This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under contract DE-AC52-07NA27344. Lawrence Livermore National Security, LLC

Main  Physical  Phenomena  in  Metal  Powder  Bed  Fusion  Q.  3  &  8  

DC Workshop Oct 7-9, 2015

LLNL-­‐PRES-­‐676079  

Lawrence Livermore National Laboratory 2

Selective laser melting process is complex: It is easy to introduce defects. Our understanding of the interplay between process parameters (laser power/speed, powder distribution/thickness…) is still lacking.

W. King, J. Mat. Proc. Tech. 2014

Defects are born at the single powder layer and may seed more defects in subsequent layers

defect and can be attributed to the latter, see Fig. 15(a). The twoneighboring defects are created at either side of a strings neckedarea by surface wave disturbance, see Fig. 15(b).

4.4. The effect of scan strategy and self-closing mechanism of defects

From Section 3.1, cross hatching strategy is more effective indefect inhibiting than zigzag strategy, and the mechanism will beanalyzed in this section. From 3D reconstructed morphologies ofsingle-layer (Defect No. 1) and multi-layer defects (Defect No. 2)it can be concluded that they have different structures as seen fromFigs. 14(a) and 16(a), respectively. The SEM image of Defect No. 2shows a discontinuity of the melt track over a previously existingsmall void, see in Fig. 16(b). The fault generated by this broken melttrack penetrates several powder layers, but has limited length andwidth within the laser scan surface. It implies that surface faults or

morphology (roughness) of a previous consolidated layer is impor-tant for the initiation of multi-layer defects. A concave area (or pre-vious void) may accumulate a locally thicker powder layer thatsometimes become thicker than the laser penetrating depth. Asthe laser beam is not sufficient to reach the bottom molten metalflow over and encapsulate powder particles (and previous defect).This is in accordance with our observations where unmelt particlesare found in the cavities of defects, see the cross sectional images inFig. 16(c–e). Actually, the structure of Defect No. 2 can be schemat-ically described as an octahedron with powder particles trappedwithin this geometry, as shown in Fig. 17(a). Considering the spec-ified laser scan strategy, where laser scan direction rotates by 67!for every single layer, the 3D structure of Defect No. 2 should be atwisted octahedron, which is seen in Fig. 17(b). Cross sectionalimages (at different depth) are present in Fig. 17(c) and (d). Thedefect direction in these two layers at different depth has an angle

Fig. 14. (a) Reconstructed 3D morphology of Defect No. 1. (b) SEM image of the same kind of defect at the top surface. (c) Cross sectional images viewed from front, (d) sideand (e) top surfaces. Some powder particles trapped in the cavity can be observed and verified, red arrow. (For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.)

Fig. 15. (a) Defects No. 4 had two paired sub-defects, which distributed in both sides of a single laser track. (b) These defects could also be observed in SEM images. The mainreason is molten track instability causing surface oscillation.

12 X. Zhou et al. / Acta Materialia 98 (2015) 1–16

gas. The gas flow, relative to the direction of beam moving, iscoflow or counterflow. In this circumstance, the melt–gas interfaceof melt track will be disturbed by gas flow shear stress, tempera-ture gradient and surface tension gradient, which then inducesthe fluid oscillation and surface waves [37–39].

As shown in Fig. 12(a), the melt track flow can be described bythe boundary layer model of falling melt films with angle h [40,41],the incline angle h represents the surface roughness in a smallregion. The simplified governing equation and boundary condi-tions are [41]:

@u@tþ u

@u@xþ t @u

@y¼ # 1

q@p@xþ g sin hþ m @

2u@y2 ð3Þ

1q@p@y¼ #g sin h ð4Þ

@u@xþ @t@y¼ 0 ð5Þ

y ¼ 0;u ¼ t ¼ 0 ð6Þ

Fig. 10. SEM image and surface topography with different hatch space, 200 lm (a), 60 lm (b). Other parameter: laser power 200 W, laser speed 700 mm/s.

Fig. 11. SEM image and surface topography with different laser speed, 1167 mm/s (a), 467 mm/s (b). Other parameter: laser power 200 W, hatch space 125 lm.

