Influence of Printing Strategies on the Microstructure and Mechanical Properties of Additively Manufactured Alloy 625 Using Directed Energy Deposition (DED-LB-p)

Abstract

Directed energy deposition (DED-LB-p) is used for the production of large components due to the high deposition rates. The large number of process parameters and printing strategies makes it difficult to optimize this process to achieve the optimal properties. Intensive post-processing is still the main obstacle to the widespread use of this process. In this work, the influence of different printing strategies and process parameters on the microstructural and tensile mechanical performance at room temperature is investigated. The porosity is measured in both printing directions. The grain orientation and size are analyzed by EBSD. A very low porosity of less than 0.4% is found in all the printed samples. The samples printed with the optimized offset printing strategy show a significant improvement in tensile strength of 1000 MPa without heat treatment compared to the other processing routes.

1. Introduction

Additive manufacturing (AM) processes enable the production of complex geometries and structures that are only possible to a limited extent with conventional manufacturing processes [1]. Powder bed fusion (PBF) is the most common industrial process used to produce dense metal parts for a variety of applications, e.g., individual medical implants, lightweight structures or parts for the aerospace industry [2,3,4]. The high costs and slow printing rates are the main limitations. Therefore, further research has been conducted to explore the advantages of directed energy deposition (DED), which enables higher printing rates. In directed energy deposition, either a powder or a wire can be applied as a starting material to a substrate onto which an energy source such as a laser beam, an electron beam or an electric arc is simultaneously focused, forming a molten pool and depositing the material continuously, layer by layer [5].

In addition to the high deposition rates, the benefits of DED include the possibility of printing multi-materials, composites and functionally graded materials or the simple integration with other manufacturing processes for repairing or coating various components [6]. Compared to laser PBF, however, a lower dimensional resolution, higher residual stresses and intensive post-processing are required, which are still the main obstacles in the DED process [5,7]. Figure 1 shows the basic principle of directed energy deposition using a laser beam as the heat source and powder as the feedstock material (DED-LB-p).

Alloy 625 is a face-centered cubic nickel-based superalloy that is strengthened by the solid-hardening effect of refractory metals such as niobium and molybdenum in a nickel–chromium matrix [8,9]. The precipitation of the fine metastable phase γ″ (Ni3Nb) after annealing in the temperature range of 550–850 °C is the main reason for the high mechanical properties [10]. In addition, various carbides, such as MC, M6C and M23C6, can also precipitate as a function of time and temperature. The intermetallic Laves- and δ-phase are also reported [11]. Because of its combination of excellent properties, such as its high temperature strength, corrosion resistance high ductility and stress corrosion cracking resistance, Alloy 625 is used for a diverse range of applications [12]. These include turbines, furnace hardware and components exposed to seawater [13].

Recently, the field of additive manufacturing has gained a lot of interest and multiple efforts are currently taking place to enhance understanding of the underlying mechanisms. The application of DED-based additive manufacturing in repair was investigated by Saboori et al. [14]. The effect of the deposition strategy and post-processing on the microstructure and mechanical properties of serviced IN625 parts repaired using DED was investigated by Chaurasia et al. [15]. Hua et al. [16] repaired 42CrMo4 substrates using IN718 powder. The microstructure and hardness comparison of as-built IN625 alloy following various additive manufacturing processes was reported by Gamon et al. [17]. Cho et al. studied the effects of processing parameters for DED manufacturing technologies [18]. Analytical and numerical modeling was conducted to improve the efficiency and to understand the effects of the processing parameters and the resulting thermal gradients and clad geometries [19,20,21,22]. Caiazzo et al. [20,21] developed a 2D numerical modeling approach for the DED of metal deposition and laser welding, predicting the melt pool temperatures of the deposition of overlapping clad tracks. Models have been developed by Ren et al. [22] in which a thermos-mechanical modeling approach was combined with thermal imaging experiments to optimize the laser scanning patterns of a single layer.

In this work, samples of Alloy 625 were printed by directed energy deposition (DED-LB-p). Single tracks were deposited with different process parameters to optimize the porosity and deposition rate. Adjacent tracks were deposited in the following experiments in which the influence of the line spacing on the porosity was analysed. A parameter study in which the porosity, track height and build rate were measured was conducted. Trials of the deposition of larger volumes used for manufacturing tensile specimens were based on these experiments. The microstructural evolution as well as the mechanical tensile properties at room temperature under varying printing strategies were investigated.

