Phase Composition, Microstructure and Mechanical Properties of Zr₅₇Cu₁₅Ni₁₀Nb₅ Alloy Obtained by Selective Laser Melting

Abstract

Zr₅₇Cu₁₅Ni₁₀Nb₅ (more known as Vit-106) is a promising zirconium-based alloy with a high glass-forming ability, and belongs to the so-called bulk metallic glasses (BMG). Workpieces with a size of around one centimeter in all three dimensions can be obtained from a BMG alloy by casting. However, further increasing the cast size decreases the cooling rate and thus induces crystallization. Selective laser melting (SLM) is a well-known technique to overcome size limitations for BMGs because a workpiece is built by the addition of multiple melt portions in which the cooling rate is kept above the critical one. Currently, BMG parts obtained by SLM suffer from partial crystallization. The present work studies the influence of SLM process parameters on the partial crystallization of Vit-106 by metallography and the influence of the microstructure on mechanical properties by microhardness and wear resistance testing. Submicron crystalline inclusions are observed in an amorphous matrix of a Vit-106 alloy obtained by SLM. The size and the concentration of the inclusions can be controlled by varying the laser scanning speed. It is shown that submicron crystalline inclusions formed in the amorphous matrix during SLM can favorably affect microhardness and wear resistance.

1. Introduction

Today, amorphous metallic alloys, also known as bulk metallic glasses (BMGs), have attracted much attention, as recent works have reported on the new ways of production, high mechanical properties such as tensile strength, proportional limits, hardnesses, good wear behaviors and corrosion resistance of these materials [1,2,3]. For example, the American company “Liquidmetal” has patented a technology for the manufacture of products from BMGs for medicine and industrial needs [4]. However, the application of BMG alloys is not widespread, especially as structural materials. These materials are very sensitive to cooling rates during casting, which is the main method of their production. Therefore, the size and shape of amorphous metals are limited.

In contrast to casting, selective laser melting (SLM) technology provides high cooling rates combined with the capability of complex-shaped part building, while the dimensions of parts are only limited to the working zone of a machine. Applying SLM for obtaining BMGs has provoked great interest in the past ten years, indicating the promising outlook of this research area [5,6,7,8,9]. Most of the alloys obtained by SLM have a crystalline and anisotropic structure, as shown in the article [10]; in the case of BMG, an amorphous matrix with possible crystalline inclusions is formed. The amorphous structure does not exhibit anisotropy. Due to the high cooling rates characteristic of SLM, reaching 10⁶ K/s, this alloy does not have time to crystallize. Nevertheless, BMG alloys are difficult materials for laser processing, as experimental works have exposed the partial crystallization of obtained parts. However, according to Ouyang et al., the content of the crystalline phase could be controlled by processing parameters [11]. The authors observed the crystalline phase in the heat-affected zone of an iron-based BMG. Similar findings were described for zirconium-based BMGs [12,13,14,15,16]. Further investigation revealed cracking of an iron-based BMGs during SLM, resulting in poor strength, while zirconium-based ones were crack-free [6,9]. Also, their reduced tendency to crack in comparison with BMG alloys of other types provides them with high impact toughness and resistance to fatigue failure. The impact toughness of zirconium OMC alloys [17] is comparable to medium-carbon low-alloy steels [18]. Due to a high vitrification ability and crack resistance during laser processing, zirconium alloys have been found to be the most suitable for SLM.

The amorphous phase is formed by quenching from a liquid state. The polycrystalline material is melted by a moving laser beam, and after moving the beam to another location, it cools rapidly due to the small size of the melt pool surrounded by cold material. During the rapid cooling, the alloy in question solidifies before it has had time to fully crystallize.

Pauly et al. experimentally proved that the SLM method can produce completely amorphous metal alloys with mechanical strengths comparable with similar cast alloys [19]. Common examination of printed Zr-based BMGs to prove their key properties include indentation and nanoindentation [5,11,20,21], compression tests [4,5,22], four-point bending [20,21], impact toughness and fatigue strength tests [21,22] and pin-on-disk wear behavior tests [5,20].

Of the known zirconium-based BMG alloys, the following commercial alloys have become widespread: Vit 1, Vit 106, Vit 106a, and AMZ 4. The composition and basic thermal properties of these alloys are summarized in

Table 1. Composition and basic thermal properties of zirconium-based BMG alloys.

	Zr, %	Ti, %	Cu, %	Be, %	Ni, %	Nb, %	Al, %	Glass Transition Temperature, K	Crystallization Temperature, K	Melting Point, K	Ref.
Vit 1	41.2	13.8	12.5	22.5	0	0	0	625	705	930	[23]
Vit 106	57	0	15.4	0	12.6	5	10	672	738	1092	[24]
Vit 106a	58.5	0	15.6	0	12.8	2.8	10.3	668	772	1110	[25]
AMZ 4	59.3	0	28.8	0	0	1.5	10.4	673	743	1203	[6]

. In this table, the composition of the alloys is listed in atomic percentages.

