*Article* **Selective Electron Beam Melting (SEBM) of Pure Tungsten: Metallurgical Defects, Microstructure, Texture and Mechanical Properties**

**Xin Ren 1, Hui Peng 2, Jingli Li 1, Hailin Liu 1, Liming Huang <sup>1</sup> and Xin Yi 1,\***


**Abstract:** Effects of processing parameters on the metallurgical defects, microstructure, texture, and mechanical properties of pure tungsten samples fabricated by selective electron beam melting are investigated. SEBM-fabricated bulk tungsten samples with features of lack of fusion, sufficient fusion, and over-melting are examined. For samples upon sufficient fusion, an ultimate compressive strength of 1.76 GPa is achieved at the volumetric energy density of 900 J/mm3–1000 J/mm3. The excellent compressive strength is higher and the associated volumetric energy density is significantly lower than corresponding reported values in the literature. The average relative density of SEBM-fabricated samples is 98.93%. No microcracks, but only pores with diameters of few tens of micrometers, are found in SEBM-ed tungsten samples of sufficient fusion. Properties of samples by SEBM and selective laser melting (SLM) have also been compared. It is found that SLM-fabricated samples exhibit inevitable microcracks, and have a significantly lower ultimate compressive strength and a slightly lower relative density of 98.51% in comparison with SEBM-ed samples.

**Keywords:** pure tungsten; selective electron beam melting (SEBM); porosity; microstructure; mechanical properties

### **1. Introduction**

Tungsten has a high melting point, a high density, a low erosion tendency, high thermal stress resistance, and high thermal conductivity, and exhibits low swelling and low tritium retention. Owing to these superior thermophysical properties, tungsten and tungsten alloys are considered the most promising candidate materials for plasma facing components in nuclear reactors [1], and have also been attractive for industrial, aerospace, and medical applications such as high temperature furnaces and X-ray shielding [2]. On the other hand, tungsten materials show processing difficulties and have limited engineering applications due to the inherent features of high-melting point, brittleness at room temperature, and high temperature oxidation behaviors [3,4].

Selective laser melting (SLM) and selective electron beam manufacturing (SEBM) are two important representatives of powder bed based additive manufacturing processes for forming near-net shaped components of metals. They not only offer new feasible processing methods for refractory metals, but also liberate the design freedom and significantly expand the engineering application scope of refractory metals.

Rapid progress has been made in understanding the processing-microstructureproperties relationships of pure tungsten samples fabricated by SLM [3–7]. For example, the SLM-fabricated bulk pure tungsten samples of high relative densities of 95–98.51% can achieve an ultimate compressive strength up to 1.01 GPa after heat treatment [7,8]. Experimental observations also indicate that microcracks owing to high residual stresses

**Citation:** Ren, X.; Peng, H.; Li, J.; Liu, H.; Huang, L.; Yi, X. Selective Electron Beam Melting (SEBM) of Pure Tungsten: Metallurgical Defects, Microstructure, Texture and Mechanical Properties. *Materials* **2022**, *15*, 1172. https://doi.org/ 10.3390/ma15031172

Academic Editors: Vadim Sufiiarov and Konda Gokuldoss Prashanth

Received: 1 January 2022 Accepted: 31 January 2022 Published: 3 February 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

induced by high temperature gradients in SLM-ed samples are inevitable [4–11]. In SLM-ed samples, one can also find pores whose formation is most likely attributed to the entrapped protective inert gas (e.g., argon) by the Marangoni effect [7,12].

In contrast to SLM with laser power on the order of hundreds of Watts, the electron beam power of SEBM can reach as high as 3 kW, much higher energy input than SLM [13]. In SEBM, preheating the substrate and every powder layer by the defocused electron beam can significantly minimize the residual stress in the fabricated components. The SEBM process works under a vacuum condition which forms a nearly perfect protection against oxidization and gas contamination [14,15]. Owing to these features, SEBM is more suitable than SLM for fabricating refractory metal components and brittle materials with affinity to gases such as oxygen at high temperature [16–19].

In comparison with studies of SLM-ed pure tungsten samples [3–7], few studies on the fabrication of pure tungsten by SEBM have been reported in the literature [20]. A processing window for SEBM-fabricated pure tungsten is preliminarily determined by characterizing the surface morphologies of the melt pool, and the sample compression strength along the building direction is reported as 1.56 GPa [20]. Nevertheless, important questions, such as how to relate the mechanical properties of SEBM-ed pure tungsten samples to the microstructure and texture, remain to be fully elucidated. Moreover, no comparative analysis has been performed on the microstructural and mechanical characterizations of pure tungsten samples manufactured by SEBM or SLM.

In this work, we focus on some key issues of SEBM-ed pure tungsten samples, such as features of microdefects including cracks and pores, microstructure, texture, and mechanical properties. A thorough comparison on the microstructure and mechanical properties of pure tungsten samples manufactured by SEBM or SLM is performed, which is helpful in understanding and optimizing the additive manufacture of tungsten samples.

### **2. Materials and Experimental Procedures**

### *2.1. Materials and SEBM Process*

Spherical powder particles of good flowability are most suitable for SEBM. However, owing to a large amount of powder required for SEBM and the corresponding expensive cost attributed to spheroidization of pure tungsten powder, polygonal pure tungsten powder of good flowability emerge as an alternative option. The Hall flow rate of polygonal pure tungsten powder particles with size of 65 μm–105 μm (Figure 1) is measured according to the ASTM B213 standard test method. Measurements with 50 g mass of powder require about 8.5 s, indicating fluidity good enough for SEBM.

**Figure 1.** Morphology of tungsten powders used in the SEBM process.

The Arcam A2XX EBM system (Mölndal, Sweden) is used to fabricate the bulk pure tungsten samples with a minimum electron beam diameter of around 250 μm. The vacuum pressure in the chamber is kept below 0.2 Pa. The 316 stainless steel substrate plate is initially preheated to 1150 ◦C by fast scanning with a defocused electron beam to decrease thermal gradients during the building process. Then, a sample is fabricated layer by layer, and each layer is subject to a four-step process of depositing-preheating-melting a powder layer and lowering platform (Figure 2a): (1) depositing a powder layer onto the substrate plate by the rake, (2) preheating and slightly sintering the powder layer with a defocused electron beam, (3) selectively scanning and melting the preheated powder layer by a focused electron beam according to a schemed computer-aided system, and (4) lowering the building platform by a nominal layer thickness and repeating from step (1) for the next layer fabrication. Step (2) is essential to prevent so-called smoke events occurring in step (3), which lead to unfavorable powder spreading within the machine and eventual termination of the SEBM process [21]. A zigzag scanning pattern with an interlayer misorientation of 90◦ between layers is adopted (Figure 2b).

**Figure 2.** Schematic of the four-step process for building each layer in SEBM (**a**) and zigzag scanning strategy (**b**). BD, building direction.

Critical processing parameters impacting the input energy density in SEBM include the electron beam power *P*, scanning velocity *v*, powder layer thickness *t*, and hatch distance *h*. A combination of them gives the volumetric energy density *E* = *P*/(*h* × *v* × *t*), which is used to estimate the energy density input into the powder layer. The beam power is given by *P* = *I* × *U*, where *I* is the beam current and *U* is the voltage. The building processes exhibit features of lack of fusion, sufficient fusion, and over-melting as *E* increases. A lower *E* with a high scanning speed may induce the incomplete spread of the liquid metal and lead to the lack of fusion. An excessive *E* with a low scanning speed could result in the disturbance of the melt pool. Values of the processing parameters are listed in Table 1. The SEBM-ed pure tungsten samples have dimension of 15 × <sup>15</sup> × 20 mm<sup>3</sup> (Figure 3).

