**Heat Treatment of Corrosion Resistant Steel for Water Propellers Fabricated by Direct Laser Deposition**

**Ruslan Mendagaliev 1,2, Olga Klimova-Korsmik 1,2,3, Vladimir Promakhov 3,\*, Nikita Schulz 3, Alexander Zhukov 3, Viktor Klimenko <sup>3</sup> and Andrey Olisov <sup>3</sup>**


Received: 16 May 2020; Accepted: 15 June 2020; Published: 17 June 2020

**Abstract:** The urgency of heat treatment of samples of maraging steel obtained by direct laser deposition from steel powder 06Cr15Ni4CuMo is considered. The structural features and properties of 06Cr15Ni4CuMo steel samples after direct laser deposition and heat treatment are studied. The work is devoted to research into the influence of thermal processing on the formation of structure and the mechanical properties of deposit samples. Features of formation of microstructural components by means of optical microscopy are investigated. Tests for tension and impact toughness are conducted. As a result, it was established that the material obtained by the direct laser deposition method in its initial state significantly exceeds the strength characteristics of heat treatment castings of similar chemical composition, but is inferior to it in terms of impact toughness and relative elongation. The increase in relative elongation and impact toughness up to the level of cast material in the deposit samples is achieved at the subsequent heat treatment, which leads to the formation of the structure of tempered martensite and reduction in its content at two-stage tempering in the structure of the metal. The strength of the material is also reduced to the level of cast metal.

**Keywords:** direct laser deposition (DLD); direct metal deposition; additive manufacturing (AM); corrosion resistant steel; heat treatment (HT); maraging steel; microstructure; mechanical characteristics

### **1. Introduction**

Currently, to increase the competitiveness of shipyards for the manufacture of parts of marine engineering, new high-tech technologies are used. Additive manufacturing methods are increasingly being used, including direct laser deposition technology (DLD). In the DLD process it is possible to obtain parts, including from shipbuilding steels used in the Arctic. Iron and its modified alloys are the most important class of metallic materials used in shipbuilding. Different grades of stainless steels can be treated as a part of traditional manufacturing techniques such as casting, machining, powder metallurgy and welding, including any combination of those. [1,2].

As a result of the research on estimation of material characteristics in 1980, this has been developed and mastered in the industry of martensitic–austenitic stainless steel (CA6NM—06Cr15Ni4CuMo) for manufacturing compressors [3], propeller blade [4–7], castings of blades, components of chemical and oil industry and other cast details of responsible purpose, and now for the manufacture of large castings for a propeller blade of blades and hub steel of 06Cr15Ni4CuMo. Currently, it is relevant to obtain parts and blanks for industrial production of this steel by the DLD method.

With the development of modern production technologies, it has become possible to manufacture parts from virtually any metal powder using additive technologies (AM). AM is characterized by quick fabrication and economical spending of expensive materials. Direct laser deposition (DLD) methods are of special interest when large workpieces need to be made using AM. The DLD technology makes it possible to fabricate large parts from stainless and cold-resistant steels [8–12]. The features of the DLD process include high temperature gradients, and repeated fast heating and fast cooling that cause residual strains and form heterogeneities in the microstructure. Microstructural features, such as grain size and morphology (as well as phase transitions), are very sensitive to the dynamic thermal history and they directly influence the microhardness, tear strength and modulus of resilience [13–18].

The mechanical properties of alloys obtained using DLD depend on their structure-phase states. For steels, an important role is played by the content of laminar low-carbon martensite (α- ) microstructure and the presence of other phases, such as delta ferrite (δ), austenite (γ) and chromium carbides. It is known that the austenite phase is preserved in the tempering process and/or it is restored as a result of the treatment of quenched material. The secondary phase of ferrite in the alloy is δ-ferrite, which forms at very high temperatures during hardening in the course of casting or the DLD process [19].

To achieve high mechanical characteristics for martensite stainless steel, heat treatment (HT) is normally used. However, the microstructure of alloys obtained using DLD is close to a cast structure and is anisotropic. Such materials are characterized by low plasticity and modulus of resilience [20–22]. Here, the material's inner structure is particularly influenced by cyclic heating during DLD due to layer-by-layer metal deposition [23–25]. During the fabrication process, the work piece is tempered. While classical casting and welding envisages a full HT cycle (quenching and tempering), parts fabricated by deposition require the development of ad-hoc HT that is different from the classical one [26,27]. The influence of these effects on the structural characteristics and mechanical properties needs research, specifically the presence of residual austenite γ, non-quenched martensite α- , and delta ferrite δ. These factors govern the final properties of the manufactured parts.

The goal of the research work was to grow a plate of 06Cr15Ni4CuMo steel and further reveal the patterns of microstructure formation and mechanical properties after high tempering and determine the maintenance regime providing the required level of ductility of steel not inferior to casting.

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

### *2.1. Materials*

We have chosen 06Cr15Ni4CuMo (an analog of CA6NM) for the material. The starting material is 06Cr15Ni4CuMo fraction 45–160 μm Figure 1 is the producer of the "Polema" powder. The chemical composition of the steel is provided in Table 1.

The impact bending tests of the 06Cr15Ni4CuMo steel were conducted on an RKP 450 (Zwick/Roell, Ulm, Germany) unit at −10 ◦C, with impact energy 150 J and tensile tests were performed on a Z100 (Zwick/Roell, Ulm, Germany) unit at room temperature. Samples made under mechanical tensile testing GOST 1497-84 (RU) and impact testing GOST 6996-66 (RU).

To show the structure, we used chemical etching with Kalling's reagent in a solution (33 mL HCl + 33 mL of ethanol + 33 mL of H2O + 1.5 g of CuCl2) over 30–60 s.

The deposited cladding layers were visually examined and instrumentally measured; then, they were investigated by optical microscopy on the DMI 500 Leica (Leica Microsystems, Wetzlar, Germany) microscopes using Thixomet (Thixomet, St.-Petersburg, Russia).

**Figure 1.** Surface of powder particles 06Cr15Ni4CuMo.

**Table 1.** The chemical composition of the steel.


### *2.2. Fabrication of Samples Using DLD and Their Heat Treatment*

We used the following equipment: a robotic complex based on an LRM-200iD\_7L (Fanuc, Oshino, Japan) industrial robot; fiber laser based on a LS-3 Yb (IRE Polus Ltd., Fryazino, Moscow Region, Russia) unit; an FLW D30 (IPG Photonics, Oxford, UK) laser deposition head with a detachable SO12 (Fraunhofer IWS, Aachen, Germany) deposition nozzle and a Twin 10C (Sulzer Metco Inc., New York, NY, USA) powder feeder. A shielding gas atmosphere was used for deposition: an air-proof chamber filled with argon at an excess pressure of 2–3 MPa. In the argon-filled chamber the content of oxygen was not more than 300 ppm. Manufacturing of samples by DLD method was carried out at power <sup>P</sup> <sup>=</sup> 2300 W, speed V = 25 mm/s and cross section of depositing bead 2 <sup>×</sup> 0.8 mm2 powder flow rate G = 35 g/min, displacement along the Δx = 1.6 mm, and Δz = 0.7 mm. Five blocks (1 in the initial state and 4 per HT) with LxWxH dimensions 130 × 80 × 16 mm in Figure 2 were deposit simultaneously: a layer was applied alternately to each of the samples, after that the transition to the next layer took place.

**Figure 2.** The scheme of cutting samples for mechanical testing.

DLD was carried out in a shielding chamber with controllable atmosphere. High purity argon was used as a transport and protective gas. Overpressure of 2–3 Mbar was maintained in the deposition chamber. The residual oxygen content in the working atmosphere did not exceed 300 ppm. The porosity in the grown plates did not exceed 2% of the total volume of the deposit sample.

The heat treatment was performed in a SNOL 30/1300 muffle furnace without, shielding gas. The heating rate was 200 ◦C per hour with subsequent exposure and cooling as shown in Figure 3. Cooling rate of samples after high-temperature tempering was 50 ◦C per hour and then air cooling was 150 ◦C.

**Figure 3.** The heat treatment of the samples fabricated by deposition.

### **3. Results**

In the course of parts fabrication from martensite grade steels using DLD, forced process thermal cycling is taking place due to extensive heat deposition. Heat deposition then impacts the presence of retained austenite and δ-ferrite [28] in the structure of the samples fabricated by DLD. Residual austenite can have a negative effect on hardness and toughness.

