A
Effect of heat treatments on the structure and mechanical properties of a laser powder fused Ti-6Al-2Sn-4Zr-6Mo alloy at room and elevated temperaturesAbstract
Ti-6Al-2Sn-4Zr-6Mo (Ti6246) is a new generation titanium alloy designed to outperform the widely used Ti-6Al-4V (Ti64), particularly in high-temperature applications. Combined with additive manufacturing (AM) technologies, such as Laser Powder Bed Fusion (LPBF), Ti6246 offers the potential to produce complex geometries and expand their application range. This study investigates the influence of post-processing heat treatments on the microstructure and mechanical performance of LPBF-processed Ti6246. It is observed that as-built specimens exhibit an ultrafine orthorhombic α″ martensite microstructure, resulting in a high mechanical strength (yield strength ~ 1270 MPa) but poor ductility (~ 5% elongation). Two heat-treatment protocols were applied to the printed alloy to improve its applicability: low temperature annealing at 600°C, followed by either a) subtransus (875°C) or b) supertransus (950°C) annealing. In both cases, structural analyses of the heat-treated alloy revealed the transformation of α″ martensite into a stable α + β duplex microstructure, with coarsening of α-laths and chemical partitioning between Al-rich α and Mo-rich β phases. Both treatments significantly enhanced the room temperature ductility (elongation up to ~ 14%) but reduced the strength (yield strength down to ~ 970 MPa), with more pronounced softening after the supertransus annealing. At 480°C, the post-treated samples maintained a good strength-ductility balance (567–577 MPa and 10–12%), but the as-built samples showed the best performance, with a strength of ~ 958 MPa and a ductility of 13%.The results underscore a trade-off between strength and ductility induced by heat treatments and highlight the need for further optimization to match or exceed the performance of conventionally processed Ti6246 for elevated temperature applications.
Keywords:
Additive manufacturing
Laser Powder Bed Fusion
Post-processing Heat Treatment
Microstructure and Mechanical Properties
Elevated temperature testing
A
1. Introduction
Titanium alloys possess specific properties such has a high strength-to-weight ratio, excellent corrosion resistance and biocompatibility, which make them attractive for major industries (aerospace, automotive, biomedical, etc.). In solid state, titanium alloys can form α (hexagonal close-packed (HCP)), β (body centered cubic (BCC)) or duplex α + β crystallographic phases and are generally grouped according to these allotropic forms. Alloying elements in these alloys act as either α (e.g., Al, N, O, Zr) or β (e.g., Fe, V, Mo, Nb) stabilizers and their contents are generally adjusted to satisfy specific manufacturing and application requirements. The most used titanium alloy is α + β Ti-6Al-4V (Ti64), which was initially developed for the aerospace industry. The use of this alloy is however limited to applications where service temperatures do not exceed 315°C, above which Ti64 starts to be outperformed by other titanium alloys, such as Ti-5.72Al-3.97Sn-3.82Zr-0.69Nb-0.57Mo-0.36Si (Ti834) and Ti-6Al-2Sn-4Zr-6Mo (Ti6246), among others. The latter normally retains its higher mechanical characteristics up to 540°C, and this advantage, while being slightly offset by a higher mass density, makes Ti6246 promising for such applications as hot sections of gas turbines, for example [1].
Gas turbine applications often require complex and customizable components produced in small series, the areas where additive manufacturing (AM) processes are particularly performant. These processes are generally based on the layer-by-layer forming principle and are preferred for small production lots of complex and multifunctional components with intricate geometries that must be reproduced with great accuracy. Among AM processes, laser powder bed fusion (LPBF), which uses fine powder feedstocks (20–50 µm), is considered particularly suitable for producing small to medium-sized parts with print resolutions as fine as 0.1–0.2 mm. These results are generally achieved at a lower cost than in concurrent metal AM technologies, such as electron beam powder bed fusion (EB-PBF) or directed energy deposition (DED).