**Table 1.** Processing parameters for the fabrication of tungsten using SEBM.


<sup>1</sup> S1 and S2, lack of fusion; S3 and S4, sufficient fusion; S5, over-melting.

**Figure 3.** Pure tungsten cube (S4) of size 15 <sup>×</sup> <sup>15</sup> <sup>×</sup> 20 mm3 fabricated by SEBM.

### *2.2. Characterization*

The SEBM-ed bulk pure tungsten samples are cut from the substrate using wire electrical discharge machining, then cleaned with acetone, alcohol, and water. For microstructure characterization, smaller specimens cut from the bulk components are mechanically ground, followed by electrolytic polishing with 2% NaOH solution at voltage of 20 V, then these specimens are examined with electron backscatter diffraction (EBSD) analysis. To detect the metallographic microstructure, the samples undergo electrolytic etching with 2% NaOH solution at voltage of 5 V and are examined using the Leica DM2700 M optical microscope (Wetzlar, Germany). Cylinder samples with length-to-diameter ratio *L*/*D* = 1.5 are cut from the fabricated components and the end faces are polished for compression tests at room temperature using an Instron 5585H universal testing equipment (Norwood, United States) at a strain rate of 10<sup>−</sup>3/s.

### **3. Results and Discussion**

### *3.1. Metallurgical Defects*

Three representative top surface morphologies and side view optical micrographs of the bulk pure tungsten samples fabricated by SEBM are shown in Figure 4. At high scanning velocity and low volumetric energy density, the liquid metal has no sufficient time to fully spread owing to the rapid solidification velocity, and evident shrinkage holes could be formed. As shown in Figure 4a, the corresponding fabricated samples show features of lack of fusion with large shrinkage holes having rough surfaces. As the scanning velocity *v* decreases to a proper value, samples of smooth surfaces and dense structures with few tiny pores are fabricated with sufficient fusion (Figure 4b). As *v* further decreases, the energy input becomes excessive, causing over-melting and unfavorable disturbance of the melt pool and leading to a higher surface periphery (Figure 4c). Although the energy inputs for the samples of sufficient fusion and over-melting are different, the pores there have similar size and number. The pore formation might be due to the trapping of gas such as residual oxygen within the powder particles, even the building process works in a high vacuum environment. A thorough and detailed mechanistic study on the formation of different surface morphologies is challenging and deserves further detailed investigations in the future. The effects of pores on the mechanical properties of the SEBM-ed samples are discussed comparing with SLM-ed samples in Section 3.3.

No microcracks are seen; only pores (irregular shrinkage holes in the case of lack of fusion, and tiny spherical pores in cases of sufficient fusion and over-melting) exist in the SEBM-ed tungsten samples (Figure 4). In SLM-ed tungsten samples, microcracks are usually inevitable [6,8–10], as shown in Figure 5. A widely adopted aspect causing the microcrack formation in the SLM-ed samples is the high thermal stress generated by the high temperature gradient during the SLM process. That high thermal stress could cause tearing along grain boundaries, which are embrittled by the oxidation and resulting impurities [4,9,22,23]. It has been reported that a high concentration of oxygen and impurities segregate to the grain boundaries during the cooling process of melted tungsten in SLM [9,24], and the strength of pure tungsten and tungsten alloys are greatly affected by the oxides distributed at grain boundaries [22,25–28].

**Figure 4.** Three representative top surface morphologies (top row) and side view optical micrographs (bottom row) of the bulk pure tungsten samples (S1, S4, and S5) fabricated by SEBM with characteristics of lack of fusion (**a**), sufficient fusion (**b**), and over-melting or excess energy input (**c**). In (**c**), the surface periphery is higher than the central region. Lateral dimension is 15 <sup>×</sup> 15 mm2.

**Figure 5.** Optical micrographs of polished cross-sections (top (**a**) and front (**b**) views) for bulk pure tungsten samples fabricated by SLM.

In SEBM, the vacuum environment prohibits the oxidation, and the high temperature gradient is lower compared to SLM. Therefore, no microcracks are formed in SEBMed samples. For example, in SEBM for tungsten fabrication the substrate is heated to 1150 ◦C by the defocused electron beam and that temperature is maintained throughout the building process. Each powder layer is also preheated. Therefore, the temperature gradient in SEBM is much lower than that in SLM for tungsten where the substrate can only be pre-heated up to 200 ◦C and there is no preheating for powder layers [8]. The lower temperature gradient in SEBM gives rise to a lower cooling rate, which in turn causes a much lower thermal stress [13,29,30] and low possibility of microcrack formation. Moreover, the ductile-to-brittle transition temperature (DBTT) of pure tungsten is in a range of 150 ◦C to 400 ◦C [1,31]. The sustained elevated temperature in SEBM could not only relieve the thermal stress [32], but also enable SEBM-ed tungsten samples above DBTT exhibiting plasticity to a certain extent. Overall, the crack formation in SEBM-ed tungsten is inhibited during the building process.

According to the Archimedes' principle, relative densities of as-fabricated SEBM samples S1–S5 are measured. Having knowledge of the volumetric energy density *E* in Table 1, the relation between the relative density and the volumetric energy density is determined (Figure 6). It is shown that samples of either sufficient fusion or over-melting have high relative densities, and an S4 sample of sufficient fusion has the highest relative density of 98.93%. Though S5 samples have high relative densities, the associated irregular surface periphery, as shown in Figure 4c, limits the usage of S5 samples of over-melting as near-net shaped components.

**Figure 6.** The relative density versus the volumetric energy density of the pure tungsten additive fabrication.

In Figure 6, we also provide the values associated with SLM-ed tungsten for references. In our previous work on the additive manufacture of pure tungsten samples by SLM with spherical powder particles of 5 μm to 25 μm in diameter, the optimized SLM processing parameters include layer thickness of 30 μm, hatch distance of 0.08 mm, laser power of 300 W and scanning velocity of 300 mm/s at a laser spot size of around 70 μm [8]. As in the SLM-ed samples there only exist small microcracks and the crack number is small (Figure 5), the relative density of the SLM-ed samples is only slightly lower than the SEBMed samples of sufficient fusion with the absence of microcracks. In SLM and SEBM, the energy absorption depends on the powder particle size and shape as well as the physical features of laser and electron beams [33,34]. As all these parameters in the present work of SEBM and our previous work of SLM are significantly different, it is not surprising that the volumetric energy densities of the SLM-ed and SEBM-ed tungsten samples with similar relative densities could be significantly different (Figure 6).

### *3.2. Microstructure and Texture*

Columnar grains are observed in all SEBM-ed samples (Figure 7). It is known that the columnar grains grow along the maximum temperature gradient direction. In most cases, the building direction is parallel to the direction of the maximum temperature gradient, and one can see that columnar grains grow epitaxially along the building direction for SEBM-ed samples (as demonstrated in Figure 7a,c). In some cases, particularly at the sample edges surrounded by thick powder layers, the heat dissipation rate of the particles is much lower than that of the bulk, resulting in an evident deviation of the maximum temperature gradient direction from the building direction. That direction deviation is gradually reduced toward the bulk region, consistent with marked arrows in Figure 7b.

**Figure 7.** Columnar grain structures of as-fabricated pure tungsten samples of lack of fusion (**a**), sufficient fusion (**b**), and over-melting (**c**).