In the course of tempering during subsequent cooling, metal plasticity deteriorates, and this is due to the formation of secondary martensite as a result of conversion of residual austenite. That is why it makes sense to conduct a second tempering of secondary martensite. This promotes an increase in the metal's relative elongation, creating a finer structure as a result of the decomposition of secondary martensite and the formation of quenched martensite.

Cast metal needs HT for quenching and dual subsequent tempering. During DLD, however, forced thermal cycling is taking place. This strengthens the samples, so we only need to conduct dual tempering to achieve the desired results. Based on the mentioned data from the literature and the results of experimental studies of the steel's characteristics, we have found HT modes for the samples. These modes envisaged high-temperature tempering that would provide the best combination of mechanical properties: high strength, elongation, and impact strength [27].

After DLD the steel has high strength characteristics and low plasticity. To achieve the required mechanical properties for the steel, we have selected several HT modes, as shown in Table 2.

Different δ-morphologies were clearly revealed after etching the fabricated sample, Figure 4a. These morphologies are created by the incomplete growth of Widmanstetten γ-grains during the solid-state δ→γ phase transformation, Figure 4b. Here, incomplete growth results in residual δ stringer inclusions being left at the borders [4]. These inclusions outline the growth front that resembles the initial orientation of the ex- Widmanstett patterns in the final microstructure. During further cooling, γ-austenite converts into α- -martensite, but some amount of δ-delta-ferrite remains in the final microstructure, as shown in Figure 4a. Afterwards, the fabrication δ-delta-ferrite was discovered in the samples, and its content did not exceed 5%.


**Table 2.** Mechanical properties for heat treatment (HT) modes.

(**c**) (**d**)

**Figure 4.** *Cont.*

**Figure 4.** The microstructure of 06Cr15Ni4CuMo (**a**) DLD; (**b**) mode 1 (Т = 750 ◦C, t = 2 h); (**c**) mode 2 (Т = 650 ◦C, t = 2 h); (**d**) mode 3 (Т = 650 ◦C, t = 4 h); (**e**) mode 4 (Т = 620 ◦C, t = 2 h); (**f**) mode 5 (Т = 620 ◦C, t = 2 h/x2).

At such high temperatures, δ-grains are growing rapidly during heating. Then, during the cooling process, they convert into a γ-phase with subsequent transition into a α structure. In Figure 4b, after a single iteration of high-temperature tempering, some amount of γ-austenite (as well as M7C3 chromium carbides) is still observed.

Figure 4c,d shows some amount of non-converted α- -martensite that is an unstable microstructure in the α matrix. It promotes the formation of carbides, interlayer boundaries at large angles and a γ-phase.

Figure 4e,f includes a structure after high-temperature tempering that is represented by quenched lath α- -martensite with fine particles of residual austenite. The particles are located between martensite laths and at the boundaries of martensite batches with large inclusions of δ-ferrite in the matrix basis.

The best mechanical properties were achieved at dual tempering at Т = 620 ◦C, t = 2 h/x2 the cycle is repeated twice Figure 4. This mode complies with the technical specifications for this steel grade while having slightly lower plasticity characteristics.

Figure 5 shows the distribution of microhardness single weld bed and their arrangement for different HT modes and a thermokinetic diagram for Fe-Cr-Ni.

The average microhardness in the fabricated sample tempered at 750 ◦C, Figure 5b, was 355 HV. Tempering at 750 ◦C was chosen because of the phase transition into the γ-austenity area that preceded the dissolution of existing chromium carbides in the fabricated sample (i.e., M23C6/M7C3). Thus, it increased the concentration of carbon in the martensite matrix at room temperature.

With temperature reduction to 650 ◦C it was possible to decrease the hardness to 280 HV with an exposure time of 2 h, and an exposure time of 4 h was required to decrease it to 301 HV. At high-temperature tempering at Т = 620 ◦C with an exposure time of 2 h, the microhardness was 260 HV, and it was 273 HV after the second tempering.

**Figure 5.** (**a**) microhardness; (**b**) the Fe–Cr–Ni equilibrium phase diagram [29].

### **4. Discussion**

It is established that the DLD process is the fastest and most convenient for creating parts. After the process, it is necessary to produce high tempering to achieve all the necessary mechanical properties. Samples obtained by the DLD method are not inferior in characteristics to casting, and in some cases are most in demand.

Thus, we can conclude that the propeller with an optimized structure has reliability characteristics close to the original solid version. We showed the manufacturing of hub and blades via DLD and built-up propeller before and after CNC-machining and manual polishing. After all producing stages, the propeller was weighted. The weight test showed that the mass is finished product is 105 kg. This is 20% less than the original cast design. A more detailed description of design analysis, optimization procedure and production process is presented in [30].

### **5. Conclusions**

The DLD process of 06Cr15Ni4CuMo steel achieves a high strength due to the forced thermal cycling process at low impact toughness and relative elongation. In order to eliminate the imbalance of the complex of mechanical properties, HT is proposed.

According to the results of a comparison of mechanical properties, it was established that the lowest structural matrices are tempered by fine martensite with a low level of residual austenite and δ-ferrite. Based on the analysis of the relationship between HT, mechanical properties and 06Cr15Ni4CuMo steel structure, the most suitable heat treatment mode for deposit samples was established, consisting of a double HT at Т = 620 ◦C, t = 2 h/x2; the cycle is repeated twice, in which a finely dispersed structure of tempered rack martensite is formed, providing a set of properties equal to the material obtained by casting.

**Author Contributions:** Conceptualization, R.M. and V.P.; methodology, V.P., V.K. and O.K.-K.; validation, A.Z. and N.S.; resources, V.K. and V.P.; writing—original draft preparation, R.M., A.O., O.K.-K., and V.P.; writing—review and editing, N.S. and R.M.; visualization, A.Z.; supervision, R.M. and A.O. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by RFBR and Tomsk Region, project number 19-48-703019, was supported by The Tomsk State University competitiveness improvement program (grant No. 8.2.04.2020), grant SP-724.2019.1.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Effect of Elevated Temperatures on the Mechanical Properties of a Direct Laser Deposited Ti-6Al-4V**

**Sergei Ivanov 1,\*, Marina Gushchina 1, Antoni Artinov 2, Maxim Khomutov <sup>3</sup> and Evgenii Zemlyakov <sup>1</sup>**

	- <sup>2</sup> Bundesanstalt für Materialforschung und -Prüfung (BAM), Unter den Eichen 87, 12205 Berlin, Germany; Antoni.Artinov@bam.de

**Abstract:** In the present work, the mechanical properties of the DLD-processed Ti-6Al-4V alloy were obtained by tensile tests performed at different temperatures, ranging from 20 ◦C to 800 ◦C. Thereby, the process conditions were close to the conditions used to produce large-sized structures using the DLD method, resulting in specimens having the same initial martensitic microstructure. According to the obtained stress curves, the yield strength decreases gradually by 40% when the temperature is increased to 500 ◦C. Similar behavior is observed for the tensile strength. However, further heating above 500 ◦C leads to a significant increase in the softening rate. It was found that the DLD-processed Ti-6Al-4V alloy had a Young's modulus with higher thermal stability than conventionally processed alloys. At 500 ◦C, the Young's modulus of the DLD alloy was 46% higher than that of the wrought alloy. The influence of the thermal history on the stress relaxation for the cases where 500 ◦C and 700 ◦C were the maximum temperatures was studied. It was revealed that stress relaxation processes are decisive for the formation of residual stresses at temperatures above 700 ◦C, which is especially important for small-sized parts produced by the DLD method. The coefficient of thermal expansion was investigated up to 1050 ◦C.

