Functionally graded materials of SS316L and IN625 manufactured by direct metal deposition

2.1 Materials and methods

SS316L and IN625 powders were used for the deposition, and their chemical compositions are listed in Table 1. The morphologies of the powders are shown in Figure 1. These powders were gas-atomized particles with a near-spherical shape, and sizes varied from 50 to 150 µm. The alloy compositions prepared for three gradient layers were 65% SS316L + 35% IN625, 50% SS316L + 50% IN625, and 30% SS316L + 70% IN625. These alloy compositions were deposited in the middle of the FGM blocks to avoid the formation of brittle phases. The metal powders were pre-heated to 60°C for 10 min before the deposition occurred. A stainless-steel 316 (SS316) substrate of size 152.5 mm length × 50 mm wide × 16.5 mm thick was utilized for depositing pure SS316L powder on it and on top of the block pure IN625. The substrate chemical composition is given in Table 1. A DMD machine, shown in Figure 2, was used for the deposition of the metal powders. It was equipped with a double heterostructure laser diode. This laser is produced by a laser-producing unit of Laserline LDF4.000-30. Nine partial FGM blocks of size 26 mm width × 34 mm thickness × 32 mm height were 3D printed vertically with bi-directional scanning using a co-axial nozzle of a commercial DMD machine by Taguchi’s L9 approach. It has a laser power capacity of 100–4,000 W, scan velocity of 10–30 mm/s, and powder feeding rate of 15–50 g/min. A co-axial nozzle having a standup distance of 23 mm with a diode laser of 5 mm beam spot was used for depositing the metal powders of 1.3 mm thick layers with 2 mm hatch spacing. The beam spot size varied from 3 to 5 mm. Its wavelength range was 902–1,030 nm. It has one powder hopper with a powder filling capacity varying from a minimum of 100 g to a maximum of 1.5 kg. The powder carrying gas (with 10–12 L/min) and shielding gas (with 12–15 L/min) used here is argon (Ar).

Figure 1

Morphology of (a) SS316L powder and (b) IN625 powder.

Figure 2

A robot-based direct metal deposition setup.

2.2 Taguchi’s method

The design of the experiments performed was based on full factorial analysis or Taguchi’s method. Therefore, in this research, Taguchi’s method was selected. It consists of eight steps to determine the optimum parameter set. A series of experiments were performed using a commercial DMD machine with the SS316L powder, IN625 powder, and three different powder compositions prepared from the powders of SS316L and IN625. The factors that affected the metal powder deposition on the DMD machine are listed in Table 2. Based on these factors, as shown in Table 2, the most required and influencing factors, such as laser power, scan velocity, and powder feed rate, were selected for powder deposition/experiments and are listed in Table 3. The rationale is the perfect joining of two dissimilar materials of SS316L and IN625 using a DMD rather than a conventional welding process. These factors were also used in many DED processes [66]. Three factors and three levels were selected. Thus, using the Taguchi L9 orthogonal array, only nine experiments were sufficient. The levels of DMD parameters selected for deposition were based on lower laser power with a low scan velocity and low powder feed rate; moderate laser power with a moderate scan velocity and moderate powder feed rate; a high laser power with a higher scan velocity; and a higher powder feed rate for obtaining high strength. However, high strength was obtained at a moderate laser power with a high scan velocity and low powder feed rate (i.e., at six sets of parameters, see Table 4). Table 4 provides the matrix of experiments for the selected values of the factor-based Taguchi L9 orthogonal array. Nine experiments were conducted for the deposition of partial FGM blocks of size 26 mm width × 34 mm thickness × 32 mm height, and one of them is shown in Figure 3. The input variables for FGM depositions are used for obtaining output responses of high strength and hardness, as shown in Table 5.

Figure 3

Partial FGM block.

2.3 Characterization

Micro-tensile tests were performed on the ASTM E8 samples in the direction perpendicular to the materials deposition using a SHIMADZU micro-tensile testing machine with a capacity of 250 kN. According to ASTM E8 standards, the load applied on the specimens was at a rate of 5 mm/min for a gauge length of 12.5 mm. Figure 4 shows the ASTM E8 tensile specimens cut from each FGM block using an Electronica CNC wire-cut EDM. The ASTM E8 specimen was 9 mm wide, 3 mm thick, and 30 mm height.

Figure 4

ASTM E8 tensile specimens.

