before elec- representative zones, i.e., high energy density zone (Zone A), suit-
54 C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64

Fig. 5. Low-magnification side-surface morphologies of the printed samples: (a) S17, (b) S18, (c) S19, and (d) S20.

able energy density zone (Zone B), and insufficient energy density ing/solidification of TA15 alloy during the SLM process, leading to
zone (Zone C), as shown in Fig. 2. High laser power coupled with the formation of this non-equilibrium phase [6].
low scan speed, resulted in high energy densities, causing over- Fig. 5 shows the low-magnification side-surface morpholo-
melted top surfaces and microscopic balling effect, which were also gies of samples S17, S18, S19, and S20. Martensite-dominated
reported in SLM-built Ti-6Al-4 V and Ti-5Al-2.5Sn alloys [22,23]. An microstructures, consisting of a substantial amount of needle-like
adequate energy density input into the TA15 powder bed formed ␣ martensite with a nano-sized width and a length below 200 ␮m
stable melt pools with proper overlap ratios, successive tracks, and embedded within the columnar prior-␤ phase were observed in all
a dense surface. An insufficient energy input resulted in melt pools samples. Other than melting the current powder layer, the laser can
with a relatively narrow width, giving rise to inadequate over- penetrate through several previously deposited layers during the
lap ratios and weak metallurgical bonding between adjacent melt layer-wise melting process, which leads to the solidification and
tracks. epitaxial growth of the prior-␤ phase along the steep temperature
The relative density of all printed samples (25 in total) was gradient in the building direction. It should be noted that there
measured by the Archimedes’ principle to further determine the was an overall increasing tendency in the width of the columnar
optimal processing parameter range, and the results are shown in prior-␤ grain boundaries from approximately 80 ␮m–200 ␮m as
Fig. 3. Printed samples S14, S17-S20, and S23-S25 had high densities laser power increased. The width of the columnar prior-␤ grains
of greater than 99.5 %. Therefore, according to the top surface mor- is mainly determined by the width of the melt pool and its heat-
phology and bulk density, a laser power from 230 W to 380 W and affected zone. It is reasonable to state that the width of the melt
a scan speed from 675 mm/s to 800 mm/s were deemed as the opti- pool and its heat-affected zone are proportional to the laser power
mized parameter ranges for SLM-built TA15 alloy. Because of the if all other parameters are kept constant. Moreover, most of the aci-
narrow optical scan speed range, the broad laser power range has a cular martensitic ␣ phase within the same prior-␤ grain showed
more significant impact on the microstructure and mechanical per- an approximate orientation of ± 45◦ to the prior-␤ grain bound-
formance of the SLM-built samples. Hence, the effect of laser power aries, because of the Burgers orientation relationship between the
on the microstructure and mechanical performance of samples S17 ␤ and ␣ phases during the fast solidification SLM process [6,24].
to S20 was studied further. To observe the microstructure in more detail, sample S17
was observed under high-resolution TEM (HRTEM), as shown in
Fig. 6. Apart from the needle-like ␣ matrix, several parallel fine
3.1.2. Effect of laser power on microstructure and mechanical twins with a nanoscale thickness could be distinctly observed and
properties were also verified by the HRTEM measurement (Fig. 6(d)). These
Fig. 4 shows the XRD spectrum of the as-received powder and nanoscale twins could be classified into two different categories:
the SLM-built samples (S17, S18, S19, and S20). XRD peaks corre- deformation twins and annealing twins. The former are induced
sponding to the ␤ phase were not detected, and overall diffraction by the residual stress from the mismatched cooling rates between
peaks of HCP ␣ -Ti could be identified as a result of the rapid melt-
C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64 55

Fig. 6. TEM results of SLM-built sample S17: (a) bright-field image, (b) magnified image of nano-size twins in Area A, (c) HRTEM image of nano-size twins in Area B, (d)
magnified image of needle-like structures, and (e), (f), and (g) correspond to the SAD patterns of points I, II, and III in (d), respectively.