**Keywords:** direct laser deposition; Ti-6Al-4V; mechanical properties; microstructure; stress relaxation; elevated temperatures

### **1. Introduction**

Direct Laser Deposition (DLD) is one of the most widely utilized additive manufacturing (AM) technologies for the production of Ti-6Al-4V alloy parts. The uneven local heating of the buildup during DLD leads to significant stresses and distortion that affect the service properties and the shape of the final parts [1–3]. In the last, decade numerous models have been proposed for the simulation of these phenomena [4–6]. Thereby, the temperature dependence of the mechanical properties is a major factor influencing the accuracy of the simulation other than the assumptions of the mathematical model. Nowadays, it is a common practice to use the material properties of wrought Ti-6Al-4V alloys for the simulation of the DLD process. Mukherjee et al. [7] conducted a thermomechanical simulation of a DLD-processed Ti-6Al-4V-alloy by considering the material properties as being temperature dependent in the temperature range between 20 ◦C and 1600 ◦C. In their study, a combination of material properties derived from the fully lamellar wrought alloy [8] and the Ti-6Al-4V metal matrix composite [9] was used rather than the properties of the DLD-processed Ti-6Al-4V alloy. Denlinger and Michaleris [10] found a significant difference between the numerically predicted and experimentally measured distortion when using the mechanical properties of the wrought Ti-6Al-4V. Lu et al. [11]

**Citation:** Ivanov, S.; Gushchina, M.; Artinov, A.; Khomutov, M.; Zemlyakov, E. Effect of Elevated Temperatures on the Mechanical Properties of a Direct Laser Deposited Ti-6Al-4V. *Materials* **2021**, *14*, 6432. https://doi.org/10.3390/ma 14216432

Academic Editor: Amir Mostafaei

Received: 28 September 2021 Accepted: 22 October 2021 Published: 27 October 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/).

revealed significant scattering of the available data on the mechanical properties of the wrought Ti-6Al-4V alloy at elevated temperatures. Furthermore, a sensitivity analysis of the mechanical properties of Ti-6Al-4V showed that the distortion and the residual stresses strongly depend on the thermal expansion coefficient and less on the Young's modulus and the elastic limit. They concluded that for the numerical analysis of the AM process, it is mandatory to use material properties that are specific to the particular manufacturing process. The present study is intended to fill the gap in the lack of data on the temperature dependence of the mechanical properties of DLD-processed Ti-6Al-4V alloys.

The poor mechanical properties of a commercially pure titanium hinder its application as a structural material. However, critical parts that require high strength and ductility, corrosion resistance in aggressive environments, heat resistance, etc., are made of titanium alloys. The Ti-6Al-4V alloy used in this study is a two-phase (α + β) alloy. Recognized as the most popular titanium alloy, Ti-6Al-4V occupies almost a half of the market share of titanium products used in the world today. The proportion of Al and V results in the material having attractive mechanical properties. Ti-6Al-4V contains 6 wt% Al, which stabilizes the α-phase of the hexagonal close-packed structure and 4 wt% V, which stabilizes the β-phase of the body-centered cubic structure. The two phases have different properties due to their structures, with α exhibiting greater strength yet lower ductility and formability [12]. The aluminum in the alloy increases the strength and heat-resistant properties, whereby vanadium increases not only the strength properties but also increases the ductility. It is well-known that two-phase titanium alloys have a lower sensitivity to hydrogen, e.g., hydrogen-induced cold cracking, compared to pseudo-α-alloys. Furthermore, they have good manufacturability and a relatively low tendency to undergo salt corrosion [13–15]. Titanium alloys have a good castability due to the short solidification interval of less than 50–70 ◦C [15,16]. Hereby, the chemical composition of the Ti-6Al-4V alloy utilized in casting does not differ from that of the wrought alloy [17]. The Ti-6Al-4V alloy utilized in additive manufacturing only has a slight difference in the content of carbon impurities from the wrought alloy [18].

It is a well-known fact that the mechanical and service properties of the alloy are determined by the microstructure. The high ductility and cyclic strength correspond to an equiaxed fine grain microstructure. On the other hand, the lamellar microstructure has a high fracture toughness and greater crack propagation resistance. Therefore, it can be said that the bimodal (duplex) microstructure offers an optimal combination of the mechanical properties of the wrought Ti-6Al-4V alloy. The control and optimization of the morphology of the α phase is one of the important issues in terms of the use of the alloy. Thermomechanical processing is a very useful method for improving the microstructure, e.g., controlling the size and the aspect ratio of the α lamellar phase, optimizing the phase ratio of the α to β phases, and controlling the morphology of the β phase [19,20]. The microstructure of the Ti-6Al-4V alloy obtained by direct laser deposition (DLD) depends strongly on the heat input and the inter-pass temperature [1,21], the variation of which results in a wide range of obtained mechanical properties. It is worth noting that the ductility of a DLD-processed alloy can fall to near zero, whereas the strength properties remain comparable to those of the wrought alloy [22].

The effect of the microstructure on the short-term strength of the Ti-6Al-4V alloy at elevated temperatures is similar to its effect on its strength at room temperature. The best combination of ductility, fracture toughness, heat resistance, and endurance is found in alloys with a 70–80% lamellar microstructure [15]. In [23], it was found that alloys with basket-weave microstructures exhibit the most obvious work hardening behavior and the highest strength during hot tensile deformation by temperatures of about 800 ◦C. The best ductility corresponds to alloys with an equiaxed microstructure. According to [24], the alloy with initial equiaxed microstructures also showed the highest ductility during tensile testing in the temperature range of 20–600 ◦C, while the material with initial full martensite microstructure showed better thermal strength. A critical analysis of the literature showed a significant spread in the experimental data of the short-term strength of the wrought alloy at elevated temperatures, as can be seen from Figure 1 [7,10,25–33]. This can be explained by the variation in the test conditions, the initial microstructure of the specimens, and the loading parameters.

**Figure 1.** Temperature dependence of the (**a**) yield stress and (**b**) Young's modulus according to [7,10,25–33].

It should be noted that there is practically no data available on the properties of additively manufactured Ti-6Al-4V alloy at elevated temperatures. In [34], an electron beam melting (EBM)-processed material showed a lower flow stress than the wrought alloy during a compression test in the temperature range of 1000–1200 ◦C. This can be attributed to the larger prior β-grain size and thickness of the α-plates in the EBM-processed alloy. In [35], the flow stress curves of the selective laser melting (SLM)- and direct energy deposition (DED)-processed and wrought alloys were compared in a compression test for temperatures ranging from 850 ◦C to 1100 ◦C. It was found that the presence of a percolating β-phase during the decomposition of martensite seems to be the reason for the reduced flow stress of the additively manufactured material compared to conventional wrought material with a lamellar microstructure. DED and SLM materials show a faster transformation to a globular microstructure compared to conventional wrought material. The temperature dependence of the tensile strength of the SLM-processed Ti-6Al-4V alloy in the temperature range between 20 ◦C and 550 ◦C was studied in [36]. SLM-processed Ti-6Al-4V alloy showed excellent ultimate tensile strength below 500 ◦C, which was 100 MPa higher than a solution-treated and aged Ti-6Al-4V alloy and 300 MPa higher than an annealed Ti-6Al-4V alloy. The effect of the strain rate and temperature on the mechanical properties of the DLD-processed alloy with a Widmanstätten microstructure was studied in [37], using a compression and tension test. However, the study lacks a description of the thermal history during the fabrication of the specimens. The presence of defects such as pores and a lack of fusion had a significant impact on the obtained results. The study presented in [38] compares the mechanical properties of a wrought alloy and a SLMprocessed alloy in a compression test utilizing strain rates of 0.001–1 s−<sup>1</sup> in the temperature range of 20–1000 ◦C. Both publications revealed that the anisotropy of the mechanical properties of the Ti-6Al-4V alloy obtained by various AM methods is insignificant.

The review presented above shows that most of the available material data for additively manufactured Ti-6Al-4V alloy are limited to the study of flow stresses at different strain rates in compression tests above 700 ◦C. These data are important for determining the parameters of the hot forging or stamping processes but are insufficient for the numerical simulation of stresses and distortion induced by the DLD. It should be noted that to the best of the authors knowledge, none of the publications contain data on the temperature dependence of the Young's modulus of the Ti-6Al-4V alloy obtained with AM methods. Moreover, the majority of studies are devoted to the investigation of the material properties of SLM-processed alloys, which differ from those of the DLD-processed alloys.

In the present paper, the mechanical properties of a DLD-processed Ti-6Al-4V alloy were obtained through a tensile test performed for different temperatures ranging from 20 ◦C to 800 ◦C. The conditions used to obtain the test specimens were close to the conditions used in the manufacturing of large-sized structures by the DLD method. The influence of the thermal history on the stress relaxation for the case of 500 ◦C and 700 ◦C maximum temperatures was revealed. In addition, the temperature dependence of the coefficient of thermal expansion was obtained. The influence of the initial microstructure of the samples on the deformation and fractures at elevated temperatures was as well analyzed. An approximation of the measured tensile curves for given temperatures using a proposed fitting function was obtained and used to describe the hardening behavior during plastic deformation.