Microstructure and crystallographic texture analysis of the SEBM-ed sample S4 is performed using EBSD analysis. As shown in Figure 8, coarse grains appear equiaxed from the top view (Figure 8a), significantly different from the scattered checkboard pattern of SLM-ed samples reported in our previous work [8]. Long columnar grain structures along the building direction are observed in SEBM-ed samples (Figure 8a), indicating a stable melt pool during the SEBM process [20]. In contrast, columnar grain structures in the SLM-ed samples are relatively small (grain diameter about 125 μm in comparison with 320 μm for the SEBM-ed samples) and discrete [8], indicating disrupted epitaxy growth of columnar grains during SLM, which might be owing to the scanning strategy of a pattern with an interlayer misorientation of around 67◦ [8,10].

**Figure 8.** EBSD analysis of the SEBM-ed sample S4. (**a**) Grain size characterization, (**b**) inverse pole figure (IPF) coloring map, and (**c**) orientation pole figure taken from top view in (**b**). Scale bars in (**a**,**b**), 200 μm.

The IPF coloring map in Figure 8b and the pole figure in Figure 8c show that the SEBMed pure tungsten S4 sample has a strong <100> grain orientations. As pure tungsten has a body-centered cubic crystal structure, the columnar grains usually prefer <100> growth direction [35], consistent with Figure 8c. In the process of melting and solidification of tungsten, the epitaxial growth of columnar grains shows a strong rotated cube {100} <110> along the building direction. An unusual (111) texture has been found in SLM-ed tungsten, which is owing to the rotation of thermal gradient caused by the 67◦ rotation of scanning direction between layers [29,36]. The scanning direction with rotation of 90◦ between layers adopted in the present SEBM process may have little influence on the direction of

maximum thermal gradient, and therefore the preferred <100> growth direction is observed in this work.

### *3.3. Mechanical Properties*

Pore formation affects not only the density but also the mechanical properties of fabricated samples. Compression tests of cylindrical tungsten samples with features of lack of fusion (S1), sufficient fusion (S4), and over-melting (S5) are conducted at room temperature (Figure 9). The cylindrical samples have diameter of 3 mm and length of 4.5 mm. The average ultimate compressive strength for each sample is given in the inset of Figure 9.

**Figure 9.** Compressive stress–strain curves for SEBM-ed pure tungsten samples S1, S4, and S5. Inset, ultimate compressive strengths of corresponding pure tungsten SEBM-ed bulk samples and previous SLM-ed sample.

Owing to the large and irregular pore structures in S1 samples (Figures 4a and 7a), the corresponding ultimate compressive strength is as small as 0.61 GPa and the fracture engineering strain is as small as around 10%. Samples S4 and S5 of dense structures containing tiny pores (Figures 4b,c and 7b,c) have a much larger ultimate compressive strength and fracture strain. Specifically, the ultimate compressive strength of the S4 sample (sufficient fusion) is 1.76 GPa with a fracture engineering strain of around 40%, both larger than reported values of 1.56 GPa and 18% in the literature [20]. Moreover, the volumetric energy density for the S4 sample is about 1000 J/mm3, significantly lower than that of 1440 J/mm3–3840 J/mm3 in ref. [20]. The compressive strength of optimized SLM-ed bulk pure tungsten samples is around 1.01 GPa [8], much lower than that achieved in the present work by SEBM, though the SLM-ed sample also has a high relative density of 98.51%. This observation suggests that the large ultimate compressive strength of SEBM-ed S4 sample mainly results from the absence of microcracks in fabricated samples, rather than the high relative density. In S4, samples the columnar grains are relatively coarse due to the long-time high temperature fabrication process, and grain refinement with addition of the carbide nanoparticles or oxide dispersion [9,22,23,26,37] could be employed to further improve the mechanical properties of fabricated samples.

### **4. Conclusions**

Pure tungsten samples of high relative density and superior ultimate compressive strength are fabricated by SEBM at different volumetric energy densities. Samples with features of lack of fusion, sufficient fusion, and over-melting are identified. The corresponding metallurgical defects, microstructure, texture, and mechanical properties of the SEBMfabricated samples are analyzed and compared with counterparts of the SLM-fabricated samples. The main conclusions are summarized as follows.


**Author Contributions:** Conceptualization, X.R. and X.Y.; methodology, X.R. and H.P.; investigation, X.R., H.P., J.L., H.L., L.H. and X.Y.; writing—original draft preparation, X.R.; writing—review and editing, X.Y.; supervision, X.Y.; and funding acquisition, X.Y. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Natural Science Foundation of China (grant No. 11988102) and National Science and Technology Major Project (2017-VI-0003-0073).

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


### *Article* **Computer Simulation of Hydrodynamic and Thermal Processes in DLD Technology**

**Gleb A. Turichin \*, Ekaterina A. Valdaytseva \*, Stanislav L. Stankevich and Ilya N. Udin**

World-Class Research Center, "Advanced Digital Technologies", State Marine Technical University, 190121 Saint Petersburg, Russia; s.stankevich@ilwt.smtu.ru (S.L.S.); youdin@ilwt.smtu.ru (I.N.U.) **\*** Correspondence: gleb@ltc.ru (G.A.T.); laser@corp.smtu.ru (E.A.V.)

**Abstract:** This article deals with the theoretical issues of the formation of a melt pool during the process of direct laser deposition. The shape and size of the pool depends on many parameters, such as the speed and power of the process, the optical and physical properties of the material, and the powder consumption. On the other hand, the influence of the physical processes occurring in the material on one another is significant: for instance, the heating of the powder and the substrate by laser radiation, or the formation of the free surface of the melt, taking into account the Marangoni effect. This paper proposes a model for determining the size of the melt bath, developed in a onedimensional approximation of the boundary layer flow. The dimensions and profile of the surface and bottom of the melt pool are obtained by solving the problem of convective heat transfer. The influence of the residual temperature from the previous track, as well as the heat from the heated powder of the gas–powder jet, taking into account its spatial distribution, is considered. The simulation of the size and shape of the melt pool, as well as its free surface profile for different alloys, is performed with 316 L steel, Inconel 718 nickel alloy, and VT6 titanium alloy

**Keywords:** direct laser deposition; heat transfer; mass transfer; hydrodynamics; simulation of the melt pool; alloys

### **1. Introduction**

The direct laser deposition (DLD) process, according to the classification considered in [1], is currently becoming a more and more promising technology for the additive production of parts for various purposes in shipbuilding, aircraft construction, mechanical engineering, and other industries. There is already a positive experience in the manufacture of ship fittings, propellers [2,3], water jet propellers [4,5], large-sized products and machine parts [6], high-pressure vessels, and others. This technology belongs to the direct metal deposition (DMD) technologies, in which the product is formed from a metal powder supplied by a gas jet directly into the area of action of focused laser radiation. In this case, the heating and melting of the powder and the substrate is controlled by equipment in order to maintain the process in the stability zone. The high power of the laser source should ensure and maintain a high melting rate of the powder in order to ensure high process productivity up to 1.5–2.5 kg/h [4]. However, this can lead both to the appearance of instability of the wall formation described in [7], and to a deviation from the specified dimensions [8]. A large number of important parameters of the processing mode, and the danger of leaving the process in the zone of unstable formation, greatly complicate the selection of technological parameters by the experimental method. Therefore, the availability of a physically adequate and fast mathematical model that allows the performance of numerical modeling will facilitate the development and optimization of technological parameters for the direct laser deposition process. A large number of papers describe the use of various numerical schemes of finite element analysis [9–11], analytical models [12], and even statistical models [13] for modeling thermal and hydrodynamic processes. However, in most of them, the case of cladding the bead on a thick and wide substrate is considered. This is not suitable

**Citation:** Turichin, G.A.; Valdaytseva, E.A.; Stankevich, S.L.; Udin, I.N. Computer Simulation of

Hydrodynamic and Thermal Processes in DLD Technology. *Materials* **2021**, *14*, 4141. https:// doi.org/10.3390/ma14154141

Academic Editor: Federico Mazzucato

Received: 1 July 2021 Accepted: 22 July 2021 Published: 25 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

for thin walls, due to differences in boundary conditions. The influence of the Marangoni effect, capillary forces, and the mutual influence of hydrodynamics and heat transfer in the melt pool are also rarely taken into account. Finite element analysis allows scientists to take these factors into account, but the calculation time can be hours, or even days. The presented work is a development of the model developed earlier in [6,7,14,15]; it was developed specifically for modeling the process of forming thin-walled structures. This model takes into account the Marangoni effect, the transfer of the powder by a gas jet, the heating of the powder by laser radiation, the interaction of the jet with the substrate, and heat transfer in the solid and liquid phases, as well as the hydrodynamics of the melt pool. This work presents the results of theoretical studies and modeling of joint thermal and hydrodynamic processes in the stationary case for the DLD process, taking into account the influence of the heated powder on the melt pool.