the top layer and the previously solidified layers during the SLM Table 1
Room-temperature and high-temperature tensile properties for the printed TA15
process, while the latter result from the previously solidified lay-
alloy samples under varied SLM processing parameters and heat treatments.
ers being re-heated or partially re-melted again during melting
of the top layer, which has a similar effect as an annealing treat- Samples Room temperature 500 ◦ C
ment. Nanoscale twins are beneficial to enhance the strength and UTS (MPa) EI (%) UTS (MPa) EI (%)
ductility of the SLM-built TA15 simultaneously, which was demon-
S17 1422.1 ± 50.2 9.5 ± 1.5 990.2 ± 2.2 15.6 ± 1.9
strated and discussed in our previous study [19]. The ␣ and ␣ phases S18 1362.6 ± 46.2 9.3 ± 0.4 949.6 ± 19.6 13.6 ± 3.2
possess a similar HCP crystalline structure and characteristic XRD S19 1349.8 ± 26.7 5.6 ± 0.1 909.4 ± 40.1 12.5 ± 0.4
diffraction peaks. It is difficult to distinguish between the needle- S20 1063.6 ± 136.9 7.5 ± 0.5 939.5 ± 20.7 15.3 ± 1.7
like ␣ and ␣ phase structures using only the XRD examination; the S26 1234.2 ± 53.1 7.3 ± 0.7 907.5 ± 4.9 11.8 ± 0.4
S26 + HT1 1207.3 ± 3.8 8.9 ± 0.1 853.2 ± 10.7 12.5 ± 1.1
␣ phase has a crystal structure with a c/a ratio of approximately S26 + HT2 1123.6 ± 14.0 11.3 ± 0.7 757.1 ± 7.0 13.4 ± 1.1
1.58, which is slightly lower than that of the ␣ phase (1.59–1.60) S26 + HT3 1024.9 ± 19.5 9.0 ± 0.4 715.3 ± 4.3 13.0 ± 0.9
[25]. From the selected area diffraction (SAD) results shown in S26 + HT4 936.5 ± 4.9 8.0 ± 1.0 639.8 ± 6.7 14.3 ± 0.9
Fig. 6(e)-(g), the c/a values of the successive lath-like structures S26 + HT5 906.9 ± 5.5 5.8 ± 0.2 602.0 ± 5.6 15.9 ± 5.0
S26 + HT6 780.6 ± 7.2 3.8 ± 0.4 596.3 ± 5.6 15.6 ± 2.5
I, II, and III were 1.581, 1.578, and 1.575, respectively, confirming
Wrought TA15 [24] 965 18.8 680 24.8
that the needle-like structures are the ␣ phase.
Top-surface EBSD crystal orientation and inverse pole figure
(IPF) maps of the four samples are shown in Fig. 7. The overall
microstructures of these four samples showed similar fine grains, ature and 500 ◦ C. Sample S17 exhibited the highest UTS (1422 MPa
presenting basket-weave morphologies consisting of interlaced and 990 MPa) and EL (9.5 % and 15.6 %) at room temperature
needle-like martensitic phase. and 500 ◦ C, respectively. The room-temperature tensile strength of
Table 1 summarizes the ultimate tensile strength (UTS) and these samples continuously decreased with increasing laser power.
elongation (EL) of samples S17, S18, S19, and S20 at room temper- The high-temperature tensile strength did not exhibit an evi-
dent declining tendency, suggesting that room-temperature tensile
56 C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64

Fig. 7. Top-surface EBSD grain orientation and IPF maps of TA15 samples: (a) S17, (b) S18, (c) S19, and (d) S20.