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

### *2.1. Specimens*

Almost all of data published on the mechanical properties of the DLD-processed Ti-6Al-4V alloy refer to samples obtained without or with very short dwell time between the deposited layers, leading to a significant overheating of the buildup. Therefore, in the present study, the tensile samples were machined from buildups with an inter-pass temperature in the range of 60–80 ◦C, which is typical for large-sized components. Note that the interpass temperature was controlled with type K thermocouples with a diameter of 0.5 mm. The DLD process parameters were as follows: a beam power of 1900 W; a beam diameter of 2.5 mm; a process speed of 20 mm s<sup>−</sup>1; a powder flow rate of 10.5 g min−1; and a gas flow rate of 25 L min−1. The specimens were made using an in-house robotic DLD machine that was developed at the St. Petersburg State Marine Technical University in St. Petersburg, Russia. The machine included a Fanuc 6500 5-axis industrial robot, a rotary table, and a processing head with a discrete coaxial powder feed. To prevent the oxidation of the specimens during the buildup, the sealed chamber of the machine was filled with argon. Hereby, the residual oxygen content in the chamber did not exceed 100 ppm. In total, 160 layers were deposited, having 12 mm in width, 0.8 mm in height, and 140 mm in length, and each layer consisted of seven passes. Spherical Ti-6Al-4V powder with a diameter of 45–90 μm, which was produced by a plasma rotating electrode method, was used for the buildups. The size distribution of the powder particles was unimodal with no visible non-metallic inclusions on the surface, as shown in Figure 2. The chemical composition was in accordance with the standard ASTM F136-02a [39].

**Figure 2.** (**a**) Scanning electron micrograph of Ti-6Al-4V powder and (**b**) size distribution of the powder particles.

### *2.2. Optical and Scanning Electron Microscopy*

Optical metallography of etched microsamples was conducted using a Leica DMI8A microscope with a magnification of up to 1000 times. For the etching, Kroll's reagent (1 mL HF + 2 mL HNO3 + 47 mL H2O) was used [40]. All metallographic cross-sections were taken from the middle of the buildup. The Vickers hardness was measured according to the ISO 6507 standard on an FM-310 hardness tester (Future Tech, Tokyo, Japan) with a load of 3 N. To determine the chemical composition and to analyze the fracture surface of the specimens after testing, a Tescan Mira3 scanning electron microscope (TESCAN, Brno, Czech Republic) with an Oxford AZtec console was used (Oxford Instruments NanoAnalysis, Wycombe, UK).

### *2.3. Tensile Tests at Elevated Temperatures*

The mechanical properties of the DLD-processed Ti-6Al-4V alloy were obtained using a Gleeble 3800 metallurgical simulation system at the National University of Science and Technology MISiS in Moscow, Russia. The setup allows sequential tensile-compression deformation with a force of up to 10 t and simultaneous heating of the sample by direct electric current transmission to be performed. Depending on the specimen configuration and size, the heating and the cooling rate can reach up to 10,000 ◦C s−<sup>1</sup> and 3000 ◦C s−1, respectively. The temperature field was controlled by the contact method using a type K thermocouple with a diameter of 0.25 mm that was fixed to the surface of the sample by discharge spot welding. A schematic of the specimen used for the uniaxial tension tests is shown in Figure 3. Thereby a heating rate of 10 ◦C s−<sup>1</sup> and a strain rate of 3 mm min−<sup>1</sup> were used. An externally mounted sensor was used for the precision recording of the transverse strain. The transverse strain was measured with a 500 Hz sampling rate in the central section of the specimen. The noise in the experimental data, which is shown in Figure 3b, significantly hampered their processing. Therefore, a robust discrete cosine transform (DCT) filter, which was implemented in the commercial software Matlab, was used to process the data [41,42].

**Figure 3.** (**a**) A schematic of the Gleeble 3800 test specimen and (**b**) an example of experimental data processing using the DCT filter.

An approximation of the measured tensile curves for given temperatures was performed to describe hardening behavior during plastic deformation. The following fitting function was proposed:

$$\sigma(\varepsilon) = \frac{p\_1 \cdot \varepsilon^2 + p\_2 \cdot \varepsilon + p\_3}{\varepsilon + p\_3} \cdot \sigma\_{0.2} \tag{1}$$

where *σ* is the stress, *ε* is the strain, *σ*0.2 is the yield strength, and *p*1, *p*2, *p*<sup>3</sup> are the fitting coefficients.

Note that a zero strain in Equation (1) corresponds to a stress level that is equal to the yield strength of the material. The fitting coefficients in Equation (1) were determined by the nonlinear least squares method [43]. Figure 4 shows an example of fitted experimental data. Note that for clarity, only a small part of the recorded experimental points is shown. It can be seen that the proposed fitting function agrees well with the experimental data. However, it should be noted that the fitting of the engineering stress–strain curves was conducted using the data included between the strain corresponding to the yield strength and the strain corresponding to the tensile strength.

**Figure 4.** An example of the measured stress–strain curves and their approximation.

### *2.4. Thermal Expansion Tests*

A DIL 805 A/D quenching dilatometer test machine was used to determine the temperature dependent coefficient of thermal expansion (CTE). A cylindrical specimen with a 4 mm diameter and 10 mm length was inductively heated up to 1050 ◦C at a rate of 3 ◦C s−1. The tests were conducted in vacuum to prevent oxidation. After holding the sample at the maximum temperature for 20 min, the specimen was cooled at a rate of 0.94 ◦C s−<sup>1</sup> by blowing it with helium. An instantaneous α and secant *α* coefficient of thermal expansion were determined, according to Figure 5. Therefore, the following expressions were used:


$$\alpha = \frac{\Delta L\_2 - \Delta L\_1}{L\_0} \cdot \frac{1}{T\_2 - T\_1};\tag{2}$$


$$
\overline{\alpha} = \frac{\Delta L(T)}{L\_o} \cdot \frac{1}{\left(T - T\_o\right)}.\tag{3}
$$

The instantaneous CTE was determined by the first derivative of the experimental thermal strain curve with respect to the temperature. Note that an irregular experimental curve can cause significant high-frequency fluctuations of the calculated derivative values. Thus, to obtain a smooth curve of the instantaneous CTE, an experimental thermal strain curve was approximated by piecewise polynomials of 9th degree.

**Figure 5.** Determination of the coefficients of thermal expansion at point A.

### *2.5. Stress Relaxation Tests*

The visco-plastic behavior of DLD-processed Ti-6Al-4V alloy was measured using a uniaxial tensile test utilized in the Gleeble 3800 machine according to the method described in [44]. Each specimen was first heated to 500 ◦C or 700 ◦C and subsequently tensioned to a specified strain value. All samples were tensioned with a strain rate of 3 mm min−<sup>1</sup> to a total strain of 2%, which is equal to a stress level of about 88% of the yield strength of the alloy at the corresponding temperature. In the next step, the applied strain was held constant, and the stress relaxation was measured as a function of time.

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

### *3.1. Microstructure of the DLD-Processed Ti-6Al-4V Alloy*

The following factors affect the microstructure of the Ti-6Al-4V buildup: (1) a high crystallization rate of the deposited metal due to low interpass temperatures [45,46]; (2) multiple short-term irregular reheating phases from subsequent passes [47,48]; (3) epitaxial crystal growth [49,50]. The microstructure consisted of a lamellar α'-phase, as shown in Figure 6, and a small amount of residual β-phase in the form of thin interlayers [51–53]. The residual β-phase cannot be detected by optical or scanning electron microscopy due to its minor content. The presence of an α -phase leads to an increase in the strength and a decrease in the ductility of the material. The nucleation of the α-phase initiates at the boundaries of the β-grains. The α-plates grow inside the grain until they meet plates growing from other boundaries during further cooling phases. As a result, colonies of unidirectional α-plates are formed in the grain. Cutting the lamellas of different colonies and grains at the different angles in the plane of the microsample results in a visible difference in the α-plate thickness. The thicknesses of such plates are close to each other. The average microhardness of the buildup alloy was 397 HV0.3.

**Figure 6.** Microstructure of a Ti-6Al-4V buildup obtained by the DLD method.

### *3.2. Effect of the Temperature on the Fracture Behavior*

A macrograph of the fracture surface of the specimens tested at different temperatures is shown in Figure 7. All of the specimens had a cup-and-cone ductile fracture. A distinctive feature of the specimens tested at 200 ◦C, as shown Figure 7a, is the presence of two zones: a fibrous zone and a shear zone. The fibrous zone corresponds to the area of slow crack growth. It is located in the center of the fracture. The shear zone is an annular fracture zone that is adjacent to the free surface of the specimen. The extent of the shear zone decreases until it disappears completely as test temperature increases. This is clearly visible in the macrographs of the specimens tested at 500 ◦C and 700 ◦C, shown in Figure 7b,c. The higher test temperature corresponds to a significantly higher reduction of the area and the presence of large pores with a diameter of approximately 350 μm.

