- Email: [email protected]
Microstructure and mechanical properties of as-processed scandium-modified aluminium using selective laser melting
CIRP Annals - Manufacturing Technology 65 (2016) 213–216
Contents lists available at ScienceDirect
CIRP Annals - Manufacturing Technology jou rnal homep age : ht t p: // ees .e lse vi er . com /ci r p/ def a ult . asp
Microstructure and mechanical properties of as-processed scandiummodified aluminium using selective laser melting Adriaan B. Spierings a,*, Karl Dawson b, Mark Voegtlin a, Frank Palm c, Peter J. Uggowitzer d a
Innovation Centre for Additive Manufacturing, INSPIRE-AG, Lerchenfeldstrasse 5, CH-9014 St. Gallen, Switzerland Centre for Materials and Structures, School of Engineering, University of Liverpool, Liverpool, UK c Airbus Group, Innovations, TX2, 81663 Munich, Germany d Laboratory of Metal Physics and Technology, Department of Materials, ETH Zurich, Wolfgang-Pauli-Str. 10, 8093 Zurich, Switzerland Submitted by Prof. Dr. Gideon Levy (1), TTA Technology Turn Around, Switzerland b
A R T I C L E I N F O
A B S T R A C T
Keywords: Selective laser melting (SLM) Aluminium Microstructure
Additive manufacturing (AM) offers significant benefits towards optimized lightweight structures for aircraft and space applications. High-strength aluminium alloys are of special interest to reach a maximum mass reduction. The paper presents the development of appropriate selective laser melting (SLM) processing windows for a scandium modified aluminium alloy (Scalmalloy1), reaching densities >99%. The mechanical properties of as-processed material are analyzed, pointing out a comparably low anisotropy with regard to the build orientation. A fine-grained microstructure is observed next to regions of coarser, elongated grains. The paper discusses the observed microstructure, and concludes with suggestions for innovative material design for AM. ß 2016 CIRP.
1. Introduction Additive manufacturing technologies such as selective laser melting (SLM) are still on the threshold to industrial application. Especially in high-tech sectors the technologies offer benefits e.g. in terms of lightweight and functionally optimized parts and applications. Thereby the degree of an achievable mass reduction in lightweight applications depends not only on structural optimisations (e.g. the integration of lattice structures or topology optimization), but also on the mechanical properties of the lightweight material used. High-strength structural aluminium alloys are therefore of special interest for the SLM technology. There is a basic need to develop aluminium alloys, designed specifically for additive manufacture, which will satisfy a broad range of industrial requirements. For lightweight applications, there are currently very few alternatives to AlSi10Mg. However, although AlSi10Mg can offer good mechanical properties, in both as-processed and heat treated condition, these properties can be influenced significantly by the selection of different SLM-processing parameters. This makes it more difficult for industry to use AM for real applications, and consequently every machine and processing window needs to be qualified separately. A potential solution to this problem is designing alloy systems being more suitable to the specific SLM process conditions, also with regard to variations of alloying elements,
main alloying element. Examples are the AlSi12 and AlSi10Mg alloys, respectively. They were originally developed for casting processes, offering good castability and reduced shrinkage imparted by the presence of a relatively large volume fraction of Al–Si eutectic with a corresponding low melting temperature. Mechanical properties of SLM-processed AlSi10Mg are shown in Table 1. Buchbinder [1,2] reports that changing the processing window (basically laser power PL) influences the mechanical properties significantly, as for higher PL and consequently higher scan speed vs the cooling rate is increased, leading to a more fine grained material. Thijs [17] investigated how microstructures could be influenced by the implementation of different scan strategies and demonstrated that fine grain structures, which were developed as sub-micron cells, grew along h100i crystal directions from the melt pool border towards the centre of the melt pool, hence along the temperature gradient. Columnar grains are further growing at the centreline of the scan tracks along the build direction. It was also reported that due to the moving heat source and scan strategy applied, the thermal gradients and growth rate vary over the melt pool, and consequently the fineness of cells and possible textures changed over the cross-section of the melt pool. Hence anisotropic mechanical behaviour is a typical result for additively processed Table 1 Mechanical properties for SLM-AlSi10Mg, serving as an age hardenable alloy for comparison to Scalmalloy1. Left column: Build orientation.
