- Email: [email protected]
Electron beam welding of AlSi10Mg workpieces produced by selected laser melting additive manufacturing technology
Additive Manufacturing 8 (2015) 63–70
Contents lists available at ScienceDirect
Additive Manufacturing journal homepage: www.elsevier.com/locate/addma
Electron beam welding of AlSi10Mg workpieces produced by selected laser melting additive manufacturing technology Moshe Nahmany a,∗ , Idan Rosenthal b , Isgav Benishti a , Nachum Frage b , Adin Stern b a b
Materials Department, Nuclear Research Center, Negev, P.O.B. 9001, Beer Sheva 8410900, Israel Materials Engineering Department, Ben Gurion University of the Negev, P.O.B. 653, Beer Sheva 8410501, Israel
a r t i c l e
i n f o
Article history: Received 11 February 2015 Received in revised form 20 June 2015 Accepted 24 August 2015 Available online 28 August 2015 Keywords: AM SLM AlSi10Mg alloy EBW
a b s t r a c t Electron beam welding (EBW) is a high-density energy (low heat input) welding technique, resulting in a narrow heat affected zone (HAZ), causing minimal metallurgical changes in the workpieces. The present research work investigates EB autogenous welded AlSi10Mg samples, produced by the selective laser melting (SLM) additive manufacturing (AM) method, with emphasis on the characterization of the joint’s macro- and microstructure. When comparing the EB welded AM parts to the EB welded cast samples two main differences were observed: weld metal porosity and a negligible HAZ in the AM joints and low porosity level but substantial HAZ in the welded cast parts. These preliminary results show for the first time the feasibility of the EBW technique on AM-SLM specimens. © 2015 Elsevier B.V. All rights reserved.
1. Introduction The concept of advanced manufacturing technology has been viewed as the utilization of computer and numerical-based processing of components. One of these technologies includes three-dimensional printing and related layer-by-layer fabrication, known as additive manufacturing (AM) [1–4]. AM technology builds a solid, often geometrically complex object from a series of layers, each one “printed” on top of the previous one. AM “printers” use a virtual, mathematical model to construct a physical artifact that can be manipulated on the computer screen. Manufacturing components (by powder-bed process, for example) offers a high geometrical flexibility and accuracy enabling the construction of components with very complex external and internal shapes (e.g. embedded cooling channels). The AlSi10Mg alloy, which was investigated in the present study, is a widely used casting alloy with near-eutectic composition, displaying good mechanical properties and weldability. The addition of small amounts of Mg (0.3–0.5 wt.%) provides precipitation hardening of the aluminum alloy by forming Mg2 Si precipitates upon natural or artificial aging treatments [5]. Typical hardness values of AlSi10Mg fabricated by AM-SLM are reported [6] as 94 ± 5 HV and up to 120 ± 5 HV (reported [7] as 120 ± 5 HBW). The engineering design of many structural components requires the manufacturing of complex shaped modules
which is quite expensive. An attractive way to achieve this goal is to fabricate the composite modules by joining two or more AM processed components. A reliable joining technology provides enhanced design flexibility and engineering solutions for AM modules that are impossible to produce due to size limitations of the additive manufacturing equipment. Power beam welding processes characterized by energy densities up to 106 W/cm2 [8], are ideally suited for producing high quality welds, both minimizing the weld heat input and reducing the weld-induced distortion of the components. The EBW process (Fig. 1a) has been chosen in this research because it is well-positioned to provide metal industries with high quality and reliable joining solutions. Moreover, the solidification rates for this process (105 –106 ◦ C/s) are significantly higher than those for conventional arc welding processes (102 –103 ◦ C/s), creating the fine microstructure of the weld metal. There are four major types of weld defects in EBW of aluminum: porosity, cracking, inclusions and loss of alloying elements [9]. The porosity is usually located in the keyhole path, whereas gas pores are homogeneously distributed with slight enrichment at the melting line. When the welding mode switches from the keyhole to conduction mode, the level of porosity is significantly reduced, because there is enough time for gases to escape from a shallower weld pool. 2. Goals of research
∗ Corresponding author. E-mail address: [email protected] (M. Nahmany). http://dx.doi.org/10.1016/j.addma.2015.08.002 2214-8604/© 2015 Elsevier B.V. All rights reserved.
• Demonstrate the feasibility of using EBW technology to weld SLM AlSi10Mg components.
64
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
Fig. 1. (a) Schematic illustration of the EBW process, presenting a deep and narrow weld metal (WM). The illustration displays the metallographic cross sections: T-CS for transverse, P-CS for planar. (b) Schematic illustration of the welded specimens, presenting the build direction and the weld direction.
• Compare the microstructure of EBW joined region of SLMproduced specimens and casting-produced specimens.
3. Experimental Welding samples were machined from AM-SLM built AlSi10Mg commercial specimens, manufactured using an EOSINT-M280 system. Pre-alloyed AlSi10Mg powder with particle size distribution between 7 and 50 m has been used [10]. All specimens were built in the vertical direction. The welding samples were machined from 10 mm diameter rods and axially aligned for the welding operation. The circumferential welds were oriented perpendicularly to the building direction (Fig. 1b). For the sake of comparison, welding specimens were prepared from A356-T6 plates fabricated by casting. Autogenous electron beam welding technique was used to join both types of specimens on a Wentgate EB machine with capabilities of 60 kV acceleration voltage and a 40 mA beam current. The machine is equipped with both rotational turntable and XY-stage. The electron beam can be consistently focused (with diameter about 0.3–0.5 mm) on the specimen’s top surface for all welds. Welds were performed with electron beam power varying from 120 to 600 W using heat inputs of 21–45 J/mm. Welding parameters are presented in Table 1. Welding parameters were selected using our experience on EBW of cast aluminum and were updated according to the preliminary results of EBW of AM parts. We have tried to keep some parameters constant and vary others to clarify their effects. The joints were characterized to visualize the development of novel and unusual macrostructures and microstructural architectures resulting from EB welding of the AlSi10Mg alloy. Welds were characterized from a transverse cross-section and from a planar cross-section noted as T-CS and P-CS respectively. Metallographic samples were prepared according to the ASTM-E3 Standard. First, the samples were hot mounted, and then a rough polish was performed. Surface preparation was performed by grinding and fine polishing (up to 0.5 m) and etched in Flick’s Reagent (10% HF–90% H2 O) to expose the microstructure. Welds were examined by an Optical Microscope (OM Zeiss, Aalen, Germany), Scanning Electron Microscope (SEM JEOL, JSM-2500) and High Resolution Scanning Electron Microscope (HRSEM, JEOL 7400F). Micro-hardness tests
Table 1 Welding parameters; all welds were performed using 60 kV acceleration voltage. Last specimens, designated with “C” are cast specimens. Heat input was calculated as (beam current × beam voltage)/welding speed. No.
