Revealing precipitation behavior and mechanical response of wire-arc directed energy deposited Mg-Gd-Y-Zr alloy by tailoring aging procedures

Mg-Gd-Y-Zr alloy, as a typical magnesium rare-earth (Mg-RE) alloy, is gaining popularity in the advanced equipment manufacturing fields owing to its noticeable age-hardening properties and high specific strength. However, it is extremely challenging to prepare wrought components with large dimensions and complex shapes because of the poor room-temperature processability of Mg-Gd-Y-Zr alloy. Herein, we report a wire-arc directed energy deposited (DED) Mg-10.45Gd-2.27Y-0.52Zr (wt.%, GW102K) alloy with high RE content presenting a prominent combination of strength and ductility, realized by tailored nanoprecipitates through an optimized heat treatment procedure. Specifically, the solution-treated sample exhibits excellent ductility with an elongation (EL) of (14.6 ± 0.1)%, while the aging-treated sample at 200 °C for 58 h achieves an ultra-high ultimate tensile strength (UTS) of (371 ± 1.5) MPa. Besides, the aging-treated sample at 250 °C for 16 h attains a good strength-ductility synergy with a UTS of (316 ± 2.1) MPa and a EL of (8.5 ± 0.1)%. Particularly, the evolution mechanisms of precipitation response induced by various aging parameters and deformation behavior caused by nanoprecipitates type were also systematically revealed. The excellent ductility resulted from coordinating localized strains facilitated by active slip activity. And the ultra-high strength should be ascribed to the dense nano-β′ hampering dislocation motion. Additionally, the shearable nano-β 1 contributed to the good strength-ductility synergy. This work thus offers insightful understanding into the nanoprecipitates manipulation and performance tailoring for the wire-arc DED preparation of large-sized Mg-Gd-Y-Zr components with complex geometries.

Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Introduction
Magnesium (Mg) alloys, as a new generation of lightweight metals, are characterized by recyclability, low density, abundant reserves, and high specific strength [1][2][3][4][5].Driven by the concepts of sustainable development and energy conservation, the demand for such environmentally-friendly materials is becoming more and more pressing.Lightweight magnesium rare-earth (Mg-RE) alloys are attracting increased attention in the automotive, aerospace and aircraft fields owing to the excellent properties that can be obtained by tailoring heat treatment schemes [6][7][8].Moreover, the aforementioned advanced equipment manufacturing industry places great emphasis on the development of large-size and monolithic lightweight components [9].However, Mg-RE alloys suffer from poor processability at ambient temperature as a result of their hexagonal close pack (hcp) crystal structure [1,10].Some pioneering studies have shown that the formability of Mg-RE alloys can be improved by proper alloying design and delicate microstructural control, e.g., tailoring bimodal-grained structures and introducing pre-twinning structures [11].However, this definitely increases the complexity and cost of processing raw materials.Consequently, it is extremely challenging to integrally fabricate wrought Mg-RE alloy components with large dimensions and complex geometries using conventional manufacturing processes including forging, rolling and extruding [1,12].Accordingly, there is a great demand to shift the research focus towards the emerging additive manufacturing (AM) process, which is known for its ability to fabricate components with complex geometries due to the layer-by-layer deposition.
So far, extensive researches have successfully applied three prevalent AM processes, including laser directed energy deposition (L-DED) [13][14][15], laser powder bed fusion (L-PBF) [1,8,[16][17][18][19], and wire-arc directed energy deposition (wire-arc DED) [20][21][22][23], to the preparation of Mg-RE alloy components.Specifically, Zheng et al [14] and Jiang et al [13] deeply revealed the influence mechanism of the multiple thermal cycles and Gd content on the precipitate type, grain size, and tensile properties of L-DED-prepared Mg-Gd-Y-Zr alloys, respectively.Although their investigations have confirmed the feasibility of the L-DED process for the preparation of Mg-RE alloys, these components suffered from poor mechanical properties with an elongation of just around 2% [13].Previous studies have demonstrated that the good strength of Mg-RE alloys such as Mg-Gd [16,17] and Mg-Y [24,25] series can be realized using the L-PBF process, but the ductility is greatly reduced due to massive oxide inclusions [24,25] and metallurgical defects [16,25].Both the L-DED and L-PBF processes face challenges in avoiding the generation of defects because of the high susceptibility of the RE elements to oxidation and the evaporation of the Mg element under the action of high-energy laser [4].Furthermore, Mg-alloy powders are susceptible to explosion due to their active nature, which introduces safety concerns for the preparation process of engineering components [26].In order to reduce the safety hazards caused by Mg-alloy powders, the oxygen content in the building chamber is usually below 100 ppm.Such low oxygen content generally requires a reduction in the space of the building chamber, which undoubtedly limits the size of the manufacturable components.Meanwhile, the manipulation of Mg-alloy powders, before and after the additive process, also requires extensive attention to avoid potential explosion.Conversely, wire-arc DED enables integrated prototyping of large-sized components with complex shapes through melting wires by arc [22,23], while subsequent machining makes fabrication of complex components possible.The specific surface area of filaments is much smaller than that of powders, which greatly inhibits oxidation during feedstock preparation and deposition process [26].Besides, the energy of the arc is gentler than that of the laser, which mitigates the evaporation intensity of the alloying elements to a certain extent [4].
The low strength and high ductility of AZ-series Mg-alloys make it easy to obtain wires.As a result, a large number of previous studies focused on the wire-arc DED of AZseries Mg-alloys [4,20,22].However, the strength of those works is limited to less than 300 MPa due to their nonheat treatability, making it difficult for practical engineering applications [20,22].With the steady improvement of wire preparation technology in recent years, some researchers have shifted their focus to the Mg-RE alloys [6,7,9,26,27].Specifically, Tong et al [26] revealed the effect of thermal profiles inherent in the quasi-wire-arc DED of WE43 alloy on the inhomogeneous microstructure and mechanical properties.Cao et al [7] and Ma et al [9] successfully prepared high-quality GW63K alloy components using ultrasonic frequency pulsed and cold metal transfer (CMT) based wirearc DED, respectively.Meanwhile, dispersed nanoprecipitates were obtained by heat treatment, thereby substantially increasing the strength to about 350 MPa [7,9].The effect of heat treatment on the phase transformation behavior and mechanical properties of wire-arc DED-prepared GWZ1031K alloy was also investigated by Cao et al [6].Those densely long period stacking ordered phases and nano-β ′ substantially enhance the mechanical properties with a strength of 337 MPa and an elongation of 2.8% [6].In summary, the strengths and elongations of wire-arc DED-fabricated Mg-RE alloys that have been reported in the literature are restricted to 330-350 MPa and 2%-7%, respectively.Moreover, the effects of various heat treatment parameters on the nanoprecipitates type (nano-β ′ , nano-β ′′ , nano-β 1 ) and precipitation behavior of Mg-Gd-Y-Zr alloys have not been clearly revealed.As such, there is a need to exploit a Mg-Gd-Y-Zr alloy with high RE content to enhance age-hardening capability and strength.
In the present study, the thin-walled component was successfully prepared by CMT based wire-arc DED using the customized Mg-10.22Gd-2.14Y-0.43Zr(wt.%,GW102K) wires with high RE content.Besides, the influences of the heat treatment schemes on the precipitation characteristics, grain sizes, mechanical properties, and failure modes were thoroughly analyzed with the help of various characterization techniques.Particularly, the evolution mechanisms of precipitation behavior and nanoprecipitation type induced by tailored aging parameters were also systematically disclosed.Moreover, the impact of the nanoprecipitates on the deformation mechanism was deeply revealed to achieve a prominent combination of strength and ductility.This study aims to explore a Mg-RE alloy with a prominent strength-ductility combination suitable for the wire-arc DED process, shedding new light on the nanoprecipitates regulation and performance tailoring of largesized Mg-Gd-Y-Zr component with complex geometries.