**Figure 7.** Fracture surface of specimens tested at (**a**) 200 ◦C, (**b**) 500 ◦C, and (**c**) 700 ◦C.

Deep dimples are clearly visible in the central region of the fracture surfaces of the specimens tested at 200 ◦C. A ductile local fracture is immediately initiated around those dimples, as seen in Figure 8a. The dimples are formed by the coalescence of micropores, which, in turn, grow and expand under a triaxial stress state [54,55]. Flat equiaxed dimples are clearly observed in the shear fracture zone shown in Figure 8b,c. They are formed due to the coalescence of micropores under the action of shear stresses. At higher test temperatures, the fracture is caused by the nucleation of the micropores at the grain boundaries, which are formed by a grain boundary slip, see Figure 9. The subsequent diffusion of vacancies or the development of local sliding leads to an enlargement of the pores. The larger dimples observed in Figure 9a correspond to triple grain boundary junctions. The small dimples seen in Figure 9b originate from the walls of the dislocation cells. At a test temperature of 700 ◦C, the pores are larger and deeper, as seen in Figure 9c. Note that the pores are elongated in the direction of plastic deformation.

**Figure 8.** SEM fractograph of a specimen tested at 200 ◦C showing (**a**) the fibrous central zone and (**b**,**c**) the shear zone.

**Figure 9.** Fracture surface of specimen tested at (**a**,**b**) 500 ◦C and (**c**) 700 ◦C.

### *3.3. Short-Term Mechanical Properties of the Ti-6Al-4V Alloy over a Wide Temperature Range*

The experimentally obtained engineering and true tensile stress curves of the Ti-6Al-4V alloy for the temperature range between 20 ◦C and 800 ◦C are shown in Figure 10. The processing of the experimental data was conducted according to the procedure described in Section 2.3. It can be seen in Figure 10a that the total strain corresponding to the ultimate strength increases when the temperature increases. The ductility of the alloy increases significantly at temperatures above 700 ◦C. It should be noted that the ductility of the DLD-processed alloy at room temperature is comparable to that of the wrought alloy [13].

In Figure 11a the yield and tensile strength are plotted as functions of the temperature. It can be observed that the yield strength decreases gradually by approximately 40% as the temperature rises to 500 ◦C. A further increase of the temperature leads to a significant increase in the softening rate. This behavior is associated with the intensification of the diffusion-controlled decomposition of the metastable α'-phase and the grain boundary slip process [56]. Thus, the yield strength decreases almost linearly from 600 MPa to 70 MPa in the temperature range between 500 ◦C and 800 ◦C. According to the published data shown in Figure 11b, a further increase in the temperature leads to complete softening of the material. The green curve corresponds to the sample with the α' -microstructure [29]. This shows a close correlation to the obtained curve for the DLD-processed alloy, especially at temperatures above 400 ◦C. The discrepancy between the curves in the temperature range of 20–400 ◦C can be explained by the differences in the size and the shape of the prior β-grain as well as by the morphology and the thickness of the α-plates [12]. On the other hand, the blue curve corresponds to the (α + β) microstructure [31] and shows lower yield

strength values. However, its behavior is almost identical to that of the obtained curve for the DLD-processed alloy.

**Figure 10.** (**a**) Engineering and (**b**) true tensile curves of the DLD-processed Ti-6Al-4V alloy.

**Figure 11.** (**a**) Yield and tensile strength of the Ti-6Al-4V alloy as functions of temperature and (**b**) comparison of the obtained yield strength with published data from [26,29,31].

In Figure 12, the temperature-dependent Young's modulus is shown. It is observed that the Young's modulus remains almost unchanged for a temperature increase of up to 500 ◦C. However, it decreases sharply by approximately 70% from 109 GPa to 26 GPa upon further heating to 800 ◦C. A comparison of the obtained curves with previously published data shows a significant discrepancy. The blue curve corresponds to a sample with a bi-modal Widmanstätten microstructure obtained from a plate that was 12 mm thick. Note that the plates were treated by annealing for 6 h at 790 ◦C [31]. The DLD-processed alloy shows a Young's modulus with greater thermal stability. At 500 ◦C the Young's modulus of the alloy is about 46% higher than that of the wrought alloy.

**Figure 12.** Comparison of the measured Young's modulus of the Ti-6Al-4V alloy with published data from [7,31].

An approximation of the obtained tensile curves for given temperatures using the proposed fitting function was performed according to the procedure described in Section 2.3. The obtained coefficients of the fitting function are given in Table 1. These data are of high importance for the numerical analysis of the residual stresses and distortion of additively manufactured parts.


**Table 1.** Mechanical properties and fitting coefficients.

### *3.4. Temperature Dependence of the Thermal Expansion Coefficient*

The experimentally obtained temperature dependence of the thermal strain is shown in Figure 13. It is not difficult to see that the heating and cooling parts of the curve have different slopes for temperatures above 600 ◦C. The temperature dependence of the thermal expansion coefficient was obtained according to the method described in detail in Section 2.4. A decrease of about 20% in the CTE occurs in the temperature range between 400 ◦C and 600 ◦C during heating, as seen in Figure 14a. According to [29], this can be explained by the diffusion-controlled phase transformation α → α + β, which is accompanied by a slight volume decrease. Note, that above 800 ◦C, the α + β → β transformation begins and that the transformation rate is not constant. However, the transformation rate is rather slow in the interval between 800–900 ◦C, which is clearly visible in the secant CTE curve shown in Figure 14b. An increase of the diffusion mobility of the atoms at temperatures above 900 ◦C leads to a sufficient increase in the rate of β-phase formation. During holding at 1050 ◦C, the β-phase content reaches 100%, which leads to a reduction of the sample's volume. However, the coefficient of thermal expansion does not change significantly during cooling.

**Figure 13.** Temperature-dependent thermal strain curve of the DLD-processed Ti-6Al-4V alloy.

**Figure 14.** Temperature dependent coefficient of thermal expansion of (**a**) an instantaneous and (**b**) secant.

### *3.5. Analysis of the Stress Relaxation*

The instability of the phase composition of the material and the relaxation of the residual stresses arising in the parts due to various technological operations may cause spontaneous changes of their size and shape over time, affecting their service properties. The conditions for relaxation are described by the following equation:

$$
\varepsilon\_0 = \varepsilon^\varepsilon + \varepsilon^p = \frac{\sigma}{E} + \varepsilon^p = \text{const at } \varepsilon^\varepsilon \neq \text{const; } \varepsilon^p \neq \text{const}, \tag{4}
$$

where *ε*<sup>0</sup> is the initial total strain, *ε<sup>e</sup>* is the elastic strain, and *ε<sup>p</sup>* is the plastic strain.

The total strain during the stress relaxation test remains constant due to the increase in plastic strain over time caused by the decrease of the fraction of the elastic strain. These processes can have a considerable effect on the shape stability of the part during DLD as well as during service. The experimentally obtained stress relaxation curves for 500 ◦C and 700 ◦C are shown in Figure 15. Note that for clarity, only a small part of the recorded experimental points is shown, as the data recording frequency was 500 Hz. The experimental data were approximated according to the following equation, which describes the intergranular diffusion relaxation processes [57]:

$$
\sigma = \sigma\_o \cdot \exp\left(-\frac{k \cdot t}{1 + p \cdot t}\right),
\tag{5}
$$

where *σ<sup>o</sup>* is the applied stress, and *k* and *p* are coefficients dependent on the temperature, the microstructure, and the phase composition.

**Figure 15.** Stress relaxation curves of the DLD-processed Ti-6Al-4V alloy for (**a**) 700 ◦C and (**b**) 500 ◦C.

To determine the instantaneous creep strain rate, the time derivative of Equation (5) is obtained:

$$
\dot{\varepsilon}\_c = \frac{d\varepsilon\_c}{dt} = -\frac{1}{E} \frac{d\sigma}{dt} \,\tag{6}
$$

where *E* is the Young's modulus at a given temperature.