Travel speed (mm/min)
Beam current (mA)
Heat input/specific energy (J/mm)
1 2 3 4 5 6 7 8 9 10 11 12 C-13 C-14
1000 1000 1000 1000 750 500 484 242 484 242 1000 500 1000 1000
6 8 10 13 6 4 4 2 3 3 8 4 5.5 16
21.7 28.8 36 46.8 28.8 28.8 29.8 29.8 22.3 44.7 28.8 28.8 19.9 57.6
were conducted using a Buehler micro-hardness tester (MMT-7), with a load of 100 g. 4. Results and discussion 4.1. Base metal composition and microstructure The microstructure of the components fabricated by the AM is the end-result of pre-alloyed powder melting/solidification, with adjacent neighboring tracks responsible for partial remelting of the solidified tracks. Average concentrations (in wt.%) of the alloying elements are 11.1 Si, 0.2 Mg, ∼0.1 Fe with traces of Cr, Cu, Mn, Ti, and Zn. the concentration of the alloying elements in the base metal was determined by a wet chemical analysis. The typical microstructure of the AlSi10Mg part (Fig. 2a and b), consisting of approximately half-cylindrical solidified melt pools, reveals that most of the defects (Fig. 2c) are located at regions where the melt pools overlap, strengthening the role of the melting/solidification cycles and their effect on both microstructure and defect formation. The fast melting process of the stochastic powder bed can induce vigorous melt pool movements which sometimes lead to faults acting as starting points for defects such as channels and/or pores between layers [6,10,11]. The microstructure morphology is
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
65
Fig. 2. AM-SLM typical microstructure of the base metal [6,10]. The arrows mark defects. (a) General view of the melt pools. (b) Half-cylindrical shape of each melt pool. (c) Enlarged view showing porosity at weld pool boundaries (marked by the dashed line), some pores are marked by arrows.
known as “fish scales”. The concentrations (in wt.%) of the alloying elements for the cast A356 T6 alloy are 7.7 Si, 0.37 Mg, 0.7 Fe, 0.13 Ti with traces of Cr, Cu, Mn, and Zn; the typical microstructure is characterized by a coarse “Chinese script” eutectic morphology.
4.2. Weld Metal microanalysis In order to clarify the effect of evaporation of the melt components the EDS microanalysis was conducted. Possible dilution of alloying element (Si, Mg) in the WM was examined. Weld metal microanalysis was performed using two configurations: perpendicular to the top surface downwards and parallel to the top surface. Base metal microanalysis exhibits values (wt.%) of Si-10.7 ± 0.5 and Mg-0.9 ± 0.2, over an average of 6 arbitrarily-located measurements. These values are slightly different from the chemical analysis shown earlier, probably due to fact that the local EDS measurements are semi-quantitative only. Weld metal, perpendicular to the top surface downwards (13 continuous measurements) exhibits no trend of dilution with values of Si-10.9 ± 0.5 and Mg0.9 ± 0.2. Weld metal, parallel to the top surface (7 continuous measurements) exhibits no trend of dilution as well, showing values of Si-11.2 ± 0.7 and Mg-0.9 ± 0.2. The semi-quantitative local analysis indicates no dilution of Silicon or Magnesium in the weld metal. The A356 cast alloy after T6 treatment displays a typical “Chinese script” microstructure, which is completely different from the microstructure of AM parts (Fig. 3).
Fig. 3. Microstructure of the A356 cast alloy after T6 treatment.
66
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
Fig. 4. WM and neighboring regions (weld metal, top left corner). (a) AM specimen. (b) A356-T6 specimen, PMZ adjacent to the WM, marked with red parallel lines.
4.3. Weld metal macrostructure The welds of the A356-T6 parts were performed utilizing a beam power of 220 W (60 kV) leading to a heat input of 20 J/mm. The A356-T6 specimens (Fig. 4) exhibit a notable partially melted zone (PMZ, marked with parallel lines) adjacent to the weld metal (WM), due to the lower melting temperature of the eutectic colonies. No HAZ was detected near the PMZ. The AM welded parts show no traceable HAZ. 4.3.1. Transverse metallographic view The WM EB joints were characterized (T-CS) by penetration depth and width which are presented in Table 2. Both keyhole (Fig. 5a and b) and conduction modes (Fig. 5c and d) were observed. Moreover, a different microstructure was detected in the uppercentral region (Fig. 5d), designated as CZ-WM (central zone weld metal). The general geometry of the weld metal depends strongly
Table 2 Weld metal dimensions in relation to the heat input. No.