Raw materials and wire-arc DED process
The customized GW102K Mg-alloy ingots are hot extruded and drawn to obtain filler wires with a diameter of 1.2 mm.The substrate is the rolled AZ31 Mg alloy.The arc is supplied by a welding machine (Fronius, CMT Advanced 4000 R) and the travel path of the torch is controlled by a six-axis robot (KUKA, KR20).Based on the previous experimental foundation and considering the surface quality and forming efficiency, the following processing parameters were used: wire supply speed of 4.5 m•min −1 , welding speed of 0.4 m•min −1 , welding current of 78 A, welding voltage of 12.4 V, and bidirectional deposition strategy.As shown in figure 1(a), the single-pass multilayer component with good surface quality was prepared using the aforementioned parameters.The dimensions of the component are 142 mm in length, 84 mm in height, and 10.5 mm in width.
The alloy compositions of the welding wires and components were analyzed by the Inductively Coupled Plasma-Atomic Emission Spectroscopy (ICP-AES, NexION 350D) method.Due to the fact that the Mg element evaporates more easily compared to the other elements, it leads to a transition of the actual composition from 10.22% Gd, 2.14% Y, and 0.43% Zr for the welding wires to 10.45% Gd, 2.27% Y, and 0.52% Zr for the thin-walled components.

Heat treatment
Considering the excellent solubility and age-hardening capabilities of Mg-RE alloy, optimization of solution (T4) and solution plus aging treatment (T6) of the as-built component were conducted.For simplicity, the sample without heat treatment was named as as-built sample, while those treated with T4 treatment and T6 treatment were named as T4 and T6 samples, respectively.According to literature, the optimum solution temperature for Mg-Gd series alloys is located around 530 • C [28].Given the content of residual precipitates and the grain size, the optimal time of T4 treatment was sought in the range of 0.5-2 h.In order to investigate the effect of aging temperature on precipitation behavior, the aging temperatures were set as 150 • C, 180 • C, 200 • C, 220 • C, and 250 • C, respectively.These samples treated at the optimum solution parameters were subjected to aging treatment at the aforementioned temperatures for 4-144 h.The optimum duration of aging treatment at each temperature was determined by the peak micro-hardness.

Microstructural characterization
As displayed in figure 1(a), the samples for microstructure characterization were taken from bottom, middle and top regions of the component by electrical discharge machining.These mechanically ground and manually polished samples were etched to characterize grain structures and precipitates.The optical microscopy (OM, Olympus) was used to view the metallographic structures in the different regions.The scanning electron microscope (SEM, Hitachi SU8230) equipped with energy-dispersive spectrometer (EDS) was applied to characterize the elemental distribution and types of precipitates.To ensure the index ratio of electron backscatter diffraction (EBSD) data to be higher than 90%, these manually polished samples were removed from the stressed surface layer by electrolytic polishing in an AC2 solution at a temperature of −20 • C (current of 0.1 A, voltage of 20 V).The step size for EBSD characterization was set to 2 µm at a voltage of 15 kV and a current of 10 nA.The grain characteristics and texture intensity were analyzed by the AztecCrystal software.The xray diffractometer (XRD, Bruker D8 Advance, Cu Kα) technique was applied at a voltage of 40 kV and a current of 40 mA to identify the phase constitution of samples with a scanning step of 0.02 • .The transmission electron microscope (TEM, Talos F200X) technique was used to investigate the eutectic phase and nano-precipitates.And these TEM samples were prepared by mechanical grinding to 80 µm and subsequent ion thinning.

Mechanical tests
The microhardness instrument (SCTMC 402SXV) was used to evaluate the micro-hardness of these samples with a loading force of 100 g and a duration of 15 s.Microhardness testing experiments were carried out on 15 random positions of the samples to ensure statistically reliable data and the average mean value was taken to define the average microhardness.As shown in figure 1(a), the tensile tests were performed along the travelling direction (TD) and building direction (BD) by the universal testing machine (CMT-5105), respectively.According to the ASTM E8M-09, the dimensions of parallel section were 6 mm × 2 mm × 30 mm (figure 1(b)).During tensile testing, an extensometer was used to monitor the deformation of the gauge length in real time.To ensure the reproducibility of the mechanical properties, each set of tensile samples subjected to different heat treatment schedules was tested at least three times.As displayed in figure 1(c), SEM and TEM were used to analyze the failure modes and deformation mechanisms of the fractured samples, respectively.