### **2. Materials and Methods**

### *2.1. Melt Flow Description*

The high-speed DLD process is characterized by the formation of a melt pool with a length "L" much larger than the width "b" and depth "H" (Figure 1).

**Figure 1.** Design diagram of the deposited wall and the melt pool.

In this case, to describe the velocity field in the melt, we can limit ourselves to a one-dimensional formulation, and use the approximation of a one-dimensional boundary layer. In this case, the longitudinal velocity *Vx*, directed along the direction of movement of the laser, is much greater than the transverse velocities *Vy* and *Vz*. In the case of a steady-state process, the Navier–Stokes fluid motion equation can be written as:

$$V\_{\mathbf{x}} \frac{\partial V\_{\mathbf{x}}}{\partial \mathbf{x}} = -\frac{1}{\rho} \frac{\partial p}{\partial \mathbf{x}} + \nu \frac{\partial^2 V\_{\mathbf{x}}}{\partial z^2} \tag{1}$$

The boundary condition at the "bottom" of the melt pool is given as:

$$V\_x|\_{z=0} = 0,$$

The boundary condition on the "top" surface is obtained from the requirement of the stress tensor continuity:

$$-\eta \frac{\partial V\_x}{\partial z}\Big|\_{z=H} = \frac{\partial \sigma}{\partial x}$$

We assume that the temperature change along the melt pool surface is much less than the average temperature of its surface. We assume that *Ts* is the maximum surface temperature, while *Tm* and *Tb* are the melting and evaporation temperatures, respectively. Assuming that the law oftemperature drop to the tail of the melt pool is linear , then we can write:

$$\eta \frac{\partial V\_x}{\partial z}\Big|\_{z=H} = \frac{\sigma}{L} \frac{T\_s - T\_m}{T\_b - T\_m} = \frac{\sigma^\*}{L} \tag{2}$$

Let the velocity distribution of the liquid in the melt have a parabolic shape. This approach will ensure that the boundary conditions and the conditions of mass flow conservation along the "x" axis are met.

$$V\_{\mathbf{x}}(z) = V\_{\mathbf{x}}\left(\mathbf{a} + \beta z + \gamma z^2\right) \tag{3}$$

Substituting Equation (3) into the boundary conditions, we obtain expressions for the coefficients of the equation:

$$\begin{aligned} \alpha &= 0\\ \gamma &= -\frac{3}{4\overline{\eta}\overline{H}} \left(\frac{2}{H} - \frac{\sigma^\*}{L\overline{\eta}V\_x}\right) \\ \beta &= \frac{2}{H} \left(1 + \frac{H}{4} \left(\frac{2}{H} - \frac{\sigma^\*}{L\overline{\eta}V\_x}\right)\right) = \frac{2}{H} \left(\frac{3}{2} - \frac{\sigma^\*}{4\overline{\eta}V\_x} \frac{H}{L}\right) = \frac{3}{H} - \frac{\sigma^\*}{2\overline{\eta}V\_xL} \end{aligned} \tag{4}$$

Then Equation (1) will look like this:

$$V\_{\mathbf{x}} \frac{\partial V\_{\mathbf{x}}}{\partial \mathbf{x}} = -\frac{1}{\rho} \frac{\partial p}{\partial \mathbf{x}} - 3\nu \frac{V\_{\mathbf{x}}}{H^2} + \frac{3\sigma^\*}{2\rho LH} \tag{5}$$

We will use the continuity equation to relate the position of the weld pool surface to the melt velocity *Vx*. Here, it is necessary to take into account that the density of the mass flow *j*(*x*) incident on the melt surface is determined by a gas–powder jet. Then the continuity equation for the flow will be written as follows:

$$\frac{\partial}{\partial \mathbf{x}} (V\_{\mathbf{x}} H) = \frac{j(\mathbf{x})}{\rho} \tag{6}$$

Considering that the pressure in the melt:

$$p = \frac{\sigma}{\mathcal{R}}$$

and when *H<b* we get:

$$R \approx b + \frac{H^2}{2b}, \; p \approx \frac{\sigma}{b} - \frac{\sigma H^2}{2b^3}$$

In this case, the terms of the Navier–Stokes equation associated with the change in the transverse radius of curvature of the surface will look like:

$$\frac{\partial p}{\partial x} = \frac{\sigma}{2b^3} \frac{H \partial H}{\partial x}$$

A change in the longitudinal radius of the curvature of the melt surface will give an additional pressure, which can be expressed as:

$$p\_{add} = \sigma \frac{\partial^2 H}{\partial x^2}.$$

Then, one can write:

$$V\_{\mathbf{x}} \frac{\partial V\_{\mathbf{x}}}{\partial \mathbf{x}} = \frac{\sigma}{\rho b^3} \frac{H \partial H}{\partial \mathbf{x}} - 3\nu \frac{V\_{\mathbf{x}}}{H^2} + \frac{3\sigma^\*}{2\rho LH} \tag{7}$$

After integration, we will get:

$$H(\mathbf{x}) = \frac{1}{\rho v\_{\mathbf{x}}} \int\_{0}^{\mathbf{x}} j(\mathbf{x}) d\mathbf{x} \tag{8}$$

The boundary conditions for the last equation can be written as:

$$\left.H\right|\_{x=0} = 0,\\
\left.\frac{\partial H}{\partial x}\right|\_{x=0} = 0,\\
\left.\frac{\partial H}{\partial x}\right|\_{x=L} = 0.$$

Using Equation (8) in Equation (7), and solving it numerically, we obtain the profile of the upper surface of the melt pool, taking into account the Marangoni effect. The parameters of the length "L" and the depth "H" of the melt pool are determined by the solution to the heat transfer problem.

### *2.2. Influence of the Powder Jet on the Heat Transfer in the Deposited Wall*

As was shown in [6], the thermal field when applying a single deposited bead to a thin wall in a steady state can be described by the equation of convective heat transfer. At the same time, since the wall width is comparable to the diameter of the laser beam spot, the temperature gradient along the y axis can be neglected.

$$V\_{\mathbf{x}} \frac{\partial T}{\partial \mathbf{x}} = \chi \left( \frac{\partial^2 T}{\partial \mathbf{x}^2} + \frac{\partial^2 T}{\partial z^2} \right) \tag{9}$$

The boundary conditions for this case can be written as:

$$-\lambda \frac{\partial T}{\partial z}\bigg|\_{z=0} = q(\mathbf{x}) , \ T|\_{z \to \infty} \to T\_0 \tag{10}$$

where λ and χ correspond to heat conductivity and thermal diffusivity coefficients, respectively; *q*(*x*) is the distribution of the total energy flow on the melt pool surface; and *T*<sup>0</sup> is the initial temperature of the substrate.