Based on this table, it was decided that the use of the Vit 1 alloy in this work would be hazardous to health. There are similarities in the composition and properties between the Vit 106, Vit 106a and AMZ 4 alloys, so it was decided to choose a more accessible material for this research. In our case, Vit 106 turned out to be the more accessible material.

Previous works revealed partial crystallization of the BMGs during SLM [26]. Research to date has not yet determined its influence on the properties of amorphous materials. Thus, this work aimed at studying the microstructure of the Vit-106 alloy after SLM and influence of the features observed on the properties of the BMG material.

2. Materials and Methods

2.1. Materials

In total, 5 Vit-106 specimens were studied within this work. Specimen No. 1 was an as-received Vit-106 alloy plate. However, since a defect layer is expected on the surface of a plate after casting in molds, specimen No. 2 was prepared by grinding an as-received plate to a depth of 30 µm. Specimen No. 3 was also a Vit-106 plate subjected to heat treatment in order achieve full crystallization. Previous studies have revealed the SLM process parameters that lead to the partial crystallization of the Vit-106 alloy [26]. These parameters were used for the SLM production of two 5-layered specimens, No. 4 and No. 5, with different laser scanning speeds V, laser powers P, powder layer thicknesses H and hatch distances S (

Table 2. Specimen specification.

No.	Specimen	Description
1	Amorphous plate	As-received
2	Amorphous plate	Ground plates to remove defect surface layers
3	Crystalline plate	Plate annealed at 700 °C for 2 h in vacuum with 10 K/min heating and cooling rate
4	Multilayered SLM	P = 160 W, V = 350 mm/s, S = 185 µm, H = 70 µm, 5 layers
5	Multilayered SLM	P = 160 W, V = 700 mm/s, S = 125 µm, H = 70 µm, 5 layers

).

To obtain crystalline Vit-106 specimens, amorphous plates were annealed in a Termionik T1 vacuum furnace (JSC Termionik, Podolsk, Russia). Powder for the SLM process was manufactured by milling an amorphous Vit-106 chip in distilled water in a planetary ball mill Retsch pm100 (Retsch, Haan, Germany). The amorphous Vit-106 chip was obtained by machining Vit-106 plates. The average size of the powder particles was controlled by optic microscopy every 15 min, because milling time is a key parameter that determines the size of powder particles. A short time results in coarse particles and a broad range of sizes while long milling leads to submicron particles. Therefore, an experimentally optimal time for powder production with an average particle size of 50 µm required for most SLM machines was found (

Table 3. Powder characterization after ball milling.

Ball milling time, min	35	50	75	100
Average particle size, µm	156	103	85	51
Standard deviation, µm	77	46	32	20

). Since it is impossible to obtain a spherical shape by ball milling, irregularly shaped 20–50 µm powder was used in the current work. After milling, the powder was dried and sieved to provide 20–50 µm particle size intervals. Figure 1 demonstrates the prepared powder.

2.2. SLM

A laboratory SLM rig (Figure 2) was used to build multilayered specimens. The rig included industrial components: the ytterbium fiber laser LK-200-B (JSC NTO IRE-Polus, IPG Photonics, Fryazino, Russia), the Raylase SuperScan-II-15 equipment (Raylase, Wessling, Germany), our self-developed powder delivery system and the weldMARK 2.0 software (Raylase, Germany); therefore the quality of the ALAM laser processing was not inferior to commercial SLP machines such as EOS. The process took place in an argon atmosphere to prevent oxidation. Amorphous Vit-106 plates were used as a substrate. For the obtained samples, the scanning speed was the only variable parameter, since it influenced the cooling rate, and consequently the crystallization, more than the other parameters.

2.3. Characterization

The microstructure of the specimens was studied by optic microscopy with an Olympus BX 51 (Olympus Corp., Tokyo, Japan) and scanning electron microscopy with a Phenom ProX (Phenom-World BV, Eindhoven, The Netherlands). Chemical composition of the alloy Zr₅₉Cu_14.3Ni_11.9Al_9.3Nb_5.5 was confirmed by energy-dispersive analysis (EDX) in [26]. In addition, the content of oxygen was measured by the method of the reductive melting of the material in a helium flow using a Leco TC-600 analyzer (LECO Corporation, St. Joseph, MI, USA). The cross sections of the specimens were polished prior to studying. To increase the contrast for the optic microscopy, some cross sections were etched with a mixture of hydrofluoric and nitric acids. X-ray diffraction (XRD) analysis was used to determine the phase composition of the samples. XRD patterns were obtained using a PANalytical Empyrean X-ray diffractometer with CuK_a radiation. Phase composition analysis was performed by PANalytical High Score Plus software (version 4.8), software [27] and ICCD PDF-2 and COD databases [28].