1.1. State of the art To date, the most commonly used aluminium alloys for SLM are casting alloys from the 4XXXer series, containing silicon as the * Corresponding author. http://dx.doi.org/10.1016/j.cirp.2016.04.057 0007-8506/ß 2016 CIRP.
08 908 08 908 a
s0.2 (MPa)
sUTS (MPa)
A (%)
Ref.
240 215 250 235
420 400 350 275
5.9 1 3.2 0.5 2.5 1.1
(Buchbinder et al., 2011)a
Without preheating of the build platform.
[18]
214
A.B. Spierings et al. / CIRP Annals - Manufacturing Technology 65 (2016) 213–216
aluminium, although reported property values are overcoming the requirements of EN-1706 for all build directions [2]. Siddique [13] investigated AlSi12, processed with different specific laser energy inputs, base plate heating and post-build stress relief combinations. He concluded that high strength can be achieved by developing a fine microstructure, but at the cost of fracture strain. Again changing the processing conditions significantly affected the mechanical properties, and a certain anisotropy is obvious. Magnesium serves as the primary alloying element in 6XXX series alloys, with smaller additions of Si. These alloys are age hardenable. Jerrard et al. [6] showed the basic SLM-processability of an Al6061 alloy, and investigated the consolidation behaviour compared to pure Al. He found a better material density as Mg and Si significantly improved wetability; however, mechanical properties were not reported. 1.2. Scandium modified aluminium (Scalmalloy1) Scalmalloy1 is a protected Sc- and Zr-modified aluminium alloy concept for AM procedures, developed by Airbus Group Innovations. The metallurgical origins can be traced back to materials developed, during the 1970s, in the US and former Soviet Union [7]. The addition of Sc to the Al alloy facilitates good levels of microstructural control, however, due to the high cost of Sc and high buy-to-fly ratios in conventional lightweight engineering, such alloys have not yet caught the attention of industry. The effects of Sc on the properties of Al-alloys have been investigated by several researchers [4,9–11,20], and can be summarized as follows: – The Al–Sc solid solution can decompose upon cooling, forming fine dispersed, fully coherent intermetallic Al3Sc precipitates of the cubic LI2 crystal structure, acting as an age hardener and leading to high strength, with an strength increment of about 40 to 50 MPa per 0.1 wt.% Sc. – The precipitation of Al3Sc takes place at higher temperatures (275–325 8C) compared to classical age hardening high strength Al-alloys (150–200 8C), leading to a high resistance against premature precipitates coarsening (e.g. strong heat affected zone occurrence). – Al3Sc precipitates can prevent recrystallization even up to high temperatures. – Al3Sc particles are known to be very effective for grain refinement, acting as seed crystals during solidification (heterogeneous nucleation). – Sc is claimed to improve general weldability in Al alloys, e.g. for alloys which are susceptible to hot cracking and extended heat affected zone formation. These effects are very welcome for AM where process-specific effects such as e.g. the formation of columnar grains and the corresponding anisotropic material behaviour is known for many alloys [14,19]. As in AM the fly-to-buy ratio is much closer to 1 and therefore the cost-influence of Sc is limited, Sc-containing alloys might be an interesting way to overcome some of these microstructure related properties. Schmidtke [12] proofed the basic SLM-processability of Scalmalloy1, and presented first mechanical properties in a post-built-up heat treated condition. 1.3. Concept of the paper The paper presents the development of an appropriate SLM processing window, and analyses the micro-structure in the as-processed condition. Static mechanical properties are presented and discussed. The paper concludes with assumptions on the formation mechanisms for the observed microstructure, which will serve as a basis for future alloy design for AM. 2. Materials and methods 2.1. SLM process The SLM-process is affected by many processing variables, including powder properties and machine setup. A ConceptLaser
M2 machine was used equipped with a 200 W Nd-YAG laser operated in cw mode. The Gaussian beam focus diameter is 100 mm [3]. Next to a given powder the main parameters affecting the process are the laser power PL, the layer thickness tL, the scan velocity vs and the hatch distance d. Due to high reflectivity of aluminium, PL was set to the maximum of 200 W to optimize process productivity, and the layer thickness tL was 30 mm. The hatch distance d was varied between 90 mm and 225 mm and scan velocities vs were varied to optimize material density. The volume specific energy density EV combines the parameters to a single main driving parameter for material density. PL ðJ=mm3 Þ EV ¼ (1) vs t L d 10 10 10 mm3 test samples were produced using an alternating scan strategy from layer to layer, and bi-directional scanning within the layers as given in Fig. 1. To avoid over-heating effects in scan reversals, no ‘‘island-scanning’’ was applied, which would be the typical scan strategy used on ConcepLaser machines. In total > 120 samples were produced, and for promising processing windows 3 samples were analyzed to assess the standard deviation of the material density.