Since the energy flux density on the pool surface includes the laser radiation intensity *I* and the convective heat flux brought by the heated powder, we can write:

$$q(\mathbf{x}) = I(\mathbf{x}) \cdot A + j(\mathbf{x}) \cdot \mathbf{c} \cdot \left(T\_p(\mathbf{x}) - T\_0\right) \tag{11}$$

To determine *Tp*, we can use a well-known analytical solution to the problem of temperature distribution in a homogeneous ball of radius *R* with an initial temperature *T*<sup>0</sup> for the case when a constant heat flow *qp* is fed into the ball through its surface [16,17]:

$$T\_i(r,t) = T\_0 + \frac{q\_p R}{\lambda} \left(\frac{3\chi t}{R^2} - \frac{3R^2 - 5r^2}{10R^2}\right) - \frac{2q\_p}{\lambda R r} \sum\_{i=1}^{\infty} \frac{\sin(\mu\_i r)}{\mu\_i^3 \cos(R\mu\_i)} e^{-\chi \mu\_i^2 t}$$

where *μ<sup>n</sup>* are the positive roots of the equation *tg*(*Rμ*) = *Rμ*; and *t* is the heating time, which is determined for each particle as the time of flight through the laser radiation zone

before it enters the melt pool. Knowing the density of the powder flow in the gas–powder jet, the trajectory of the particles, their size, and the amount that got into the melt pool for example, from [18]—it is possible to obtain the temperature distribution *Tp*(*x*) in the gas—powder jet at the time of meeting with the melt surface.

Furthermore, using the method of solving the convective heat transfer equation described in [6], for the temperature field during surfacing of the i-th layer, we obtain:

$$T(\mathbf{x},\mathbf{z}) = \frac{e^{-\frac{V\_{\mathbf{r}}}{2\zeta}}}{\lambda} \int\_{-\infty}^{\infty} \left( A \cdot I(\mathbf{x'}) + j(\mathbf{x'})c\left(T\_P(\mathbf{x'}) - T\_0\right) \right) e^{-\frac{V\_{\mathbf{r}'}}{2\zeta}} K\_0\left(\frac{V}{2\chi}\sqrt{\mathbf{z}^2 + \left(\mathbf{x} - \mathbf{x'}\right)^2}\right) d\mathbf{x'} + T\_W \tag{12}$$

where *Tw* is the residual temperature of the previous layer when applying the bead, determined by the product construction strategy.

### **3. Results and Discussion**

The proposed model was tested at the Institute of Laser and Welding Technologies of St. Petersburg State Marine Technical University, for various products and materials. A comparison of the calculation results and experimental data showed that the error does not exceed 20%; this is a good indicator of the performance of a fast, semi-analytical model.

For the calculations, data on physical and optical properties of the works [19–26] were used. The data shown in Table 1 were averaged.


**Table 1.** Thermophysical and optical properties of the chosen materials.

Examples of surface profile calculations for different values of motion speed and powder feed rate are shown in Figure 2. It is evident that the surface profile depends on the melt pool length "*L*" and depth "*H*".

A comparison of the melt pool surface profiles for different materials is shown in Figure 3. It can be seen from the figure that with the same mode parameters, the VT6 alloy gives a greater increment in height when the bead is deposited. At the same time, the process efficiency for steel and nickel alloy is approximately the same.

The impact of the heated powder jet on the surface temperature distribution along melt pool surface is shown in Figure 4.

Calculations show that even a small addition of energy flux with the heated powder leads to an increase in the melt pool's size (Figure 5).

For all materials, the contribution of heat from the heated powder was significant. In addition, due to the greater absorption capacity of the titanium alloy, not only was the melt pool length increased, but also the depth. For Inconel 718 and 316 L steel, this effect was not so significant. The figure shows that for 316 L steel and Inconel 718 alloy, there is a more elongated, but shallow melt pool, while for titanium alloy, on the other hand, there is a shorter and deeper melt pool. In the presence of a smaller pool, the amount of powder entering the melt will be less, and the efficiency of the process will decrease. It turns out that despite the higher absorption coefficient, having a melt pool with a smaller surface area, the efficiency of the deposition process for titanium alloy may be less than expected. This fact should be taken into account when selecting processing modes.

**Figure 2.** Melt pool top surface profile for 316 L steel for different powder feed rates (**a**) and motion speeds (**b**); laser beam radius on the surface was 2.5 mm, beam power was 2000 W, and powder jet diameter was 3 mm.

/RQJLWXGLDOFRRUGLQDWHPP

**Figure 3.** Melt pool top surface profiles for Inconel718, VT6, and 316 L steel in comparison with one another. Motion speed was 20 mm/s, laser beam power was 2000 W, laser beam radius on the surface was 2.5 mm, powder jet diameter was 3 mm, and powder feed rate was 2 kg/h.

ORQJLWXGLDOFRRUGLQDWHPP

**Figure 4.** Temperature distribution along the melt pool length. Laser power was 2000 W, motion speed was 20 mm/s, laser beam radius on the surface was 2.5 mm, powder jet diameter was 3 mm, and powder feed rate was 2 kg/h.

**Figure 5.** Melt pool shape. Laser power was 2000 W, motion speed was 20 mm/s, laser beam radius on the surface was 2.5 mm, powder jet diameter was 3 mm, and powder feed rate was 2 kg/h.

### **4. Conclusions**

Direct laser deposition is a complex physical process. When developing new modes, a number of experiments can be replaced by a physically adequate mathematical modeling. If possible, the mutual influence of different processes on one other should be taken into account. The hydrodynamics of the melt are inextricably connected to the temperature field formed in the substrate under the action of laser radiation. In addition to the radiation itself, the final values of the temperature field in general, and the surface temperature in particular, are significantly affected by the heated powder entering the melt via a gas– powder jet. The effect of additional heat input from the powder on the pool length is noticeable for all of the materials considered—the simulation results clearly demonstrated this. Such an increase can be predicted with high probability for any metal materials. The quantitative contribution of heat from the powder depends on both the thermophysical and optical properties of the material. For some materials, it may be significant to increase not only the length of the melt pool, but also its depth—as, for example, it turned out to be for a titanium alloy. However, it should be borne in mind that reducing the size of the pool at the

same mass flow density will lead to a decrease in the efficiency of the deposition process. With a significant variability in the properties of materials, the choice of parameters should mainly be carried out using mathematical modeling, since the experimental selection of modes can be extremely time-consuming.

**Author Contributions:** Conceptualization, G.A.T.; methodology, G.A.T. and E.A.V.; software, E.A.V.; validation, S.L.S. and I.N.U.; formal analysis, E.A.V.; investigation, G.A.T., E.A.V., S.L.S. and I.N.U.; resources, G.A.T. and E.A.V.; data curation, S.L.S. and I.N.U.; writing—original draft preparation, G.A.T. and E.A.V.; writing—review and editing, G.A.T. and E.A.V.; visualization, S.L.S. and I.N.U.; supervision, G.A.T.; project administration E.A.V.; funding acquisition, G.A.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Ministry of Science and Higher Education of the Russian Federation as part of the World-Class Research Center Program: Advanced Digital Technologies (contract No. 075-15-2020-903 dated 16.11.2020).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **Nomenclature**


### **References**


### *Article* **Features of Heat Treatment the Ti-6Al-4V GTD Blades Manufactured by DLD Additive Technology**

**Marina Gushchina 1,\*, Gleb Turichin 1, Olga Klimova-Korsmik 1, Konstantin Babkin <sup>1</sup> and Lyubov Maggeramova <sup>2</sup>**


**\*** Correspondence: gushcina\_mo@corp.smtu.ru

**Abstract:** Additive manufacturing of titanium alloys is one of the fastest growing areas of 3D metal printing. The use of AM methods for parts production in the aviation industry is especially promising. During the deposition of products with differently sized cross-sections, the thermal history changes, which leads to non-uniformity of the structure and properties. Such heterogeneity can lead to failure of the product during operation. The structure of deposited parts, depending on the thermal cycle, may consist of α', α + α' + β', and α + β in different ratios. This problem can be solved by using heat treatment (HT). This paper presents research aimed towards the determination of optimal heat treatment parameters that allows the reception of the uniform formation of properties in the aftertreatment state, regardless of the initial structure and properties, using the example of a deposited Ti-6Al-4V gas turbine blade.