strength was more sensitive to laser power than high-temperature strate a significant difference among the four samples. All samples
tensile strength for SLM-built TA15 alloy. presented a ductile fracture behavior with coarse and deep dimples
The fracture surfaces of the tensile specimens at room temper- observed on the fracture surface.
ature are shown in Fig. 8. The fracture surface of sample S17 was
almost fully covered by near-equiaxed dimples, indicative of a duc-
3.1.3. Effect of layer thickness on microstructure and mechanical
tile fracture feature and good ductility. For sample S18, the fracture
properties
surface was also mainly decorated by dimples, along with some
The layer thickness increment is a more effective method to
near-spherical pores, suggesting a relatively weaker metallurgi-
shorten printing time and improve forming efficiency in contrast to
cal bonding. This is one of the reasons why the tensile strength
increasing the scan speed or the hatch spacing. Hence, in the follow-
of sample S18 was slightly lower than that of sample S17. With
ing study, the layer thickness was increased from 30 ␮m to 60 ␮m
the increase of laser power to 330 W (sample S19), the fracture
and the remaining parameters were kept the same as those used
morphology transformed into a brittle rupture; a high number of
for S17, as this sample possessed the best tensile properties among
typical river-like cleavage patterns could be seen. When the higher
the other printed parts manufactured using optimized parameters.
laser power of 380 W was applied, the fracture surface of sample
The sample printed with 60 ␮m layer thickness is denoted as S26
S20 consisted of cleavage steps, tear ridges, fine spherical pores,
hereafter.
and near-equiaxed dimples, implying a brittle-ductile mixed-mode
The side-surface OM images of the microstructures of samples
fracture behavior. All fracture morphologies were consistent with
S17 and S26 are shown in Fig. 10. The length of prior-␤ phase
the tensile results presented in Table 1.
boundaries spanned the entire surface along the build direction for
The high-temperature fracture surfaces of the four specimens
sample S17, while the length in sample S26 was less than 200 ␮m.
are shown in Fig. 9. Compared with the room-temperature fracture
This occurs because the layer-wise tracks break off the link of the ␤
surfaces, the high-temperature fracture features did not demon-
phase along the printing direction. This implies that for sample S26,
C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64 57

Fig. 8. Fracture surfaces of the four samples: (a) S17, (b) S18, (c) S19, and (d) S20 after tensile testing at room temperature.

Fig. 9. Fracture surfaces of the four samples: (a) S17, (b) S18, (c) S19, and (d) S20 after tensile testing at 500 ◦ C.
58 C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64

Fig. 10. Side-surface OM microstructures of samples under various layer thicknesses: (a) S17 with 30 ␮m layer thickness and (b) S26 with 60 ␮m layer thickness.

3.2. Effect of heat-treatment temperature

The room-temperature and high-temperature tensile strength


values of sample S26 were slightly lower than those of sample S17,
yet were 27.9 % and 33.4 % higher than those of the wrought sample
[26], respectively. More importantly, the manufacturing time can
be shortened by half when the layer thickness is increased from
30 ␮m to 60 ␮m. In this regard, the S26 parameters are more time-
saving and are more likely to be applied in industrial production.
Consequently, S26 samples were selected for the heat-treatment
temperature optimization study.
Martensite begins to transform into the ␣/␤ structure at 575 ◦ C
[7], and the complete decomposition of martensite is obtained at
temperatures above 800 ◦ C [27]. The ␤tr temperature of TA15 alloy
was calculated using the linear regression analysis proposed by Guo
et al., given as follows [28]:

Fig. 11. X-ray patterns of untreated S26 sample and S26 samples with heat treat- ␤tr = 882 + 21.1[Al]–6.9[Zr]–9.5[Mo]–11.8[V]
ments HT1 (650 ◦ C), HT2 (750 ◦ C), HT3 (850 ◦ C), HT4 (950 ◦ C), HT5 (1000 ◦ C), and
HT6 (1100 ◦ C). Here [M] is the weight fraction of the alloy element M. For the
chemical composition of Ti-6Al-2Zr-1Mo-1 V, the ␣tr temperature
of 973.5 ◦ C was determined by this equation. Based on the ini-
tial martensite decomposition temperature of 575 ◦ C, the complete
decomposition temperature of 800 ◦ C, and the ␤tr temperature
the top layer absorbs most of the laser energy, resulting in insuf- of 973.5 ◦ C, six different temperatures in the three representative
ficient laser energy available to penetrate through the previously temperature ranges were chosen and applied in the following heat-
deposited layers. It is also difficult for this laser energy to gener- treatment experiments.
ate a sufficient temperature gradient for ␤ phase growth along the
vertical direction.
(i) Heat-treatment temperatures of 650 ◦ C and 750 ◦ C in the
The tensile strength of sample S26 at room temperature and
martensite partial decomposition temperature range.
500 ◦ C was slightly lowered by 13.2 % and 8.4 %, respectively,
(ii) Heat-treatment temperatures of 850 ◦ C and 950 ◦ C in the
in comparison to those of sample S17 (Table 1). It is notewor-
martensite complete decomposition temperature range.
thy that the tensile strength of sample S26 was still superior to
(iii) Heat-treatment temperatures of 1000 ◦ C and 1100 ◦ C in the
its wrought counterpart at both room temperature and 500 ◦ C,
range above the ␤tr temperature.
whereas the ductility of both S17 and S26 was lower than that
of the wrought one [26]. High tensile strength coupled with poor
ductility is a common phenomenon for SLM-built metallic parts, The heat-treated samples at 650, 750, 850, 950 1000, and
especially for titanium alloys. The ␣ martensitic phase is a hard and 1100 ◦ C are denoted as HT1, HT2, HT3, HT4, HT5, and HT6, respec-
brittle non-equilibrium phase that is weak in resisting crack initi- tively. X-ray diffraction patterns of untreated S26 and heat-treated
ation and propagation, leading to SLM-built titanium alloys with samples are shown in Fig. 11. All diffraction peaks of ␣ /␣ phases
superb strength but low elongation. The undesired ␣ martensitic could be detected in all samples, and a weak diffraction peak for
phase can decompose into a more stable and ductile lath-like ␣/␤ the (110) plane of the ␤ phase at 38.4 ◦ C was found in the post-
structure above its decomposition temperature. Accordingly, six treated samples HT3, HT4, HT5, and HT6. This result indicated that
different heat-treatment temperatures were chosen to realize ␣ (i) the minor amount of ␤ phase in samples HT1, and HT2 was
martensitic phase transformation into an ␣/␤ mixed structure with insufficient to produce observable diffraction peaks in the XRD
tunable length and to obtain an attractive combination of strength examination, owing to the incomplete decomposition of the ␣
and ductility for SLM-built TA15, as discussed in the following sec- phase; (ii) the HCP-Ti phase was still the dominant phase, even
tion. for samples treated above the ␤tr temperature.
C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64 59

Fig. 12. Microstructures of untreated S26 sample and heat-treated samples in the partial martensitic decomposition temperature range: (a, b) untreated S26, (c, d) HT1
(650 ◦ C), and (e, f) HT2 (750 ◦ C). White and yellow arrows indicate needle-like structure and lath-like structures, respectively.

The microstructures of untreated S26 sample and heat-treated the ␣ /␣ phases, resulting in the nano-sized ␤ phase forming within
samples HT1 and HT2 are shown in Fig. 12. Partial needle-like struc- the lath-like structures.
tures were retained and the decomposed needle-like ␣ martensitic When the heat-treatment temperatures were further increased
phase transformed into fine lath-like ␣/␤ structures with hundred- to the martensite complete decomposition temperature range, the
nanometer thickness in HT1 sample. When the heat treatment needle-like ␣ structure disappeared, and lath-like ␣/␤ structures
temperature was increased to 750 ◦ C, most of the needle-like struc- were homogeneously distributed within samples HT3 and HT4, as
tures faded away and turned into a mixture of microstructures shown in Fig. 13. When the heat-treatment temperatures were
comprised of the nano-sized ␤ phase embedded within the lath- increased to above the ␤tr temperature, as shown in Fig. 14, the
like ␣/␤ structure spacing in sample HT2. The transformation from microstructures of samples HT5 and HT6 showed a typical Wid-
the ␤ phase into the ␣ phase during the SLM process is a diffu- manstatten structure composed of coarse lamellar and lath-like
sionless transformation because of the super high cooling rate of structures with greater than 10-micrometer thickness, which was
104 -106 ◦ C/s, causing the supersaturated ␤-phase stabilizing ele- much coarser than that of the heat-treated samples below the
ments (Zr, Mo, and V) to become trapped in the ␣ phase. During ␤tr temperature. This is because the atom diffusion coefficient in
the heat treatment, the ␣ phase will nucleate along the needle-like HCP ␣-Ti is two orders of magnitude lower than that in BCC ␤-Ti,
␣ boundaries, and ␤-phase stabilizing elements are expulsed from and the grain will be much easier to grow once the heat temper-
60 C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64