Fig. 1. Scan strategy for SLM test samples. z-direction indicates the build orientation.
2.2. Scalmalloy1 powder raw material The chemical composition of the powder used during these trials is presented in Table 2. The particle size distribution was measured by optical particle analysis using PowderShape1. Thereby, the particles are filtered for an area-equivalent diameter < 120 mm; more details on the measurement principle are given by Spierings [16]. Table 2 Chemical composition (in wt.%) of the material according to supplier information. D10/D50/D90 are the quantiles for number (q0) and volume (q3) weighted particles size distributions. Mg
Sc
4.6
0.66 0.42 q0 D10 3.9 mm q3 D10 38 mm
Zr
Mn
Fe [12]
Others [12]
0.49
0.068 D50 6.2 mm D50 58 mm
0.05 Bal. D90 49.9 mm D90 83 mm
Al
2.3. Measurement techniques The density (r) of consolidated material was determined by the Archimedes method, using a theoretical maximum density of 2.67 g/cm3, as described by Spierings et al. [15]. Density measurements were corroborated by optical measurements showing minor residual porosity in the builds. The microstructures of as-processed samples were analyzed using a FEI Helios dual beam FIB instrument equipped with an EDAX-EBSD system. Electro-polished surfaces were prepared using a Struers Tenupol 3 twin jet electro-polisher, a 5 vol.% perchloric acid in 95 vol.% methanol electrolyte, cooled to 50 8C, voltage of 25 V. High resolution EBSD maps were collected using a step size of 80 nm in areas of fine grained material (10K magnification), and 250 nm steps were used to map large areas of coarse grained material at lower 750 magnification. All EBSD maps were recorded using beam conditions of 20 kV, 5.5 nA and a 12 mm working distance. The cutting plane orientations for SEM measurements are given in the respective figure captions, using the coordinate system shown in Fig. 1.
A.B. Spierings et al. / CIRP Annals - Manufacturing Technology 65 (2016) 213–216
215
2.4. Static mechanical testing Static tensile specimens according to DIN-500 125-A525 were manufactured in the horizontal (908) and vertical (08) direction. The samples were turned to a surface finish N5 in the as-built condition (without heat treatment), and mechanically tested in accordance with EN-10002/ISO-6892. From the measured ‘‘engineering’’ stress– strain curves, the true stress–strain curves and true ultimate strength (Rm, T) is calculated according to Hosford [5] by
s T ¼ s ð1 þ eÞ;
Rm;T ¼ Rm ð1 þ eÞ;
2 T ¼ lnð1 þ eÞ
(2)
3. Results 3.1. SLM-processing window The processing window shows good attainable material density >99% in a range of EV between 75 J/mm3 and 240 J/mm3 and for d between 135 mm and 165 mm, as shown in Fig. 2. The high thermal conductivity of the Al-alloy leads to scan tracks widths of 260 20 mm, allowing hatch distances d larger than the laser spot diameter. Smaller d are leading to overheating effects resulting in uneven scanning surfaces, hence to irregular next powder layers and consequently to a bad build morphology. Higher d reduces EV to values too small to reach sufficient density. The general reproducibility is acceptable with a mean standard deviation of 0.09% for the considered d values and corresponding densities >99%. The size and type of defects per mm2, taken from xy-micrographs of cube samples, is presented in Fig. 3 showing acceptable values for EV > 120 J/mm3. Only for very high EV > 230 J/mm3 the number of defects can further be minimized, but on the costs of scanning productivity (Fig. 3). Two different EV-levels were selected for the production of tensile test specimens: EV1 = 238 J/mm3 and EV2 = 135 J/mm3, to analyze the impact of varying EV on mechanical properties.