Weld penetration (mm)
Weld width (mm)
Heat input/specific energy (J/mm)
1 2 3 5 6 7 8 9 10
1.5 2.4 2.8 1.4 0.8 0.63 0.4 0.4 0.55
1.2 2.0 2.1 1.3 1.3 0.85 0.55 0.7 0.9
21.7 28.8 36 28.8 28.8 29.8 29.8 22.3 44.7
on the heat input and the penetration depth is mainly affected by the travel speed (Fig. 5a–d). It has to be pointed out, that the shallow penetration joint has been achieved when EBW was conducted using a relatively low weld velocity, when compared to the typical
Fig. 5. Overall metallographic (T-CS) view of EB welded AM parts. (a) High speed weld (1000 mm/min, 28.8 J/mm), keyhole mode predominant showing deep penetration and high porosity. Some pores are marked by arrows. (b) Lower heat input weld (1000 mm/min, 21.7 J/mm). (c, d) Much slower welds, conduction mode predominant showing shallow penetration (500 mm/min, 29.8 J/mm and 500 mm/min, 28.8 J/mm respectively). Horizontal line in (d) indicates the height of the planar metallography (P-CS), ∼0.2 mm below the part surface, as discussed later.
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
67
Fig. 6. Planar metallography (P-CS); (a) General view, yellow squares represent the regions of higher magnification. (b) Lower-right corner of a. (c, d) Enlargement of the lower-right corner of (b), as marked in (b). (e) BM/WM interface, indicating the absence of a HAZ.
velocity used in practice. Moreover, relatively low levels of weld metal porosity have been achieved when using low speed welds at different heat inputs. It seems that the porosity (soundness) of the weld metal mostly depends on the porosity of the AM base-metal.
4.3.2. Planar metallographic view The sample was polished about 0.2 mm below surface (see Fig. 5d) and the macrostructure is presented in Fig. 6a. Banding and rippling originated from variations in welding speed and in heat input or rather by solidification halts that can be clearly observed (Fig. 6a); these phenomena were displayed and explained in details by Kou in [8]. Fig. 6b–d shows specific regions at higher magnification as indicated in Fig. 6a and b. The morphology of the Al–Si eutectic microstructure (Fig. 6c and d) reflects heat-dissipation outwards to the BM and toward the earlier solidified WM (marked by arrows). Fig. 6e depicts the smooth continuous interface between the BM and WM, with no pores and cracks present. No visible indication of a HAZ has been identified. Several solidification modes are clearly observed in the WM TCS micrographs (Fig. 7a–d). Columnar and equiaxed dendrites and a typical Al–Si eutectic morphology can be distinguished (Fig. 7a, b and d). Adjacent the fusion line, the dendrites appear to be aligned parallel to the local heat-flow direction. A slightly modified microstructure was observed in the upper-center region of the WM, designated as Central Zone Weld Metal (CZ-WM) and
illustrated in Fig. 7c. An almost abrupt transition from the CZWM to the WM outer region has been clearly observed and the microstructure evolution is schematically illustrated in Fig. 8. The macrostructure of the CZ-WM is oriented parallel to the weld direction, with the red arrow representing heat flow backwards to the colder, previously solidified metal. The orientation of the macrostructure of the WM outer regions matches the direction of the heat flow toward the BM as indicated by the black arrows (Fig. 8).
4.4. SEM microstructural characterization Subject to the location within the WM, different microstructure morphologies have been observed by SEM examination. The weld root (Fig. 9a) exhibited a very thin, hardly observed HAZ, without pores or cracks (Fig. 9b). The typical Al–Si eutectic morphology is displayed in Fig. 9c, the solidification mode is mainly cellular, but occasionally, some side branches can be seen. The gray cellular features are primary Al decorated with white fibrous Si particles. The transition from a columnar-eutectic to an equiaxed-eutectic morphology, adjacent the fusion line, is shown in Fig. 9d (marked as CE-WM and EXE-WM respectively). The transition is probably due to the local changes in the temperature gradients and growth rates near the fusion line [8]. Slower welds exhibit a wider columnareutectic zone near the fusion line (Fig. 9e); CE-WM for slow welds
68
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
Fig. 7. Solidification modes; arrows mark heat dissipation toward the BM. (a, b) Typical microstructure adjacent to the fusion line. (c) Transition from the CZ-WM to WM outer zone. (d) slow velocity weld microstructure.
is about 40 m wide, while CE-WM width of fast welds is around 10 m. A closer look into the microstructure of the AM BM in Fig. 9d and e, reveals the presence of a much finer cellular–dendritic solidification structure with a size smaller than 1 m. Thermal gradients as high as in the order of 106 K/m can occur during SLM building [5] explaining the very small cell size observed in this AlSi10Mg base metal when compared to the WM solidification microstructure. 4.5. Effect of weld speed and current on WM geometry The main effect of the weld speed on the WM’s depth and weld area is shown in Fig. 5. The weld speed, altered together with the beam current, results in a deeper and narrower weld metal. The depth and weld area were plotted vs. weld speed (Fig. 10). Higher weld speed leads to deeper penetration and to a high fraction of porosity due to the preferred key-hole welding mode. On the other
hand, slower and lower current welds yield a shallow penetration and lead to lower porosity of the WM, due to conductive welding mode. It has to be pointed out that the samples were relatively small, thus having limited ability to absorb energy from the electron beam resulting in low penetration welds. WM’s depth as a function of weld heat input for various speeds (500, 100 mm/min) is shown in Fig. 11. For both weld speeds the welding mode is mainly keyhole and deeper penetration was obtained with increasing current.
4.6. Avoiding porosity in welds Substantial porosity could be observed occasionally in the BM (Fig. 12a and b). The WM usually displays smaller volume fraction porosity than the BM (excluding specimens marked #3, as discussed later). Volume fraction porosity has a tendency to decrease at lower weld speeds resulting in sounder welds. The slowest welds (250–500 mm/min) display practically no porosity. The key-hole mode specimens, exhibiting deep-penetration welds (Fig. 12), displayed a relatively porous microstructure. This outcome is not typical for EB welded wrought aluminum alloys, prepared under similar welding conditions. The porosity, found in the WM of the AM specimens, may suggest that the AM material has low weldability when using relatively high heat inputs; this outcome may be associated with the inherent small porosity found in the BM.