The micro-hardness of nine partial FGM samples was evaluated using an ECONOMET hardness tester. It has a diamond pyramid indenter with an applied load capacity ranging from 5 g to 1 kg. A load of 500 g was applied to the indenter for 15 s at each location perpendicular to the deposition direction. These locations have 0.5 mm interval distances along the center-line of the samples. The micro-hardness indentations on compositional layers of the sample-9 are shown in Figure 5. To understand the gradient zones in a better way, the macro-structure of sample-6 was taken using an Olympus optical microscope, as shown in Figure 6. Micro-structural examination was performed to understand the distribution of phases in various layers of the gradient zones. The morphologies of the fractured surfaces of the broken specimens were captured using a GeminiSEM500 at various magnifications. Chemical compositional analysis was performed using Octane Elite EDAX with the help of GeminiSEM500.

Figure 5

Micro-hardness indentations on compositional layers of sample-9.

Figure 6

Macrostructure of ASTM E8 sample-6.

3.1 Macro-structure

Here, gradient path partial FGM indicates three different compositional layers in the middle of the ASTM E8 specimen, shown in Figure 6, captured by an Olympus optical microscope. It is the macrostructure of sample-6. These layers are known as gradient zones. The chemical compositions of the first, second, and third layers are 65% SS316L + 35% IN625, 50% SS316L + 50% IN625, and 30% SS316L + 70% IN625, respectively. These samples were polished with emery papers of 200, 600, and 1,000 using a METCO BAINAPOL polishing machine for macrostructure analysis. However, emery papers of 1, 500, and 2,000 were used along with the above polishing papers for observing the microstructures of the samples. The etchant used for these samples is a modified Frys reagent. The measured thickness of each gradient layer is 1.3 mm, and the step-over is 1.9 mm. There is a small variation in the thickness of individual gradient layers due to the metal powders swirling by the atmospheric air in an open part building chamber. The swirling powders cannot flow toward the melt pool but mix with atmospheric air. This indicates the loss of metal powders, and may occur by improper focusing of the laser beam at the focusing point in the melt pool during melting of metal powders. However, the gradient layer thickness and step-over meet the desired values.

3.2 Microstructure

The microstructures of the three compositional layers of specimen-2 taken by an Olympus optical microscope are shown in Figure 7. Crack propagation perpendicular to the deposition direction is observed in the first compositional layer, as shown in Figure 7a. Fine, small columnar grains are shown in Figure 7b. However, large coarse columnar grains observed in the third compositional layer are shown in Figure 7c. The white regions in all compositional layers of the specimen indicate an austenitic phase, and dark spots are unmelted powder particles. The reasons for the formation of the unmelted powder particle regions were due to insufficient laser power and improper focusing of the laser beam at the melt pool. Sometimes, impurities were also present in the unmelted powder particle regions. However, crack propagation was generated, which may be due to the existence of tensile residual stresses in the sample.

Figure 7

Microstructures of (a) layer 1, (b) layer 2, and (c) layer 3 of sample-2.

The microstructures of specimen-6 were similar to those of sample-2 and are shown in Figure 8. No microcracks or micropores were observed in the three different compositional layers of the specimen during microscopic examination. This is because of the lower powder feed rate, high scan velocity, and sufficient laser power for the melting of the SS316L and IN625 powder particles. However, coarse parallel columnar grains in the first compositional layer, fine columnar grains in the second compositional layer, and equiaxed and coarse columnar grains in the third compositional layer were observed in the specimen and are shown in Figure 8a–c, respectively. The microstructures of these three compositional layers also had white regions, indicating the austenite phase, dark spots of unmelted powder particles, and light–dark regions of compounds of SS316L and IN625. In general, conventional microstructural analysis indicates that the presence of coarse parallel columnar grain microstructures can decrease the mechanical properties and fine grain microstructures can increase the mechanical properties. However, sample-6 had high UTS because of the mixed presence of coarser and finer columnar grains and equiaxed grains microstructures. The remaining specimens exhibited coarse columnar grain microstructures with microcracks and micropores.

Figure 8

Microstructures of (a) layer 1, (b) layer 2, and (c) layer 3 of sample-6.

3.3 Tensile tests

Three tensile tests were performed for each experiment. The averages of the three UTS values of the partial FGM specimens are listed in Table 6. The fractured specimens after the microtensile tests are shown in Figure 9. From the micro-tensile test results (see Table 6), it was found that sample-6 had a high UTS of 532.6 MPa compared with other specimens. This indicated that the use of medium laser power (i.e., 2,500 W) with high scan velocity and with a lower powder feed rate can produce specimens with high tensile strength because this medium laser power can melt powder particles completely and cause strong metallurgical bonds. The three trial stress–strain curves of the micro-tensile tested sample-6 are depicted in Figure 10. The UTS obtained from the thick layered (i.e., thickness more than 1 mm) gradient path partial FGM sample-6 in this research work was almost close to the thin layered (i.e., thickness less than 1 mm) multilayer material specimens of previous studies [26,33]. Therefore, in the present research, the UTS results indicate that for joining two dissimilar materials to obtain the desired properties, either the thin-layer multilayer materials method or the thick-layer gradient path method (for partial FGM) should be selected. However, it is better to choose the latter because it requires less time for part deposition and is more economical. If the UTS of the conventional welded joints (399–522 MPa) of SS316L and IN625 are considered, it was less than that of the partial FGM sample-6.

