Effect of Laser Power on the Microstructure and Fracture of Notched IN718 Specimens Fabricated by Laser Powder Bed Fusion

Abstract

This study examines the impact of laser power on the microstructure and fracture behavior of IN718 specimens fabricated using laser powder bed fusion. Single-edge notched bend specimens were fabricated with varying laser power from 140 W to 260 W, and their fracture behavior was analyzed following the ASTM E1820-23b standard. The porosity and grain morphology remained unaffected by the presence of a notch parallel to the build direction. An elastic–plastic fracture mechanics approach was used to measure J-R curves, which quantify the energy required for crack propagation. Crack initiation and growth during quasistatic loading were monitored using image analysis. The results revealed a strong correlation between crack initiation and propagation, type of porosity, and relative density. The specimen printed with the optimal laser power of 180 W demonstrated the highest relative density and the greatest resistance to crack propagation. Large non-spherical defects formed due to lack-of-fusion at lower laser power are more detrimental to the crack propagation resistance.

1. Introduction

Additive manufacturing (AM) has the potential to revolutionize the metal manufacturing industry by enabling the production of near-net-shaped components with tunable properties and minimal waste [1,2,3]. Among various metal AM techniques, Laser Powder Bed Fusion (L-PBF) is particularly well-suited for fabricating complex parts with high dimensional accuracy and superior surface finish. However, the broad applicability of L-PBF requires process optimization to achieve reproducibility comparable to conventional manufacturing techniques. The properties of L-PBF manufactured components are highly sensitive to process parameters, which must be optimized for each material. Numerous studies have focused on understanding the effects of L-PBF parameters on the microstructure and properties of various alloy systems, including stainless steel [4], Ni alloys [5], Ti alloys [6], Al alloys [7], and high-entropy alloys [8]. These advancements have led to the development of new L-PBF strategies, enabling site-specific microstructure engineering through the optimization of processing parameters during printing [5,9].

Key L-PBF process parameters include laser power, scan speed, hatch spacing, layer thickness, and scan strategy [10]. Many of these parameters are often combined into linear or volumetric energy density terms to explain their effect on the quality, microstructure, and properties of L-PBF specimens. The energy input influences melt pool dynamics, thermal gradients, and the solidification during L-PBF [1]. Improper process parameters can lead to the formation of defects due to insufficient melting, excessive evaporation, splatter, gas entrapment, and thermal stress. Excessive volumetric energy density achieved through higher laser power or lower scan speed leads to the formation of keyhole porosity. In contrast, lower volumetric energy density results in insufficient melting, forming lack-of-fusion defects in L-PBF specimens. Previous studies have demonstrated that optimizing energy density by adjusting process parameters can reduce defects and produce full-density (>99%) L-PBF parts [11,12]. In addition to porosity, process parameters also influence grain size, crystal texture, sub-grain structure, and phase formation [13,14]. Depending on the processing parameters, columnar and equiaxed grains with a cellular or dendritic sub-grain structure can be formed [15].

Processing-structure–property studies on L-PBF manufactured alloys have primarily focused on mechanical properties such as hardness, yield strength, ultimate tensile strength, and ductility. Extreme energy density values lead to the formation of large irregular shaped pores, which degrade most of the desirable mechanical properties [16]. In the intermediate processing window, mechanical properties can be tailored by optimizing the microstructure through adjustments to laser power, scan speed, and scan strategy [17,18,19]. Wang et al. demonstrated that the maximum hardness, tensile strength, and elongation for SS316L were achieved at a scan speed of 1000 mm/s and a laser power of 300 W [20]. Luo et al. studied the effects of processing parameters on the tensile properties of L-PBF Ti-6Al-4V and found a linear correlation between the tensile properties and linear energy density [21]. Similar parametric studies have been conducted on IN718 [22] and Al alloys [23].