Microhardness was measured by the indentation method with loads of 50 g and 500 g for 10 s using a Tykon 2500 microhardness tester (TUKON, Norwood, MA, USA). Each specimen was measured 9 times on the preliminary polished surface and cross section. The indentation was located 70 µm away from the edge.

Tribological pin-on-disk tests were performed on a Ducom tribometer (DUCOM, Bangalore, India) at a load of 10 N and sliding velocity of 200 min⁻¹ for 900 s. A ceramic 6 mm Al₂O₃ ball was used a counterbody. The depth of the wear tracks was measured on a Hommel Tester 8000 prophilograph (Jenoptik, Jena, Germany). The roughness of each specimen was measured in two perpendicular directions.

3. Results

3.1. Microstructure and Phase Composition

Figure 3 shows the cross sections of the specimens No. 4 and No. 5 at different magnifications. Common SLM microstructure features such as melt pool fragments were observed. Material flows are clearly seen in the SEM images (Figure 4), which are attributed to high viscosity. Similar phenomena were found in silicon glasses subjected to SLM processing [29]. In contrast to specimen No. 5, specimen No. 4 had crystalline phases at the layer–substrate interface (Figure 4). It is possible that a lower scanning speed V provoked the crystallization of the sublayers. In specimen No. 4, coarser inclusions were detected at the boundary of the vortex (Figure 4b,d; Zone I) than inside (Figure 4b,d; Zone II). Meanwhile, specimen No. 4 had areas free from such inclusions. In addition, Zone III, with clear white inclusions around black ones, was distinguished in Figure 4b. Being bright, the inclusions presumably were lighter phases with the deterioration of aluminum. Specimen No. 5 demonstrated no such precipitations as dark inclusions were surrounded by a homogenous grey area. Dark areas observed in the microstructure are presumably either intermetallic Al inclusions or oxide crystals, as the raw material contains 0.058% wt. of oxygen.

Figure 5 shows the XRD patterns of the raw alloy, powder and specimens No. 3–5. The powder and raw plate share a very similar profile, indicating that both materials are in an amorphous state. After annealing, the material became crystalline as the “amorphous halo” was no longer observed. Multilayered specimens were found to be partially crystallized as the “amorphous halo” appeared as a smooth part of the curve, with its highest intensity at 2θ = 39°. Along with those peaks identified as Zr₃O and Al, Zr and Cu intermetallic compounds were revealed. Therefore, the obtained specimens were partially crystalline.

3.2. Mechanical and Tribological Properties

Figure 6 shows the results of the microhardness tests on the surface of the specimens at 50 g and 500 g. Since no significant difference between the results of the specimens No. 1 and No. 3 was observed, the 50 g tests for specimen No. 2 were not performed. The microhardness of the cross sections was measured only for the multilayered specimens made from powder (Figure 7). The highest value of 883 HV was demonstrated by the annealed crystalline plate, while the lowest (530 HV) value was measured for the raw amorphous alloy. The microhardness of the SLM specimens was closer to the amorphous ones. Being obtained at the fastest scanning speed, specimen No. 5 had the closest-to-amorphous state hardness value. This may be attributed to the size of crystalline inclusions that grow at low scanning speeds and, hence, low cooling rates [30]. The hardness results of the fully crystalline specimen No. 3 showed a large mean square deviation, which was possible due to the size of the crystalline phase which can be comparable to the indentation size (Figure 8). The deviation range reduces with the growth of load, which increases the size of the dint and averages the contribution of the main material and crystalline inclusions.

The results of tribological tests and roughness measurements are given in

Table 4. The results of wear tests.

Specimen No.	Max. Wear Track Depth, µm	The Coefficient of Friction	Surface Roughness, Ra, µm	Surface Roughness, Rz, µm
1	33	0.5	0.684	4.351
3	28	0.7	0.272	2.134
4	19	0.52	0.533	2.950
5	17	0.5	0.310	4.237

. Average roughness was neglectable compared to a wear track depth and it did not affect the results. The maximum profile depth was measured, as shown in Figure 9.