Microstructure of as-built samples
Figure 2 illustrates the metallographic structure, phase composition, SEM and EDS mapping results of the as-built GW102K samples.As shown in the OM images in figures 2(a) and (b), the metallographic structure of the samples consists mainly of the α-Mg matrix and the eutectic phases along the grain boundary.The eutectic phase can be inferred from XRD results and confirmed by TEM results, respectively.Moreover, the metallographic structure at the bottom region remains almost the same as that at the top region of the thinwalled component, indicating that the microstructure remains almost constant with increasing deposition layer.The XRD characterization result of figure 2(c) verifies that the phase composition in the as-built sample is mainly composed of α-Mg and β-Mg 24 (Gd, Y) 5 .The characterization results displayed in figures 2(d)-(f) suggest that the precipitates in the asbuilt sample are composed mainly of large-size reticulated β-Mg 24 (Gd, Y) 5 , cuboid phase and Zr clusters.In addition, RErich regions enriched in Gd and Y can also be observed along the grain boundaries.The development of RE-rich regions can be attributed to the following two aspects: (1) the high solubilities of RE in the α-Mg matrix in the high temperature state [29].(2) The high cooling and solidification rates inherent to wire-arc DED process improve the supersaturation level of the alloying elements [30][31][32].
Figure 3 presents the TEM results of micro-precipitates and nano-precipitates in the as-built sample.The selected area electron diffraction (SAED) pattern and EDS mapping results shown in figures 3(b) and (c) can reconfirm that these micron-scale precipitates are β-Mg 24 (Gd, Y) 5 , which is mainly enriched in RE.The SAED pattern in figure 3(c) indicates that the β-Mg 24 (Gd, Y) 5 eutectic phase is a face-centered cubic (fcc) lattice structure.In order to further discover possible nano-precipitates, the high resolution TEM (HRTEM) images and SAED pattern were taken along the [0001] α and [2 11 0] α zone axis, respectively.Some small black fuzzy particles, with diameters located between a few nano-meters and a dozen nano-meters, can be faintly observed from the HRTEM images shown in figures 3(d) and (e).Besides, the SAED pattern illustrated in figure 3(f) shows that the diffraction spots have both the high bright spots of the α-Mg matrix and the weak bright spots coherent with the matrix.These weak bright spots indicate the presence of some nanoprecipitates within the as-built sample.These nanoprecipitates are metastable nano-β ′′ precipitates, which is in agreement with the observations of the wire-arc DED preparation of Mg-6Gd-3Y-0.5Zralloy conducted by Cao et al [7].

Microstructure evolution during T4 treatment
We subsequently conducted solution treatment on the as-built samples in order to figure out the evolution of second-phase particles and solute atom levels.As shown in figures 4 and 5, the effect of T4 treatment on grain size, grain boundary characteristics and texture was investigated using EBSD technique.As can be seen in figures 4(a) and (b), figures 5(a)and (b), the grain size distribution within as-built sample is relatively more uniform compared to T4 sample.The inhomogeneous grain size distribution in the T4 samples is mainly reflected in the significantly larger grain size of the heat affected zones (HAZ).This is due to the fact that the HAZ has already undergone in-situ heat treatment during deposition, thus the tendency of grain growth in the HAZ is more significant for the same duration of T4 treatment.However, a gradient distribution of grain size can still be faintly observed from both the as-built and T4 samples.Specifically, coarse grains (∼31 µm) are distributed in the HAZ, refined grains (∼23 µm) are distributed in the melt pool boundary (MPB) and medium-sized grains (∼26 µm) are distributed in the melt pool center (MPC) of the as-built sample, as shown in figure 5(c).After T4 treatment, as shown in figure 5(d), the grain of these samples underwent some degree of growth, with the grain sizes in the HAZ region, the MPB region and the MPC region increasing to 36 µm, 26 µm and 28 µm, respectively.Additionally, figures 4(b) and 5(b) also reveal that the T4 treatment caused some grains to grow abnormally in the HAZ region.The maximum texture intensity for both the as-built and T4 samples lies between 1 and 3, indicating a close-to-random distribution of crystallographic orientations (figures 4(c) and (d)), regardless of the location of melt pool.The statistical results shown in figure 5(e) indicate that the percentage of grain boundary with varying misorientation in different regions is almost the same, suggesting that the T4 treatment exerted little effect on the grain boundary characteristics.Particularly, the percentage of grain boundaries with misorientation located from 2 • to 15 • , from 15 • to 45 • and greater than 45 • are approximately 3%, 23% and 74%, respectively.
Figure 6 displays the metallographic structure, phase composition, SEM and EDS mapping results of the as-built and T4 samples.When the duration of T4 treatment is 0.5 h, the continuous network RE-rich region in the sample disappears and only a limited amount of β-Mg 24 (Gd, Y) 5 eutectic phases survives, as displayed in figure 6(b).When the duration is extended to 1 h, as can be seen in figure 6(c), the large-size reticulated β-Mg 24 (Gd, Y) 5 is likewise totally solubilized into the α-Mg matrix, leaving only minor amounts of the cuboid phase.However, when the duration is increased to 2 h, the abnormal growth of some grains occurs, as illustrated in figure 6(d).The above phenomena indicate that the optimal solution parameters for wire-arc DED-fabricated GW102K Mg-alloy are at a temperature of 530 • C and a duration of 1 h.In addition, it is difficult to distinguish the grain boundaries from the SEM and OM images in figures 6(c) and (e), which can be attributed to the supersaturated matrix.This phenomenon can also be identified in solution treated GW63K samples of [7].From the EDS mapping results in figures 6(g)-(i), it can be noticed that the distribution state of all the alloying elements is roughly homogeneous, except for the presence of elemental enrichment in the vicinity of some precipitates with high solution temperatures.
We then used the TEM technique to further characterize these nanoprecipitates.As shown in figures 7(a) and (b),   these cuboid phases are mainly enriched in Gd and Y, with some Zr clusters wrapped around them.The SAED results in figure 7(c) further reveal that these cuboid phases are characterized by a fcc lattice structure.Previous studies have shown these cuboid phases are (Gd, Y)H 2 compounds [17], which is frequently observed in the Mg-RE alloy.As the content of these cuboid phases is minor, the XRD pattern shown in figure 6(f) only detects the α-Mg matrix without peaks related to (Gd, Y) H 2 compounds.These flocculent nano-precipitates are enriched in Zr and thus identified as Zr clusters, as displayed in figures 7(d)-(f).These Zr clusters could hinder the grain growth by pinning the grain boundaries during T4 treatment [33].HV, while that of the sample after T4 treatment is decreased to (79.1 ± 2.7) HV.A significant increase in micro-hardness can be observed after only a short aging treatment of about 5 h, indicating that the customized GW102K alloys in this study possess excellent age-hardening capabilities.In order to reveal the reason for the large differences in the peak micro-hardness corresponding to each aging parameter, we adopted the TEM technique to characterize the aging nano-precipitates of 150-96 h, 200-58 h, and 250-16 h samples.Figure 9 depicts the nano-precipitates of the 150-96 h sample.The BF and HAADF images shown in figures 9(a) and (b) indicate the existence of two kinds of nano-precipitates within 150-96 h the sample.The morphology of one type presents a globular shape, while the other appears as a spindlelike shape.As shown in figure 9(c), according to previous studies [7,9,28], these globular nano-precipitates with a diameter of 6.6 nm are identified as nano-β ′′ (Mg 3 RE), while these spindle-like nano-precipitates with a width of 5 nm and a length of 13 nm are identified as nano-β ′ (Mg 7 RE).The SAED pattern of figure 9(a) shows that these extra diffraction spots are located at 1/2(01 10) α .Additionally, the nano-β ′′ is characterized by a DO 19 lattice structure, while the nano-β ′ is characterized by a base-centered orthorhombic (bco) lattice structure [7].Furthermore, the orientation relationships between these nano-β ′′ and matrix are (10 10) β ′′ //(10 10) α and