Compared to hardness and tensile properties, the fracture toughness of L-PBF samples has received less attention [24,25,26,27,28,29]. Porosity is an inherent characteristic of L-PBF samples due to the layer-by-layer melting and bonding of powder particles. Depending on their size or shape, pores can promote crack initiation by concentrating stress and facilitating propagation. Vieille et al. compared the fracture behavior of IN718 specimens prepared by L-PBF and casting [24]. They observed a significant difference in fracture toughness values based on the orientation of the notch with respect to the build direction. Most prior fracture toughness studies were conducted on specimens prepared using a single set of L-PBF parameters. Recently, Wilson et al. studied the effect of laser power and scan speed on the fracture toughness of SS316 specimens, but the corresponding microstructure was not characterized [27]. The objective of this study is to investigate the correlation between the laser power, microstructure, and fracture behavior of L-PBF specimens. To achieve this, we use IN718 single-edge notched bend (SENB) specimens fabricated at different laser powers. This resulted in a systematic variation in both volumetric and linear energy densities as all other L-PBF parameters were kept constant.

2. Materials and Methods

2.1. L-PBF Process

The SENB specimens were fabricated using SLM 125 HL (SLMSolutions Group AG, Lübeck, Germany) L-PBF system equipped with a 400 W Ytterbium fiber laser. The feedstock material was IN718 alloy powder with an average particle size of 28 µm. The powder was acquired from SLM Solutions, and its chemical composition is provided in

Table 1. The chemical composition of IN718 powder.

Element	Fe	Cr	Ni	Mo	Nb + Ta	Mn	Si	P	C	Co	Al
Mass fraction (%)	Bal.	17–21	50–55	2.8–3.30	4.75–5.50	0.35	0.35	0.015	0.08	1.0	0.2–0.8

. The build plate was preheated to 100 °C to reduce thermal stress and improve part adhesion. The printing was performed in an argon environment to prevent oxidation. The SENB specimens were designed following ASTM E1820-23b standard [30], with each sample measuring 50 mm in length, 10 mm in width (W), 10 mm in thickness (B), and containing a 5 mm notch (a_o), as illustrated in Figure 1a. The specimens were printed on a stainless-steel build plate, and the notch was oriented parallel to the build direction (Figure 1b). This orientation was selected to minimize the roughness in the upward-facing notch surfaces. The roughness of the notch surfaces for all specimens was in the range of 6–10 µm.

The processing conditions were controlled by varying the laser power (LP) while keeping the scan speed (SS), the hatch spacing (HS), and the layer thickness (LT) constant.

Table 2. The L-PBF process parameters used to fabricate SENB IN718 samples and the corresponding relative density values.

Sample No.	LP (W)	SS (mm/s)	HS (mm)	LT (mm)	VED (J/mm3)	Relative Density (%)
1	140	600	0.12	0.03	64.8	98.40
2	180	600	0.12	0.03	83.3	99.83
3	220	600	0.12	0.03	101.8	99.65
4	260	600	0.12	0.03	120.4	99.34

summarizes the L-PBF process parameters used for each sample. The laser power varied from 140 W to 260 W for the four SENB specimens. A stripe scanning strategy was employed with a 67° rotation angle of the laser vector between successive layers. The volumetric energy density (VED) was calculated using the following [31]:

$$VED={\displaystyle \frac{LP}{SS·LT·HS}}$$

(1)

2.2. Microstructural Characterization

The SENB samples were removed from the build plate using a metal cutting bandsaw. The cut specimens were sonicated in isopropyl alcohol to remove residual powder particles. The relative density of each sample was measured using Archimedes’ principle. For microstructural analysis, the specimens were mechanically polished and chemically etched using Carpenter’s etchant (8.5 g FeCl₃, 2.4 g CuCl₂, 122 mL hydrochloric acid, 6 mL Nitric acid, 122 mL Ethanol) acquired from Pace Technologies (Tucson, AZ, USA). The melt pools, grains, and sub-grain features were analyzed using a Keyence digital optical microscope (Itasca, IL, USA) and a Zeiss SIGMA 500 VP scanning electron microscope (Oberkochen, Germany).

2.3. Mechanical Testing

The mechanical properties of the SENB specimens were evaluated by three-point bending using an Instron machine equipped with a 100 kN load cell. The crack length was calculated from the camera images using ImageJ software (1.54 P, Bethesda, MD, USA). The bending tests were performed in a displacement-controlled mode at a rate of 0.6 mm/min. The load–displacement data from the Instron machine were used to generate bending stress ($\sigma$)–strain ($ϵ$) curves using the following relations:

$$\sigma ={\displaystyle \frac{3PS}{2B{W}^2}}$$

(2)

$$ϵ={\displaystyle \frac{6Wv}{S^2}}$$

(3)

where P is the applied load, S is the support span, B is the thickness, W is the width of the specimen, and v is the applied displacement (see Figure 1a).