X. Zhou et al. / Acta Materialia 98 (2015) 1–16 9

X. Zhou, Act. Mat. 2015

Pores in Keyhole-mode Rough surface, bad wetting Incomplete melting Low density part, bad quality

Page 13 of 26

Accep

ted M

anus

cript

13

Figure 8 Hydrogen pore density of SLM samples depending on the scan speed; ds,2 = 1 mm,

PL = 910 W

3.3 Influence of the laser beam diameter on the hydrogen pore density

For SLM samples built up with the two different beam diameters – ds,1 = 0.3 mm and

ds,2 = 1 mm – the latter show an approx. 10-fold higher pore density (Figure 9).

Figure 9 Hydrogen pores in AlSi10Mg SLM samples built up with dried powder (Theat = 90 °C) and a laser beam diameter of ds,1 = 0.3 mm (vs = 2250 mm/s��ȡP = 0.4%) (left)

and ds,2 = 1 mm (vs = 250 mm/s��ȡP = 9.2%) (right)

The temperature distribution of the melting process in SLM was calculated with an FEM

simulation. The results are shown in Figure 10. When the laser beam diameter ds,1 is used, a

higher local maximum temperature Tmax,1 § 3200 °C is reached. This is caused by the higher

C. Weingarten J. Mat. Proc. Tech. 2015

Lawrence Livermore National Laboratory 3

200W 1500mm/s 1000x300x50µm3

Laser power Laser scan speed

Substrate dimensions

What is the driving physics? How do pores form and evolve? Guidance for better parameter choice?

Physics Pore Generation

Recommendations

Lawrence Livermore National Laboratory 4

power W and full radius R, which is approximately equal to thefull width of the beam at half maximum !FWHM".

6 Results and DiscussionThe experimental results are obtained with SLM machine

“Phenix Systems” employing 1.075 !m laser of 50 W nominalpower and 70 !m nominal beam diameter. Taking into accountabsorption by the optical system, the maximum laser power isestimated as 45 W. The pumping power of the laser can be re-duced producing laser radiation of reduced power, which is mea-sured in fractions of the maximum power.

To validate the model, heat transfer in the substrate withoutpowder is modeled first. Figure 6 compares experimental crosssection micrographs !on the bottom" with the calculated projec-tions of the melt pool on the plane perpendicular to the scanningdirection !on the top". The absorbed fraction of laser energy 1−"=0.3 used in calculations is expected to be the most uncertainmodel parameter because it is estimated from the reflectivity "measured at the room temperature, while the absorptivity of met-als generally tends to increase with temperature. Another ques-tionable point of the model is the negligence of the Marangoniconvection in the melt pool. Both the increase in absorptivity withtemperature and the convection enlarge the melt pool. The ob-

served agreement between the experiment and the model shown inFig. 6 indicates that the absorptivity does not significantly in-crease with temperature and the contribution of the Marangoniconvection to heat transfer is low.

The diameter of the laser beam is evaluated by its traces atreduced laser power. The full diameter estimated from such ex-periments is around 100 !m. However, a slow dependence on thelaser power is possible. Therefore, two values of FWHM diameterof 50 !m and 60 !m were tested in the calculations. The differ-ence was not principal. However, the shape of the experimentalmelt pool is reproduced better by 50 !m FWHM at 50% power!22.5 W" and by 60 !m FWHM at 100% power !45 W".

Figure 7 presents the typical modeling results for a 50 !mpowder layer on a substrate. The isotherms in diagrams !c" and !d"show the melting point Tm=1700 K. The level in diagram !e" isthe powder-substrate interface. The incident laser power of 30 Wchosen for this calculation is less than the maximum of 45 W totake into account heat losses by evaporation. According to thepreliminary calculations, the beam of 45 W incident power con-siderably overheats the powder layer above the boiling point be-cause of low thermal conductivity of powder. Indeed, while noevaporation is experimentally observed from the noncovered sub-strate, the traces of evaporation as ejection of droplets are noticed

(a) (b) (c)

-40 -20 0 20 40y (µm)

-40-20

z(µm)

22.5 W 15 cm/s50 µm FWHM

-40 -20 0 20 40y (µm)

-40-20

z(µm)