Microstructure evolution during T6 treatment
[0001] β ′′ //[0001] α , as can be seen in figure 9(a).Figures 9(d)-(f) show the BF, HAADF and HRTEM images of nanoprecipitates with the incident electron beam parallel to [2 11 0] α zone axis.Although it is difficult to distinguish the aforementioned nanoprecipitates by morphology from the [2 11 0] α zone axis, the existence of the nano-β ′′ and nano-β ′ can be judged from the extra diffraction spots in figure 9(d).The morphology of figures 9(d) and (e) also verifies the existence of nano-γ ′ precipitates.As shown in figure 9(f), these nano-γ ′ precipitates are similar to the basal plane stacking faults (SF) with a 2 H structure, which is in accordance with the experimental results of wire-arc DED-fabricated GW63K alloy conducted by Ma et al [9].The EDS mapping results shown in figure 9(g) indicate that these nano-γ ′ precipitates are enriched in RE.Figures 10(g)-(i) illustrates the BF, HAADF, HRTEM images and corresponding SAED patterns of nano-precipitates in the 250-16 h sample with the incident electron beam parallel to [0001] α zone axis.It could be noticed from the morphology shown in figures 10(g) and (h) that there are two types of nano-precipitates in the 250-16 h sample.The morphology of one type presents a spindle-like shape consisting of striped bands, while the other appears as solid lenticular shape.These nano-precipitates with spindle-like shape are identified as nano-β ′ , which differed from the nano-β ′ in 200-58 h sample by a significant increase in their size.It can be inferred from the SAED pattern in figure 10(g) and the HRTEM image in figure 10(i) that these nanoprecipitates with solid lenticular shapes are nano-β 1 precipitates.Besides, the HRTEM image indicates the nano-β 1 extends along the ( 2110) α planes (figure 10(i)).In fact, prior studies [28] have illustrated that this nano-β 1 exhibits a fcc lattice structure and the orientation relationships between these nanoβ 1 , nano-β ′ and the matrix are ( 1

Tensile properties and fracture morphologies
Figure 12 illustrates the tensile properties of this study and a comparison with those Mg-RE alloys prepared by wire-arc DED, L-DED, L-PBF and conventional processes.The representative tensile engineering stress-strain curves and room temperature tensile properties for samples along TD are presented in figures 12(a) and (b).The yield strength (YS), ultimate tensile strength (UTS), and elongation (EL) of the asbuilt sample along TD are found to be (149 ± 3.2) MPa, (247 ± 4.0) MPa, and (8.1 ± 0.7)%, respectively.Following T4 treatment, the YS and UTS of the samples show a slight decrease to (129 ± 0.6) MPa and (238 ± 1.5) MPa, respectively, while the EL increased significantly to (14.6 ± 0.1)%, implying an 80.2% improvement in the EL.All the five groups of aging parameters set in this study achieve a remarkable increase in strength, but at the expense of the ductility to some extent.The different combinations of strength and ductility can be achieved by controlling the type and content of the nanoprecipitates developed under different aging parameters.It is noteworthy that not only did the strength of 150-96 h and 250-16 h samples increase significantly, but their EL is also comparable to that of the as-built sample.In particular, the 250-16 h sample along TD exhibits YS of (211 ± 1.5) MPa, UTS of (316 ± 3.1) MPa and EL of (8.5 ± 0.1)%, which increased YS, UTS, and EL by 41.6%, 27.9%, and 4.9%, respectively, when compared to the as-built samples.Among all the samples, 200-58 h sample exhibits the most excellent strength, yet accompanied by poor EL.Specifically, compared to the as-built sample, the 200-58 h sample along TD exhibits a significant enhancement in strength, whose YS and UTS increase up to (239 ± 1.0) MPa and (371 ± 1.5) MPa, signifying a 60.4% and 50.2% improvement in YS and UTS, respectively.Figures 12(c) and (d) exhibit the tensile properties for samples along BD.Generally, the tensile properties of the samples along BD are almost identical to those of the samples along TD, thus they will not be repeated here.The phenomenon should be ascribed to the highly equiaxial grains and the nearly random texture.Figure 12(e) presents the strain hardening rate (SHR) curve of the samples.It is obvious that the SHR of the T4 sample is lower than that of the other samples and remains to be steady, thus achieving an ultra-high ductility of up to 14.6%.In contrast, the SHR of the T6 samples is much higher than that of the T4 sample, but the rate of decline in SHR is sharp, making it difficult to sustain excellent workhardening ability.Such unstable work-hardening state can be attributed to the fact that its dense nanoprecipitates effectively hinder dislocations and induce localized stress concentration, which ultimately leads to premature failure of the samples.The deformation mechanism of these samples will be discussed in detail in section 4.4.Notably, as shown in figure 12(f), the 180-90 h, 200-58 h and 220-24 h samples of this study achieved a good combination of strength and ductility, which is much better than that of the Mg-RE alloys prepared by L-DED [13,15] and conventional processes [13,17,29,[34][35][36][37][38], and slightly higher than that of the WE43 [26], GW63K [7,9], and GWZ1031K [6] Mg-alloys deposited by wire-arc DED, as well as comparable to that of Mg-RE alloys fabricated by L-PBF [25,37,[39][40][41].Furthermore, as shown in the red circle of figure 12(f), the desired strength and ductility of the wire-arc DED-fabricated GW102K alloy in present study can be tailored by customizing the aging parameters.
Since the mechanical properties of the as-built, T4, 150-96 h, 200-58 h and 250-16 h samples in this study are significantly different, only these five types of samples were subsequently characterized in terms of fracture morphologies and deformation behavior.