Fig. 4. Static mechanical properties for as-processed Scalmalloy.
These values indicate a comparably very low anisotropic mechanical behaviour of the SLM-processed alloy, which is correlated to the material microstructure. 3.3. Material microstructure The microstructure of the consolidated material was analyzed for two EV-levels EV1 and a second level EV3 = 115 J/mm3, which is roughly half of EV1, and close to level EV2. Fig. 5 shows SEM microstructure processed using EV1, showing a very fine grained (FG) microstructure, next to regions of coarser grains (CG). In Fig. 5, a high amount of bright particles can be found in the FG material, most of them being located at grain boundaries. Additional EDS analysis on TEM foils reviled that these are Mg–O particles with varying composition, probably also MgAl2O3 spinel. Fig. 6 presents the microstructure of material processed with ½EV1 showing a similar structure with regard to FG and CG material. Due to the weld-lines depth of 100 mm and the corresponding high amount of re-melting of previous layers, the perpendicular and longitudinal weld-line cross-sections cannot be identified very clearly (Fig. 6a). But it shows that solidification takes place either in a columnar structure with elongated grains along the build (z-) direction, or a very fine grained structure without any preferential grain orientation, as the corresponding discrete pole figure (Fig. 6c) does not indicate any preferential grain orientation. In contrast, Fig. 6d shows that the CG material displayed a preferential crystallographic h100i orientation approx. parallel to the build (z-) axis of the samples, which is typical for AM-processed materials.
Fig. 2. Material density in dependence of EV and hatch distances d. Dotted lines represent too small d values.
Fig. 5. CBS-images taken in a xz-/yz-plane for EV1. Fine (FG) and coarse (CG) grains are indicated.
Fig. 3. Number, size and type of defects per mm2 for different EV-levels.
3.2. Static mechanical properties High static mechanical properties, overcoming the values reported for AlSi10Mg (Table 1), are reached in the as-processed condition (Fig. 4), with true Rm,T > 400 MPa and RP0.2 > 277 MPa for all build orientations and EV. The E-modulus is 68.0 0.8 GPa for EV1 and 73.6 1.2 GPa for EV2. The anisotropy between horizontal (908) and vertical (08) build orientation in both EV-levels is <2% for true Rm and <3.4% for RP0.2.
Fig. 6. Microstructure processed with EV3 showing FG and CG regions. (a) Edched and EBSD (b) microstructure of the same region. (c) EBSD of FG-material and [001] Pole figures for FG (left) and CG region (right).
216
A.B. Spierings et al. / CIRP Annals - Manufacturing Technology 65 (2016) 213–216
An image analysis revealed that the FG, as-built structure accounts for approx. 52–56% of the total cross-section. Similar to this processing window shown in Fig. 6, Fig. 7(left) presents the microstructure of material processed with EV1, showing a comparable amount of 49–56% of the total cross-sections as FG. The distribution of the misorientation angles between single grains (Fig. 7(right)) in the FG-region supports that the fine grains are rather randomly oriented, as the distribution follows nicely the ideal distribution for random orientation [8]. With regard to grain sizes Fig. 8 presents its distributions for FG and CG regions of samples processed with the two energy levels EV1 and EV3. Only small differences in FG-regions from both processing energy levels EV can be observed. Assuming a normal distribution of grain sizes, with mean xEV1 ¼ 0:6 mm and xEV2 ¼ 1:05 mm. The ¯ ¯ CG-material thereby shows coarser grains, with sizes up to 13 mm. Even such grain sizes are considered as ‘‘fine’’ compared to other manufacturing processes.