The fracture behavior of the SENB specimens was evaluated using crack growth resistance (J-R) curves, which are typically used in elastic–plastic fracture mechanics [32]. The J-R curves represent the energy required per unit fracture area to initiate and propagate a crack. The procedure for obtaining the J-R curves was followed from the ASTM E1820-23b standard [27,30]. The fracture toughness (J_i) at a given point in time is the sum of the elastic fracture toughness (J_el) and the plastic fracture toughness (J_pl), which is calculated as follows:

$$J_i=J_{el, i}+J_{pl, i}$$

(4)

The $J_{el, i}$ is calculated from the stress intensity factor $K_{ i}$ and the elastic constants of the material (Young modulus E = 210 GPa and Poisson ratio ν = 0.30) through the following equations:

$$J_{el,i}={\displaystyle \frac{K_i^2\left(1-{\nu }^2\right)}{E}}$$

(5)

$$K_i=\left({\displaystyle \frac{P_{i}S}{B{W}^{3/2}}}\right)f\left({\displaystyle \frac{a_i}{W}}\right)$$

(6)

where $f\left({\displaystyle \frac{a_i}{W}}\right)$ can be expressed as follows:

$$f\left({\displaystyle \frac{a_i}{W}}\right)={\displaystyle \frac{3{\left(\frac{a_i}{W}\right)}^{1/2}\left[1.99-\frac{a_i}{W}\left(1-\frac{a_i}{W}\right)\left(2.15-3.93\frac{a_i}{W}+2.7{\left(\frac{a_i}{W}\right)}^2\right)\right]}{2\left(1+2\frac{a_i}{W}\right){\left(1-\frac{a_i}{W}\right)}^{3/2}}}$$

(7)

Equation (7) depends on the crack-extension-to-specimen-width ratio (${\displaystyle \frac{a_i}{W}}$). The crack extension values were measured as the perpendicular distance from the base surface to the crack tip using ImageJ software.

For a load-line displacement test, $J_{pl,i}$ is determined using the following equation:

$$J_{pl,i}=\left[J_{pl,i-1}+{\displaystyle \frac{1.9}{b_{i-1}}}\left({\displaystyle \frac{A_{pl,i}-A_{pl,i-1}}{B}}\right)\right]\left[1-0.9\left({\displaystyle \frac{a_i-a_{i-1}}{b_{i-1}}}\right)\right]$$

(8)

where $A_{pl,i}$ represents the plastic area associated with the force ($P_i$) and displacement (${\mathrm{v}}_{\mathrm{i}}$) at the specified time. It is computed using the following equation:

$$A_{pl,i}=A_{pl,i-1}+{\displaystyle \frac{\left(P_i+P_{i-1}\right)\left({\mathrm{v}}_{\mathrm{pl},\mathrm{i}}-{\mathrm{v}}_{pl,i-1}\right)}{2}}$$

(9)

where $v_{pl,i}$ denotes the plastic component of the load–line displacement at the specified time. It is determined using the following equation:

$$v_{pl,i}=v_i-\left(P_i{C}_i\right)$$

(10)

where $C_i$ represents the compliance of the material, which is calculated as follows:

$$C_i={\displaystyle \frac{1}{EB}}{\left({\displaystyle \frac{S}{W-a_i}}\right)}^2\left[1.193-1.98{\displaystyle \frac{a_i}{W}}+4.478{\left({\displaystyle \frac{a_i}{W}}\right)}^2-4.443{\left({\displaystyle \frac{a_i}{W}}\right)}^3+1.739{\left({\displaystyle \frac{a_i}{W}}\right)}^4\right]$$

(11)