2. Materials and Methods

Gas-atomized powders of Alloy625 were supplied by Inopowders (Paris, France).

Table 1. Chemical composition of Alloy625 powder, Inopowders; wt.%.

Alloy	Cr	Mo	Nb	Fe	Ti	Si	Cu
625	22.0	9.0	3.7	1.1	0.03	0.14	0.06
Alloy	Al	Co	C	P	O	N	Ni
625	0.1	0.007	0.008	0.005	0.016	0.15	Bal. (63.7)

shows the chemical compositions. The analysis of the particle size distribution was conducted using a dynamic image analysis instrument (Camsizer_XT, Microtrac RETSCH, Haan, Germany) and showed that 10% of the powder measured less than 37.0 µm in diameter, 50% less than 59.3 µm in diameter, and 90% less than 93.9 µm [23].

The flow rate was determined by the Hall flow method to be 16.1 s/50 g. Figure 2a,b present the morphology of the powders used for the experiments. The micrographs were obtained using a scanning electron microscope (SEM, Zeiss 1540 XB, Zeiss, Oberkochen, Germany). It can be observed that the majority of the particles have a very spherical shape with small satellites attached to the surface.

The DED system used for the experiments consisted of a disk laser (TruDisk 2000, Trumpf, Ditzingen, Germany) emitting at a wavelength of 1030 nm and a maximum power of 2000 W, a coaxial powder feed nozzle with three nozzles, a six-axis robot (Kuka KR30, Kuka, Augsburg, Germany) and a two-axis positioner (DKP 400, Kuka) on which the substrate was mounted. Figure 3a–c show the setup used for the experiments.

Sandblasted rolled 304L steel plates with dimensions of 6 × 60 × 100 mm³ were used as the substrate. Prior to deposition, the substrates were degreased and cleaned using acetone and ethanol. A pneumatic powder feeder with four independent lines (GTV4, Trumpf, Ditzingen, Germany) was used to transport the powder to the powder nozzle. Argon gas with >99.999% purity (with O₂ < 2 ppm, N₂ < 5 ppm, H₂O < 3 ppm and hydrocarbon < 0.2 ppm) was used as the carrier gas and shielding gas for protecting the melt pool from oxidation and subsequent contamination during the experiments. The flow rates were experimentally found to be in the range of 5 L min⁻¹ for the carrier gas and 9 L min⁻¹ for the shielding gas. For carrier gas rates less than 5 L min⁻¹, the powder flow is not stable and reproducible. Carrier gas flow rates higher than 6 L min⁻¹ result in higher gas velocity, which leads to a loss in efficiency caused by powder ricocheting at the impingement point. These observations are in good accordance with Jinoop’s [24] work in which thin walls of Inconel 718 were manufactured by DED.

A number of single tracks were deposited under variation of the laser power P_L, laser scan speed v and powder feed rate m: The working distance (WD from the end of the nozzle to the deposited surface) was generally set at 12 mm, and the laser was in-focus. In three following test runs, samples were deposited using a laser power of 300, 600 and 700 W, laser scanning speeds in the range of 2 up to 10 mm s⁻¹ and powder feed rates from 3.2 up to 9.1 g min⁻¹. Metallographic cross-sections were prepared and examined by light optical microscopy (Zeiss Axioplan 2, Zeiss, Oberkochen, Germany). The primary goal of these test runs with single tracks was to optimize the laser deposition parameters for the minimum porosity and the maximum deposition rate. Porosity was measured by binarization of the optical microscope images from the cross-sections using ImageJ software, Version 1.54f. The ISO50 threshold value was selected and porosity determined in accordance with DIN EN ISO 15708-3 [25]. Figure 4a,b show a light optical image of a cross-section and the corresponding binarized image. The porosity for this example is 0.1%.

After these trials, test runs with four adjacent tracks were conducted. The laser power, scan speed and powder feed rates were identical to the deposition of single tracks, and the overlap Φ of adjacent tracks was 10 to 60% in 5% steps. In these tests, the influence of the line spacing and thus the overlap on the porosity was analyzed.

The results of the deposition of single and adjacent tracks were used for the deposition of larger volumes in a parameter study in which samples with a geometry of 15 × 15 mm² and 10 layers were fabricated. The scan strategy was offset unidirectional, as shown in Figure 5a. Based on the optimized parameter set determined in the preliminary tests, samples with varying laser power, scan speed and energy density were deposited.