layer from the powder raw material, which is broken up during processing. In FG material, such particles are embedded at the grain boundaries, whereas in CG material they are expected to be pushed at the solidification front, leading to a much lower concentration of such particles. 5. Outlook Work is proceeding with regard to the full understanding of the formation mechanisms for the observed microstructure, different types of particles and precipitates. Appropriate heat treatments will be developed, and related static and dynamic mechanical properties have to be measured. Full understanding of the ‘‘working principle’’ of this alloy system might pave the way to the development of other AM-designed alloys. Acknowledgments The authors gratefully acknowledge Prof. Dr. K. Wegener, Mr. Demont, Mr. Frauchiger and Mr. Vincenz for their support of this project. Part of this project is co-financed by the Swiss Commission for Technology and Innovation, CTI, No. 17365.1.
Fig. 7. EBSD in selected area (10K magnification) in FG as-built material processed by EV1 (left), grain orientation angle distribution in this FG area (right).
Fig. 8. Grain size distribution for FG material (EV1 and EV3) and CG material (EV1).
4. Discussion and conclusion The Sc-/Zr-modified Al-alloy (Scalmalloy1) is well processable with SLM, reaching densities >99%. Suitable processing windows require a laser energy input significantly higher compared to other Al-, Fe- or Ni-based alloys. The processing window is robust with regard to changes in EV, as long as the EV > 120 J/mm3. Both material density and general microstructure is not significantly affected by variations of EV, although lower energy input leads to a further refinement of the fine grains. This ensures stable mechanical properties, overcoming the properties of other typical SLM Al-alloys, e.g. AlSi10Mg. Thereby a higher EV reduces the number of defects, but with almost no effect on mechanical performance except for E-modulus, which is reduced for a slightly higher defect level. The mechanical performance is related to some very interesting microstructural features, making Scalmalloy1 an ideal candidate as a structural alloy for AM. Due to the Hall–Petch relationship, the high content of a very fine-grained microstructure without any preferential grain orientation supports high material strength, with low anisotropic properties. The reasons for the ‘‘bimodal’’ grain size distribution are not yet fully understood. However, it is assumed that a high number density of Al–Sc seed crystals are formed during consolidation, favouring the formation of FG-material, whereas for the CG-region, the density of such seed crystals is reduced. This results in a grain growth along the cooling gradient, and leading to coarser, elongated grains, as typical for AM-processed materials. Next to the formation of the microstructure, fine dispersed particles are found, basically Al–Mg-oxides with varying composition. Their origin has to be found in the oxide
References [1] Buchbinder D, et al (2011) High Power Selective Laser Melting (HP SLM) of Aluminum Parts. Physics Procedia 12(Part A):271–278. [2] Buchbinder D, Meiners W, Wissenbach K, Propawe R (2014) Selective Laser Melting of Aluminium Die-Cast Alloy. in Demmer A, (Ed.) Fraunhofer Direct Digital Manufacturing Conference 2014, 6. [3] Cloots M, Uggowitzer PJ, Wegener K (2016) Investigations on the Microstructure and Crack Formation of IN738LC Samples Processed by Selective Laser Melting Using Gaussian and Doughnut Profiles. Materials & Design 89:770– 784. [4] Davydov VG, Elagin VI, Zakharov VV, Rostoval D (1996) Alloying Aluminum Alloys With Scandium and Zirconium Additives. Metal Science and Heat Treatment 38(8):347–352. [5] Hosford WF (2005) Mechanical Behavior of Materials, Cambridge University Press, New York447. [6] Jerrard PGE, Hao L, Dadbakhsh S, Evans KE (2010) Consolidation Behaviour and Microstructure Characteristics of Pure Aluminium and Alloy Powders Following Selective Laser Melting Processing. 