By differentiating Equation (5) and substituting it into Equation (6) the instantaneous creep strain rate is obtained:

$$\varepsilon\_c^\bullet = \frac{\sigma\_o}{E} \cdot \exp\left(-\frac{k \cdot t}{1 + p \cdot t}\right) \cdot \left[\frac{k}{\left(1 + p \cdot t\right)^2}\right].\tag{7}$$

It is a well-known fact that stress relaxation is a thermally activated process, which is particularly effective at high temperatures. Two regions can be distinguished in the curves shown in Figure 15. The first is characterized by an abrupt stress drop, and the second is characterized by a slow stress drop. As noted in [56], the sharp stress decrease at the beginning of the relaxation process is associated with the elimination of a large number of lattice defects. Over time, the amount of lattice distortions decreases, causing the relaxation rate to become slower. In addition, it can as well be explained by the fact that at the beginning of the relaxation phase, the value of the applied stress is high and thus closer to the yield strength of the individual crystallites and mosaic blocks. In the second region, the relaxation curve is asymptotic to a straight line that is parallel to the abscissa axis and shifts from it by the value of the peak stress at which relaxation will not occur. The kinetics of these processes are well illustrated by the creep rate curves shown in Figure 16. The creep rate at 700 ◦C is 0.12 × <sup>10</sup>−<sup>3</sup> % s−<sup>1</sup> at the beginning of the relaxation process, which is 23 times higher than the creep rate at 500 ◦C, as seen in Figure 15a. A 50% reduction of the stress occurs during the first 60 s at 700 ◦C. After a sharp decrease, the stress continues to decrease, but at a considerably lower rate. In the first 600 s, the stress is reduced by a total of 86% at 700 ◦C and by 25% at 500 ◦C, as seen in Figure 15. Hence, it can be concluded that the overheating of the buildup due to its size and/or absence of inter-pass dwell time will lead to a significant reduction of the residual stresses. Therefore, published data on experimentally measured residual stresses without a detailed

description of the process parameters affecting the temperature field cannot be used to analyze and verify the accuracy of simulation procedures.

**Figure 16.** Creep rate curves of the DLD-processed Ti-6Al-4V for 500 ◦C and 700 ◦C.

### **4. Conclusions**

The mechanical properties of the DLD-processed Ti-6Al-4V alloy were obtained by a tensile test performed in the temperature range between 20 ◦C and 800 ◦C. The influence of the thermal history on the stress relaxation for the cases with maximum temperatures of 500 ◦C and 700 ◦C were studied. In addition, the temperature dependence of the coefficient of thermal expansion was obtained. The influence of the initial microstructure of the samples on the deformation and the fractures at elevated temperatures was analyzed. An approximation of the measured tensile curves for given temperatures using a proposed fitting function was performed to describe the hardening behavior during plastic deformation. The following conclusions are drawn:


**Author Contributions:** Conceptualization, S.I., M.G., and A.A.; methodology, S.I., A.A., M.G., and M.K.; formal analysis, investigation, S.I., M.G., and M.K.; data curation, S.I., A.A., and M.K.; writing original draft preparation, S.I., M.G., and A.A.; writing—review and editing, S.I., M.G., A.A., and M.K.; visualization, S.I.; project administration, E.Z.; funding acquisition, E.Z. 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 November 2020).

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

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

### **References**


### *Article* **Preparation of W-C-Co Composite Micropowder with Spherical Shaped Particles Using Plasma Technologies**

**Andrey Samokhin \*, Nikolay Alekseev, Aleksey Astashov, Aleksey Dorofeev, Andrey Fadeev, Mikhail Sinayskiy and Yulian Kalashnikov**

> Baikov Institute of Metallurgy and Materials Science of the Russian Academy of Sciences, 49, Leninskiy prosp., 119334 Moscow, Russia; nvalexeev@yandex.ru (N.A.); aastashov@imet.ac.ru (A.A.); adorofeev@imet.ac.ru (A.D.); afadeev@imet.ac.ru (A.F.); sinaisky@imet.ac.ru (M.S.); ulian1996@inbox.ru (Y.K.) **\*** Correspondence: asamokhin@imet.ac.ru

> **Abstract:** The possibility of obtaining composite micropowders of the W-C-Co system with a spherical particle shape having a submicron/nanoscale internal structure was experimentally confirmed. In the course of work carried out, W-C-Co system nanopowders with the average particle size of approximately 50 nm were produced by plasma-chemical synthesis. This method resulted in the uniform distribution of W, Co and C among the nanoparticles of the powder in the nanometer scale range. Dense microgranules with an average size of 40 microns were obtained from the nanopowders by spray drying. The spherical micropowders with an average particle size of 20 microns were received as a result of plasma treatment of 25.36 microns microgranule fraction. The spherical particles obtained in the experiments had a predominantly dense microstructure and had no internal cavities. The influence of plasma treatment process parameters on dispersity, phase, and chemical composition of spherical micropowders and powder particles microstructure has been established.

> **Keywords:** tungsten carbides; cobalt; nanopowder; synthesis; granulation; spheroidization; DC thermal plasma

### **1. Introduction**

Hard metals based on tungsten carbide are widely used in the production of cutting tools for metalworking, tools for rock drilling, wear-resistant parts, and coatings, etc. [1–3].

Hard metal products are manufactured by various methods, including powder metallurgy methods, but the production of complex-shaped parts encounters significant difficulties. Nowadays, intensively developing additive technologies make it possible to overcome the problems of manufacturing parts with complex shapes, so the attention of researchers and developers is drawn to the development of additive technologies for the manufacture of hard metal compacts. An overview of current research in this area can be found in [4].

Metal powders used in the layer-by-layer growth of products by methods of additive technologies should have good flowability and provide the highest possible packing density of particles when creating a powder layer [5,6]. These requirements may be achieved by using powders with a spherical particle shape having a size in the range from 20 to 60 μm.

An effective method for producing powders with a spherical particle shape is the processing of powders consisting of particles with an irregular shape in a flow of thermal plasma of electric discharges where the particles melt and become spherical due to surface tension forces. Plasma processes of spheroidization, the studies of which were started in the second half of the last century [7,8], are now widely used for processing metal powders [9–14]. However, the production of powders of WC-Co hard metals with a spherical particle shape in plasma processes has been studied only in a very limited number of works.

**Citation:** Samokhin, A.; Alekseev, N.; Astashov, A.; Dorofeev, A.; Fadeev, A.; Sinayskiy, M.; Kalashnikov, Y. Preparation of W-C-Co Composite Micropowder with Spherical Shaped Particles Using Plasma Technologies. *Materials* **2021**, *14*, 4258. https:// doi.org/10.3390/ma14154258

Academic Editor: Federico Mazzucato

Received: 19 June 2021 Accepted: 27 July 2021 Published: 30 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/).

The authors of [15] obtained spherical powders by processing composite microgranules WC-12 wt.% Co in a thermal plasma flow generated by an electric arc plasma torch. The initial microgranules had an average size of 38 μm and were obtained by spray drying a suspension of WC and Co powders. Plasma treatment of microgranules made it possible to reduce the porosity in composite microparticles and ensure their spheroidization; however, high-temperature treatment led to undesirable effects—the transformation of some parts of WC carbide into W2C and the partial evaporation of cobalt. The size of WC grains in the obtained spherical particles was in the micron range.

The authors of the paper [16] showed the possibility of obtaining dense spherical microparticles of the hard alloy WC-Co as a result of processing the corresponding porous microgranules in a thermal plasma flow followed by a heat treatment. Initial WC-30 wt.% Co microgranules with an average size of 87 μm were obtained by spray drying a suspension of a mixture of initial WC and Co powders. As a result of the processing of porous microgranules, dense spherical particles with a changed phase composition, represented by the phases C, Co, W2C, and Co3W3C, were obtained. Subsequent heat treatment of this powder at a temperature of 950 ◦C provided a return to the original phase composition WC-Co, while the particles retained their spherical shape, and the powder had the fluidity required for its use in additive technologies.

In the above-mentioned works [15,16], the obtained spherical WC-Co particles had a microstructure with micron-sized WC grains.

The particle size of tungsten carbide powder is one of the critical factors determining the mechanical properties of WC-Co hard metals and the transition to nanostructured hard metals is a means to significantly improve these properties [17,18].

The aim of this research was to experimentally determine the possibility of obtaining composite W-C-Co system micropowders with a spherical particle shape having a submicron/nanoscale internal structure, using an approach involving three successive stages:


In some cases, the final classification of dense spherical particles obtained after plasma treatment is required with the purpose of removing nano- and submicron particles formed during condensation of vaporized material.

The proposed approach can make it possible to obtain micropowders of a WC-Co hard metal with a spherical particle shape and a submicron microstructure. To date, such powders have not been produced, but they are of interest for the manufacture of hard metal parts of complex shapes with an ultrafine microstructure using modern methods of additive technologies.