**Keywords:** Ti-6Al-4V; direct energy deposition; thermal history; annealing; phase composition; microstructure; tensile properties

### **1. Introduction**

The transition of the world aircraft industry to innovative technologies, including the replacement of metal structures with composite materials, the development of additive manufacturing, and the introduction of new artificial intelligence systems in the aircraft control system, is becoming an increasingly relevant trend.

Computer engineering is widely used to create new materials in the aviation industry, which reduces the cost of the production of expensive full-scale prototypes by using virtual models. Additive technologies, most commonly based on the use of virtual models, are widely available due to the fact that they allow the manufacturing of products with complex geometric shapes and profiles. The use of such advanced technologies will significantly reduce the time introduction of products to the market and their cost, reduce material consumption, and reduce the product failure, which is a clear advantage for use in most industries [1–3].

The aviation industry is characterized by increased requirements for structural materials, and in some cases is a main customer and consumer of new materials and technologies [4]. A range of AM techniques are now available. The following additive manufacturing methods have become particularly popular in the aircraft industry for metal parts: selective laser melting (SLM), electron beam melting (EBM), and direct laser deposition (DLD). In SLM and EBM, thin powder layers are consolidated layer by layer using electron or laser beam scanning, and a layer is formed along the path of the corresponding usercreated model [5,6]. In DLD, the formation of a layer occurs by the coaxial supply of laser radiation and powder through special nozzles. Each of the methods has its advantages and disadvantages [7]. Due to the design features of the machines, SLM and EBM are more suitable for small- and medium-sized products, while DLD is attractive due to the possibility of producing large-sized products [7,8].

**Citation:** Gushchina, M.; Turichin, G.; Klimova-Korsmik, O.; Babkin, K.; Maggeramova, L. Features of Heat Treatment the Ti-6Al-4V GTD Blades Manufactured by DLD Additive Technology. *Materials* **2021**, *14*, 4159. https://doi.org/10.3390/ma14154159

Academic Editor: A. Javier Sanchez-Herencia

Received: 28 June 2021 Accepted: 23 July 2021 Published: 27 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

DLD is applicable to producing products from alloys based on iron, titanium, and nickel, as well as composite materials or compositionally graded materials. The α + β titanium alloy Ti-6Al-4V is widely used in aerospace applications, and much research has been conducted on AM with this alloy. There are many works devoted to the study of the structure and properties of additively manufactured Ti-6Al-4V in the building as well as the post-processing state [9–12]. Heat treatments (HTs) are solved problems of structural and phase inhomogeneity, residual stresses, and anisotropy. Hot isostatic pressing (HIP) of deposited samples allows the removal of defects, such as pores and non-fusion [13].

The fundamental possibility of achieving high strength and fatigue properties of model samples deposited from Ti-6Al-4V powder is shown in a number of works [14–18]. However, in most studies, the represented data for model samples have the form of plates, which are far from real parts. For example, [19,20] presented an investigation regarding the influence of product shape and thickness on other properties, and it was found that the shape has a more significant effect on thin-walled samples than on thick-walled samples. The overall dimensions of the samples also affected the properties: samples with a thick wall had higher strength, which, according to the authors, is associated with the size and morphology of the initial β-grain [21,22]. Thus, we can assume that the combination of thin-walled and thick-walled elements in one part can lead to significant heterogeneity of properties and structures throughout the product, which can negatively affect the performance of the entire part.

In this particular case, this problem can be solved by careful and accurate selection of modes, which requires a significant amount of work. On a global scale, the problem can be solved by developing the optimal heat treatment parameters that, regardless of the initial phase composition and grain size, the uniform structure and properties can provide.

This paper presents studies on the developing and selection of optimal heat treatment modes that ensure the uniform formation of properties in the deposited compressor gas turbine engine (GTE) blades of a Ti-6Al-4V after heat treatment, regardless of the initial structure and properties. In this work, we investigated a wide range of HT temperatures for laser deposited Ti-6Al-4V to establish the relationships between heating temperature and the mechanical properties. The selected mode was tested for samples with different initial microstructures and phase compositions.

### **2. Materials and Methods**

All the compressor gas turbine engine blades investigated in this research were built using the Ti-6Al-4V titanium alloy powder, with a fraction of 45–90 microns produced by PREP (plasma rotating electrode process). Samples were produced by the robotic complex developed at St. Petersburg State Marine University. The complex includes an LS-5 fiber laser, an anthropomorphic robot, a 6 m3 protecting chamber, a two-axis positioner, a powder feeder, and a control system. The path was generated in the Additive Control 1.0 program (ILWT, Saint Petersburg, Russia) using a 3D model of the blade. Figure 1a shows the trajectory of the deposition of the blade. The technological parameters for a Ti-6Al-4V alloy are presented in Table 1. The process of blade deposition and the final result are shown in Figure 1b,c. The process was carried out in a chamber with an argon atmosphere, and the oxygen level in the chamber was 2000 ppm.

**Table 1.** Technological parameters of the Ti-6Al-4V DLD process.


**Figure 1.** The blades of a gas turbine compressor; (**a**) the path generated in Additive Control; (**b**) the process of blade deposition; (**c**) for a Ti-Al-4V deposited blade.

The effect of heat treatment on the structural change was studied on the 5 × <sup>5</sup> × 10 mm3 pieces, which were cut from the central part of the deposited blades. The optimal heat treatment conditions for the Ti-6Al-4V titanium alloy were determined by varying the furnace heating temperature and holding time. The temperature varied within the range of 600–950 ◦C. The hold time was 2 h.

An air atmosphere was chosen for heat treatment in an SNOL furnace, without additional argon protection. This choice was made on the basis of preliminary studies conducted by the authors of this article. It was shown that at a temperature HT of 900 ◦C and a holding time of 4 h, the maximum thickness of the alpha case layer in Ti-6Al-4V, which consists of TiO2 and a Ti (O) solid solution, does not exceed 200 μm (Figure 2a).

At temperatures above 600 ◦C, titanium actively interacts with oxygen, and forms a solid solution of up to 10 at.% in α-Ti. The diffusion rate of oxygen atoms increases with increasing temperature and holding time, but the formation of TiO2 oxide on the surface decreases the diffusion rate. Thus, the effect of oxygen on the properties was insignificant when we used intervals of temperatures and time considered in the article for Ti-6Al-4V heat treatment. The assessment was based on hardness changes (Figure 2b).

**Figure 2.** The formation of an oxide layer in deposited Ti-6Al-4V during heat treatment in an air atmosphere; (**a**) an oxide layer of Ti-6Al-4V after HT at 900 ◦C for 4 h; (**b**) the dependence of the oxide layer hardness of the holding temperature.