22.5 W 3 cm/s50 µm FWHM

-40 -20 0 20 40y (µm)

-40-20

z(µm)

45 W 30 cm/s60 µm FWHM

50% power 15 cm/s 50% power 3 cm/s 100% power 30 cm/s

Fig. 6 Cross-sections of laser tracks on the stainless steel substrate with-out powder: calculated phase diagrams „upper row… and experimental mi-crographs „lower row…

(a) (b) (d)

(c) (e)

Fig. 7 Laser beam scanning of a 50 !m powder layer with optical thickness "=2 at the incident laser power of 30W, the laser beam FWHM of 60 !m, and the scanning velocity of 20 cm/s. Distributions at the symmetry planey=0: „a… normal component of net laser radiation flux density Q, „b… volumetric heat source due to absorption oflaser radiation U, „c… temperature T, „d… temperature T at the top powder surface z=0, and „e… melt pool shape.

072101-6 / Vol. 131, JULY 2009 Transactions of the ASME

Downloaded 01 Jun 2009 to 193.50.200.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

power W and full radius R, which is approximately equal to thefull width of the beam at half maximum !FWHM".

6 Results and DiscussionThe experimental results are obtained with SLM machine

“Phenix Systems” employing 1.075 !m laser of 50 W nominalpower and 70 !m nominal beam diameter. Taking into accountabsorption by the optical system, the maximum laser power isestimated as 45 W. The pumping power of the laser can be re-duced producing laser radiation of reduced power, which is mea-sured in fractions of the maximum power.

To validate the model, heat transfer in the substrate withoutpowder is modeled first. Figure 6 compares experimental crosssection micrographs !on the bottom" with the calculated projec-tions of the melt pool on the plane perpendicular to the scanningdirection !on the top". The absorbed fraction of laser energy 1−"=0.3 used in calculations is expected to be the most uncertainmodel parameter because it is estimated from the reflectivity "measured at the room temperature, while the absorptivity of met-als generally tends to increase with temperature. Another ques-tionable point of the model is the negligence of the Marangoniconvection in the melt pool. Both the increase in absorptivity withtemperature and the convection enlarge the melt pool. The ob-

served agreement between the experiment and the model shown inFig. 6 indicates that the absorptivity does not significantly in-crease with temperature and the contribution of the Marangoniconvection to heat transfer is low.

The diameter of the laser beam is evaluated by its traces atreduced laser power. The full diameter estimated from such ex-periments is around 100 !m. However, a slow dependence on thelaser power is possible. Therefore, two values of FWHM diameterof 50 !m and 60 !m were tested in the calculations. The differ-ence was not principal. However, the shape of the experimentalmelt pool is reproduced better by 50 !m FWHM at 50% power!22.5 W" and by 60 !m FWHM at 100% power !45 W".

Figure 7 presents the typical modeling results for a 50 !mpowder layer on a substrate. The isotherms in diagrams !c" and !d"show the melting point Tm=1700 K. The level in diagram !e" isthe powder-substrate interface. The incident laser power of 30 Wchosen for this calculation is less than the maximum of 45 W totake into account heat losses by evaporation. According to thepreliminary calculations, the beam of 45 W incident power con-siderably overheats the powder layer above the boiling point be-cause of low thermal conductivity of powder. Indeed, while noevaporation is experimentally observed from the noncovered sub-strate, the traces of evaporation as ejection of droplets are noticed

(a) (b) (c)

-40 -20 0 20 40y (µm)

-40-20

z(µm)

22.5 W 15 cm/s50 µm FWHM

-40 -20 0 20 40y (µm)

-40-20

z(µm)

22.5 W 3 cm/s50 µm FWHM

-40 -20 0 20 40y (µm)

-40-20

z(µm)

45 W 30 cm/s60 µm FWHM

50% power 15 cm/s 50% power 3 cm/s 100% power 30 cm/s

Fig. 6 Cross-sections of laser tracks on the stainless steel substrate with-out powder: calculated phase diagrams „upper row… and experimental mi-crographs „lower row…

(a) (b) (d)

(c) (e)

Fig. 7 Laser beam scanning of a 50 !m powder layer with optical thickness "=2 at the incident laser power of 30W, the laser beam FWHM of 60 !m, and the scanning velocity of 20 cm/s. Distributions at the symmetry planey=0: „a… normal component of net laser radiation flux density Q, „b… volumetric heat source due to absorption oflaser radiation U, „c… temperature T, „d… temperature T at the top powder surface z=0, and „e… melt pool shape.