Deformation behavior
Aiming to investigate the deformation behavior of the samples during tensile loading, we analyzed the microstructure of the fractured positions by TEM. Figure 14 presents the TEM characterization results of the as-built and T4 samples after fracture.It is clear in figure 14(a) that the fracture first occurs in these continuous rod-shaped β-phases, and then the cracks extend rapidly along the interface between the β-phases and the matrix.Since these β-phases with a large aspect ratio are hard and brittle compared to the matrix, they are highly susceptible to stress concentrations during loading.In addition, a large quantity of dislocations is obstructed in the vicinity of the β-phases and dislocation substructures such as dislocation walls are developed (figure 14(a)).The stress concentration induced by these dislocation walls further exacerbates the fracture of the β-phases.Besides, as displayed in figures 14(a)-(c), massive unidirectional slip bands can be identified on the α-Mg matrix and terminate near the grain boundaries.Further, as clearly demonstrated in figure 14(c), the dislocation motion is effectively impeded by the slip bands to generate a planar dislocation array.Figures 14(d  dislocation wall, no slip bands are identified.In fact, slip bands are formed by localized dislocation slip during plastic deformation, while the nano-β ′ precipitates originally existed within the material.This means the nano-β ′ precipitates already hinder dislocation movement at the beginning of deformation before the formation of slip bands.Accordingly, it can be inferred from the above phenomena that the nanoβ ′ precipitates are more effective than slip bands in hindering dislocation motion, which leads to the development of massive dislocation substructures around those nano-β ′ precipitates.It is the absence of slip bands during tensile loading that leads to severe dislocation plugging and stress concentration, thus leading to premature failure of the 200-58 h sample with EL of only 4%.As shown in figure 15(g), a large amount of dense nano-β 1 precipitates and dislocation slip bands are present on 250-16 h sample.Moreover, fluctuating slip bands passing through these nano-β 1 precipitates can be clearly observed from the HAADF images presented in figures 15(h) and (i).In other words, these shearable nano-β 1 precipitates are sheared by massive dislocation slip bands during tensile loading.As a result, this nanoprecipitates shear phenomenon effectively mediates dislocation glide [42], thus facilitating the achievement of an excellent combination of strength and ductility.

Unique microstructure of wire-arc DED
As is well-known, it is the complex thermal history of the AM process that contributes to the unique microstructure [43][44][45][46][47][48][49].The microstructure of GW-series alloys prepared by conventional casting and AM processes such as wire-arc DED and L-DED is compared in figure 16.Apparently, the grain sizes of the GW-series Mg-alloys manufactured by wire-arc DED (26.5 µm) are far smaller than those of as-cast (114 µm) and slightly higher than those of L-DED-prepared ones (11 µm).
The medium grain size of this study should be explained by the fact that the cooling rate of the centimeter-scale molten pool developed by the wire-arc DED is intermediate between that of the decimeter-scale melt pool of the conventional casting and the millimeter-scale melt pool of L-DED [9,50].Although the variation in cooling rate did not change the type of eutectic phase, the volume fraction and size of the β-Mg 24 (Gd, Y) 5 formed by the AM process is smaller than those of the conventional casting process.This phenomenon is attributable to the high cooling rate which reduces the primary dendrite arm spacing and results in a high solidification rate, which promotes the solubility of the RE elements and inhibits the elemental precipitation to form eutectic phases [13].As a result, the area of RE-rich regions follows the following pattern: L-DED > wire-arc DED > as-cast.