4. Discussion

The obtained multilayered SLM specimens No. 4 (Multilayered SLM, P = 160 W, V = 350 mm/s, S = 185 µm, and H = 70 µm; five layers) and No. 5 (Multilayered SLM, P = 160 W, V = 700 mm/s, S = 185 µm, and H = 70 µm; five layers) consisted of an amorphous matrix with crystalline inclusions that spread inhomogeneously and formed a vortex structure. A lower scanning speed V = 350 mm/s resulted in coarser inclusions in the substrate–alloy interface and a vortexless structure (Figure 3b). The vortices in the remelted zone were measured to be around 100 μm (Figure 3b,d), which corresponds to melt pool size under these conditions [23]. It can be suggested that either crystalline inclusions or their nuclei are transported by convective melt flows. The surface of the melt material could serve as the source for the nucleates, given that a small content of oxides is expected due to residual oxygen in the shield gas. Oxidation was earlier reported to provoke crystallization in Zr-based BMGs [31]. Zr₃O and other oxides were found in the specimens studied within this work (Figure 6).

Dark inclusions in the BSE SEM figures are presumably enriched with light elements which may be attributed to the oxides (Figure 4). Meanwhile, there is a bright halo around the inclusions, which suggests the deterioration of the matrix with light elements (Figure 4b,d). Thus, it can be concluded that during solidification, the inclusions grow due to the diffusion of light elements from the matrix to them. The list of such elements includes oxygen dissolved in the matrix and the lighter elements of the alloy, Al, Ni, and Cu, which form a CuZr₂ intermetallic compound in raw material and Al₂Zr₃ и NiZr₂ compounds after annealing (Figure 5). However, the corresponding intermetallic lines of the specimens are not clear in the XRD patterns (Figure 5), but the phase composition for this alloy in its original annealed state was studied earlier [32]. Therefore, the observed inclusions mainly consist of ZrO₃.

The inclusions in the remelted zone are submicron, and the higher the laser scanning speed the smaller the inclusions (Figure 4b,d). The number of inclusions per unit area also decreases. It follows the model for homogeneous nucleation and nuclei growth. Indeed, the cooling rate increases with an accelerating laser scanning speed during solidification, which reduces the time of homogeneous nucleation and inclusion growth. Therefore, the concentration and the size of the inclusions are anticipated to be smaller.

An amorphous plate with CuZr₂ intermetallides was taken for the multilayered specimens 4 and 5 [23]. These inclusions are observed in the nontreated area in Figure 4a,c. Part of the substrate material close to the remelted layers was exposed to the repeated thermal effects of a laser beam, which may have caused the growth of these inclusions. This is evidenced by the increased volume of the CuZr₂ particles (Figure 4b). The size of the typical inclusion reached 20 μm (Figure 10). A higher laser scanning speed increased the cooling rate, reducing the time of thermal exposure in the range where intensive crystal growth was possible. Therefore, a faster scanning speed results in less intense growing of the inclusions.

The coefficient of friction (CoF) of the multilayered specimens remained almost the same as the raw material. At the same time, the wear track depth was significantly less, indicating the positive influence of the submicron crystalline inclusions on wear resistance. Interestingly, that wear resistance slightly improves when increasing the laser scanning speed from 350 mm/s to 700 mm/s. Obviously, smaller inclusions are beneficial for better wear resistance.

The microhardness of the SLM specimens was found to be higher than for the raw material (Figure 7). This is also explained by the strengthening effect of submicron crystalline inclusions. At the same time, with increasing scanning speed, the microhardness decreases slightly. This trend is opposite to the effect of scanning speed on wear resistance and may be due to a decrease in the volume fraction of the crystalline phase with an increasing scanning speed.

The results obtained show that submicron crystalline inclusions formed in an amorphous matrix during SLM can favorably affect mechanical properties such as microhardness and wear resistance. Correlations were revealed between these properties and such structural parameters as the size, concentration and volume fraction of the inclusions. The influence of such an important process parameter as scanning speed on the forming amorphous-crystalline structure was also shown. All these results can find practical application in the manufacture of structural elements from Zr-based BMG using SLM. At the same time, further systematic study of the influence of the oxygen content in the starting material and possible partial oxidation from the gas phase on the forming microstructure, as well as the influence of the microstructure on its properties, is required.

5. Conclusions

• Within this study, multilayered SLM specimens from Vit-106 were obtained, their microstructure was investigated, and their microhardness and tribological behavior were tested.
• The SLM specimens consisted of an amorphous matrix with two types of crystalline inclusions. Submicron CuZr₂ particles were found in the remelted vortex structure, while coarser CuZr₂ of up to 20 µm were observed in the heat-affected zone of the substrate. The nature of the distribution of submicron inclusions shows that they can grow from oxide nuclei transported by convection in the melt pool.
• Large crystals in the thermally affected zone are most likely formed during the growth of intermetallic inclusions initially contained in the substrate material.
• Submicron crystalline inclusions increase wear resistance and can increase microhardness. The role of oxygen in the formation of the amorphous–crystalline structure, as well as the influence of structural parameters on mechanical properties requires further research.