The energy density was calculated according to Equation (1).

$${\rho }_{\mathrm{E}}={\displaystyle \frac{P_{\mathrm{L}}}{v}}$$

(1)

With $P_{\mathrm{L}}$ laser power and $v$ scan speed.

Table 2. Parameter study: parameters and quantified characteristics (porosity, track height and build rate).

Varied Parameter	Sample No. Laser Power_Scan Speed_Powder Feed Rate	Porosity ꓕ SD in %	Porosity || SD in %	Track Height in µm	Build Rate ${\dot{\mathit{B}}}_{\mathbf{R}}$ in g min−1
Laser power	1 500 W_8 mm s−1_4 g min−1	0.24 ± 0.04	0.27 ± 0.06	380 ± 20	1.7
	2 (Optimized Parameter) 600 W_8 mm s−1_4 g min−1	0.16 ± 0.05	0.12 ± 0.03	410 ± 23	1.8
	3 700 W_8 mm s−1_4 g min−1	0.26 ± 0.1	0.26 ± 0.07	450 ± 25	2.0
Feed rate	4 600 W_8 mm s−1_6 g min−1	0.39 ± 0.1	0.36 ± 0.05	620 ± 33	2.8
	5 600 W_8 mm s−1_8 g min−1	0.19 ± 0.03	0.14 ± 0.02	880 ± 36	3.9
	6 600 W_8 mm s−1_10 g min−1	0.76 ± 0.31	0.57 ± 0.25	1060 ± 62	4.7
Scan speed, Feed rate	7 600 W_8.8 mm s−1_4.4 g min−1	0.16 ± 0.03	0.17 ± 0.04	403 ± 32	1.8
	8 600 W_9.6 mm s−1_4.8 g min−1	0.14 ± 0.06	0.15 ± 0.05	400 ± 24	1.7
	9 600 W_10.4 mm s−1_5.2 g min−1	0.22 ± 0.09	0.17 ± 0.03	380 ± 33	1.9
Laser power, Scan speed, Feed rate	10 660 W_8.8 mm s−1_4.4 g min−1	0.18 ± 0.03	0.09 ± 0.04	425 ± 30	1.85
	11 720 W_9.6 mm s−1_4.8 g min−1	0.11 ± 0.03	0.19 ± 0.04	415 ± 13	1.9
	12 780 W_10.4 mm s−1_5.2 g min−1	0.22 ± 0.05	0.26 ± 0.08	430 ± 23	2.0

summarizes the process parameters used for the deposition and the resulting porosity values, track height and build rates. The build rates were calculated using Equation (2).

$${\dot{B}}_{\mathrm{R}}={\displaystyle \frac{v{·w}_{\mathrm{t}}·h_{\mathrm{T}}·\rho }{60}\hspace{1em}}\left[{\displaystyle \frac{\mathrm{g}}{\min }}\right]$$

(2)

With scan speed $v,$ track width $w_{\mathrm{t}}$, track height $h_{\mathrm{T}}$ and density $\rho$.

For the preparation of tensile specimens, samples with a geometry of 40 × 25 × 10 mm³ were fabricated by depositing 20 layers of 25 adjacent tracks, each with a length of 40 mm. The process parameters used were those with the lowest porosity in both printing directions (see Table 2, sample number 2, optimized parameters). Figure 5a shows the movement of the laser for three different scan strategies that were conducted. In the first case, the laser scan direction (SD) for all the tracks was identical and the tracks were stacked without any offset from layer to layer (stacked unidirectional). For the second case, the laser scan direction was identical within a layer and the offset of the track in the following layer was 50% of the track width (offset unidirectional). In the third variant, the laser scan direction of each track changed by 180°, with 50% offset from layer to layer (offset bidirectional). Figure 5b shows an optical micrograph of a cross-section obtained from the offset unidirectional scan strategy.

Tensile specimens were fabricated using wire electrical discharge machining (AgieCharmilles Cut 200SP, GF AgieCharmilles, Losone, Switzerland). For each build up strategy, at least five samples were eroded parallel and perpendicular to the scan direction. The samples used for the tensile testing were 25 × 2 × 1 mm³ in size, with a gauge length of 8 mm. Figure 6a shows the specimen geometry for the tensile tests, while Figure 6b shows the specimens after testing.