Mechanisms of microstructure evolution during heat treatment
Figure 17 schematically illustrates the microstructure evolution of the non-equilibrium solidification and the heat treatment process.The melting of the wires in response to the arc energy results in the development of a melt pool, as shown in figure 17(a).Nevertheless, these second-phases such as cuboid phase and Zr particles are already present before solidification due to their melting point higher than the temperature of melt pool [15,17].In fact, the introduction of water vapor during the ingot customization and filament production leads to the combination of RE elements and H element to form hydrides (cuboid phase, (Gd, Y)H 2 ) [17,52].Furthermore, the residual H element in the melt pool reacts with the RE elements under the high temperature of the arc, leading to a further increase in the content of cuboid phase.These cuboid phases deteriorate the mechanical properties of Mg-RE alloys, especially the ductility.In this study, the content of cuboid phase in the deposited samples was minimized by vacuum packing and drying the wire feedstock prior to wire-arc DED.Notably, some of the Zr particles dissolve in the high-temperature melt pool to form Zr solutes that inhibit the continuous growth of the grains [26], thus maintaining a refined grain with an average size of 26.5 µm.Those undissolved Zr particles act as efficient sites to promote nucleation, thus inhibiting the formation of columnar grains [26,53].During non-equilibrium solidification, RE elements tend to be pushed to the front of the solidification interface.As shown in figure 17(b), some RE elements segregate to form eutectic phase, while the remaining ones are solidified into the matrix to form RE-rich regions distributed along the grain boundaries [7,9].In particular, metastable nano-β ′′ (Mg 3 RE) precipitated during the early stages of non-equilibrium solidification process is preserved in the as-built component due to the solidification rate up to 100 K•mm −1 inherent in the wire-arc DED process [30].
As shown in figure 17(c), after solution treatment, these eutectic phases dissolve into the matrix and the metastable nano-β ′′ within the grains disappear completely.Furthermore, the RE-rich region has completely disappeared due to the atomic diffusion promoted in the high-temperature state, which in turn achieves a homogeneous distribution of the alloying elements within the matrix.Some stable secondphases such as the cuboid phase and Zr particles still exist in the T4 samples owing to the fact that their stability and dissolution temperatures are much higher than those of the eutectic phase [16,17].These residual second-phases pinned down the migration of grain boundaries to a certain extent [26], thus inhibiting grain coarsening to achieve an increase in grain size of only 11%.
As depicted in figures 17(d)-(f), the aging-treated samples in this study all precipitated nano-γ ′ resembling basal plane stacking faults.As is well-known, Mg-alloys possess low stacking fault energy and the RE elements further reduce that of the base plane [9,54].Consequently, the RE elements in the supersaturated matrix precipitate to form RE-rich nano-γ ′ during aging treatment.
As illustrated in figure 17(d), two types of nanoprecipitates including nano-β ′′ and nano-β ′ were formed when the aging treatment was carried out at 150 • C for 96 h.The nanoβ ′′ precipitated in such aging parameters can be attributed to the following three aspects: (1) the crystal structure of the nano-β ′′ (DO 19 ) is similar to that of the matrix (hcp) [55], so the Mg atoms can be easily replaced by RE atoms during the aging process to promote the precipitation of nano-β ′′ .(2) The coherent interface between the nano-β ′′ and the matrix facilitates precipitation by reducing the nucleation resistance [56].(3) The low aging temperature with a value of 150 • C is beneficial for the stabilization of the nano-β ′′ [28], thus allowing only part of the nano-β ′′ to evolve into nano-β ′ .Previous studies [9,28,52] suggested that the chemical compositions of nano-β ′′ and nano-β ′ are Mg 3 RE and Mg 7 RE, respectively.Besides, the crystal structures of nano-β ′′ and nano-β ′ are DO 19 and bco [28], respectively.As a result, the transition from nano-β ′′ to nano-β ′ is characterized by both elemental diffusion and the evolution of the isomorphic lattice structure.
Figure 17(e) demonstrates that those dense nano-β ′ of the 200-58 h sample are without the presence of nano-β ′′ .It can be concluded that it is hard to maintain the presence of metastable nano-β ′′ at an aging temperature of 200 • C and a duration of 58 h.Further, the size of the nano-β ′ in the peak-aging state at 200 • C is much larger than that at 150 • C. The growth of nano-β ′ can be attributed to higher aging temperatures promoting more active atomic diffusion [28,56,57].
When the aging treatment was carried out at 250 • C for 16 h, as shown in figure 17(f), the nanoprecipitates within the 250-16 h sample transformed into nano-β ′ and nano-β 1 .Compared to the 200-58 h sample, the size of the nano-β ′ within the 250-16 h sample is further increased.Actually, these nano-β ′ are also unstable [55], thus allowing some of them to be further transformed into nano-β 1 .The mechanism of transformation from nano-β ′ to nano-β 1 can be ascribed to the following two aspects: (1) Nano-β ′ (bco) and nano-β 1 (fcc) are characterized by a similar lattice structure [55,56].(2) The interfacial energy during the transition is reduced due to the semi-coherent relationship between nano-β ′ and nano-β 1 [28].
(1) Hall-Petch strengthening (σ gb ).The σ gb due to grain boundaries is generally estimated using the following equation [58]: where d is the average grain size, and k is the stress concentration factor (303 MPa•µm 1/2 ) for Mg-RE alloy [58].Based on the EBSD date shown in figure 5, the d of as-built and T4 samples counted by using the equivalent circle diameter method are 26.5 µm and 29.5 µm, respectively.Considering that aging treatment does not cause grain growth [52], it is assumed that the average grain size of the T6 sample is 29.5 µm.Consequently, the σ gb is estimated to be 58.9MPa, 55.8 MPa, and 55.8 MPa for the as-built, T4, and T6 (150-96 h, 200-58 h, 250-16 h) samples, respectively.
(2) Load-bearing strengthening (σ sp ).SEM and TEM characterization results indicate the existence of a large number of eutectic phases within the as-built sample.Furthermore, fracture of the eutectic phase can be observed in figure 14(a), which indicates that the load transfer from the matrix to the eutectic phase.The load-bearing strengthening (σ sp ) due to eutectic phase can be determined as follows [18]: where f sp is the volume fraction of the eutectic phase, σ YS is the YS of the as-built sample (150 MPa).Based on the results of metallographic structures and SEM images, the volume fraction of β-Mg 24 (Gd, Y) 5 is estimated to be about 5.2%.Accordingly, the contribution of σ sp is about 3.9 MPa for the as-built sample.(3) Solid-solution strengthening (σ ss ).As illustrated in figure 2(d), there are substantial RE-enrich regions in the as-built samples, indicating the presence of Gd and Y elements within the α-Mg matrix.The EDS results of SEM and TEM showed that the concentrations of Gd and Y in the matrix of asbuilt sample are 0.46 at.% and 0.18 at.%, respectively, while that of T4 sample is increased to 1.48 at.% and 0.57 at.%, respectively.The increase in the YS due to lattice distortion caused by solution of the element can be calculated by the following equation [59]: where B Gd (737 MPa (at.%) 2/3 ) and B Y (683 MPa (at.%) 2/3 ) [9] are the strengthening factor for Gd and Y, respectively.X Gd and X Y are the solid solubilities of Gd and Y in the matrix, respectively.Therefore, the contributions owing to solution of RE are determined to be 24.8MPa for the as-built sample and 53.7 MPa for the T4 sample, respectively.(4) Dispersion strengthening.The deformation microstructure of the fractured 150-96 h sample illustrated in figures 15(c)-(f) demonstrates that these β ′′ and β ′ nanoprecipitates were not sheared by dislocations.Consequently, it can be concluded that these nanoprecipitates are bypassed by dislocations due to their non-shear ability.The strengthening contribution due to non-shearable nano-β ′′ can be determined based on the following equation [33,60]: where M is the Taylor factor (2.5), G is the shear modulus (16.5 GPa), b is the burger vector (0.32 nm), v is the Poisson's ratio of matrix (0.35) [23], r β ′′ and f β ′′ are the average radius and volume fraction of nano-β ′′ , respectively.As shown in the TEM results of figures 3(c), (d) and 9(a)-(c), the r β ′′ in the as-built and 150-96 h samples are approximately 2.7 nm and 3.3 nm, respectively.The f β ′′ is estimated to be 0.09% for the as-built sample and 0.16% for the 150-96 h sample, respectively.Consequently, the contribution of dispersion strengthening due to non-shearable nano-β ′′ is determined to be 41.4 MPa for the as-built sample and 48.7 MPa for the 150-96 h sample.
Similarly, the increment in YS caused by dislocations bypassing these non-shearable nano-β ′ to develop Orowan looping can be estimated by [33,60]: (5) Shear-resistant strengthening.Figures 15(h) and (i) indicate that these nano-β 1 inside the 250-16 h sample can be sheared by dislocations.Previous studies [42,61,62] have revealed that anti-phase boundaries induced by slip bands will develop as the nanoprecipitates are sheared by dislocations.The shear-resistant strengthening induced by nano-β 1 can be estimated by the following equation [42]: Where d β 1 , t β 1 , and f β 1 are the average diameter, average thickness, and volume fraction of nano-β 1 , respectively [42].According to the TEM results of figures 10(g)and (h), the d β 1 , t β 1 , and f β 1 of nano-β 1 are about 114 nm, 24 nm, and 4.8%, respectively.As such, the contribution of shearresistant strengthening due to shearable nano-β 1 is estimated to be about 98.2 MPa for the 250-16 h sample.
The theoretical and actual values of strengthening contribution for all samples are statistically presented in figure 18.Specifically, the total theoretical contributions of the as-built, T4, 150-96 h, 200-58 h, and 250-16 h samples are 150 MPa, 130.5 MPa, 182.2 MPa, 241 MPa, and 210.1 MPa, respectively, which are in good agreement with their actual YS.