3. Results and Discussion

3.1. Porosity and Relative Density

The commonly observed defects in L-PBF parts are lack-of-fusion (LOF), keyholes, and metallurgical micro-pores [1,15,17,18]. Figure 2 shows representative optical images revealing different types of pores and the corresponding size distribution in the four SENB samples printed at various laser powers. The distribution of pore size (equivalent diameter) was determined using image analysis of five micrographs taken from different areas (near and away from the notch) in each sample covering 5 mm × 5 mm areas. Sample 1, printed at the lowest laser power of 140 W, exhibits a large fraction (~11%) of irregularly shaped LOF pores larger than 10 µm, which result from insufficient energy to fully melt and solidify the powder particles (Figure 2a). In addition, numerous micro-pores with varying sizes and shapes are also observed in sample 1. Consequently, sample 1 has the lowest relative density of 98.40% among the four samples (Table 2). With an increase in laser power to 180 W, sample 2 shows a reduction in all defects (Figure 2b) and an increase in relative density, reaching 99.83%. Predominantly spherical micro-pores (~98%), which are formed due to gas entrapment, are observed, indicating that the laser energy density is sufficient to achieve proper melting and fusion of powder particles. However, with further increasing the laser power to 220 W and 260 W (samples 3 and 4) the fraction of large pores increases and the relative density follows a downward trend (Figure 2c,d). It is important to note that a large fraction of micro-pores in samples 3 and 4 (Figure 2c,d) are non-spherical, which are more detrimental as they can act as crack initiation sites. Pores with equivalent diameters higher than 10 μm were found in samples processed at (1) low laser power due to insufficient energy input, and (2) high laser power due to keyholing and spatter. The intermediate laser power of 180 W was optimal for the fabrication of nearly fully dense IN718 samples with small spherical gaseous pores.

3.2. Melt Pools

The melt pools of notched IN718 samples were characterized by using digital optical microscope after polishing and etching. The width and depth of the melt pools show significant differences across four samples printed at different laser powers (Figure 3). The average melt pool width and depth values measured using ImageJ software are listed in

Table 3. Melt pool dimensions extracted by ImageJ software for the fabricated samples.

Sample No.	Width, X (µm)	Depth, Z (µm)	Depth to Half Width Ratio (Z/0.5X)
1	138 ± 16	61 ± 11	0.88
2	156 ± 14	66 ± 8	0.85
3	192 ± 28	91 ± 17	0.95
4	192 ± 29	135 ± 21	1.41

. Only statistically significant values confirmed using the ANOVA method are presented from multiple measurements. Sample 1, which was made at the lowest laser power of 140 W, displays narrow and shallow melt pools. The melt pools become wider and deeper with increasing laser power. The increase in melt pool depth is larger compared to the width. The average melt pool width increases from 138 ± 16 μm at 140 W laser power to 192 ± 29 μm at 260 W laser power. The corresponding melt pool depth increases from 61 ± 11 μm to 135 ± 21 μm. Higher laser power or energy density results in more heat penetration into the material leading to the formation of deeper and wider molten pools. Additionally, the melt pools become less uniform at higher laser power, forming a combination of deep and shallow melt pools (Figure 3c,d). Non-uniform melt pools at higher laser power form because of unstable fluid dynamics and potential keyholing effects. The large depth-to-half-width ratio of melt pools is an indicator of a pronounced keyholing phenomenon [33].

3.3. Microstructure

Figure 4 shows the Scanning Electron Microscopy (SEM) images revealing the grain morphology near the notch in IN718 specimens. The grain boundaries are distinguishable due to variation in the orientation of sub-grain features. Samples 2, 3, and 4 exhibit a higher fraction of coarse columnar grains oriented parallel to the build direction. In contrast, sample 1 displays randomly oriented finer grains. The change in grain size (

Table 4. The average grain size, PDAS values, and the estimated cooling rates.

Sample No.	Grain Size (µm)	PDAS (µm)	Cooling Rate, 106 (K/s)
1	22 ± 4	0.46	6.21
2	32 ± 3	0.64	2.22
3	36 ± 5	0.68	1.89
4	38 ± 6	0.78	1.23

) and morphology with laser power are attributed to variation in thermal gradient and cooling rate. The rapid cooling process at lower laser power creates numerous nucleation sites for new grains, leading to a high density of small grains with random crystallographic orientations. Samples printed at higher laser powers experience a stronger thermal gradient, facilitating epitaxial growth in the direction of heat flow. The grain size and morphology showed no significant location dependence within the samples.