LPBF Ti6246 alloys have recently been widely studied [1–3] and the dependence of their mechanical properties on as-printed microstructures revealed. It has been observed that under the rapid cooling conditions inherent to the process, this alloy forms an orthorhombic α’’ martensite phase ([2, 4–6]). The presence of a metastable martensitic phase makes the as-built Ti6246 unsuitable for structural applications and requires the use of post-treatments. In contrast to LPBF Ti64, for which post-treatment strategies have been extensively developed through decades of research and application [7], only a limited number of studies have addressed the post-treatments of LPBF-produced Ti6246 alloys. To bridge this gap, Carrozza et al., 2022 [8] focused their study on the effects of the annealing temperature on the microstructure, microhardness and tensile properties of this material. By applying subtransus (875°C) and supertransus (950°C) heat treatments, they found that the former leads to a bi-lamellar microstructure (columnar prior-β and primary α grains combined with secondary α needles in β grains), while the latter forms an equiaxed duplex α + β microstructure in the material. Both structures produce an excellent combination of room temperature mechanical properties: a yield strength of ⁓1000 MPa and an elongation to failure of 17–20%.
While the effects of these post-treatments on the room temperature mechanical properties of LPBF processed Ti6246 alloys have already been made public, their impacts on the mechanical properties of these alloys at elevated temperatures, where their application is the most appealing, are yet to be addressed in the literature. The present work aims to contribute to ongoing research efforts and to the understanding of the effects of post-treatments on the structural characteristics and mechanical properties of Ti6246 at both room and elevated temperatures.
2. Material and methods
Plan of experiments
The Ti-6Al-2Sn4Zr-6Mo powder used in this study is a gas-atomized pre-alloyed powder provided by Eckart TLS GmbH (Bitterfeld-Wolfen, Germany). The D10 = 26, D50 = 43 and D90 = 64 (µm) particle size distribution (PSD) was measured using the water module of an LS13 320 XR (Beckman Coulter, Brea, CA, USA) particle size analyzer (Fig. 1a). Then, the particle size and morphology distributions were attested using a TM3000 scanning electron microscope, SEM (Hitachi, Tokyo, Japan) (Fig. 1b).
Fig. 1
Ti6246 powder a) Particle size distribution and b) SEM observations.
Ten cylindrical (10 mm diameter, 21 mm high) and eighteen prismatic (17 x 8 x 81 mm3) specimens were printed using the following set of printing parameters: laser power P = 139 W, scanning speed v = 741 mm/s, hatching distance h = 75 µm and layer thickness t = 25 µm, obtained in a previous study on the LPBF printability of Ti6246 alloys [2]. The printing was carried out on a Ti-6Al-4V baseplate without preheating using a TruPrint 1000 system (TRUMPF GmbH, Ditzingen, Germany). All the specimens were oriented along the build direction and contained 5 mm thick non-solid supports. They were removed from the plate using a chisel and divided into three groups, each containing three cylindrical and six prismatic specimens. The first group of specimens was kept in the as-built (AB) condition, while the two others were subjected to two heat treatment sequences under vacuum (10− 6 hPa) atmosphere using a WEBB 120 furnace (R.D. WEBB COMPANY INC., Rhode Island, U.S.A).
These heat treatment sequences included a low-temperature annealing at 600°C (S600) followed by either subtransus annealing at 875°C (P875) or supertransus annealing at 950°C (P875), as illustrated in Fig. 2. For each of these treatments, the heating and cooling rates were respectively set to 5 and 2°C/min, and the treatment duration, to 2h. S600 annealing allowed safe handling of the specimens. Preliminary experiments showed that the as-printed specimens were too fragile and prone to distortions, which complicated their removal from the building plate and subsequent machining. This phenomenon was attributed to an extreme finesse of the as-built microstructure (see the Results section), which contributed to a high level of residual stresses and significant fragility of printed specimens. This distinctive feature stemmed from the fact that, contrary to most studies covering LPBF of Ti6246 alloys, the building plate was not preheated in the present study, which increased both the thermal gradients and cooling speeds, thus significantly refining the as-built microstructure.
Fig. 2
Two post-treatment sequences involving low-temperature annealing at 600oC (S600) followed by two annealing conditions: subtransus 875oC (P-875) and supertansus 950oC (P-950); all heat treatments lasted 2h, with heating and cooling rates of 5 and 2oC/min, respectively.