36th International MATADOR Conference, Springer London, Manchester UK487–490. [7] Kolobnev NI (2002) Aluminum–Lithium Alloys with Scandium. Metal Science and Heat Treatment 44(7–8):297–299. [8] Morawiec A, Szpunar JA, Hinz DC (1993) Texture Influence on the Frequency of Occurrence of CSL-Boundaries in Polycrystalline Materials. Acta Metallurgica et Materialia 41(10):2825–2832. [9] Norman AF, Prangnell PB, McEwen RS (1998) The Solidification Behaviour of Dilute Aluminium–Scandium Alloys. Acta Materialia 46(16):5715–5732. [10] Roder O, Schauerte O, Lu¨tjering G, Gysler A (1996) Correlation Between Microstructure and Mechanical Properties of Al–Mg Alloys Without and With Scandium. Materials Science Forum 217–222(3):1835–1840. [11] Røyset J, Ryum N (2005) Scandium in Aluminium Alloys. International Materials Reviews 50(1):19–44. [12] Schmidtke K, Palm F, Hawkins A, Emmelmann C (2011) Process and Mechanical Properties: Applicability of a Scandium modified Al-alloy for Laser Additive Manufacturing. Physics Procedia 12(Part 1):369–374. [13] Siddique S, et al (2015) Influence of Process-Induced Microstructure and Imperfections on Mechanical Properties of AlSi12 Processed by Selective Laser Melting. Journal of Materials Processing Technology 221:205–213. [14] Spierings AB, Herres N, Levy G (2011) Influence of the Particle Size Distribution on Surface Quality and Mechanical Properties in Additive Manufactured Stainless Steel Parts. Rapid Prototyping Journal 17(3):195–202. [15] Spierings AB, Schneider M, Eggenberger R (2011) Comparison of Density Measurement Techniques for Additive Manufactured Metallic Parts. Rapid Prototyping Journal 17(5):380–386. [16] Spierings AB, et al (2015) Processing ODS Modified IN625 Using Selective Laser Melting. in Bourell DL, (Ed.) Proceedings of the Annual International Solid Freeform Fabrication Symposium, . [17] Thijs L, Kempen K, Kruth J-P, Van Humbeeck J (2013) Fine-Structured Aluminium Products with Controllable Texture by Selective Laser Melting of PreAlloyed AlSi10Mg Powder. Acta Materialia 61(5):1809–1819. [18] Vilaro T, Abed S, Knapp W (2008) Direct Manufacturing of Technical Parts Using Selective laser Melting: Example of Automotive Application. Assises Europe´ennes de Prototypage Rapide. [19] Vilaro T, et al (2012) Microstructural and Mechanical Approaches of the Selective Laser Melting Process Applied to a Nickel-base Superalloy. Materials Science and Engineering: A 534(0):446–451. [20] Zakharov VV (2003) Effect of Scandium on the Structure and Properties of Aluminum Alloys. Metal Science and Heat Treatment 45(7):246–253.
Contents lists available at ScienceDirect
CIRP Annals - Manufacturing Technology jou rnal homep age : ht t p: // ees .e lse vi er . com /ci r p/ def a ult . asp
Microstructure and mechanical properties of as-processed scandiummodified aluminium using selective laser melting Adriaan B. Spierings a,*, Karl Dawson b, Mark Voegtlin a, Frank Palm c, Peter J. Uggowitzer d a
Innovation Centre for Additive Manufacturing, INSPIRE-AG, Lerchenfeldstrasse 5, CH-9014 St. Gallen, Switzerland Centre for Materials and Structures, School of Engineering, University of Liverpool, Liverpool, UK c Airbus Group, Innovations, TX2, 81663 Munich, Germany d Laboratory of Metal Physics and Technology, Department of Materials, ETH Zurich, Wolfgang-Pauli-Str. 10, 8093 Zurich, Switzerland Submitted by Prof. Dr. Gideon Levy (1), TTA Technology Turn Around, Switzerland b
A R T I C L E I N F O
A B S T R A C T
Keywords: Selective laser melting (SLM) Aluminium Microstructure
Additive manufacturing (AM) offers significant benefits towards optimized lightweight structures for aircraft and space applications. High-strength aluminium alloys are of special interest to reach a maximum mass reduction. The paper presents the development of appropriate selective laser melting (SLM) processing windows for a scandium modified aluminium alloy (Scalmalloy1), reaching densities >99%. The mechanical properties of as-processed material are analyzed, pointing out a comparably low anisotropy with regard to the build orientation. A fine-grained microstructure is observed next to regions of coarser, elongated grains. The paper discusses the observed microstructure, and concludes with suggestions for innovative material design for AM. ß 2016 CIRP.