The rapid heating and cooling cycles in L-PBF lead to the formation of cellular or dendritic structures within the grains due to elemental segregation. Figure 5 compares the sub-grain microstructure for four IN718 samples. A mixture of cells and dendrites is observed in all samples where the cell wall thickness increases with increasing laser power. Samples 1 and 4 show more columnar dendrites, whereas samples 2 and 3 show interconnected cellular microstructure. The EDS analysis confirmed Nb enrichment in the cell and dendrite walls which indicates the formation of γ′/γ″ phases, as reported in the literature [34]. It is well established in L-PBF that the solute segregation and the sub-grain microstructure are controlled by the thermal gradient and the cooling rate [15,17,18]. Therefore, microstructural features such as primary dendritic arm spacing (PDAS) can be used to estimate the local cooling rates. The PDAS values were measured using the following area method reported in previous studies [5,35]:

$$PDAS={\displaystyle \frac{1}{M}}{\left({\displaystyle \frac{A}{N}}\right)}^{{\displaystyle \frac{1}{2}}}$$

(12)

where M is the magnification in the SEM micrograph, A is the selected area, and N is the number of sub-grain structures in the area. PDAS values were measured at multiple locations and the average values are listed in Table 4. The average PDAS increases from 0.46 µm in sample 1 to 0.78 µm in sample 4 with increasing laser power. The large PDAS values at higher laser power stem from longer solidification times or slow cooling rates. The cooling rate (q) can be estimated from PDAS using the following semi-empirical relationship [5,36,37]:

$$PDAS=80·q^{-0.3}$$

(13)

The cooling rate decreases by a factor of five with increasing laser power from 140 W to 260 W (Table 4). Higher laser power input creates larger melt pools leading to slower heat dissipation because of increased thermal mass. Thus, the observed microstructural variations in PDAS and grains in IN718 SENB specimens can be explained by the change in cooling rate with laser power. These results confirm the inverse correlation between cooling rate and laser energy density, which is consistent with previous studies [38].

3.4. Fracture Behavior

The bending stress–strain values calculated according to Equations (2) and (3) for the SENB specimens are plotted in Figure 6. The crack initiation points were determined based on the detection of a crack in the recorded images. All samples show elastic and plastic deformation before crack formation. During crack propagation, the bending stress decreased to zero, although complete fracture was not observed for any sample. There is significant variation in the peak bending stress, the total strain, and the crack behavior among L-PBF samples printed at different laser powers. Sample 2 exhibits the highest peak stress of 503 MPa and the maximum bending strain of about 26%. In contrast, sample 4 shows the lowest peak stress of 414 MPa, whilst sample 1 exhibits the lowest total strain of about 22%. Furthermore, the crack initiation occurs at lower strains in samples 1 and 4 compared to samples 2 and 3. There is a clear difference between the bending stress–strain curves for sample 1 and the other three samples during the crack propagation stage. Stress drops rapidly for sample 1, whereas significant resistance to crack propagation is observed for the other three samples (Figure 6). These results align well with the porosity and microstructural observations. Samples 1 and 4 contain a higher fraction of large non-spherical pores which can act as crack initiation sites. Furthermore, the columnar dendrites in samples 1 and 4 are less effective in deflecting the crack compared to the interconnected cellular microstructure in samples 2 and 3.

The crack propagation in SENB samples can be further examined by comparing the crack lengths as a function of applied displacement (Figure 7). The crack lengths were measured from images captured at constant displacements. Steady crack growth is observed at small displacements up to approximately 5 mm, after which the crack growth rate gradually decreased for all samples. The results indicate the presence of significant plastic deformation at the crack tip which slows down the crack propagation. The decrease in crack growth rate is most pronounced in sample 2 and least in sample 1. The crack extension is maximum for sample 1 at each applied displacement. Sample 2 demonstrates the highest resistance to crack propagation, while sample 1 offers the least resistance. The observed variation in crack propagation resistance strongly correlates with the pore size distribution and relative density. Sample 1, printed at 140 W, contains a large fraction of LOF pores and exhibits the lowest relative density. In contrast, sample 2, printed at 180 W, shows the highest relative density and contains only spherical micro-pores. Irregularly shaped LOF pores with sharp edges can promote crack initiation and propagation by inducing stress concentration and facilitating the coalescence of microcracks. At higher laser powers (samples 3 and 4), crack propagation resistance decreases slightly due to the increased presence of intermediate and large pores, primarily resulting from keyholing and spatter. Increase in grain size and columnar dendrites further decrease the crack propagation resistance in specimens printed at higher laser power. These findings underscore the importance of optimizing the L-PBF parameters and its microstructure to minimize defect formation and improve crack resistance.