072101-6 / Vol. 131, JULY 2009 Transactions of the ASME

Downloaded 01 Jun 2009 to 193.50.200.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

power W and full radius R, which is approximately equal to thefull width of the beam at half maximum !FWHM".

6 Results and DiscussionThe experimental results are obtained with SLM machine

“Phenix Systems” employing 1.075 !m laser of 50 W nominalpower and 70 !m nominal beam diameter. Taking into accountabsorption by the optical system, the maximum laser power isestimated as 45 W. The pumping power of the laser can be re-duced producing laser radiation of reduced power, which is mea-sured in fractions of the maximum power.

To validate the model, heat transfer in the substrate withoutpowder is modeled first. Figure 6 compares experimental crosssection micrographs !on the bottom" with the calculated projec-tions of the melt pool on the plane perpendicular to the scanningdirection !on the top". The absorbed fraction of laser energy 1−"=0.3 used in calculations is expected to be the most uncertainmodel parameter because it is estimated from the reflectivity "measured at the room temperature, while the absorptivity of met-als generally tends to increase with temperature. Another ques-tionable point of the model is the negligence of the Marangoniconvection in the melt pool. Both the increase in absorptivity withtemperature and the convection enlarge the melt pool. The ob-

served agreement between the experiment and the model shown inFig. 6 indicates that the absorptivity does not significantly in-crease with temperature and the contribution of the Marangoniconvection to heat transfer is low.

The diameter of the laser beam is evaluated by its traces atreduced laser power. The full diameter estimated from such ex-periments is around 100 !m. However, a slow dependence on thelaser power is possible. Therefore, two values of FWHM diameterof 50 !m and 60 !m were tested in the calculations. The differ-ence was not principal. However, the shape of the experimentalmelt pool is reproduced better by 50 !m FWHM at 50% power!22.5 W" and by 60 !m FWHM at 100% power !45 W".

Figure 7 presents the typical modeling results for a 50 !mpowder layer on a substrate. The isotherms in diagrams !c" and !d"show the melting point Tm=1700 K. The level in diagram !e" isthe powder-substrate interface. The incident laser power of 30 Wchosen for this calculation is less than the maximum of 45 W totake into account heat losses by evaporation. According to thepreliminary calculations, the beam of 45 W incident power con-siderably overheats the powder layer above the boiling point be-cause of low thermal conductivity of powder. Indeed, while noevaporation is experimentally observed from the noncovered sub-strate, the traces of evaporation as ejection of droplets are noticed

(a) (b) (c)

-40 -20 0 20 40y (µm)

-40-20

z(µm)

22.5 W 15 cm/s50 µm FWHM

-40 -20 0 20 40y (µm)

-40-20

z(µm)

22.5 W 3 cm/s50 µm FWHM

-40 -20 0 20 40y (µm)

-40-20

z(µm)

45 W 30 cm/s60 µm FWHM

50% power 15 cm/s 50% power 3 cm/s 100% power 30 cm/s

Fig. 6 Cross-sections of laser tracks on the stainless steel substrate with-out powder: calculated phase diagrams „upper row… and experimental mi-crographs „lower row…

(a) (b) (d)

(c) (e)

Fig. 7 Laser beam scanning of a 50 !m powder layer with optical thickness "=2 at the incident laser power of 30W, the laser beam FWHM of 60 !m, and the scanning velocity of 20 cm/s. Distributions at the symmetry planey=0: „a… normal component of net laser radiation flux density Q, „b… volumetric heat source due to absorption oflaser radiation U, „c… temperature T, „d… temperature T at the top powder surface z=0, and „e… melt pool shape.

072101-6 / Vol. 131, JULY 2009 Transactions of the ASME

Downloaded 01 Jun 2009 to 193.50.200.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

91 w 380 mm/s

! "(!