1. Introduction Additive manufacturing technologies such as selective laser melting (SLM) are still on the threshold to industrial application. Especially in high-tech sectors the technologies offer benefits e.g. in terms of lightweight and functionally optimized parts and applications. Thereby the degree of an achievable mass reduction in lightweight applications depends not only on structural optimisations (e.g. the integration of lattice structures or topology optimization), but also on the mechanical properties of the lightweight material used. High-strength structural aluminium alloys are therefore of special interest for the SLM technology. There is a basic need to develop aluminium alloys, designed specifically for additive manufacture, which will satisfy a broad range of industrial requirements. For lightweight applications, there are currently very few alternatives to AlSi10Mg. However, although AlSi10Mg can offer good mechanical properties, in both as-processed and heat treated condition, these properties can be influenced significantly by the selection of different SLM-processing parameters. This makes it more difficult for industry to use AM for real applications, and consequently every machine and processing window needs to be qualified separately. A potential solution to this problem is designing alloy systems being more suitable to the specific SLM process conditions, also with regard to variations of alloying elements,
main alloying element. Examples are the AlSi12 and AlSi10Mg alloys, respectively. They were originally developed for casting processes, offering good castability and reduced shrinkage imparted by the presence of a relatively large volume fraction of Al–Si eutectic with a corresponding low melting temperature. Mechanical properties of SLM-processed AlSi10Mg are shown in Table 1. Buchbinder [1,2] reports that changing the processing window (basically laser power PL) influences the mechanical properties significantly, as for higher PL and consequently higher scan speed vs the cooling rate is increased, leading to a more fine grained material. Thijs [17] investigated how microstructures could be influenced by the implementation of different scan strategies and demonstrated that fine grain structures, which were developed as sub-micron cells, grew along h100i crystal directions from the melt pool border towards the centre of the melt pool, hence along the temperature gradient. Columnar grains are further growing at the centreline of the scan tracks along the build direction. It was also reported that due to the moving heat source and scan strategy applied, the thermal gradients and growth rate vary over the melt pool, and consequently the fineness of cells and possible textures changed over the cross-section of the melt pool. Hence anisotropic mechanical behaviour is a typical result for additively processed Table 1 Mechanical properties for SLM-AlSi10Mg, serving as an age hardenable alloy for comparison to Scalmalloy1. Left column: Build orientation.
1.1. State of the art To date, the most commonly used aluminium alloys for SLM are casting alloys from the 4XXXer series, containing silicon as the * Corresponding author. http://dx.doi.org/10.1016/j.cirp.2016.04.057 0007-8506/ß 2016 CIRP.
08 908 08 908 a
s0.2 (MPa)
sUTS (MPa)
A (%)
Ref.
240 215 250 235
420 400 350 275
5.9 1 3.2 0.5 2.5 1.1
(Buchbinder et al., 2011)a
Without preheating of the build platform.