The fracture behavior of SENB specimens is typically quantified using J-R curves, which are plotted as a function of crack extension. The strain energy release rate, J, was calculated using Equations (4)–(11), and the crack length was measured from the images. As shown in Figure 8, J increases rapidly with small crack extensions, indicating crack blunting due to plastic deformation. The slope of the J-R curves decreases as the crack length increases, indicating a transition to stable ductile tearing. The curves for samples 2–4 are steeper, indicating a higher resistance to further crack extension compared to sample 1. The crack behavior can be described based on the pore size and shape distributions presented in Figure 2. Sample 2, which has predominantly spherical micro-pores, exhibits the highest J values at every crack length. Samples 3 and 4 containing a small fraction of non-spherical keyhole pores exhibit slightly reduced J values. Similarly, the transition from crack blunting to stable tearing is also influenced by porosity, with sample 2 showing the highest J value of 493 kJ/m². In contrast, sample 1, which contains large lack-of-fusion pores, exhibits the lowest J values across the entire range.

Figure 9 summarizes the effect of laser power on the porosity and fracture behavior of the IN718 SENB specimens. The error bars in J_max correspond to the maximum error of ±3.25% possible due to an error in crack length measurement from camera images. The error bars in strain and peak stress values were determined based on the resolution of a mechanical testing machine. As the laser power increased, the mode of porosity transitioned from lack-of-fusion to keyhole. The pore size distribution varied with laser power as larger pores formed at both low and high laser powers, while only small gaseous micro-pores were observed at intermediate laser power. The micro-pore regime, characterized by the lowest porosity and best mechanical performance, was achieved at 180 W under the current L-PBF conditions. Despite higher relative density values compared to the sample printed at lower laser power, the peak bending stress decreased more rapidly with increasing laser power. This difference can be attributed to the coarsening of grains and sub-grain dendrites at higher laser power. Both increasing and decreasing laser power from the optimal value resulted in a reduction in the crack initiation strain. The crack initiated at a lower strain in the sample printed at a higher laser power. However, the maximum strain energy release rate (J_max) showed the largest drop (~30%) when the laser power was decreased. In contrast, J_max decreased by only 2–6% when the laser power was increased beyond the optimal value. These trends suggest that large pores resulting from lack-of-fusion are more detrimental to crack propagation resistance.

The failure of notched specimens due to bending contains two regions: the flat fracture surface (plane strain conditions) and outer shear lips (plane stress conditions). The size of the shear lips is an indicator of the ductility and fracture toughness of the specimens. Shear lips form as the crack propagates at an angle to the tensile stress direction near the specimen’s surfaces, resulting in slanted fracture surfaces. This angled propagation is a result of the material yielding and undergoing plastic deformation under the combined effect of tensile and shear stresses, demonstrating the material’s ductility. In the context of flexural strength, the presence and characteristics of shear lips are indicators of the material’s ability to absorb energy and deform plastically before failure. The shear lips of four specimens were compared (Figure 10). The shear lip heights for samples 1–4 were 0.69 mm, 1.15 mm, 1.01 mm, 0.91 mm, respectively. The large shear lips for samples 2 and 3 indicate higher ductility, which is consistent with the J-R curves.

4. Conclusions

In this study, we investigated the microstructure and fracture behavior of single-edge notched bend IN718 specimens fabricated by laser powder bed fusion using different laser powers. The results were analyzed to assess the effects of porosity, density, and grain structure on crack initiation and propagation. Relative density increased with laser power, reaching a maximum at 180 W before decreasing again. The density values exhibited a strong correlation with the observed porosity. The cooling rate decreased with increasing laser power, leading to coarser columnar grains and larger primary dendritic arm spacing. The presence of an upward-facing notch parallel to the build direction had no significant effect on microstructure and porosity. The sample printed at 180 W, with spherical micro-pores, exhibited the highest bending stress and strain. Lower laser power negatively affected ductility, while excessive laser power reduced strength. The J-R curves revealed that resistance to both crack initiation and propagation decreases as the laser power deviates from the optimal value. This effect is more pronounced in samples printed at lower laser power than in those printed at higher laser power. The results suggest that a higher volume fraction of irregular-shaped pores formed due to lack-of-fusion at lower laser power are more detrimental to fracture toughness than the keyhole pores formed at higher laser power. The micro-pores formed at intermediate laser power have the least negative effect on the fracture toughness.