4.7. Microhardness tests of the weld metal
Fig. 8. Schematic illustration of the macrostructure evolution in the WM; transition from CZ-WM to WM outer zone morphology.
Microhardness tests indicate slightly lower hardness values in the weld metal (i.e. 103 ± 5 HV) when compared to the BM hardness (111 ± 5 HV). The relatively high hardness values of the WM indicate that the EB welding process hardly impair the mechanical properties of the joint. Base metal hardness values are slightly higher but fairly close to the typical hardness values of AM-SLM material and reported as 94 ± 5 HV in our previous research [6].
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
69
Fig. 9. SEM characterization. (a) Weld root. (b) WM-BM interface, showing very thin HAZ, weld velocity is high: 750 mm/min. (c) Typical eutectic microstructure. (d) Transition from narrow (∼10 m) columnar-eutectic (marked as CE-WM) to equiaxed-eutectic (marked as EXE-WM) microstructure for a high speed weld. (e) Transition from a wide (∼40 m) columnar-eutectic to equiaxed-eutectic microstructure, weld velocity is low: 240 mm/min.
Fig. 10. Weld penetration and estimation area vs. weld speed, weld heat input of ∼29 J/mm.
Fig. 11. Weld penetration as a function of heat input, for various speeds.
70
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
Fig. 12. (a) Keyhole mode weld displaying a substantial amount of pores, specimen #3 (non-typical WM). Some of the pores in the BM and WM are marked with arrows. (b) EB welded AA6061 joint, shown for comparison.
5. Summary References Autogenous electron beam welding technique has been used to join rotund parts produced by the AM process using the SLM technology, and for the sake of comparison to join flat plates fabricated by casting. Weld metal soundness (porosity) was greatly influenced by the AM base-metal’s amount of porosity. When working under the keyhole mode regime, deep penetration welds have been obtained at high heat inputs while working at relatively high speeds. Weldability of the AM material was improved and partially optimized by switching to the conduction mode regime; at slower welding speed sound WM was obtained. AM welds were compared to A-356-T6 welds; due to the fine microstructure, no PMZ was observed for the AM material. Furthermore, insignificant HAZ was observed for the AM welds, resulting in improbable degradation of the joint’s mechanical properties. Weld metal solidification patterns were partially investigated and reported. No depletion of the alloying elements was detected in the WM, suggesting that EB welding did not change the composition in the joint. Finally, it is worth mentioning that to the best of our knowledge, this investigation provides the first evidence for the feasibility of EB welding for AlSi10Mg components produced by SLM additive manufacturing technology. And, despite the further studies required in this specific field, the initial results are extremely promising. Acknowledgements The authors would like to thank E. Millionshckik and R. Golan of the Ilse Katz Institute for Nanoscale Science & Technology, BenGurion University of the Negev, Israel, for the valuable technical assistance. The authors would also like to especially thank Y. Sharon of “Sharon Tuvia (1982)” LTD, Nes-Tziona, Israel, for supplying the samples.
[1] D.D. Gu, W. Meiners, K. Wissenbach, R. Poprawe, Laser additive manufacturing of metallic components: materials, processes and mechanisms, Int. Mater. Rev. 57 (3) (2012) 133–164, http://dx.doi.org/10.1179/1743280411Y. 0000000014. [2] E. Brandl, U. Heckenberger, V. Holzinger, D. Buchbinder, Additive manufactured AlSi10Mg samples using selective laser melting (SLM): microstructure, high cycle fatigue and fracture behavior, Mater. Des. 34 (2012) 159–169, http://dx. doi.org/10.1016/j.matdes.2011.07.067. [3] M. Krishnan, E. Atzeni, R. Canali, F. Calignano, D. Manfredi, E.P. Ambrosio, L. Iuliano, On the effect of process parameters on properties of AlSi10Mg parts produced by DMLS, Rapid Prototyp. J. 20 (6) (2014) 449–458, http://dx.doi.org/ 10.1108/RPJ-03-2013-0028. [4] N. Read, W. Wang, K. Essa, M.M. Attallah, Selective laser melting of AlSi10Mg alloy: process optimization and mechanical properties development, Mater. Des. 65 (2015) 417–424, http://dx.doi.org/10.1016/j.matdes.2014. 09.044. [5] L. Thijs, K. Kempen, J.P. Kruth, J.V. Humbeeck, Fine-structured aluminium products with controllable texture by selective laser melting of pre-alloyed AlSi10Mg powder, Acta Mater. 61 (2013) 1809–1819, http://dx.doi.org/10. 1016/j.actamat.2012.11.052. [6] I. Rosenthal, A. Stern, N. Frage, Microstructure and mechanical properties of AlSi10Mg parts produced by the laser beam additive manufacturing (AM) technology, Metallogr. Microstruct. Anal. 3 (4) (2014) 448–453, http://dx.doi.org/ 10.1007/s13632-014-0168-y. [7] EOS Aluminium AlSi10Mg for EOSINT M270, Material Data Sheet, EOS Gmbh. [8] S. Kou, Welding Metallurgy, 2nd ed., John Wiley & Sons, 2003, ISBN 0-4713491-4. [9] J.L. Murphy, R.A. Huber, W.E. Lever, Joint preparation for electron beam welding thin aluminum alloy 5083, Weld. J. 69 (1990) 125s–132s. [10] I. Rosenthal, E. Tiferet, M. Ganor, A. Stern, Selective laser melting additive manufacturing: AlSi10Mg powder characterization, in: The Annals of Dunarea de Jos University of Galati. Fascicle XII, Welding Equipment and Technology, 2014, pp. 35–40, ISSN 1221-4639. [11] R.S. Rosenthal, Z. Shwartzman, A. Stern, Hatching strategy: 3D visualization model for powder bed based additive manufacturing with focused beams, in: The Annals of Dunarea de Jos University of Galati. Fascicle XII, Welding Equipment and Technology, 2014, pp. 13–18, ISSN 1221-4639.