The tensile testing was performed using a Zwick Z100 (ZwickRoell, Kennesaw, GA, USA) with a strain rate of 0.001 s⁻¹, the elongation was measured with a video extensometer, which continuously determines the distance between corresponding ridges (which is 8 mm at the beginning), and the strain was calculated using Superstrain software, version 1.1, as described in [26].

A scanning electron microscope (SEM; Zeiss 1540 XB, Zeiss, Oberkochen, Germany) combined with energy dispersive X-ray spectroscopy (EDX) and electron backscatter diffraction (EBSD) was used for the microstructural investigation.

3. Results and Discussion

Figure 7 shows the samples obtained in the parameter study in which four different test series were conducted. The standard parameter set was 600 W laser power, 4 g min⁻¹ powder feed rate and 8 mm s⁻¹ scan speed. In the first step, the laser power was varied from 500 to 700 W with a constant powder feed rate and scan speed (a); in the second step, the powder feed rate was increased by 50, 100 and 150% with fixed laser power and scan speed (b). In the third test series, the laser power was fixed to 600 W while the scan speed was increased by 10, 20 and 30% (c), and in the fourth test series, the scan speed was increased while keeping the energy density ${\mathsf{\rho }}_{\mathrm{E}}$ constant (d).

The results for the porosity are calculated as the mean value and standard deviation from five cross-sections for each orientation. The evaluation is carried out using the bright field of the light microscope perpendicular (ꓕ) and parallel (||) to the scan direction (Table 2). Apart from sample nr. 6, all the printed samples have a low porosity of less than 0.4%. No trend could be observed here when varying the process parameters. This could be attributed to the high depth of the melt pool in the DED-LB-p process as a reason for the repetitive melting process. However, the process parameters for sample number 2 were selected for further microstructural characterization and mechanical testing. This is because the lowest porosity was achieved with these parameters in both printing directions (perpendicular (ꓕ) and parallel (||)).

Figure 8a,b show SEM images from cross-sections of deposited samples in the backscattered mode. The samples were manufactured with the so-called “optimized parameters” (denoted Sample 2 in Table 2). The building direction (BD) was vertical. Figure 6a shows grains and melt pool tracks, of which one is schematically marked with a dotted line.

The element mappings show segregation between the dendritic and interdendritic regions. Figure 9 shows a SEM image and the corresponding element distributions in gray-scale maps.

The element mappings show that iron, chromium and nickel are located in the dendritic regions, whereas molybdenum and niobium segregate in the interdendritic regions. The investigation of the segregation effects of niobium in additively manufactured Ni-based superalloys is described in [27].

The process parameters for depositing larger volumes in the tensile tests were 600 W laser power, 8 mm s⁻¹ scan speed and 4 g min⁻¹ powder feed rate (denoted sample 2, optimized parameter in Table 2). Figure 10 shows the stress–strain curves for samples deposited with different scan strategies and the load applied normal to and in the scan direction. At least five samples were tested for each scan strategy and applied load direction.

There is only a small scattering within the samples. All the curves exhibit ductile material behavior. For each scan strategy, the curves achieve higher strengths in the scan direction (Figure 10e,f) than normal to the scan direction (Figure 10a–c). The elongation to failure when the load was applied normal to the scan direction for the scan strategies stacked (Figure 10a) and offset unidirectional (Figure 10b) is significantly higher compared to the samples with the load applied in the build direction. The anisotropy for the scan strategy offset bidirectional is lower than for the other strategies.

Table 3. Mechanical properties for different scan strategies and porosities.

Ultimate Tensile Strength in MPa			Elongation to Failure in %	Yield Strength in MPa	Porosity in %
S ↑↑	ꓕ	817 ± 24	38.6 ± 1.5	556 ± 17	0.39 ± 0.05
	||	915 ± 24	30.2 ± 2.0	624 ± 29	0.32 ± 0.06
O ↑↑	ꓕ	896 ± 9	40.4 ± 2.1	598 ± 12	0.49 ± 0.04
	||	998 ± 29	32.3 ± 2.7	674 ± 30	0.38 ± 0.05
O ↑↓	ꓕ	936 ± 23	31.7 ± 2.9	667 ± 14	0.22 ± 0.04
	||	1001 ± 19	30.7 ± 3.1	711 ± 27	0.24 ± 0.04
LMD [28]	ꓕ	882 ± 7	36 ± 5	480 ± 20	-
	||	1000 ± 10	24 ± 5	656 ± 14	-
LENS [29]	ꓕ	835 ± 5	43 ± 1.2 [28]	492 ± 5	-
	||	882 ± 7	30 ± 2	598 ± 7	-
Wrought alloy [30]		955 ± 6	41 ± 1	482 ± 42	-

provides an overview of the ultimate tensile strength, yield strength, elongation to failure and porosity for the different scan strategies and compares them with the literature.