To validate the selection of these heat treatment conditions, one as-built sample was subjected to a DSC analysis in the 25 to 1025°C temperature range (heating rate of 5K/min) using a NETZSCH DSC 404F3 (NETZSCH Gmbh, Germany. One clear exothermic peak can be distinguished in the 700–900°C temperature range of the DSC graph (Fig. 3), and can be attributed to stress-relaxation [8]. In fact, because of the high cooling speed inherent to the LPBF process, as-printed samples generally contain orthorhombic α’’ martensite. Thus, before this exothermic peak, in the 200–700°C temperature range, the transformation that occurs corresponds most likely to the decomposition of α’’ martensite into a mixture of stable α + β phases. Therefore, low-temperature annealing (S600) corresponds to the advanced α’’ phase decomposition phase. Finally, an absolute minimum on the DSC curve at 900 ± 15°C corresponds to the β-transus temperature.
Fig. 3
DSC graph of the as-built sample; the dashed lines highlight the endothermic peaks.
After the heat treatment, the cylindrical specimens were partitioned to obtain samples for structural analyses and microhardness measurements (Fig. 4a), while the prismatic specimens were machined to obtain tensile testing samples (Fig. 4b).
Fig. 4
Schematic representation of the a) cylindrical specimen divided into samples used for various analyses and b) tensile testing sample; all dimensions are in mm
Microstructure and phase analyses
The Y-Z cross-sections of distant parts of the cylindrical specimens (Fig. 4a) were analyzed using an X’Pert3 X-Ray diffractometer (Malvern Panalytical Ltd, Malvern, UK), equipped with a cobalt source (Kα Co = 1.79026 Å). Acquisitions were made in the Bragg Brentano configuration, with a step size of 0.017° and an 2θ range between 38 and 50° (this reduced 2θ range encompasses the main reflections of each possible phase in this material, i.e., α’’, α’, α, and β). Then, the samples were mounted in carbon-dopped resin, mirror-polished and etched (2 min) with Kroll reagent (%vol: 2:5:93 HF/HNO3/H2O). Firstly, low-resolution observations (x20) were carried out using a LEXT OLS4100 (Lext Olympus Corp., Japan) confocal microscope. Then, higher magnification (x1.k, x5k and x30k) images were captured using a secondary electron detector (acceleration voltage 10 kV, magnification 30k) of an SU-8230 Field Emission STEM (Hitachi, Tokyo, Japan). Finally, energy dispersive spectroscopy (EDS) analyses were performed using the SU-8230 STEM to detect the potential occurrence of chemical segregation. A certified reference material for 6Al-4V grade titanium alloy (BS T-5A, Brammer Standard Company, Inc., Houston, TX, USA) was used to quantify the elements.
Mechanical testing
Microhardness measurements were performed on the etched Y-Z cross-sections of all the cylindrical samples using a Struers Duramin-40 M1 (Struers, Ballerup, Denmark) microhardness tester. For each sample, 10 measurements were realized in the middle of the surface with an applied force of 300 gF and a dwell time of 15 s. Next, six coupons of each group were subjected to tensile testing using an MTS 810 load frame (MTS, Eden Prairie, MN, USA), with the force measured by a 100 kN MTS load cell and displacement measured by an LVDT: three tests at room temperature (RT) and three at 480°C (ET, for elevated temperature testing). This testing temperature was situated between the upper limits of the Ti64 and Ti6246 temperature application ranges, i.e., between 315 and 540°C, respectively. The strain rate was set to 0.405 mm/min (RT) and 0.135 mm/min (ET), as per ASTM E8-24 and ASTM E21 standards [9, 10]. Elevated temperature testing was performed under a constant argon flow of 27.5 ft3/h in an infrared (IR) furnace. The heating rate was set to 10°C/min and a 3-min prior-to-testing dwell time was applied after reaching the target temperature. Finally, strain-stress diagrams were plotted to obtain the following metrics of interest: the Ultimate Strength (US, MPa), defined as the maximum stress reached during the test, the Yield Strength (YS, MPa), calculated by moving the σ-ε slope from the origin to 0.2% of strain, and the elongation to failure (δ, %).
3. Results and Discussion
Physical and structural analyses
The X-ray diffractograms obtained for each post-treated sample are plotted in Fig. 5 alongside those of the powder feedstock and the as-built sample. It can be observed that while the powder and the as-built sample exhibit orthorhombic α’’ martensite, low-temperature annealing (S600) and both high-temperature annealing conditions (P875 and P950) lead to its decomposition into stable α (hcp) and β (bcc) phases of titanium. These observations are in agreement with the rapid cooling conditions taking place during both the powder manufacturing and 3D printing stages, leading to the formation of martensite, and its subsequent decomposition into a stable α + β phase mixture during the heat treatments.