7-ACKOWLEDGEMENTS We would like to thank Al Nichols III for his help with the thermal package in ALE3D. Also, we acknowledge fruitful discussions with Sasha Rubenshick, Wayne King and Bob Ferencz. This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. This work was funded by the Laboratory Directed Research and Development Program at LLNL under project tracking code 13-SI-002.

!"#"!"$%"&'S#?"#TF!=/64!b@*.267*5!a6,7C25!B6/!R;D*7C0;!Q*7,B*C.,/278bZ!:::F7*CB*4F6/8!RFQF!X,-07C12K9!OF!HF!S7F;FTF!bI04<0/*.,/0!;0<07;07.!()?W74!5*+0/!*-+6/<.267!-E!0782700/278!8/*;0!*5,427,49!.2.*72,4!*7;!+.005!*556E!+,/B*C0+bF!RFcF!d,+*/6D9!GF!OF!S#?"?TF!bQ6;05278!.10!27.0/*C.267!6B!5*+0/!/*;2*.267!:2.1!<6:;0/!-0;!*.!+050C.2D0!5*+0/!405.278bF!%&'()*(+%,-*./)0!1+"9!$)"W$>%F!RFcF!d,+*/6D9!GF!eF!S#??>TF!bQ6;05!6B!X*;2*.267!*7;!M0*.!I/*7+B0/!27!P*+0/WV6:;0/!G7.0/*C.267!f670!*.!O050C.2D0!P*+0/!Q05.278bF!2-3,405+-6+7.08+9,04(6.,!1+:;:9!?(#"?"F!RFcF!d,+*/6D9!gFWVF!hF!S#??&TF!bQ6;055278!6B!/*;2*.267!./*7+B0/!27!40.*552C!<6:;0/+!*.!5*+0/!./0*.407.bF!<48.,408)-405+2-3,405+-6+7.08+04/+=0((+9,04(6.,!1+$>9!$%#$W$%$%F!^/*770/9!QF!=F!S#?"?TF!G7D0+.28*.267+!67!/0+2;,*5!+./0++0+!*7;!;0B6/4*.267+!27!+050C.2D0!5*+0/!405.278F!%,-/3*8)-4+?4@)4..,)4@1+A.(.0,*&+04/+B.C.5-DE.48!1+$9!$&W%&F!a*/6527!h6/70/9!RF!^F!S#?"$TF!b=,7;*407.*5!C67+652;*.267!40C1*72+4+!;,/278!+050C.2D0!-0*4!405.278!6B!<6:;0/+bF!=-/.5)4@+04/+F)E3508)-4+)4+=08.,)05(+04/+F*).4*.+04/+?4@)4..,)4@!1+G:9!?)&?""F!a*/6527!h6/70/9!iF!RF!S#?""TF!bQ0+6+C6<2C!+24,5*.267!6B!+050C.2D0!-0*4!405.278!</6C0++0+bF!2-3,405+-6+=08.,)05(+%,-*.(()4@+9.*&4-5-@'!1+G::9!>()W>)(F!a1*7;/*+0K1*/9!OF!S">'"TF!H7'/,-/'40E)*(+04/+7'/,-E0@4.8)*+F80I)5)8'HJ!NAB6/;Z!a5*/07;67F!j*+9!OF!S#??$TF!bV1E+2C*5!*+<0C.+!6B!</6C0++!C67./65!27!+050C.2D0!5*+0/!+27.0/278!6B!40.*5+bF!K/C04*./+?4@)4..,)4@+=08.,)05(!1+"9!(?"W(""F!=/20;4*79!dF!S#?""9!@6DF!"'TF!bV*.2C50V*CK!k+0/+l!Q*7,*5bF!LLMLNF=N$">#;:!F!P2D046/09!aR9!kORZ!P*:/07C0!P2D0/46/0!@*.267*5!P*-6/*.6/EF!d,9!jF!QF!S#?"#TF!bP*+0/!*;;2.2D0!4*7,B*C.,/278!6B!40.*552C!C64<6707.+Z!4*.0/2*5+9!</6C0++0+!*7;!40C1*72+4+bF!<48.,408)-405+=08.,)05(+A.C).O!1+"!9!"$$W"'%F!d,/.50/9!=FWgF!hFWMF!S#?"$TF!bO24,5*.267!6B!5*+0/!-0*4!405.278!6B!+.005!<6:;0/+!,+278!.10!.1/00W;2407+267*5!D65,40!6B!B5,2;!40.16;bF!%&'()*(+%,-*./)0!1+$:9!)(%W)(>F!h/,.19!gFWVF!PF!S#??(TF!HP-4(-5)/08)-4+D&.4-E.40+)4+50(.,+04/+D-O/.,NI./+I0(./+50'.,./+E04360*83,)4@H!Sc65F!&'TF!aGXV!R77*5+!W!Q*7,B*C.,/278!.0C176568EF!P0D2C19!cF!^F!S">'#TF!H%&'()*-*&.E)*05+7'/,-/'40E)*(HJ!V/07.2C0!M*55F!QF!X64-6,.+!*7;!PF!=/6E079!RF!dF!S#??&TF!bV16.6<E/6050C./2C!40*+,/0407.!6B!.10/4*5!C67;,C.2D2.E!6B!40.*552C!<6:;0/+bF!2-3,405+-6+KDD5)./+%&'()*(!1+Q!9!?#%>?&F!QCa*55079!aF!S#?"#9!],5E!")TF!bRPi$jZ!R/-2./*/E!P*8/*780!i,50/2*7!I1/00W!*7;!I:6!j2407+267*5!Q6;05278!*7;!O24,5*2.67!a*<*-252.EbF!LLMLNKRFN"S"G:G!F!P2D0/46/09!aR9!kORZ!P*:/07C0!P2D046/0!@*.267*5!P*-6/*.6/EF!