Contents lists available at ScienceDirect
Additive Manufacturing journal homepage: www.elsevier.com/locate/addma
Electron beam welding of AlSi10Mg workpieces produced by selected laser melting additive manufacturing technology Moshe Nahmany a,∗ , Idan Rosenthal b , Isgav Benishti a , Nachum Frage b , Adin Stern b a b
Materials Department, Nuclear Research Center, Negev, P.O.B. 9001, Beer Sheva 8410900, Israel Materials Engineering Department, Ben Gurion University of the Negev, P.O.B. 653, Beer Sheva 8410501, Israel
a r t i c l e
i n f o
Article history: Received 11 February 2015 Received in revised form 20 June 2015 Accepted 24 August 2015 Available online 28 August 2015 Keywords: AM SLM AlSi10Mg alloy EBW
a b s t r a c t Electron beam welding (EBW) is a high-density energy (low heat input) welding technique, resulting in a narrow heat affected zone (HAZ), causing minimal metallurgical changes in the workpieces. The present research work investigates EB autogenous welded AlSi10Mg samples, produced by the selective laser melting (SLM) additive manufacturing (AM) method, with emphasis on the characterization of the joint’s macro- and microstructure. When comparing the EB welded AM parts to the EB welded cast samples two main differences were observed: weld metal porosity and a negligible HAZ in the AM joints and low porosity level but substantial HAZ in the welded cast parts. These preliminary results show for the first time the feasibility of the EBW technique on AM-SLM specimens. © 2015 Elsevier B.V. All rights reserved.
1. Introduction The concept of advanced manufacturing technology has been viewed as the utilization of computer and numerical-based processing of components. One of these technologies includes three-dimensional printing and related layer-by-layer fabrication, known as additive manufacturing (AM) [1–4]. AM technology builds a solid, often geometrically complex object from a series of layers, each one “printed” on top of the previous one. AM “printers” use a virtual, mathematical model to construct a physical artifact that can be manipulated on the computer screen. Manufacturing components (by powder-bed process, for example) offers a high geometrical flexibility and accuracy enabling the construction of components with very complex external and internal shapes (e.g. embedded cooling channels). The AlSi10Mg alloy, which was investigated in the present study, is a widely used casting alloy with near-eutectic composition, displaying good mechanical properties and weldability. The addition of small amounts of Mg (0.3–0.5 wt.%) provides precipitation hardening of the aluminum alloy by forming Mg2 Si precipitates upon natural or artificial aging treatments [5]. Typical hardness values of AlSi10Mg fabricated by AM-SLM are reported [6] as 94 ± 5 HV and up to 120 ± 5 HV (reported [7] as 120 ± 5 HBW). The engineering design of many structural components requires the manufacturing of complex shaped modules
which is quite expensive. An attractive way to achieve this goal is to fabricate the composite modules by joining two or more AM processed components. A reliable joining technology provides enhanced design flexibility and engineering solutions for AM modules that are impossible to produce due to size limitations of the additive manufacturing equipment. Power beam welding processes characterized by energy densities up to 106 W/cm2 [8], are ideally suited for producing high quality welds, both minimizing the weld heat input and reducing the weld-induced distortion of the components. The EBW process (Fig. 1a) has been chosen in this research because it is well-positioned to provide metal industries with high quality and reliable joining solutions. Moreover, the solidification rates for this process (105 –106 ◦ C/s) are significantly higher than those for conventional arc welding processes (102 –103 ◦ C/s), creating the fine microstructure of the weld metal. There are four major types of weld defects in EBW of aluminum: porosity, cracking, inclusions and loss of alloying elements [9]. The porosity is usually located in the keyhole path, whereas gas pores are homogeneously distributed with slight enrichment at the melting line. When the welding mode switches from the keyhole to conduction mode, the level of porosity is significantly reduced, because there is enough time for gases to escape from a shallower weld pool. 2. Goals of research
∗ Corresponding author. E-mail address: [email protected] (M. Nahmany). http://dx.doi.org/10.1016/j.addma.2015.08.002 2214-8604/© 2015 Elsevier B.V. All rights reserved.
• Demonstrate the feasibility of using EBW technology to weld SLM AlSi10Mg components.
64
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
Fig. 1. (a) Schematic illustration of the EBW process, presenting a deep and narrow weld metal (WM). The illustration displays the metallographic cross sections: T-CS for transverse, P-CS for planar. (b) Schematic illustration of the welded specimens, presenting the build direction and the weld direction.
• Compare the microstructure of EBW joined region of SLMproduced specimens and casting-produced specimens.
3. Experimental Welding samples were machined from AM-SLM built AlSi10Mg commercial specimens, manufactured using an EOSINT-M280 system. Pre-alloyed AlSi10Mg powder with particle size distribution between 7 and 50 m has been used [10]. All specimens were built in the vertical direction. The welding samples were machined from 10 mm diameter rods and axially aligned for the welding operation. The circumferential welds were oriented perpendicularly to the building direction (Fig. 1b). For the sake of comparison, welding specimens were prepared from A356-T6 plates fabricated by casting. Autogenous electron beam welding technique was used to join both types of specimens on a Wentgate EB machine with capabilities of 60 kV acceleration voltage and a 40 mA beam current. The machine is equipped with both rotational turntable and XY-stage. The electron beam can be consistently focused (with diameter about 0.3–0.5 mm) on the specimen’s top surface for all welds. Welds were performed with electron beam power varying from 120 to 600 W using heat inputs of 21–45 J/mm. Welding parameters are presented in Table 1. Welding parameters were selected using our experience on EBW of cast aluminum and were updated according to the preliminary results of EBW of AM parts. We have tried to keep some parameters constant and vary others to clarify their effects. The joints were characterized to visualize the development of novel and unusual macrostructures and microstructural architectures resulting from EB welding of the AlSi10Mg alloy. Welds were characterized from a transverse cross-section and from a planar cross-section noted as T-CS and P-CS respectively. Metallographic samples were prepared according to the ASTM-E3 Standard. First, the samples were hot mounted, and then a rough polish was performed. Surface preparation was performed by grinding and fine polishing (up to 0.5 m) and etched in Flick’s Reagent (10% HF–90% H2 O) to expose the microstructure. Welds were examined by an Optical Microscope (OM Zeiss, Aalen, Germany), Scanning Electron Microscope (SEM JEOL, JSM-2500) and High Resolution Scanning Electron Microscope (HRSEM, JEOL 7400F). Micro-hardness tests
Table 1 Welding parameters; all welds were performed using 60 kV acceleration voltage. Last specimens, designated with “C” are cast specimens. Heat input was calculated as (beam current × beam voltage)/welding speed. No.