Fig. 5
XRD diffractograms for different Ti6226 states including powder, as-built and heat-treated samples (S600, P875 and P950): a) 35–90° 2θ range and b) zoom in the 42–50° 2θ range.
Figure 6 presents microstructures related to the as-built, S600, P875 and P950 post-treated states. As expected, the as-built microstructure contains very thin (49 ± 17 nm) α’’ needles within columnar prior-β grains (barely visible after etching) commonly observed after the LPBF process. The melt pool borders are also visible (Fig. 6a). After low-temperature annealing (S600), samples turn brown when in contact with the etching solution, revealing columnar prior-β grains (Fig. 6b) in which 57 ± 26 nm-width α lamellae can be observed using higher resolution images (Fig. 6j). After P875 annealing, coarser (828 ± 240 nm) α lamellae surrounded by a thin β-layer are encountered. The α lamellae mostly share the same crystallographic orientations, forming colonies within the columnar prior-β grains formed during printing. Following P950 annealing, the α lamellae and surrounding β-layer become even coarser (2495 ± 1121 nm) than after P875 annealing. Although less pronounced, the columnar grain aspect persists, meaning that the formation of an α + β mixture with secondary β grains is not completed.
Fig. 6
Observations with confocal microscope (a-d) and SEM (e-l) of: (a, e, i) as-built sample with orthorhombic α’’ martensite; (b, f, j) samples after 600°C annealing showing the beginning of α’’→ α + β decomposition; (c, g, k) and (d, h, l) P875 and P950 annealed samples, respectively, with duplex α + β phase.
Because of the very fine microstructure features, the as-built and S600 samples could not be subjected to EDS analysis, whereas it was possible with the P875 and P950 samples. In both latter samples, elemental maps revealed some Ti- and Al-rich zones in the dark areas, and Mo- and Zr-rich zones, in the light areas (Fig. 7). Points measurements (Table 1) confirmed significant differences in the elemental concentrations between dark and light areas. Given that Ti and Al are α-stabilizers while Mo is β-stabilizer, light areas can be identified as belonging to β phase, while dark areas, where the opposite phenomenon is observed, can be identified as α phase [4, 11]. Sn was homogeneously distributed in both phases and some very light features were randomly observed at the α/β grains boundaries as depicted in Fig. 7d, h, which were associated with the Sn precipitates (Fig. 8).
Fig. 7
SEM observations and EDS maps: (a,b,c,d) P875 sample and (e,f,g,h) P950 sample
Sample | Region | C (wt%) | O (wt%) | Al (wt%) | Ti (wt%) | Zr (wt%) | Mo (wt%) | Sn (wt%) |
|---|---|---|---|---|---|---|---|---|
P875 | Dark | 0.0 | 0.7 ± 0.0 | 7.1 ± 0.0 | 91.0 ± 0.1 | 0.2 ± 0.0 | 0.1 ± 0.0 | 0.9 ± 0.1 |
Light | 0.0 | 1.0 ± 0.1 | 4.2 ± 0.3 | 91.6 ± 0.4 | 0.3 ± 0.0 | 1.9 ± 0.2 | 1.0 ± 0.3 | |
P950 | Dark | 0.0 | 0.6 ± 0.1 | 6.8 ± 0.5 | 91.3 ± 0.5 | 0.2 ± 0.0 | 0.1 ± 0.0 | 0.9 ± 0.1 |
Light | 0.0 | 1.0 ± 0.1 | 4.2 ± 0.0 | 92.0 ± 0.4 | 0.3 ± 0.0 | 1.9 ± 0.0 | 0.5 ± 0.0 |
Fig. 8
Superposition of the SEM observations and the EDS Sn map showing Sn precipitates in the a) P875 and b) P950 samples
Mechanical characterization
A
Microhardness measurements performed on the Y-Z sample cross-sections revealed that S600 led to a slight increase in hardness (528 ± 30 HV0.3 after S600 versus 514 ± 19 HV0.3 in the as-built state) while P875 and P950 caused a significant hardness decrease (down to 397 ± 32 HV0.3 for the former and 365 ± 20 HV0.3 for the latter). These microhardness variations can be related to the described microstructural changes, since it has been established [4, 6, 8] that decomposing α’’ martensite into duplex α + β phase increases the material hardness, whereas coarsening α-laths decreases it (Fig. 9).Typical tensile stress-strain curves of the as-built and post-treated samples are compared in Fig. 10a (20 oC) and in Fig. 10b (480 oC). At room temperature, the as-built samples manifested a relatively fragile behavior with an elongation to failure of ~ 5%, while the heat-treated samples showed much more ductile behavior, with a strain to failure reaching 13–14%, but this gain was obtained at the expense of a lower yield strength: 967 MPa (P875) and 972 MPa (P950) versus 1270 MPa (AB) (Table 2)).