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

### *2.1. Obtaining Nanopowder of W-C-Co System*

W-C-Co system nanopowder was obtained by processing of mixture of tungsten oxide WO3 and cobalt Co powders with methane in thermal plasma flow generated in electric arc plasma torch with self-adjusting arc length with nominal power of 30 kW. Detailed description of the setup and processes of nanoparticles formation in the thermal plasma flow are presented in [9].

Powder mixture consisted of tungsten oxide WO3 particles with average particle size less than 50 μm and Co particles with particle size less than 5 μm. All experiments were performed at a constant mass ratio of elements W/Co = 11 in the raw mix.

Nitrogen, as a part of plasma-forming and transport gases, was supplied from the air separation unit, the oxygen content in nitrogen was no more than 0.01 vol.%. Hydrogen, with purity not less than 99.95 vol.%, was added to plasma-forming gas to ensure reduction of tungsten oxide. Compressed methane, with a purity of 99.9 vol.%, was used as a carbidizer.

### *2.2. W-C-Co System Nanopowder Microgranules Production*

Nanopowder microgranules of W-C-Co system were obtained by spray drying of aqueous suspensions of W-C-Co nanopowder on Buchi Mini Spray Dryer B-290 (Flawil, Switzerland) equipped with an ultrasonic nozzle.

The production of microgranules of the W-C-Co system based on spray drying includes three main stages:


### *2.3. Plasma Processing of W-C-Co Nanopowders*

Separated fraction of nanopowder granules was treated in a thermal argon-hydrogen (3 vol.%) plasma jet, generated in an electric arc plasma torch with a rated power of 30 kW. A detailed description of the plasma setup for powders spheroidization used is given in [10]. The formation of nano- and submicron particles formed due to partial evaporation and subsequent condensation of a raw material. Destruction of some granules by collision and removal of thermal gasification products of an organic binder from granules enhances the evaporation processes during plasma treatment of nanopowder microgranules. Sedimentation in a distilled water after ultrasonic dispersion was used to remove nano- and submicron particles from plasma treated powder. In this process, spheroidized microparticles are separated from the resulting suspension by sedimentation and subsequent drying in vacuum, while nano- and submicron particles are removed with water as a suspension.

A comprehensive analysis of the physicochemical properties of powders included:


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

### *3.1. Preparation of W-C-Co System Nanopowders*

Plasma-chemical synthesis of W-C-Co system nanopowders was carried out in the plasma reactor with the confined jet flow. The power of the plasma torch was in the range of 19.24 kW. A mixture of nitrogen and hydrogen (20 vol.%) was used as plasmaforming gas. Dispersed raw material was transported with a feeding rate of 5 g/min by a hydrogen–methane mixture.

According to the results of electron microscopy (Figure 1), obtained nanopowders represent a polydisperse system consisting of aggregated nanoparticles in the size range from 10 to 100 nm, with the nanoparticles predominantly close to spherical in shape.

(**a**) (**b**)

**Figure 1.** SEM (**a**) иTEM (**b**) images of W-C-Co nanopowder.

The specific surface area of nanopowders obtained ranged from 15 to 20 m2/g. The main process parameters affecting the values of the specific surface area are the feed rate of dispersed raw materials and the methane flow rate. With increasing feed rate of dispersed raw materials, the specific surface area decreases due to a boost in the concentration of condensed vapors, and an increase in methane flow rate leads to a boost in the specific surface area of the nanopowder due to an increase in the content of free carbon with a high specific surface area.

Phase composition of the obtained W-C-Co system nanopowder containing 4.7 wt.% of carbon is characterized by the predominance of the W2C tungsten carbide phase with the presence of WC and W phases (Figure 2a).

**Figure 2.** XRD patterns of W-C-Co (**a**) and W-C (**b**) nanopowders produced by plasma synthesis.

It is noted that during the plasma-chemical synthesis of nanopowders of the W-C and W-C-Co systems, the phase composition of the resulting nanopowders is noticeably different (Figure 2). Introduction of a cobalt into the process leads to a significant decrease in a cubic tungsten carbide phase WC1-x and an increase in hexagonal tungsten monocarbide phase WC content as well as an increase in W2C phase.

The results of the EDS microanalysis indicate that the elements W, Co, and C are evenly distributed among the nanoparticles of the powder of the W-C-Co composition with a uniformity scale in the nanometer size range (Figure 3). The high level of uniformity in the elemental composition of the obtained nanopowder is determined by the mechanism of its formation based on the co-condensation of components from the gas phase. Therefore, when carrying out this process, special attention was paid to creating structural and technological conditions to ensure the most complete evaporation of the initial disperse raw materials and the implementation of the targeted chemical reactions. The yield of the W-C-Co composition nanopowder reached 98.0–99.5 wt.%. The content of micron-sized particles in the nanopowders did not exceed 0.5–1.5 wt.%. We attribute their content mainly to incomplete evaporation of the raw material particles. The oxygen content in this fraction was 3–5 wt.%.

For the following stages of nanopowder granulation and microgranule spheroidization, a batch of W-C-Co system nanopowder containing 7.7 wt.% cobalt and 4.7 wt.% carbon was produced. Total oxygen content of the powder was 0.5 wt.%. The carbon content in the nanopowder was reduced in relation to the value corresponding to tungsten monocarbide WC since an organic binder was used for making microgranules which pyrolysis during plasma treatment leading to the formation of carbon in the microgranule volume.

**Figure 3.** Results of the EDS microanalysis with element distribution maps (W, C, Co) among particles of W-C-Co composite nanopowder.

### *3.2. Preparation of W-C-Co microgranules*

The granulation process of the obtained nanopowders was carried out in a spray drying unit at drying gas temperatures of 40–150 ◦C with a flow rate of 20 m3/h. The nozzle cooling gas flow rate was 0.3 m3/h. Flow rate of the suspension was in the range from 3 to 12 g/min. The power of the ultrasonic nozzle in all cases was 3 W.

The microgranules obtained in the spray drying process were subjected to sieving with the extraction of fractions of 25–63 microns, the yield of which was about 50%. Microgranules have mostly irregular shapes (Figure 4), determined by drying conditions, as well as the use of a low-boiling liquid-ethanol—as a dispersion medium. When distilled water and watersoluble organic binder were used as a dispersion medium we obtained spherical granules, but their strength was insufficient, and the yield of the target fraction was low. This is largely due to the fact that nanopowders of the W-C-Co system have a pyrocarbon layer on the surface that is formed as a result of hydrocarbons pyrolysis [22] and, therefore, the water suspensions of the W-C-Co nanopowder system have a poor stability.

The dispersed composition of separated granule fraction was characterized by the values of distribution parameters D10 = 23 μm, D50 = 37 μm, and D90 = 60 μm (Figure 5). The average particle size of microgranules was 39 μm.

**Figure 4.** SEM image of W-C-Co microgranules obtained in the process of spray drying.

**Figure 5.** Differential and integral microgranules size.

According to the analysis results, the carbon content in the microgranules was 6.4 wt.%, oxygen-1.0 wt.%, the apparent density of the microgranules was 2.6 g/cm3, and the flowability was 29 s/50 g. An increase in carbon content by 1.7 wt.% and oxygen content by 0.5 wt.% was due to the use of an organic binder. An additional channel for increasing oxygen in the granules was the interaction of nanoparticles with active oxygen-containing components formed during ultrasonic dispersion of an alcohol-based suspension at the stage of preparing a suspension for spray drying.

Selected granulation mode in combination with the use of an alcohol-based dispersion medium and polyvinyl butyral made it possible to ensure the strength of microgranules sufficient for their transportation without destruction from the powder feeder into the thermal plasma flow by the carrier gas.

### *3.3. Plasma Treatment of W-C-Co System Microgranules*

Fraction of nanopowders with a size range of 25 to 63 microns containing 6.4 wt.% carbon was treated in a thermal plasma jet with a mixture of Ar + 5 vol.% H2. Use of hydrogen in the plasma-forming gas allowed to increase the intensity of heating of granules due to its high thermal conductivity and, as a result, the degree of spheroidization of the obtained particles. At a plasma-forming gas flow rate of 2.0 m3/h, the power input of the plasma torch in the conducted experiments was 20–30 kW. The enthalpy of the plasma jet varied in the range from 2.4 to 4.9 kW·h/m3. Microgranules with a flow rate of 6 g/min were transported from the feeder to the plasma reactor with argon at a flow rate of 0.5 m3/h.