Deformation mechanisms
Among all the samples, the as-built sample shows relatively inferior strength and moderate ductility, the T4 sample exhibits the most excellent ductility, the 150-96 h and 250-16 h samples present moderate strength and ductility, and the 200-58 h sample possesses ultra-high strength and poor ductility.Accordingly, based on the deformation behaviors and precipitate's characteristics in figures 14 and 15, the variation of the deformation mechanism induced by various heat treatment schemes is revealed in detail.Figure 19 schematically illustrates the deformation mechanisms of the five typical samples described above, including stress concentrations around precipitates, fractured or sheared precipitates, dislocation substructures, slip bands, and crack propagation.
During the deformation process of Mg-alloys, dislocation slip activity often coordinates a significant portion of the plastic strain within the α-Mg matrix [63,64].It is generally accepted that the slip direction lies in the slip plane and the stress reaches the critical resolved shear stress (CRSS), which will activate the slip system and form a slip band [63,65,66].Figures 19(a-i)-(a-iii) display the substructure evolution and crack extension of the as-built sample.During plastic deformation, as displayed in figure 19(a-ii), the dislocation motion is hindered by the precipitates to generate dislocation walls, thereby leading to stress concentration around these precipitates.Once the stress reaches the CRSS, localized strain concentrations will be coordinated by the development of slip bands [63].These nano-β ′′ , which are coherent with the matrix, do not cause interfacial debonding under stress concentration.Nevertheless, these eutectic phases with large aspect ratios are highly susceptible to fracture, thereby leading to crack initiation [33,67,68].Although substantial slip bands can be identified in figures 14(a)-(c), only a moderate ductility is achieved in the as-built sample due to these fractured eutectic phases leading to premature failure.Besides, only an inferior strength is obtained as a result of the minor nano-β ′′ .
For the T4 sample, only submicron precipitates such as cuboid phases Zr particles are which greatly reduces the risk of premature failure due to void nucleation and crack sprouting.As a result, as shown in figures 14(d)-(f) and figure 19(a-ii) and (a-iii), a large number of dense slip bands can be observed on the matrix.Additionally, previous studies have indicated that the solution of element Y in α-Mg matrix not only strongly impedes basal <a> slip ({0001} <11 20>), but also promotes the nucleation of second order pyramidal <c + a> slip ({11 22} <113>) [65,69].This phenomenon can be attributed to the fact that the incorporation of element Y reduces the stacking fault energy and thus affects the CRSS of pyramidal slip [65].In other words, the nucleation of nonbasal slip will further promote dislocation slip activity, thus substantially enhancing ductility [65].This is the reason for the fact that the T4 sample exhibits the most excellent ductility.
Basal nano-γ ′ phases were precipitated out for all the T6 samples, which can act as barriers to hinder the slip activity of dislocations containing <c> components to some extent [55,70].Since the content of these nano-γ ′ is extremely minor, their effect on the mechanical properties can be neglected.As illustrated in figure 19(c-ii), these nano-β ′′ and nano-β ′ precipitated at the prismatic planes would increase the strength of the 150-96 h sample by effectively impeding the basal dislocation [52,71,72].Owing to the limited content of nanoprecipitates, as displayed in figures 15(a)-(c), partial dislocation slip activity is still allowed to proceed.Since these prismatic nano-precipitates significantly increase the CRSS of dislocation slip [63], the number density of slip traces in the 150-96 h sample is far less than that in the T4 sample.As a result, the 150-96 h sample achieved the moderate strength and ductility.
For the 200-58 h sample, as depicted in figure 19(d-ii), a large number of dense nano-β ′ precipitated on the prismatic planes, which effectively blocked the movement of the basal dislocations to achieve ultra-high strength [7,9].Research carried out by Nie [73] indicated that these spindlelike nano-β ′ impacted the most striking hindering effect on the basal <a> slip.Consequently, as shown in figure 19(d-iii), massive plugged dislocations and dislocation interactions caused the most significant stress concentrations in the 200-58 h sample.The absence of slip bands leads to the unavailability of coordination of localized stress concentrations, ultimately causing premature failure to yield poor ductility.

X Li et al
The nanoprecipitates in the 250-16 h sample consisted predominantly of nano-β ′ and nano-β 1 .As displayed in figures 15(g)-(i) and figure 19(e-ii) and (e-iii), these nano-β 1 are sheared by basal slip of <a> dislocations.It is the shearability of these nano-β 1 that enables the formation of substantial slip bands to coordinate stress concentrations due to dislocation obstruction [42,61,62].As a result, compared to the 200-58 h sample, although the strength of 250-16 h sample decreased by 15%, the elongation increased by 109%.

Conclusion
In the present work, a thin-walled component was successfully prepared via wire-arc DED using the customized GW102K wires with high RE content.The impact of the heat treatment scheme on the microstructure evolution and precipitation behavior was thoroughly investigated.Eventually, prominent combinations of strength and ductility are realized by manipulating the nanoprecipitates, and the effects of the nanoprecipitates on the deformation mechanism are revealed in depth.The following conclusions can be drawn: (1) The unique microstructure of wire-arc DED-fabricated GW102K alloy is characterized by fine equiaxed α-Mg with an average size of 26.5 µm, metastable nanoβ ′′ cuboid phases, Zr clusters, reticulated β-Mg 24 (Gd, Y) 5 eutectic phases and RE-rich regions distributed along the grain boundaries.Following T4 treatment at 530 • C for 1 h, both the eutectic phases and RE-rich regions in the T4 sample completely disappear, accompanied by slight grain growth.
(2) Notably, the manipulation of nanoprecipitates type and content was realized by tailoring the aging parameters.
(3) The T4 sample displays sound ductility with an EL of (14.6 ± 0.1)%, while those T6 samples achieve prominent combinations of strength and ductility, which are comparable to or much better than that of the Mg-RE alloys prepared by AM and traditional processes.Specifically, the 200-58 h sample exhibits ultra-high strength and poor ductility with a UTS of (371 ± 1.5) MPa and an EL of (4 ± 0.2)%, while the 150-96 h and 250-16 h samples achieve the moderate strength and ductility with the UTS ranging from 290 MPa to 320 MPa, and the EL ranging from 6% to 9%.(4) The minor nano-β ′′ and fractured eutectic phases are responsible for the inferior strength and moderate ductility of the as-built sample, respectively.The excellent ductility of the T4 sample can be attributed to the active dislocation slip activity.The serious dislocation blocking induced by the dense nano-β ′ and the absence of slip activity should account for the ultra-high strength and poor ductility of the 200-58 h sample.Substantial slip bands arising from shearable nano-β 1 contribute to the good strength-ductility synergy in the 250-16 h sample.