Travel speed (mm/min)
Beam current (mA)
Heat input/specific energy (J/mm)
1 2 3 4 5 6 7 8 9 10 11 12 C-13 C-14
1000 1000 1000 1000 750 500 484 242 484 242 1000 500 1000 1000
6 8 10 13 6 4 4 2 3 3 8 4 5.5 16
21.7 28.8 36 46.8 28.8 28.8 29.8 29.8 22.3 44.7 28.8 28.8 19.9 57.6
were conducted using a Buehler micro-hardness tester (MMT-7), with a load of 100 g. 4. Results and discussion 4.1. Base metal composition and microstructure The microstructure of the components fabricated by the AM is the end-result of pre-alloyed powder melting/solidification, with adjacent neighboring tracks responsible for partial remelting of the solidified tracks. Average concentrations (in wt.%) of the alloying elements are 11.1 Si, 0.2 Mg, ∼0.1 Fe with traces of Cr, Cu, Mn, Ti, and Zn. the concentration of the alloying elements in the base metal was determined by a wet chemical analysis. The typical microstructure of the AlSi10Mg part (Fig. 2a and b), consisting of approximately half-cylindrical solidified melt pools, reveals that most of the defects (Fig. 2c) are located at regions where the melt pools overlap, strengthening the role of the melting/solidification cycles and their effect on both microstructure and defect formation. The fast melting process of the stochastic powder bed can induce vigorous melt pool movements which sometimes lead to faults acting as starting points for defects such as channels and/or pores between layers [6,10,11]. The microstructure morphology is
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
65
Fig. 2. AM-SLM typical microstructure of the base metal [6,10]. The arrows mark defects. (a) General view of the melt pools. (b) Half-cylindrical shape of each melt pool. (c) Enlarged view showing porosity at weld pool boundaries (marked by the dashed line), some pores are marked by arrows.
known as “fish scales”. The concentrations (in wt.%) of the alloying elements for the cast A356 T6 alloy are 7.7 Si, 0.37 Mg, 0.7 Fe, 0.13 Ti with traces of Cr, Cu, Mn, and Zn; the typical microstructure is characterized by a coarse “Chinese script” eutectic morphology.
4.2. Weld Metal microanalysis In order to clarify the effect of evaporation of the melt components the EDS microanalysis was conducted. Possible dilution of alloying element (Si, Mg) in the WM was examined. Weld metal microanalysis was performed using two configurations: perpendicular to the top surface downwards and parallel to the top surface. Base metal microanalysis exhibits values (wt.%) of Si-10.7 ± 0.5 and Mg-0.9 ± 0.2, over an average of 6 arbitrarily-located measurements. These values are slightly different from the chemical analysis shown earlier, probably due to fact that the local EDS measurements are semi-quantitative only. Weld metal, perpendicular to the top surface downwards (13 continuous measurements) exhibits no trend of dilution with values of Si-10.9 ± 0.5 and Mg0.9 ± 0.2. Weld metal, parallel to the top surface (7 continuous measurements) exhibits no trend of dilution as well, showing values of Si-11.2 ± 0.7 and Mg-0.9 ± 0.2. The semi-quantitative local analysis indicates no dilution of Silicon or Magnesium in the weld metal. The A356 cast alloy after T6 treatment displays a typical “Chinese script” microstructure, which is completely different from the microstructure of AM parts (Fig. 3).
Fig. 3. Microstructure of the A356 cast alloy after T6 treatment.
66
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
Fig. 4. WM and neighboring regions (weld metal, top left corner). (a) AM specimen. (b) A356-T6 specimen, PMZ adjacent to the WM, marked with red parallel lines.
4.3. Weld metal macrostructure The welds of the A356-T6 parts were performed utilizing a beam power of 220 W (60 kV) leading to a heat input of 20 J/mm. The A356-T6 specimens (Fig. 4) exhibit a notable partially melted zone (PMZ, marked with parallel lines) adjacent to the weld metal (WM), due to the lower melting temperature of the eutectic colonies. No HAZ was detected near the PMZ. The AM welded parts show no traceable HAZ. 4.3.1. Transverse metallographic view The WM EB joints were characterized (T-CS) by penetration depth and width which are presented in Table 2. Both keyhole (Fig. 5a and b) and conduction modes (Fig. 5c and d) were observed. Moreover, a different microstructure was detected in the uppercentral region (Fig. 5d), designated as CZ-WM (central zone weld metal). The general geometry of the weld metal depends strongly
Table 2 Weld metal dimensions in relation to the heat input. No.