Fig. 13. Microstructures of heat-treated samples in the complete martensitic decomposition temperature range: (a, b) HT3 (850 ◦ C) and (c, d) HT4 (950 ◦ C).

Table 2 4. Discussion
Microstructure type and constituent phase of the S26 sample and heat-treatment
samples in this study.
4.1. Laser power-microstructure-tensile property relationships
Sample Microstructure type and constituent phase

S26 Full needle-like ␣ martensitic phase From the tensile results (Table 1), it can be concluded that
HT1 Fine lath-like (␣ + ␤) with needle-like ␣ martensitic phase sample S17 exhibited the best tensile strength and elongation com-
HT2 Fine lath-like (␣ + ␤) with minority needle-like ␣ martensitic phase bination compared to the three other optimized samples (S18,
HT3 Full lath-like (␣ + ␤)
S19 and S20). The room-temperature tensile strength gradually
HT4 Full lath-like (␣ + ␤)
HT5 Lath-like (␣ + ␤) with minority coarse lamellar (␣ + ␤) decreased with increasing laser power, even though all the sam-
HT6 Coarse lamellar (␣ + ␤) with minority lath-like (␣ + ␤) ples possessed an identical martensitic phase and a similar average
grain size (Fig. 7). This result can be attributed to three aspects as
follows.
ature exceeds the ␤tr temperature [15]. This can be verified by
EBSD results, as shown in Fig. 15. The average grain size of sam- (i) Relative density
ple HT5 was approximately twice as large as that of sample HT4
even though the heat-treatment temperature was only increased A denser SLM-built part with less porosity is favorable to obtain
by 100 ◦ C. The variation of microstructure types and phase con- a better combination of strength and ductility. The relative den-
stituents for heat-treated samples is summarized in Table 2. sities for samples S17, S18, S19 and S20 (Fig. 3) were 99.97 %,
A comparison diagram of tensile properties for the S26 sam- 99.81 %, 99.44 %, and 99.36 %, respectively, Additionally, the poros-
ples with and without heat treatments at room temperature and ity of samples S17, S18, S19, and S20 calculated from OM images
500 ◦ C is shown in Fig. 16. For the room-temperature tensile proper- via Image-Pro Plus 6.0 software was 0.01, 0.024, 0.071, and 0.145,
ties, the tensile strength of the heat-treated samples was inversely respectively. The OM porosity results corresponded well with the
proportional to the heat-treatment temperature. The elongation Archimedes’ results and with room-temperature tensile strength
of the heat-treated samples increased and reached its highest results.
value with the increase of heat-treatment temperature at the HT2
temperature, but then began to decrease for all the heat-treated (ii) Metallurgical quality
samples gradually. The high-temperature tensile strength of the
post heat-treated samples exhibited a linear improvement as the From the room-temperature fracture surfaces (Fig. 8), many
heat-treatment temperature increased. In contrast, the ductility of dozens of micron-sized pores were observed in the S18 fracture
the heat-treated samples showed an opposite trend compared to surface, which was a typical inter-particle facture feature stemming
the high-temperature tensile strength. from the weak metallurgical bonding among the melted powders.
C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64 61

Fig. 14. Microstructures of heat-treated samples above the ␤tr temperature: (a, b) HT5 (1000 ◦ C) and (c, d) HT6 (1100 ◦ C). Yellow arrows indicate the lamellar structure.