Figure 9

ASTM E8 tensile specimens after failure.

Figure 10

Stress versus strain curves of sample-6 for trial 1, trial 2, and trial 3.

Specimen-2, specimen-4, specimen-5, and specimen-6 were fractured at 17.6, 17.9, 16.6, and 17.1 mm, respectively, from the lower side of the pure SS316L region. This indicated that they failed in the third compositional layer. In contrast, samples-1 and samples-3 broke into two pieces in the second compositional layers at a distance of 15.3 and 14.5 mm, respectively. The specimens that failed in the second and third compositional layers have high strength because of the absence of molybdenum (Mo) and niobium (Nb) brittle phases. Specimen-7 failed in the first compositional layer (i.e., at 13 mm), and sample-8 and sample-9 broke in the regions of pure IN625. The failure specimens are depicted in Figure 9. It is concluded that specimen-3 showed the maximum percentage of reduction in area (i.e., 76.8%) and also the percentage of strain (i.e., 64.6%). Therefore, it had the highest ductility compared to other specimens.

3.4 Hardness

Hardness is the surface property of a material to resist plastic deformation. Two locations/indentation spots were selected for the evaluation of micro-hardness from each compositional layer of the samples. There were three compositional layers in each sample. Therefore, a total number of six locations/indentation spots were considered for the measurement of micro-hardness from every sample. For more details regarding micro-hardness indentation spots, see Figure 5. The micro-hardness values measured using an ECONOMET hardness tester for all nine samples are shown in Table 7.

The micro-hardness of each gradient zone increases with increasing content of IN625. However, the micro-hardness of the gradient zones is not lower than the hardness of the pure SS316L region and not higher than the hardness of the pure IN625 region in all samples. The micro-hardness graph of sample-5 is shown in Figure 11a. Its gradient zone micro-hardness (i.e., 178.2 HV) is very low compared to the those of the remaining samples. However, the highest gradient zone micro-hardness (202.5 HV) is observed in sample-9, as shown in Figure 11b. Hence, it is concluded that sample-9 printed at a high laser power has the highest gradient zone micro-hardness, and this specimen also has coarse columnar and equiaxed grain microstructure for exhibiting high hardness. Considering the micro-hardness variations in the entire sample-9, the micro-hardness of the pure SS316L region increased from its bottom layer (175.2 HV) to a maximum value of 192.4 HV in it, then decreased to that of the first compositional layer (167.4 HV). Similarly, for its pure IN625 region, the micro-hardness increased from the third compositional layer (240.2 HV) to its top surface (258.5 HV). In sample-5, as depicted in Figure 11(a), the hardness of the pure SS316L portion increases from its bottom layer (170.4 HV) to the first gradient layer (187.3 HV). However, the hardness of its pure IN625 region also increases from the third compositional layer (228.3 HV) to its top surface (254.2 HV). Therefore, for comparing the micro-hardness of the top surfaces of these samples, the ninth sample has the highest value due to the formation of fine grain microstructure by the absence of repetitive heating and cooling.

Figure 11

Micro-hardness of partial FGM (a) sample-5 and (b) sample-9.

3.5 Failure surface morphology and chemical composition analysis

Failure surface morphology is known as fractography. The morphologies of the fractured layer of sample-2, sample-6, and sample-7 were captured using a GeminiSEM500. These three specimens were printed at a lower energy parameter set (1,500 W, 20 mm/s, and 25 g/min), medium energy parameter set (2,500 W, 30 mm/s, and 20 g/min), and high energy parameter set (3,500 W, 10 mm/s, 30 g/min). Specimen-2 and specimen-6 failed at the third compositional layer. However, sample-7 broke in the first gradient zone. The surface morphologies of sample-2 and sample-6 at different magnifications are shown in Figures 12 and 13. Both microholes and microcracks of the samples are shown in Figures 12a and 13a. However, microcracks and microholes are shown individually in Figures 12b and 13b, respectively. In contrast, many simple shallows with uniform sizes are shown in Figure 12c, and smeared or angled shallows with uniform morphology are shown in Figure 13c. Therefore, the surface morphologies with shallows were mainly ductile, with very little deformation in specimen-2 and specimen-6. The surface morphology of fractured sample-7 at various magnifications exhibits the same surface morphology as compared to the morphologies of sample-2 and sample-6. The required three gradient zones of the partial FGM samples were printed with the desired amount of IN625 in the metal powder mixture of SS316L and IN625. Therefore, it can avoid the formation of microcracks and micropores. Hence, the mechanical properties of the FGM samples can be improved [38,39].