Weld penetration (mm)
Weld width (mm)
Heat input/specific energy (J/mm)
1 2 3 5 6 7 8 9 10
1.5 2.4 2.8 1.4 0.8 0.63 0.4 0.4 0.55
1.2 2.0 2.1 1.3 1.3 0.85 0.55 0.7 0.9
21.7 28.8 36 28.8 28.8 29.8 29.8 22.3 44.7
on the heat input and the penetration depth is mainly affected by the travel speed (Fig. 5a–d). It has to be pointed out, that the shallow penetration joint has been achieved when EBW was conducted using a relatively low weld velocity, when compared to the typical
Fig. 5. Overall metallographic (T-CS) view of EB welded AM parts. (a) High speed weld (1000 mm/min, 28.8 J/mm), keyhole mode predominant showing deep penetration and high porosity. Some pores are marked by arrows. (b) Lower heat input weld (1000 mm/min, 21.7 J/mm). (c, d) Much slower welds, conduction mode predominant showing shallow penetration (500 mm/min, 29.8 J/mm and 500 mm/min, 28.8 J/mm respectively). Horizontal line in (d) indicates the height of the planar metallography (P-CS), ∼0.2 mm below the part surface, as discussed later.
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
67
Fig. 6. Planar metallography (P-CS); (a) General view, yellow squares represent the regions of higher magnification. (b) Lower-right corner of a. (c, d) Enlargement of the lower-right corner of (b), as marked in (b). (e) BM/WM interface, indicating the absence of a HAZ.
velocity used in practice. Moreover, relatively low levels of weld metal porosity have been achieved when using low speed welds at different heat inputs. It seems that the porosity (soundness) of the weld metal mostly depends on the porosity of the AM base-metal.
4.3.2. Planar metallographic view The sample was polished about 0.2 mm below surface (see Fig. 5d) and the macrostructure is presented in Fig. 6a. Banding and rippling originated from variations in welding speed and in heat input or rather by solidification halts that can be clearly observed (Fig. 6a); these phenomena were displayed and explained in details by Kou in [8]. Fig. 6b–d shows specific regions at higher magnification as indicated in Fig. 6a and b. The morphology of the Al–Si eutectic microstructure (Fig. 6c and d) reflects heat-dissipation outwards to the BM and toward the earlier solidified WM (marked by arrows). Fig. 6e depicts the smooth continuous interface between the BM and WM, with no pores and cracks present. No visible indication of a HAZ has been identified. Several solidification modes are clearly observed in the WM TCS micrographs (Fig. 7a–d). Columnar and equiaxed dendrites and a typical Al–Si eutectic morphology can be distinguished (Fig. 7a, b and d). Adjacent the fusion line, the dendrites appear to be aligned parallel to the local heat-flow direction. A slightly modified microstructure was observed in the upper-center region of the WM, designated as Central Zone Weld Metal (CZ-WM) and
illustrated in Fig. 7c. An almost abrupt transition from the CZWM to the WM outer region has been clearly observed and the microstructure evolution is schematically illustrated in Fig. 8. The macrostructure of the CZ-WM is oriented parallel to the weld direction, with the red arrow representing heat flow backwards to the colder, previously solidified metal. The orientation of the macrostructure of the WM outer regions matches the direction of the heat flow toward the BM as indicated by the black arrows (Fig. 8).
4.4. SEM microstructural characterization Subject to the location within the WM, different microstructure morphologies have been observed by SEM examination. The weld root (Fig. 9a) exhibited a very thin, hardly observed HAZ, without pores or cracks (Fig. 9b). The typical Al–Si eutectic morphology is displayed in Fig. 9c, the solidification mode is mainly cellular, but occasionally, some side branches can be seen. The gray cellular features are primary Al decorated with white fibrous Si particles. The transition from a columnar-eutectic to an equiaxed-eutectic morphology, adjacent the fusion line, is shown in Fig. 9d (marked as CE-WM and EXE-WM respectively). The transition is probably due to the local changes in the temperature gradients and growth rates near the fusion line [8]. Slower welds exhibit a wider columnareutectic zone near the fusion line (Fig. 9e); CE-WM for slow welds
68
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
Fig. 7. Solidification modes; arrows mark heat dissipation toward the BM. (a, b) Typical microstructure adjacent to the fusion line. (c) Transition from the CZ-WM to WM outer zone. (d) slow velocity weld microstructure.
is about 40 m wide, while CE-WM width of fast welds is around 10 m. A closer look into the microstructure of the AM BM in Fig. 9d and e, reveals the presence of a much finer cellular–dendritic solidification structure with a size smaller than 1 m. Thermal gradients as high as in the order of 106 K/m can occur during SLM building [5] explaining the very small cell size observed in this AlSi10Mg base metal when compared to the WM solidification microstructure. 4.5. Effect of weld speed and current on WM geometry The main effect of the weld speed on the WM’s depth and weld area is shown in Fig. 5. The weld speed, altered together with the beam current, results in a deeper and narrower weld metal. The depth and weld area were plotted vs. weld speed (Fig. 10). Higher weld speed leads to deeper penetration and to a high fraction of porosity due to the preferred key-hole welding mode. On the other
hand, slower and lower current welds yield a shallow penetration and lead to lower porosity of the WM, due to conductive welding mode. It has to be pointed out that the samples were relatively small, thus having limited ability to absorb energy from the electron beam resulting in low penetration welds. WM’s depth as a function of weld heat input for various speeds (500, 100 mm/min) is shown in Fig. 11. For both weld speeds the welding mode is mainly keyhole and deeper penetration was obtained with increasing current.
4.6. Avoiding porosity in welds Substantial porosity could be observed occasionally in the BM (Fig. 12a and b). The WM usually displays smaller volume fraction porosity than the BM (excluding specimens marked #3, as discussed later). Volume fraction porosity has a tendency to decrease at lower weld speeds resulting in sounder welds. The slowest welds (250–500 mm/min) display practically no porosity. The key-hole mode specimens, exhibiting deep-penetration welds (Fig. 12), displayed a relatively porous microstructure. This outcome is not typical for EB welded wrought aluminum alloys, prepared under similar welding conditions. The porosity, found in the WM of the AM specimens, may suggest that the AM material has low weldability when using relatively high heat inputs; this outcome may be associated with the inherent small porosity found in the BM.
4.7. Microhardness tests of the weld metal
Fig. 8. Schematic illustration of the macrostructure evolution in the WM; transition from CZ-WM to WM outer zone morphology.