The ultimate tensile strength and yield strength for all the tested scan strategies are in the scan direction higher than in the normal to scan direction. The elongation to failure is, for the scanning strategy offset bidirectional in the scan direction and normal to the scan direction, comparable. For the scanning strategies offset and stacked unidirectional, the elongation to failure normal to the scan direction is at least 22% higher in comparison to in the scan direction. In order to obtain a high tensile strength and isotropic deformation, the scanning strategy offset bidirectional is recommended.

Electron backscattered diffraction (EBSD) analysis was conducted using a Zeiss Sigma 300 SEM equipped with an EDAX system. The grain size fractions and grain orientations were measured using Orientation Imaging Microscopy (OIM) software (version 8.6.0101) for the three different scan strategies. Figure 11 shows the grain orientation maps for different scan strategies.

Figure 11a shows large grains, which are larger in the horizontal direction than in the vertical direction, being surrounded by smaller grains. Figure 11b shows the corresponding grain orientation map in the scan direction, in which columnar grains that are oriented at an angle to the build direction are visible. These columnar grains are larger than the track height, which was 410 µm (sample 2, Table 2). Figure 11b,c show that the grains for the scanning strategies offset uni- and bidirectional are smaller than for the scanning strategy stacked unidirectional. Figure 11e,f also show columnar grains being oriented at an angle to the build direction, but in this case, the size is limited to the track height.

Based on the grain orientation map data, the grain sizes in the load direction were measured.

Table 4. Grain sizes in the load direction for different scan strategies.

Scan Strategy	Stacked Unidirectional		Offset Unidirectional		Offset Bidirectional
Direction	ꓕ	||	ꓕ	||	ꓕ	||
Grain size in load direction in µm	228 ± 38	188 ± 7	118 ± 14	119 ± 8	111 ± 10	88 ± 6

lists these values.

Depending on the scan strategy, the grain size decreases from stacked unidirectional via offset unidirectional to offset bidirectional. The grains in the scan direction are smaller than the grains normal to the scan direction. The mechanical properties of the samples deposited with the offset bidirectional scan strategy show the best mechanical properties, which can be attributed to the Hall–Petch relationship.

The mechanical strength properties of the samples printed with the offset bidirectional scanning strategy are better than those of the wrought alloy (Table 3), which is in good accordance with [31]. However, the ductility seems to be slightly lower, which can be attributed to the smaller tensile samples and different cooling rates [32]. M. Rombouts et al. [28] have obtained the same mechanical properties and process parameters depending on the printing direction and the printing strategies. J. Nguejio [30] and F. Zafar [33] attempted to improve the tensile strength of printed samples of Alloy 625 (DED-LB-p) by post-heat treatment and the addition of TiC as reinforcement. A comparable tensile strength of around 1000 MPa was achieved.

Further improvements concerning the deposition efficiency of additively manufactured Alloy 625 can be realized by using hot-wire laser metal deposition. Su et al. [34] manufactured defect-free thin-walled structures with a deposition rate of 1.72 kg h⁻¹. The UTS of the walls was 826 MPa and therefore only slightly lower.

4. Conclusions

The properties of the Alloy625 obtained by DED were studied, and the influence of the printing parameters and scanning strategies were considered. The main conclusions of this study are as follows:

• Porosities less than 0.5% for all the scanning strategies are possible.
• Deposition volumes of 33 cm³ h⁻¹ (which is 0.28 kg h⁻¹) are deposited.
• The scan strategy offset bidirectional shows the highest ultimate tensile strength of 1001 ± 19 MPa and yield strength of 711 ± 27 MPa, the smallest grain size and the lowest anisotropy concerning strength and elongation to failure under different loads.
• The tensile properties of the printed samples (tensile strength and yield strength) are better than those of wrought parts, but their ductility is lower, which can be attributed to the different sample geometries.
• An ultimate tensile strength of 1000 MPa and a yield strength over 710 MPa can be realized without any heat treatment.