Figure 1 .
Figure 1.Information about the experimental samples.(a) Macro-morphology of the component and extraction locations of those samples.(b) Sizes of uniaxial stretching samples.(c) TEM characterization region of the fractured location.

Figure 2 .
Figure 2. Microstructure of the as-built sample.OM located in the (a) bottom, and (b) top regions.(c) XRD pattern.(d)-(f) SEM and EDS mapping results.

Figure 4 .
Figure 4. EBSD results in the cross-section (XOZ plane) of the bottom region.(a) Inverse pole figure (IPF) of as-built sample.(b) IPF of T4 sample.(c) Pole figure (PF) of as-built sample.(d) PF of T4 sample.

Figure 5 .
Figure 5. Grain boundary characteristics in the cross-section (XOZ plane) of the bottom region.(a) Distribution map of grain boundaries in the as-built sample.(b) Distribution map of grain boundaries in the T4 sample.(c) Grain size of the as-built sample.(d) Grain size of the T4 sample.(e) Percentage of grain boundary with varying misorientation.

Figure 8
Figure 8 displays metallographic structure, aging hardening curves, XRD patterns, SEM and EDS results of the T6 samples treated with various aging parameters.As shown in figure 8(b), the micro-hardness of the as-built sample is about (90.8 ± 3.1)HV, while that of the sample after T4 treatment is decreased to (79.1 ± 2.7) HV.A significant increase in micro-hardness can be observed after only a short aging treatment of about 5 h, indicating that the customized GW102K alloys in this study possess excellent age-hardening capabilities.Figure8(b) also Figure8displays metallographic structure, aging hardening curves, XRD patterns, SEM and EDS results of the T6 samples treated with various aging parameters.As shown in figure8(b), the micro-hardness of the as-built sample is about (90.8 ± 3.1) HV, while that of the sample after T4 treatment is decreased to (79.1 ± 2.7) HV.A significant increase in micro-hardness can be observed after only a short aging treatment of about 5 h, indicating that the customized GW102K alloys in this study possess excellent age-hardening capabilities.Figure 8(b) also suggests that the time required to reach peak micro-hardness decreases as the aging temperature increases.The microhardness of the over-aged samples all shows a gentle downward trend with increasing aging time, which can be attributed to the fact that the over-aged state will increase the size of nano-precipitates.It is apparent that the optimum aging temperature is 200 • C with aging time of 58 h, where the peak micro-hardness reaches (134.8 ± 3.2) HV.For simplicity, these T6 samples are named using the aging temperature plus the peak aging time, such as 150-96 h, 180-86 h, 200-58 h, 220-24 h, and 250-16 h.The peak micro-hardness of the 150-96 h, 180-86 h, 220-24 h, and 250-16 h samples are (103.6 ± 3.2) HV, (124.0 ± 4.4) HV, (123.8 ± 4.9) HV, and (110.7 ± 7.5) HV, respectively.Compared to the metallographic structure of the T4 sample (figure 6(e)), the grain boundaries in the 200-58 h sample are more clearly defined (figure 8(a)).Although the EDS mapping results shown in figures 8(d)-(f) are quite similar to the T4 sample, the XRD pattern in figure 8(c) detects a large number of intense peaks corresponding to the nano-precipitates.In order to reveal the reason for the large differences in the peak micro-hardness corresponding to each aging parameter, we adopted the TEM technique to characterize the aging nano-precipitates of 150-96 h, 200-58 h, and 250-16 h samples.Figure9depicts the nano-precipitates of the 150-96 h

Figure 6 .
Figure 6.Microstructure of the T4 sample.(a) SEM image of the as-built sample.SEM image of the sample solution treated at 530 • C for (b) 0.5 h, (c) 1 h, (d) 2 h.(e) OM images, (f) XRD pattern, and (g)-(i) EDS mapping of the sample solution treated at the optimal solution parameters.
Figure 13 displays the fractographies to analyze the fracture mode.Massive ductile tearing ridges, rough cleavage planes, and limited secondary cracks can be observed on the fractographies shown in figures 13(a)-(c).As a result, the as-built sample can be recognized as hybrid ductile-brittle failure mode.Conversely, massive deep dimples can be identified in the T4 samples, implying a typical ductile fracture mode (figures 13(d)-(f)).As shown in figures 13(g)-(i) and (m) -(o), there are substantial rough cleavage planes and some ductile tear ridges on the fractographies of 150-96 h and 250-16 h samples, which should be identified as typical hybrid ductile-brittle failure mode.In comparison to the fractographies of 150-96 h and 250-16 h samples, 200-58 h sample exhibits smoother cleavage planes and tear ridges (figures 13(j)-(l)).Consequently, the 200-58 h sample belongs to the typical brittle failure mode, which is in agreement with the poor ductility of 4%.
)-(f) show the deformation microstructure of the T4 sample, where only deformation microstructures, such as slip bands and dislocation substructures, are observed.As shown in figure14(d), numerous dislocations and slip bands terminate near grain boundaries.Besides, as illustrated in figures 14(e) and (f), substantial dislocations are hampered by slip bands to develop dislocation walls.When compared to the as-built sample, T4 sample does not suffer from premature failure due to the fracture of the β-phases and exhibits more active slip motion.

Figure 15 Figure
Figure 15 presents the microstructure in the fractured positions of the T6 samples.Similar to the as-built and T4 samples, as shown in figures 15(a) and (b), a large number of unidirectional slip bands are formed on the matrix of 150-96 h sample, and some dislocations are effectively hindered by these slip bands to develop the visible dislocation walls.Moreover, as depicted in figure 15(c), the dislocation motion is hindered by the dense nano-β ′′ and nanoβ ′ precipitates.For 200-58 h sample, as the deformation microstructure shown in figures 15(d)-(f), although massive dislocations are blocked by nano-β ′ precipitates to form a

Figure 14 .
Figure 14.TEM results of fractured (a)-(c) as-built samples and (d)-(f) T4 samples.(a) Fractured β-phases.(b) Slip bands in HAADF image.(c) Planar dislocation array in the dark field image.(d) Slip bands terminate at grain boundaries.(e) and (f) Interaction between dislocation and slip bands.The purple arrow represents the slip direction.

Figure 19 .
Figure 19.Schematic diagrams illustrating the deformation mechanism concerning substructure evolution and crack propagation in those different types of samples.(a) As-built sample.(b) T4 sample.(c) 150-96 h sample.(d) 200-58 h sample.(e) 250-18 h sample.The (i), (ii) and (iii) correspond to the elastic stage, plastic stage, fracture modes of the deformation process for each type of sample, respectively.