Fig. 10
Representative tension stress-strain diagrams of the as-built and post-treated samples at: a) 20 oC and b) 480 oC; mechanical properties of the as-built and post-treated samples at c) room temperature and d) elevated temperature. The UTS values obtained for wrought Ti6246 and Ti64 alloys [12] are given as target values.
Protocol | Temperature | YS at offset = 0.2% (MPa) | UTS (MPa) | Elongation at failure (%) |
|---|---|---|---|---|
As-built | 20°C | 1270 ± 64 | 1287 ± 60 | 5.2 ± 0.4 |
480°C | 558 ± 29 | 958 ± 48 | 13.6 ± 3.4 | |
P875 | 20°C | 967 ± 6 | 1123 ± 29 | 13.1 ± 1.1 |
480°C | 502 ± 27 | 567 ± 26 | 11.8 ± 4.4 | |
P950 | 20°C | 972 ± 19 | 1111 ± 9 | 14.3 ± 0.3 |
480°C | 512 ± 27 | 577 ± 26 | 10.0 ± 1.3 |
Note that the orthorhombic α’’ martensite formed in the present study had a significantly higher yield strength (YS) than seen in a previous study [2]: 1270 MPa in the former as compared to 480 MPa in the latter. This difference is attributed to the one-order-of-magnitude finer α” martensite microstructure in the present study as compared to the reference study [2]: 50–70 nm wide α” laths in the former as compared to 550–700 nm wide α” laths in the latter. Similar observations have been made for the LPBF-processed Ti64 alloys [13], where it was shown that the presence of very fine α’ martensite needles formed in the as-built samples increased the room-temperature strength at the expense of a lower ductility.
Furthermore, the annealing post-treatment realized in the present work led to an increase in the room-temperature ductility and a decrease in strength, with YS ranging between 1000 and 1050 MPa and δ ranging between 15 and 20%. Similar mechanical properties were observed previously by other authors after different post-treatment strategies, including annealing at 875 and 950°C [8]. These changes were related to the decomposition of metastable α” martensite into a stable duplex α + β phase and microstructure coarsening, where the higher the annealing temperature, the larger the α-lath and β-grain sizes.
The elevated-temperature testing carried out in the present study showed that, contrary to room temperature testing, P950 samples showed higher strength but lower ductility than their P875 counterparts, which can be attributed to microstructure coarsening [14]. Note also that for RT applications, the heat-treated alloy of this study showed a more appealing strength-ductility combination than did its as-built counterpart (Fig. 10c). For elevated temperature applications however, the as-built alloy largely outperformed its heat-treated counterparts (Fig. 10d): it is similarly ductile (∼14%), but significantly more resistant (∼960 vs ∼580 MPa). This observation leads to the conclusion that more work is needed to maximize the mechanical properties of LPBF Ti6246 alloys for both low and high temperature applications. Unfortunately, among the few studies that have examined the high-temperature mechanical properties of LPBF-processed titanium alloys, none has yet targeted Ti6246 alloys.
Finally, the LPBF Ti6246 alloy of the present study is more mechanically resistant in the 20-480oC temperature range than the wrought Ti64, but lower than the wrought Ti6246 (Fig. 10c). To make this alloy competitive with its conventionally produced equivalents, new post-treatment strategies must be explored. Let us take for example the work of Pirro et al. [6], where a bi-lamellar structure composed of large primary α and ultrafine secondary α laths combined with Ti3Al precipitates resulting from subtransus annealing at 825oC followed by ageing at 500°C for 24h enabled to reach UTS = 1500 MPa and δ = 15% (room temperature testing). Pending verifications of whether these outstanding properties can be transposed to elevated temperatures, such a combined post-treatment appears to be a new avenue to explore.