We match the bare plate melt pool dimensions well, although we consider an average value for the material absorptivity.

LLNL experiment (uncertainty +/-5µm)

Khairallah, S.A., Anderson, A., 2014. “Mesoscopic Simulation Model of Selective Laser Melting of Stainless Steel Powder”, Journal of Materials Processing Technology.

Lawrence Livermore National Laboratory 5

Height 26µm Width 70µm Depth 20µm

Experiment (uncertainty +/-5µm)

! >!

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

pieces. This phenomenon is described in the SLM literature (Kruth, 2007) and is closely related to Plateau-Rayleigh instability of a long cylindrical fluid jet breaking up into droplets (Levich, 1962). The physics governing this behavior is due to surface tension driven surface energy minimization. The surface energy of a cylinder is not at a minimum. As the bead breaks up into droplets, the surface energy decreases since it approaches its optimal shape, which is a sphere. This instability is caused by unstable perturbations with wavelengths bigger than the cylindrical circumference. The fastest growing perturbation mode for a low viscous fluid has a wavelength close to !~9.02R where R is the radius of the cylinder. If we consider the most abundant particle sizes and take R= 13.5µm, then we get ! ~125µm. If we consider a larger radius R=35µm, corresponding to the melt width (see Fig. 6-b), then we get !~315µm. Gusarov et-al (A.V. Gusarov I. S., 2010) solved for the case of a segmental cylinder of a liquid on a solid substrate. The liquid cylinder is attached to the solid by a contact band with a fixed width. For a cylinder of radius R=35µm and a contact width specified by an angle of 3"/4, we get !~190µm. The simulation in =28,/0!&-c) gives !~150µm which is bracketed by these estimates. Of course, the analogy with a perfect liquid cylinder or a segmental

The simulation can predict well the main characteristics of the laser powder track

Simulation and experiment showing Plateau-Rayleigh instability

Melt pool profile

H 26um D 20um W 72um

H

W

D

Khairallah, S.A., Anderson, A., 2014. “Mesoscopic Simulation Model of Selective Laser Melting of Stainless Steel Powder”, Journal of Materials Processing Technology.

Lawrence Livermore National Laboratory 6

Depth and width calculations improved due to added physics

Melt pool profile

If Absorptivity is 0.4, then Depth is 60um Width is 80um

If Absorptivity is 0.3, then

Depth is 50um Width is 74um

Lawrence Livermore National Laboratory 7

I would like to apologize for taking out some slides. The reason is these were submitted for publication in a journal.