Figure 12

(a–c) Tensile failure surface morphology of third compositional layer of sample-2.

Figure 13

(a–c) Morphology of tensile failure surface of third compositional layer of sample-6.

In general, columnar grains with uniform morphology were generated in the DMD-processed depositions. Microholes were formed owing to the escape of powder-entrapped gases and other gases during the deposition process or melt-pool solidification. This is known as process-induced porosity. In DMD deposition, the presence of residual stress is common. Tensile residual stresses create microcracks. However, the compressive residual stresses close them. Debris means the generation of unwanted material during powder deposition (see Figure 12a). This may be due to various factors, such as the quality of the powder feedstock used, improper selection of deposition parameters, and the size of the part to be built. Smeared or angular shallows were formed because of the rapid solidification of the molten pool during the deposition process.

Fractured specimen-2, specimen-6, and specimen-7 were used for microstructure and chemical compositional analyses using GeminiSEM500 and Octane Elite EDAX. Two locations were selected on the fractured surfaces of the samples for analysis. From the EDAX report, molybdenum was present in sample-2, sample-6, and sample-7 at 3.1, 2.8, and 2.1%, respectively, in their respective fractured layers. However, niobium was present in specimen-2 and specimen-7 at 1 and 1.4%, respectively, in their fractured layers. The detected molybdenum (Mo) and niobium (Nb) content by EDAX was within the range of 2–9% and 3.1–4.1% in the fractured layers in those specimens, as compared to SEM powder morphology and powder supplier reports. This indicates Mo and Nb brittle phases were not formed and did not allow for microcracks and micropores. The scanning electron beam microstructure and elemental distribution shown by EDAX for sample-6 is shown in Figure 14. There was no presence of Nb (at the selected spots; see Figure 14a) observed in the EDAX report (Figure 14b) in the fractured surface of sample-6. However, it may be present in other layers of the sample in lower or higher quantities. It was revealed from the EDAX reports of these samples that iron (Fe) was observed in the maximum quantity, followed by chromium (Cr) and nickel (Ni). The elements observed by EDAX in sample-6 are shown in Figure 14b.

Figure 14

(a) SEM image of specimen-6 at the fracture and (b) its chemical composition distribution by EDAX.

3.6 Optimization of process parameters

3.6.1 Ultimate tensile strength

To calculate signal-to-noise (S/N) ratios, an objective function (i.e., ultimate tensile strength) larger-the-better was utilized. Therefore, the S/N ratio for this function is given by −10 × log (Σ(1/y _i ²)/n), where y _i indicates the responses (i.e., ultimate tensile strengths) for the given factor level combination and n is number of trials in the factor level combination. Table 8 shows the S/N ratios of the nine experiments, which were solved analytically, and the same values were obtained when validated with Minitab statistical 21.2.0.0 software. The S/N ratios for the individual control factors validated by the Minitab software are listed in Table 9. Among the selected parameters for UTS, the scan velocity played a prominent role (i.e., 1st rank), followed by the powder feed rate and laser power. Figure 15 shows the deposition parameter levels versus S/N ratios for determining the optimum parameter values.

Table 8

Minitab-validated S/N ratios for UTS

Experiment no.	S/N ratio
1	54.1
2	53.2
3	54.2
4	53.6
5	53.9
6	54.5
7	52.9
8	54.0
9	54.3

Table 9

S/N ratios for the individual control factors for UTS: larger the better

Level	Laser power (P), W	Scan velocity (V), mm/s	Powder feed rate (M), g/min
1	53.8	53.5	54.2
2	54.0	53.7	53.7
3	53.7	54.3	53.7
Delta	0.2	0.8	0.5
Rank	3	1	2

Figure 15

Deposition parameter levels versus S/N ratios for UTS.

To calculate UTS, a larger-the-better type of objective function is utilized. The highest S/N ratio for the corresponding factor levels was selected to optimize the conditions. From the linear graph (see Figure 15), the optimum values of the factors were the following: laser power of 2,500 W, scan velocity of 30 mm/s, and powder feed rate of 20 g/min at the levels of (2, 3, 1). This indicated that the sixth set of parameters had optimum values for obtaining high UTS in this DMD printing experiment. For more details, refer to Table 4. In this research, the time required for deposition of each FGM block using the selected commercial DMD machine was 1 hour 30 min. In the case of a non-commercial DMD machine, the deposition time may be 8 h or more for depositing one FGM block.