Plates with a size of 75 × <sup>15</sup> × 35 mm3 for mechanical tests were deposited in accordance with the thermal cycles corresponding to blade deposition. For the definition of the mechanical characteristics of deposited products, uniaxial tensile tests were performed. Mechanical tests were performed on a universal testing machine, the Zwick/Roell Z250

Allround series (Zwick/Roell, Ulm-Einsingen, Germany). The standard cylindrical samples were cut from the printed parts according to geometry from the ASTM E8, with a gauge diameter of 6.0 mm and a gauge length of 24.0 mm. The sample displacement was recorded using an extensometer. The optical microscope DMI 5000 (Leica, Wetzlar, Germany) with the Axalit software (Axalit, Moscow, Russia) was used for microstructural analysis. The sample surfaces were polished with grit SiC papers up to 2500 grits, with further polishing by an aluminum oxide suspension of 1 μm and final polishing with colloidal silica. As a final step, the etching procedure was conducted. A solution of 93 mL of H2O + 2 mL of HF + 5 mL of HNO3 was applied to the polished sample surfaces for 40 s.

A Bruker Advance D8 diffractometer (Bruker, Billerica, USA) with CuKα radiation (wavelength = 1.5418 Å) was used to perform the XRD analysis. The detector was a LynxEye linear position-sensitive detector (PSD) with a capture angle of 3.2 degrees 2θ. The cross-sectional surface of the samples was polished with a grit SiC paper of 2500 grits, and, after etching, measured in the 2θ range of 30◦–60◦ with a step size of 0.05 and an incremental time of 0.02.

Microhardness measurements were performed with an FM-310 microhardness tester (Future Tech, Kawasaki, Japan). The sample surface for testing was polished with 500 and 2500 grit SiC abrasive papers. Microhardness was measured on the central part of the deposited blades.

### **3. Results and Discussion**

The recrystallization temperature range for the Ti-6Al-4V alloy was 850–950 ◦C. Including this, temperatures of 850, 900, and 950 ◦C were investigated for heat treatment deposited samples. The pre-crystallization annealing modes were also used, since these temperatures may be sufficient to relieve stresses and metastable phase decomposition.

At temperatures above 1000 ◦C, recrystallization and significant growth of alpha plates occurs. This leads to a decrease in the mechanical properties for AM samples; therefore, heat treatment above 1000 ◦C was not considered in this work [23]. The effect of the cooling rate on the microstructure after heat treatment was also not considered, since, in previous works, it was found that the cooling rate during HT does not have a significant effect [24]. Due to the very fine martensite, the kinetics were completely different compared to treatment of equiaxed or heavily deformed microstructures. Consequently, the application of standard heat treatments shows that these treatments do not lead to the usual or expected results [24].

Before 700 ◦C, a layer structure was still observed, which is typical for as-deposited samples. Annealing for 2 h at a temperature of 700 ◦C led to the formation of new equiaxed grain near the boundaries between the layers (suggesting that, in these areas, the greatest deformation of the primary grain occurred). Thus, a shift in the temperature of the onset of the recrystallization process was observed.

This may be due to a high level of initial grain deformation due to high internal stresses (Figure 3a). Formation at the boundaries between the layers of secondary grains was also observed upon a heating temperature of 800 ◦C and a holding time of 2 h (Figure 3b). As the temperature rose to 900 ◦C, a more intense recrystallization process occurred, and the formation of new equiaxed grains occurred not only at the boundaries between the layers, but also inside the layer (Figure 3c,d). From the deformed grains, the growth of secondary under-formed grains was observed. With increasing holding time, the size of the secondary grains increased at 800–900 ◦C. Inside the grain, the morphology of the α-plates changed.

**Figure 3.** Microstructure of deposited Ti-6Al-4V samples after heat treatment (50× magnification): (**a**) 700 ◦C, (**b**) 850 ◦C, (**c**) 900 ◦C, 2 h, (**d**) 950 ◦C, 2 h.

The temperature of 950 ◦C was the boundary of the end of the recrystallization process (for α + β titanium alloys produced by conventional methods [25]). At this temperature, a morphology of the α and β plates inside the grain changed (Figure 3a). In addition, due to the closer temperature of the allotropic transformation, partial recrystallization occurred due to the transition from the low-temperature α to the high-temperature β-phase, and vice versa, upon cooling in the air. The anisotropy of the alloy decreased because of the change in grain size.

In the process of heating to 600 and 650 ◦C and holding for 2 h, only residual stresses were removed, there were no visible structural changes, and the form of the α/α' phase was the same, with a characteristic needle structure. The absence of the α' decomposition was also evidenced by the high hardness of the samples similar to the initial state. Due to the diffusion process, increasing to a temperature to 700 ◦C led to the decomposition of the metastable phase, and a thickness increase of the α phase lamellae began (Figure 3a). The morphology also changed slightly; the needle structure was replaced by a lamellar structure.

A heat treatment temperature of 800 ◦C with a holding time of 2 h led to the secondary α phase, which began to form along the grain boundaries, and the needles transformed into the plates (Figure 3b). There were both long and short plates located perpendicular to each other, as well as in separate packages of parallel plates.

During heat treatment with a temperature of 850 ◦C for 2 h, the secondary α phase was formed, and the metastable α' phase was decomposed with the formation of stable composition a + β phases. The width plates increased, and the amount of secondary α increased as well (Figure 3c). At a temperature of 900 ◦C and a holding time of 2 h, the structural components grew. At an annealing temperature of 950 ◦C, partial recrystallization occurred due to allotropic transformation, and a structure of the "basket weaving" type was formed. The thickness of the plates increased significantly in comparison with the annealing temperature of 850 ◦C.

At low annealing temperatures (below 600 ◦C), the decomposition of a' was incomplete, as is shown by the low value of the hardness.

**Figure 4.** Experimental microhardness measurements for different α-lath thickness values; the tendency line was plotted using polynomial regression analyses (**a**); experimental data of α-lath thickness versus an aging time graph for different temperatures (**b**); typical stress–strain curves of DLD with the Ti-6Al-4V alloy performed with different HT modes (**c**).

### *3.1. Heat Treatment Temperature Effect on Mechanical Properties*

Mechanical test results are presented in Table 2. It can be seen that temperature and time increasing led to YS reduction. On the contrary, an increase in elongation occurred up to a temperature of 900 ◦C; above this temperature, with a holding time of 2 h, a decrease in the relative elongation occurred. H. Galarraga et al. presented the influence of heat treatment by using different modes that consisted of several stages of electron beam-melted Ti-6Al-4V. However, for DLD, it is possible to obtain high mechanical properties using a one-stage heat treatment [12].

**Table 2.** The results of mechanical testing of deposited Ti-6Al-4V cylindrical samples with different HT modes.


The graph of Figure 4a shows the dependence of the plates' α/α' thickness on temperature and holding time during heat treatment. As expected, an increase in the temperature and holding time led to an increase in the structure components size for the Ti-6Al-4V alloy. In addition, since, at a holding temperature of 600 ◦C, the hardness does not change and corresponds to the as-deposition state, the initial phase composition α + α' + β was probably retained. At temperatures of 700–750 ◦C, even when the holding time was 4 h, only a partial transition of α' to equilibrium α + β occurred. Starting from temperatures of 800 ◦C, a more intense decomposition was observed, and above 850 ◦C, the growth of the structural components already had a prevailing effect on microhardness.

Based on the experimental data, the Hall–Petch relationship was plotted for the deposited titanium alloy Ti-6Al-4V using polynomial regression analyses that corresponded to the data presented in [12].

The microhardness gradually decreased with an increase in the α phase plate size (Figure 4a). The microhardness measurement confirmed the decomposition of the metastable α' phase, beginning at the temperature of 700 ◦C. A further decrease in hardness with increasing temperature and holding time was associated with an increase of α lamellae size (Figure 4). The Hall–Petch relationship can also be observed for temperature and time variation in Figure 4b. The results of the α lath thickness measurement for all the aging temperatures and times are plotted in Figure 4b. The graph shows that α lath size grew with temperature and time. The lamellae coarsening increased with temperature. The results for various temperatures correlated with the data obtained for the SLM-ed Ti-6Al-4V alloy [26].