In addition, a large number of pores with several micrometers in effect of hierarchical martensitic structures and a decrease in the
size, similar to the as-received powder size were distributed on the tensile strength.
S19 and S20 facture surfaces. These findings imply that many met-
allurgical defects existed in samples S18, S19, and S20. Gu et al. 4.2. Heat treatment-microstructure-tensile property
reported that high energy density would cause overheating and relationships
instability in the melt pool as well as a reduction in the surface
energy of the melt pool, resulting in the splashing of small-sized (i) Reasons for the variation in the tensile strength of the heat-
liquid droplets from the front of the liquid melt pool [22]. After treated samples
these small-sized droplets solidify, they will impede the uniform
spread of fresh powder on the previous layer and tend to induce
A decreasing trend in the tensile strength under both room-
further defects such as porosity and cracks. This is also the reason
temperature and high-temperature tensile testing conditions for
why the porosity of the printed samples continuously increases as
the heat-treated samples was shown in Fig. 16. This trend is mainly
the laser power increases.
associated with ␣ martensitic decomposition and grain growth.
The reduction in tensile strength for the heat-treated samples HT1
(iii) Hierarchical martensitic structures and HT2 originates from the hard and brittle martensite partial
decomposition into ␣ and ␤ phases. When heat-treatment tem-
From the high-magnification morphologies of printed samples peratures exceeded 800 ◦ C, the martensite completely decomposed
S17, S18, S19, and S20 (Fig. 17), hierarchical ␣ martensitic struc- into ␣/␤ phases, and the average grain size became coarser with the
tures could be observed. Small colonies composed of numerous increase in the heat-treatment temperature, which resulted in the
aligned fine ␣ martensite (white arrows) were embedded into decrease in tensile strength of samples HT3, HT4, HT5, and HT6.
the neighboring relatively coarser ␣ martensite spacing (yellow
arrows). These fine ␣ martensite colonies tended to form per- (ii) Reasons for the variation in the ductility of the heat-treated
pendicular to the coarser ␣ martensite. Yang et al. proposed that samples
the formation of hierarchical martensitic structures in SLM-built Ti
alloys were attributed to the previously deposited layers experienc- Room-temperature and high-temperature ductility showed dif-
ing multiple thermal cycles during the melting process of the top ferent trends in the six heat-treated samples (Fig. 16).
layer(s) [29]. They argued that this interlaced hierarchical structure In the room-temperature tensile results, the ductility initially
could be a strong barrier to the movement of dislocations and thus improved in samples HT1 and HT2, as most of the brittle needle-like
enhances the strength of printed samples. However, the density ␣ martensitic phase had been replaced by soft ␣/␤ lath structures
of fine ␣ martensite gradually decreased with increasing the laser after heat treatments. Then, the ductility continuously decreased
power, as shown in Fig. 17, causing a decrease in the strengthening for the samples treated below the martensite complete decom-
62 C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64

Fig. 15. Top-surface EBSD grain orientation and grain size distribution maps: (a, b) untreated S26, (c, d) HT2, (e, f) HT4, and (g, h) HT5 samples.

position temperature range and for those treated above the ␤tr improvements in ductility brought about by the ␣ phase transfor-
temperature. It should be noted that although the ␣ martensitic mation into ␣ and ␤ phases sample HT3. In other words, sample
phase completely transformed into the desirable soft ␣/␤ lath HT2 obtained an excellent combination of strength and ductility
structure in sample HT3, its ductility was still lower than that of because of the synergistic effect of fine lath-like (␣ + ␤) and minor-
sample HT2. HT2 was heat treated at a temperature of 750 ◦ C, which ity needle-like ␣ martensitic structures. When the heat-treatment
is close to the martensite complete decomposition temperature of temperature further increased, a reduction in ductility resulted
800 ◦ C, which means that most of the ␣ martensite has decom- from the grain growth.
posed into ␣/␤ lath and less ␣ martensite is retained in sample The high-temperature ductility showed a similar decreasing
HT2 (Fig. 12). With the increase of heat-treatment temperature, trend for the room-temperature test results of samples HT1, HT2,
the average grain size can increase, which is detrimental to elon- and HT3. However, when the heat-treatment temperature was
gation. Thus, it can be speculated that grain coarsening offsets the above 950 ◦ C, the high-temperature ductility presented an opposite
C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64 63