The degrees of freedom (DOF) were evaluated for the selected L9 orthogonal array of experiments. There were two degrees of freedom for each factor. There are four factors, e.g., the powder feed rate, scan velocity, laser power, and error. Thus, it has a total of eight degrees of freedom. The response magnitude in the percentage of each parameter was determined and validated using the ANOVA module of the Minitab software. It was used to identify and quantify the sources of various microtensile test trial results obtained from various parameter sets of powder depositions. The total sum of the squares (SS _T ) can be calculated as the sum of squares (SS) of all the factors and errors based on Eqs. (1) and (2), and the values are listed in Table 10.

(1) SS T = ∑ ( y i – avg . of y ) 2 ,

(2) SS T = SS P + SS V + SS M + S S e .

Table 10

Analysis of variance for UTS

Source	DOF	Adj. SS	Adj. MS	F-Value	P-Value	% SS = Adj. SS/Total
Laser power (P), W	2	0.12	0.06	0.25	0.80	5.8
Scan velocity (V), mm/s	2	1.05	0.52	2.06	0.32	48.1
Powder feed rate (M), g/min	2	0.49	0.24	0.97	0.50	22.7
Error, e	2	0.51	0.25			23.3
Total	8	2.19	1.09			100

Table 10 shows that for UTS, the contribution of scan velocity (48.1%) was a more significant factor than the powder feed rate (22.7%). These two factors were more significant than the laser power (5.8%). However, the effect of noise factors on UTS was 23.3%. The reasons for obtaining more errors are due to vibrations of the commercial DMD machine, use of manually mixed powder compositions, specimens not printed with different layer thicknesses, hatch spacing, and the use of an open part building chamber of the machine.

3.6.2 Micro-hardness

A larger-the-better objective function (i.e., micro-hardness) was utilized to evaluate signal-to-noise (S/N) ratios. Table 11 shows S/N ratios of nine micro-hardness samples, and their S/N ratios for the individual control factors are listed in Table 12. The powder feed rate is an important influencing parameter (i.e., standing at first rank) on micro-hardness, followed by laser power and scan velocity. The deposition parameter levels versus S/N ratios for determining the optimum parameter values are shown in Figure 16. The highest S/N ratio for the corresponding factor levels was considered to optimize the conditions. From the linear graph (Figure 16), the optimum values of the factors were as follows: laser power of 3,500 W, scan velocity of 30 mm/s, and powder feed rate of 25 g/min at the levels of (3, 3, 2). Table 4 shows the details about it. The ninth set of parameters had optimum values for obtaining high micro-hardness.

Table 11

Minitab-validated S/N ratios for micro-hardness

Experiment no.	S/N ratio
1	45.4
2	45.1
3	45.3
4	45.6
5	44.8
6	45.0
7	45.1
8	45.6
9	46.0

Table 12

S/N ratios of the individual control factors for micro-hardness: larger the better

Level	Laser power (P), W	Scan velocity (V), mm/s	Powder feed rate (M), g/min
1	45.3	45.4	45.3
2	45.2	45.2	45.6
3	45.6	45.5	45.1
Delta	0.4	0.2	0.5
Rank	2	3	1

Figure 16

Deposition parameters levels versus S/N ratios for micro-hardness.

ANOVA was utilized to identify and quantify the sources of various micro-hardness test trial results obtained from the different parameter sets of 3D-printed deposits. Using Eqs. (1) and (2), the total sum of the squares (SS _T ) determined as the sum of squares (SS) of all factors and errors are given in Table 13.

Table 13

Analysis of variance for micro-hardness

Source	DOF	Adj. SS	Adj. MS	F-value	P-value	% SS = Adj. SS/total
Laser power (P), W	2	0.28	0.14	1.06	0.48	26.9
Scan velocity (V), mm/s	2	0.11	0.05	0.44	0.69	10.5
Powder feed rate (M), g/min	2	0.39	0.19	1.48	0.40	37.5
Error, e	2	0.26	0.13			25.0
Total	8	1.04	0.51			100

The contribution of powder feed rate (37.5%) was a significant parameter on micro-hardness than laser power (26.9%). However, the lowest significant factor was scan velocity (10.5%) as compared to the above factors. However, the effect of noise factors on micro-hardness was 25%. From Table 13, it can be concluded that metal powder melts completely as the powder feed rate increases with increasing laser power simultaneously. However, this condition is satisfied at a lower scan velocity.