Microhardness tests indicate slightly lower hardness values in the weld metal (i.e. 103 ± 5 HV) when compared to the BM hardness (111 ± 5 HV). The relatively high hardness values of the WM indicate that the EB welding process hardly impair the mechanical properties of the joint. Base metal hardness values are slightly higher but fairly close to the typical hardness values of AM-SLM material and reported as 94 ± 5 HV in our previous research [6].
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
69
Fig. 9. SEM characterization. (a) Weld root. (b) WM-BM interface, showing very thin HAZ, weld velocity is high: 750 mm/min. (c) Typical eutectic microstructure. (d) Transition from narrow (∼10 m) columnar-eutectic (marked as CE-WM) to equiaxed-eutectic (marked as EXE-WM) microstructure for a high speed weld. (e) Transition from a wide (∼40 m) columnar-eutectic to equiaxed-eutectic microstructure, weld velocity is low: 240 mm/min.
Fig. 10. Weld penetration and estimation area vs. weld speed, weld heat input of ∼29 J/mm.
Fig. 11. Weld penetration as a function of heat input, for various speeds.
70
M. Nahmany et al. / Additive Manufacturing 8 (2015) 63–70
Fig. 12. (a) Keyhole mode weld displaying a substantial amount of pores, specimen #3 (non-typical WM). Some of the pores in the BM and WM are marked with arrows. (b) EB welded AA6061 joint, shown for comparison.
5. Summary References Autogenous electron beam welding technique has been used to join rotund parts produced by the AM process using the SLM technology, and for the sake of comparison to join flat plates fabricated by casting. Weld metal soundness (porosity) was greatly influenced by the AM base-metal’s amount of porosity. When working under the keyhole mode regime, deep penetration welds have been obtained at high heat inputs while working at relatively high speeds. Weldability of the AM material was improved and partially optimized by switching to the conduction mode regime; at slower welding speed sound WM was obtained. AM welds were compared to A-356-T6 welds; due to the fine microstructure, no PMZ was observed for the AM material. Furthermore, insignificant HAZ was observed for the AM welds, resulting in improbable degradation of the joint’s mechanical properties. Weld metal solidification patterns were partially investigated and reported. No depletion of the alloying elements was detected in the WM, suggesting that EB welding did not change the composition in the joint. Finally, it is worth mentioning that to the best of our knowledge, this investigation provides the first evidence for the feasibility of EB welding for AlSi10Mg components produced by SLM additive manufacturing technology. And, despite the further studies required in this specific field, the initial results are extremely promising. Acknowledgements The authors would like to thank E. Millionshckik and R. Golan of the Ilse Katz Institute for Nanoscale Science & Technology, BenGurion University of the Negev, Israel, for the valuable technical assistance. The authors would also like to especially thank Y. Sharon of “Sharon Tuvia (1982)” LTD, Nes-Tziona, Israel, for supplying the samples.
[1] D.D. Gu, W. Meiners, K. Wissenbach, R. Poprawe, Laser additive manufacturing of metallic components: materials, processes and mechanisms, Int. Mater. Rev. 57 (3) (2012) 133–164, http://dx.doi.org/10.1179/1743280411Y. 0000000014. [2] E. Brandl, U. Heckenberger, V. Holzinger, D. Buchbinder, Additive manufactured AlSi10Mg samples using selective laser melting (SLM): microstructure, high cycle fatigue and fracture behavior, Mater. Des. 34 (2012) 159–169, http://dx. doi.org/10.1016/j.matdes.2011.07.067. [3] M. Krishnan, E. Atzeni, R. Canali, F. Calignano, D. Manfredi, E.P. Ambrosio, L. Iuliano, On the effect of process parameters on properties of AlSi10Mg parts produced by DMLS, Rapid Prototyp. J. 20 (6) (2014) 449–458, http://dx.doi.org/ 10.1108/RPJ-03-2013-0028. [4] N. Read, W. Wang, K. Essa, M.M. Attallah, Selective laser melting of AlSi10Mg alloy: process optimization and mechanical properties development, Mater. Des. 65 (2015) 417–424, http://dx.doi.org/10.1016/j.matdes.2014. 09.044. [5] L. Thijs, K. Kempen, J.P. Kruth, J.V. Humbeeck, Fine-structured aluminium products with controllable texture by selective laser melting of pre-alloyed AlSi10Mg powder, Acta Mater. 61 (2013) 1809–1819, http://dx.doi.org/10. 1016/j.actamat.2012.11.052. [6] I. Rosenthal, A. Stern, N. Frage, Microstructure and mechanical properties of AlSi10Mg parts produced by the laser beam additive manufacturing (AM) technology, Metallogr. Microstruct. Anal. 3 (4) (2014) 448–453, http://dx.doi.org/ 10.1007/s13632-014-0168-y. [7] EOS Aluminium AlSi10Mg for EOSINT M270, Material Data Sheet, EOS Gmbh. [8] S. Kou, Welding Metallurgy, 2nd ed., John Wiley & Sons, 2003, ISBN 0-4713491-4. [9] J.L. Murphy, R.A. Huber, W.E. Lever, Joint preparation for electron beam welding thin aluminum alloy 5083, Weld. J. 69 (1990) 125s–132s. [10] I. Rosenthal, E. Tiferet, M. Ganor, A. Stern, Selective laser melting additive manufacturing: AlSi10Mg powder characterization, in: The Annals of Dunarea de Jos University of Galati. Fascicle XII, Welding Equipment and Technology, 2014, pp. 35–40, ISSN 1221-4639. [11] R.S. Rosenthal, Z. Shwartzman, A. Stern, Hatching strategy: 3D visualization model for powder bed based additive manufacturing with focused beams, in: The Annals of Dunarea de Jos University of Galati. Fascicle XII, Welding Equipment and Technology, 2014, pp. 13–18, ISSN 1221-4639.