5. Conclusions

Highly-dense SLM-built TA15 parts with a desirable combina-


tion of tensile strength and ductility were successfully fabricated
via the optimization of printing parameters and subsequent heat
treatments in this work. The following conclusions can be drawn.

(1) The SLM processing window for TA 15 alloy could be divided


into three representative zones, i.e., high energy density zone,
suitable energy density zone, and insufficient energy density
zone. The laser power (230 W to 380 W) and the scan speed
(675 mm/s to 800 mm/s) were the optimized processing param-
eter range for the SLM-built TA15 alloy because of suitable
melting-pool overlap ratios, dense surface morphologies, and
high relative densities.
Fig. 16. Comparison diagram of room-temperature and high-temperature tensile (2) In the laser power study, the SLM-built sample with 230 W laser
properties for the S26 samples without and with heat treatments. The blue frames power (S17) showed the best tensile strength and elongation
show the corresponding tensile testing temperatures. The five-pointed stars and combination than compared to the three other optimized-
squares represent tensile strength and elongation, respectively.
parameter samples (S18, S19, and S20), which resulted from
the joint effect of relative density, metallurgical quality, and
hierarchical martensitic structure.
trend compared to the room-temperature ductility. This is perhaps (3) A slight decrease in the room-temperature and high-
because HT4, HT5, and HT6 samples possess a higher amount of temperature tensile strength values were observed for sample
␤ phase, as these three heat treatment temperatures are close to S26 with 60 ␮m layer thickness compared to those of sample
or above the ␤tr temperature. The BCC-␤ phase with 12 slip sys- S17 with 30 ␮m layer thickness when other processing param-
tems has better plastic deformation capacity than the HCP-␣ phase eters were kept constant. However, the tensile strength of the
with 3 slip systems. Furthermore, the influence of grain growth on S26 sample was still higher than that of the wrought counter-
weakening ductility will decrease at high-temperature conditions, parts.
because the slip or climb of dislocations is intensified and enhanced (4) The microstructure types and phase constituents of heat-
at these high temperatures. treated samples under three representative temperature

Fig. 17. High-magnification morphologies of the printed samples: (a) S17, (b) S18, (c) S19, and (d) S20. The yellow and white arrows indicate coarse and fine ␣ martensite,
respectively.
64 C. Cai et al. / Journal of Materials Science & Technology 57 (2020) 51–64