4. Conclusion
In this study, two post-treatment protocols were applied to LPBF-processed Ti-6Al-2Sn-4Zr-6Mo alloy, and their effects on the structure and mechanical properties of the alloy were studied. Both protocols allowed the decomposition of α’’ martensite initially present in the as-built samples, and α (enriched in Al) and β (enriched in Mo) grains were obtained. Tin precipitates were also observed at the grain boundaries. Mechanical testing revealed that the increase in the α-lath width obtained after two post-treatments led to a significant decrease in microhardness and a significant increase in ductility at the expense of a lower mechanical strength at both room and elevated temperatures.
A
Author Contribution
Conceptualization, A.L. and V.B.; data curation, A.L. and T.M.; formal analysis, A.L. and T.M.; funding acquisition, V.B.; investigation, A.L. and T.M.; methodology, A.L. and T.M and V.B.; software, A.L. and T.M; project administration, V.B.; resources, V.B.; supervision, V.B.; validation, V.B.; visualization, A.L. and T.M and V.B.; writing—original draft preparation, A.L. and T.M.; writing—review and editing, V.B.;
A
Data Availability
Data are contained within the article.
A
Acknowledgement
The authors acknowledge the financial support provided by CRIAQ (Consortium de Recherche et d’Innovation en Aérospatiale au Québec) in the framework of the Exploring Innovation – INNOVR program, NSERC (Natural Sciences and Engineering Research Council of Canada) and Exonetik Turbo inc. The authors also acknowledge the contributions of Emma Bisserié in preparing the samples used in the present work.
References
1.
Aversa A et al (2025) A Comparison Between the Residual Stresses of Ti6Al4V and Ti-6Al-2Sn-4Zr-6Mo Processed by Laser Powder Bed Fusion. Mater (Basel), 18(3)
2.
Carrozza A et al (2021) A study on the microstructure and mechanical properties of the Ti-6Al-2Sn-4Zr-6Mo alloy produced via Laser Powder Bed Fusion. J Alloys Compd, 870
3.
Cobbinah PV et al (2024) Peculiar microstructural evolution and hardness variation depending on laser powder bed fusion-manufacturing condition in Ti–6Al–2Sn–4Zr–6Mo. Smart Materials in Manufacturing, 2
4.
Peng H et al (2023) Rapid hardening response of ultra-hard Ti-6Al-2Sn-4Zr-6Mo alloy produced by laser powder bed fusion. Scripta Mater, 226
5.
Hassanin H et al (2021) Laser Powder Bed Fusion of Ti-6Al-2Sn-4Zr-6Mo Alloy and Properties Prediction Using Deep Learning Approaches. Mater (Basel), 14(8)
6.
Pirro G et al (2024) A novel solution treatment and aging for powder bed fusion-laser beam Ti-6Al-2Sn-4Zr-6Mo alloy: Microstructural and mechanical characterization. Int J Min Metall Mater 32(2):414–424
7.
Zou Z et al (2021) Microstructure and tensile properties of additive manufactured Ti-6Al-4V with refined prior-β grain structure obtained by rapid heat treatment. Mater Sci Engineering: A, 814
8.
Carrozza A et al (2022) Towards customized heat treatments and mechanical properties in the LPBF-processed Ti-6Al-2Sn-4Zr-6Mo alloy. Materials & Design, p 215
9.
E28 (2020) A.C., Standard Test Methods for Elevated Temperature Tension Tests of Metallic Materials
10.
E28 (2024) A.C., Standard Test Methods for Tension Testing of Metallic Materials
11.
Alluaibi MHI et al (2024) Optimizing Structural and Mechanical Properties of an Industrial Ti-6246 Alloy below beta-Transus Transition Temperature through Thermomechanical Processing. Mater (Basel), 17(5)
12.
Boyer RR (1998) Properties, Compositions, and Applications of Selected Titanium Alloys, in Metals Handbook Desk Edition. pp. 586–588
13.
Bartolomeu F et al (2022) Mechanical Properties of Ti6Al4V Fabricated by Laser Powder Bed Fusion: A Review Focused on the Processing and Microstructural Parameters Influence on the Final Properties. Metals, 12(6)
14.
Zhao J et al (2022) Effect of grain size on the yield stress and microscopic mechanism of a near-α titanium alloy during non-superplastic hot deformation. Mater Sci Engineering: A, 840