In all experiments performed, within the specified range of process parameters variation, the degree of spheroidization of microgranules was at least 90% (Figure 6a). Apparent density of obtained spheroidized powder (after removal of nano- and submicron particles) changed from 8.8 to 9.6 g/cm<sup>3</sup> with an increasing enthalpy of the plasma jet (Table 1).

**Figure 6.** SEM images of W-C-Co micropowder after the spheroidization and removal of nano- and submicron particles (**a**) and metallographic cross-section of this powder (**b**).



The experiments performed showed that an increase in the enthalpy of the plasma jet in the range from 2.4 to 4.9 kW h/m<sup>3</sup> leads to nanoparticles content increasing in the treated powder from 8 to 13 wt.%. The formation of nanoparticles occurs due to the evaporation and subsequent condensation of microgranule components. The elemental composition of the nanoparticles is characterized by a cobalt content of 70 wt.% and a tungsten content of 30 wt.%. The predominant content of cobalt in nanoparticles relates to a lower boiling point of this metal and, as a consequence, its intensive evaporation from the surface of spheroidized granules in plasma. The presence of tungsten in nanoparticles may be caused by the removal of tungsten carbide nanoparticles from the surface of microparticles by the cobalt vapor during evaporation, as well as by the destruction of nanopowder microgranules during thermal decomposition of the organic binder with an active gas release.

The particle size distribution in the spheroidized powder after removal of nano- and submicron particles formed is shown in Figure 7. The disperse composition of the spheroidized powder is characterized by values D10 = 8 microns, D50 = 15 microns, and D90 = 28 microns. Compared with initial microgranules, the spherical particles are smaller (Figures 5 and 7), which may be related not only to the acquisition of a more compact spherical shape of particles but also to the destruction of microgranules in the plasma flow as a result of gas release at thermal decomposition of the organic binding used at granulation.

**Figure 7.** Differential and integral particle size distribution of W-C-Co micropowder after treatment in plasma and removal of nano- and submicron particles.

Thermal plasma jets generated by electric arc plasma torches are characterized by significant enthalpy and gas velocity gradients. In the plasma jet processing of polydisperse powders, the conditions of thermal interaction of particles of the processed material with the high-temperature gas are different. This fact predetermines possible differences in directions and rates of phase transformations occurring in individual particles, so that, under those conditions, particles in polydisperse powders may have different internal microstructures.

The change in phase composition in the process of nanopowder transformation into spherical microparticles is presented in Figure 8. The main phase in both nanopowder and spherical microparticles remains the tungsten carbide phase W2C. When processing granules in a plasma flow the carbide phase WC(1−x) formation occurs and its content considerably increases with an increase in plasma stream enthalpy: by the results of XRD relative intensity of 100% reflections increases from 0.3 to 0.85. The content of tungsten monocarbide WC, obviously, also increases, as the ratio of relative intensity of 100% reflections WC and W2C increases from 0.28 to 0.43. Increasing the enthalpy of the plasma jet during microgranules processing leads to an increase in the average mass temperature of the gas-dispersed stream and the microparticles present in it, which, in turn, may contribute to a rise in the content of the higher-temperature carbide phase WC(1−x).

In addition to the plasma enthalpy, there are several other factors that should significantly influence the phase composition of the W-C-Co spheroidized powder. These include the total carbon content of the nanopowder, the value of the specific surface area of the nanopowder, and the residence time of nanopowder granules in the high temperature region during plasma spheroidization. Within the framework of this work, the study of the influence of these factors was not carried out since the main task was to demonstrate the fundamental possibility of implementing the process of obtaining dense spherical microparticles of a composition based on tungsten carbides and cobalt with a submicron microstructure. A separate study is also required to determine the change in the phase composition and the structure of the resulting material in the L-PBF process using spheroidized micropowder of the W-C-Co system.

Spherical particles obtained in the experiments had predominantly dense microstructure and had no internal cavities, although some particles had small pores (Figures 6b and 9). Grain size of the particles in most cases was in the submicron range.

**Figure 8.** XRD patterns of initial W-C-Co nanopowder (**a**) and W-C-Co powder spheroidized at different plasma enthalpy: (**b**) 2.8 kW·h/m3, (**c**) 4.8 kW·h/m3.

**Figure 9.** SEM images of W-C-Co spheroidized micropowder cross-section.

According to the EDS results of individual micropowder particles cross-sections with the most characteristic morphology revealed that the cobalt in them is uniformly distributed at submicron level in accordance with the structure of the particle (Figure 10).

As a result of elemental microanalysis of a cross-sectional area on ground spheroidal particles of W-C-Co composition (Figure 11), it is found that the amount of cobalt in particles of different size and structure varies in a wide range from 2.2 to 7.2 wt.% and averages about 5 wt.% for W+Co calculation (without carbon), which is approximately 4.7 wt.% for the W-Co system.

Intense evaporation of cobalt in the process of plasma spheroidization of granules led to a noticeable decrease in its total content in the obtained spherical powder. In initial microgranules the cobalt content (when analyzed by the XRF method) was 7.7 wt.%, and, after processing in a thermal plasma stream with enthalpy of 2.4 kW·h/m3, the concentration of cobalt decreased to 4.6 wt.%. The treatment of granules in plasma at an enthalpy of 4.9 kW·h/m3 reduced cobalt content in powder to 3.7 wt.% (Table 2).

**Figure 10.** SEM image of W-C-Co spheroidized micropowder cross-section and map of Co distribution in this cross-section.

**Figure 11.** SEM image of W-C-Co micropowder cross-section and the results of EDS elemental microanalysis showing the Co and W content in the powder particles with different structures.



When processing microgranules in a plasma flow, the carbon content in them also decreases. If the initial carbon content in granules before plasma treatment was 6.4 wt.%, after treatment in the plasma flow with enthalpy 2.4 kW·h/m3, the carbon content decreased to 4.7 wt.% and for enthalpy 4.9 kW·h/m3 to 3.9 wt.% (Table 2). Carbon in microgranules is present in the form of tungsten carbide phases, in a free state, and in PVB, which was used as a binder in the production of microgranules. The synthesized nanopowder had a carbon content of 4.7 wt.%, which suggests that when the microgranules were treated in a plasma flow, the carbon carryover was mainly due to the formation of gaseous carbon compounds during pyrolysis of PVB. Thus, the carbon introduced into the microgranules by the binder should not be accounted for in the carbon balance, which is involved in the chemical transformation of tungsten carbide phases during treatment/spheroidization of nanopowder microgranules in a plasma stream.

The oxygen content in the spheroidized powder was at the level of 0.03–0.05 wt.% and was determined by the intensity of initial granules interaction with plasma flow and its enthalpy level. The radical decrease in oxygen in powder from 1.0 wt.% in granules to 0.03 wt.% in spheroidized powder is determined first of all by the reductive chemical reactions of hydrogen with oxygen-containing granules components. The contribution of reactions in which the carbon of the W-C-Co composition and the carbon-containing products of thermal decomposition of the organic binder cannot be excluded.

### **4. Conclusions**

The performed set of experimental studies has shown the principal possibility of obtaining dense spherical microparticles based on the composition of tungsten carbides and cobalt having a submicron structure by the consecutive use of plasma-chemical synthesis of a W-C-Co system nanopowder, spray drying of suspension based on a W-C-Co system nanopowder with obtaining nanopowder microgranules and spheroidization of microgranules in a thermal plasma flow.

It is found that the treatment of nanopowder microgranules in a thermal plasma stream leads to a change in their chemical composition: a reduction of carbon, oxygen, and cobalt, and the formation of nano- and submicron particles. To eliminate these negative effects, it is necessary to carry out further research aimed at investigating the development of methods to control the structure and chemical and phase composition of the obtained spherical microparticles of the tungsten carbide-cobalt system at the various stages in the process of their production in order to optimize their properties for the use of the micropowders in the manufacture of hard metal products by methods of additive technologies.

The study was supported by a grant from the Russian Science Foundation (project No. 19–73-00275).

**Author Contributions:** Conceptualization, A.S.; Data curation, A.A. and A.D.; Formal analysis, A.S., N.A. and M.S.; Investigation, A.A., A.F., M.S. and Y.K.; Methodology, A.A., A.F. and Y.K.; Resources, A.A., A.F. and Y.K.; Visualization, A.D.; Writing—original draft, N.A. and A.A.; Writing—review & editing, A.S., N.A and A.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Russian Science Foundation, grant number 19-73-00275.

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

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

### **References**


*Article*