Analysis of the fracture surface showed that an increase in the HT temperature led to a facet size increase, which also corresponded to an increase in ductility and a decrease in tensile strength. Heat treatment at a temperature of 700 ◦C did not significantly affect the change in the structure; the fracture of this sample was more similar to the fracture of the as-deposited samples, also characterized by inter-crystalline fracture (Figure 5a,b,e,f). The use of heat treatment temperatures above 800 ◦C led to a partial or complete decomposition of metastable structures. Above 900 ◦C, the growth of structural components occurred, which also affected the facet size on the fracture surface (Figure 5c,d,g,h).

**Figure 5.** Fracture surface of as-deposited Ti-6Al-4V tensile samples performed in the horizontal direction (*x*-axis) for (**a**,**e**) as-deposited; (**b**,**f**) HT = 700 ◦C; (**c**,**g**) HT = 850 ◦C; and (**d**,**h**) HT = 900 ◦C.

A comparative analysis of the diffraction patterns of the as-deposited sample and after heat treatment using different temperatures allowed the determination of the features of the change in the intensity phases' diffraction lines. As-deposited samples had a slight shift of lines that, as noted previously, indicated the formation of a metastable α' phase due to dissolving more alloying elements in its lattice, which explained the shift [27] (Figure 6). The metastable α' was partially present in the Ti-6Al-4V up to 900 ◦C, and diffraction peaks of the α phase had some shifts. Complete α' phase decomposition was observed above 900 ◦C. As-deposited XRD-patterns also had peak broadening that indicated internal stresses. After heat treatment, the internal stress level decreased, and the peak shape became narrower. This corresponded to the data presented in the work of T. Ungar [28]. In addition, a change of (101)α and (200)α intensity could be traced. Shunyu Liu et al. associated the intensity of alpha planes variation with decomposition of martensite and the preferred grain orientation changing [29]. Moreover, corresponding with optical micrograps, Figure 4 shows that the growth of structural components is observed with increasing HT temperature. The increasing of alpha plates also influenced peak intensity.

The diffraction line intensity of the β-phase increase for deposited Ti-6Al-4V samples after heat treatment. In addition, for all heat treatments, the β-phase diffraction lines were broadened. Based on this, it can be concluded that the β-phase had some inhomogeneity in composition and stress level, which was typical for the phase in which decomposition occurred. The position of the β-phase lines was not constant, and indicate a slightly different degree of decomposition depending on the holding temperature. This was especially pronounced on the sample after heat treatment with a holding temperature of 850 ◦C. This indicated that at a given temperature, the holding time of 2 h was not enough to complete the phase transition processes.

**Figure 6.** XRD patterns of as-deposited Ti-6Al-4V alloy and after heat treatment.

### *3.2. Heat Treatment of Depositede Ti-6Al-4V with Different Structures*

In the deposit samples, depending on thermal cycles, heating and cooling rates, and modes parameters, different types of structures typical for Ti-6Al-4V can be observed. For different types of structures, there can be different ratios of phases: α + β, α + β, and α + α' + β. The size, distance from the substrate, and the thermal cycle, as previously indicated for various AM technologies of Ti-6Al-4V, had a significant effect on the final structure, phase composition, and the distribution of alloying elements. [30,31].

To study the effect of the selected heat treatment mode on three types of initial Ti-6Al-4V structures, samples deposited under different conditions were tested (Table 3). Changes in microstructure and properties depending on conditions have been shown in previous work [32]. It was shown that higher cooling rates were observed at the bottom of the blade near the substrate, which led to the partial or full α' formation. Above the 40th layer, heat

accumulated and an equilibrium structure α + β was formed. In accordance with the above, the selected heat treatment mode was tested on three samples to show good applicability, regardless of the initial structure (Figure 7, Table 3).

Samples with nonuniform structure have different mechanical properties (Table 4). The heat treatment parameters must be selected in such a way that the properties in the initial state do not decrease if they have the required level comparable to the mechanical properties of this alloy in the rolled condition, but, at the same time, so that they can be improved if the required level is not achieved.

**Table 3.** Samples for testing selected heat treatment mode.


**Figure 7.** Microstructure of deposited Ti-6Al-4V with different initial structure (**a**) sample 1 (α + α'), (**b**) sample 2(α + β), and (**c**) sample 3(α + α' + β).

**Table 4.** Mechanical properties of samples deposited with different thermal cycles before and after HT.


The Table 3 shows that heat treatment led to a similar plasticity level for deposited Ti-6Al-4V samples after HT, although, before there was a significant difference in properties, which was most likely associated with the ratio of equilibrium and non-equilibrium phases in the alloy.

### **4. Conclusions**

During a direct laser deposition process from a Ti-6Al-4V powder alloy of aviation parts, which have variable geometric dimensions of the sections due to the thermal cycle's changes, a non-uniform structure can form that decreases the mechanical properties. As a result, the mechanical properties of the part become heterogeneous, which can lead to premature failure during deposition or decrease working properties. In this paper, the optimal heat treatment mode is selected, which, regardless of the structure, gives a good result and evens out the mechanical properties in the part.

(a) The stabilization of the a' phase had a strong influence on its decomposition and grain growth during subsequent heat treatment. Lamellae size showed some relation with heating temperature, illustrating the significance of the initial microstructure and heating temperature applied in post-DLD heat treatment. Above the heating temperature 850 ◦C, α lamella width growth became more intensive (>2.5 μm).

(b) The metastable α' is partially present in the Ti-6Al-4V up to 900 ◦C. Complete α' phase decomposition is observed above 900 ◦C. XRD results show that the β phase also has some inhomogeneity in composition and stress level that is typical for the phase in which decomposition occurs. Full-phase transformation for DLD-ed Ti-6Al-4V alloy when the temperature of heat treatment is above 900 ◦C occurs.

(c) An elongation increase an decrease in the tensile strength occurs with a growth in the holding temperature from 700 ◦C (σ<sup>t</sup> = 1100 MPa, el = 8.27%) to 900 ◦C (σ<sup>t</sup> = 1026 MPa, el = 14.2%). Above 900 ◦C, a decrease in elongation begins with a simultaneous decrease of tensile strength (σ<sup>t</sup> = 990 MPa, el = 13% for 950 ◦C) that is associated with an increase in the α lamellae and beta grain size. The results for all properties are well above ASTM standards for forged (ASTM F1472) and cast Ti6-Al-4V (ASTM F1108).

The best mechanical characteristics of laser-deposited Ti-6Al-4V are ensured by an HT with a temperature of 900 ◦C and a holding time of 2 h. The selected HT parameters allow the homogeneity of properties and microstructure in DLD-ed Ti-6Al-4V parts with variable thickness and complexity, particularly in deposited GTA blades.

**Author Contributions:** Conceptualization, G.T. and O.K.-K.; Methodology, K.B.; formal analysis, M.G.; investigation, M.G. and O.K.-K.; resources, G.T. and L.M.; data curation, G.T., O.K.-K. and G.T.; writing—original draft preparation, M.G.; writing—review and editing, O.K.-K. and L.M.; visualization M.G, supervision, O.K.-K. and K.B.; project administration, G.T.; funding acquisition, G.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research is funded by the Ministry of Science and Higher Education of the Russian Federation as part of the World-class Research Center program: Advanced Digital Technologies (contract No. 075-15-2020-903 dated 16.11.2020).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


*Article*