ranges experienced a successive evolution: fine lath-like (␣ + ␤) [7] W. Xu, M. Brandt, S. Sun, J. Elambasseril, Q. Liu, K. Latham, K. Xia, M. Qian, Acta
with needle-like ␣ martensitic phase → full lath-like (␣ + ␤) → Mater. 85 (2015) 74–84.
[8] R. Sabban, S. Bahl, K. Chatterjee, S. Suwas, Acta Mater. 162 (2019) 239–254.
coarse lamellar (␣ + ␤) with lath-like (␣ + ␤). The heat-treated [9] W. Xu, S. Xiao, X. Lu, G. Chen, C. Liu, X. Qu, J. Mater. Sci. Technol. 35 (2019)
sample at 750 ◦ C obtained an attractive combination of strength 322–327.
and ductility due to the synergistic effect of the soft lath-like [10] K. Wei, Z. Wang, G. Huang, J. Deng, M. Liu, J. Mater. Process. Technol. 271
(2019) 368–376.
␣/␤ structure and hard needle-like ␣ martensitic phase. [11] L.C. Zhang, D. Klemm, J. Eckert, Y.L. Hao, T.B. Sercombe, Scr. Mater. 65 (2011)
21–24.
Acknowledgements [12] L. Zhou, T. Yuan, R. Li, J. Tang, M. Wang, F. Mei, Mater. Sci. Eng. A 725 (2018)
329–340.
[13] J.C. Sabol, T. Pasang, W.Z. Misiolek, J.C. Williams, J. Mater. Process. Technol.
This work was financially supported by the National Key 212 (2012) 2380–2385.
Research and Development Program of China “the Clinical Appli- [14] A.K. Gogia, Def. Sci. J. 55 (2005) 149–173.
[15] C. Cai, B. Song, P. Xue, Q. Wei, C. Yan, Y. Shi, Mater. Des. 106 (2016) 371–379.
cation of Personalized Implant Prosthesis Additive Manufacturing
[16] A. Zhang, D. Liu, H. Wang, Mater. Sci. Eng. A 562 (2013) 61–68.
Process Research” (No. 2016YFB1101303), the National Natural Sci- [17] L. Yang, J. Liu, J. Tan, Z. Chen, Q. Wang, R. Yang, Mater. Sci. Technol. 30 (2014)
ence Foundation of China (Nos. 51705170 and 51905192) and the 706–709.
[18] Q.J. Sun, X. Xie, Mater. Sci. Eng. A 724 (2018) 493–501.
China Postdoctoral Science Foundation (Nos. 2017M620312 and
[19] X. Wu, C. Cai, L. Yang, W. Liu, W. Li, M. Li, J. Liu, K. Zhou, Y. Shi, Mater. Sci. Eng.
2018T110756). A 738 (2018) 10–14.
[20] D. Gu, Y.C. Hagedorn, W. Meiners, G. Meng, R.J.S. Batista, K. Wissenbach, R.
Poprawe, Acta Mater. 60 (2012) 3849–3860.
References
[21] C. Wang, X. Tan, E. Liu, S.B. Tor, Mater. Des. 147 (2018) 157–166.
[22] E. Liverani, S. Toschi, L. Ceschini, A. Fortunato, J. Mater. Process. Technol. 249
[1] C. Veiga, J.P. Davim, A.J.R. Loureiro, Rev. Adv. Mater. Sci. 32 (2012) 133–148. (2017) 255–263.
[2] H. Attar, K.G. Prashanth, L.C. Zhang, M. Calin, I.V. Okulov, S. Scudino, C. Yang, J. [23] K.W. Wei, Z.M. Wang, X.Y. Zeng, J. Mater. Process. Technol. 244 (2017) 73–85.
Eckert, J. Mater. Sci. Technol. 31 (2015) 1001–1005. [24] K. Wei, Z. Wang, X. Zeng, Mater. Sci. Eng. A 709 (2018) 301–311.
[3] D. Banerjee, J.C. Williams, Acta Mater. 61 (2013) 844–879. [25] P. Kumar, O. Prakash, U. Ramamurty, Acta Mater. 154 (2018) 246–260.
[4] F.H. Froes, D. Eylon, H.B. Bomberger, Titanium Technology: Present Status and [26] Z.C. Sun, J. Zhang, H. Yang, H.L. Wu, J. Mater. Process. Technol. 222 (2015)
Future Trends, Titanium Development Association, 1985, pp. 191. 234–243.
[5] L.E. Murr, S.M. Gaytan, D.A. Ramirez, E. Martinez, J. Hernandez, K.N. Amato, [27] T. Vilaro, C. Colin, J.D. Bartout, Metall. Mater. Trans. A 42 (2011) 3190–3199.
P.W. Shindo, R.B. Wicker, J. Mater, Sci. Technol. 28 (2012) 1–14. [28] Z. Guo, S. Malinov, W. Sha, Comput. Mater. Sci. 32 (2005) 1–12.
[6] C. Qiu, N.J.E. Adkins, M.M. Attallah, Mater. Sci. Eng. A 578 (2013) 230–239. [29] J. Yang, H. Yu, J. Yin, M. Gao, Z. Wang, X. Zeng, Mater. Des. 108 (2016) 308–318.

You might also like