[18]
214
A.B. Spierings et al. / CIRP Annals - Manufacturing Technology 65 (2016) 213–216
aluminium, although reported property values are overcoming the requirements of EN-1706 for all build directions [2]. Siddique [13] investigated AlSi12, processed with different specific laser energy inputs, base plate heating and post-build stress relief combinations. He concluded that high strength can be achieved by developing a fine microstructure, but at the cost of fracture strain. Again changing the processing conditions significantly affected the mechanical properties, and a certain anisotropy is obvious. Magnesium serves as the primary alloying element in 6XXX series alloys, with smaller additions of Si. These alloys are age hardenable. Jerrard et al. [6] showed the basic SLM-processability of an Al6061 alloy, and investigated the consolidation behaviour compared to pure Al. He found a better material density as Mg and Si significantly improved wetability; however, mechanical properties were not reported. 1.2. Scandium modified aluminium (Scalmalloy1) Scalmalloy1 is a protected Sc- and Zr-modified aluminium alloy concept for AM procedures, developed by Airbus Group Innovations. The metallurgical origins can be traced back to materials developed, during the 1970s, in the US and former Soviet Union [7]. The addition of Sc to the Al alloy facilitates good levels of microstructural control, however, due to the high cost of Sc and high buy-to-fly ratios in conventional lightweight engineering, such alloys have not yet caught the attention of industry. The effects of Sc on the properties of Al-alloys have been investigated by several researchers [4,9–11,20], and can be summarized as follows: – The Al–Sc solid solution can decompose upon cooling, forming fine dispersed, fully coherent intermetallic Al3Sc precipitates of the cubic LI2 crystal structure, acting as an age hardener and leading to high strength, with an strength increment of about 40 to 50 MPa per 0.1 wt.% Sc. – The precipitation of Al3Sc takes place at higher temperatures (275–325 8C) compared to classical age hardening high strength Al-alloys (150–200 8C), leading to a high resistance against premature precipitates coarsening (e.g. strong heat affected zone occurrence). – Al3Sc precipitates can prevent recrystallization even up to high temperatures. – Al3Sc particles are known to be very effective for grain refinement, acting as seed crystals during solidification (heterogeneous nucleation). – Sc is claimed to improve general weldability in Al alloys, e.g. for alloys which are susceptible to hot cracking and extended heat affected zone formation. These effects are very welcome for AM where process-specific effects such as e.g. the formation of columnar grains and the corresponding anisotropic material behaviour is known for many alloys [14,19]. As in AM the fly-to-buy ratio is much closer to 1 and therefore the cost-influence of Sc is limited, Sc-containing alloys might be an interesting way to overcome some of these microstructure related properties. Schmidtke [12] proofed the basic SLM-processability of Scalmalloy1, and presented first mechanical properties in a post-built-up heat treated condition. 1.3. Concept of the paper The paper presents the development of an appropriate SLM processing window, and analyses the micro-structure in the as-processed condition. Static mechanical properties are presented and discussed. The paper concludes with assumptions on the formation mechanisms for the observed microstructure, which will serve as a basis for future alloy design for AM. 2. Materials and methods 2.1. SLM process The SLM-process is affected by many processing variables, including powder properties and machine setup. A ConceptLaser
M2 machine was used equipped with a 200 W Nd-YAG laser operated in cw mode. The Gaussian beam focus diameter is 100 mm [3]. Next to a given powder the main parameters affecting the process are the laser power PL, the layer thickness tL, the scan velocity vs and the hatch distance d. Due to high reflectivity of aluminium, PL was set to the maximum of 200 W to optimize process productivity, and the layer thickness tL was 30 mm. The hatch distance d was varied between 90 mm and 225 mm and scan velocities vs were varied to optimize material density. The volume specific energy density EV combines the parameters to a single main driving parameter for material density. PL ðJ=mm3 Þ EV ¼ (1) vs t L d 10 10 10 mm3 test samples were produced using an alternating scan strategy from layer to layer, and bi-directional scanning within the layers as given in Fig. 1. To avoid over-heating effects in scan reversals, no ‘‘island-scanning’’ was applied, which would be the typical scan strategy used on ConcepLaser machines. In total > 120 samples were produced, and for promising processing windows 3 samples were analyzed to assess the standard deviation of the material density.
Fig. 1. Scan strategy for SLM test samples. z-direction indicates the build orientation.
2.2. Scalmalloy1 powder raw material The chemical composition of the powder used during these trials is presented in Table 2. The particle size distribution was measured by optical particle analysis using PowderShape1. Thereby, the particles are filtered for an area-equivalent diameter < 120 mm; more details on the measurement principle are given by Spierings [16]. Table 2 Chemical composition (in wt.%) of the material according to supplier information. D10/D50/D90 are the quantiles for number (q0) and volume (q3) weighted particles size distributions. Mg
Sc
4.6
0.66 0.42 q0 D10 3.9 mm q3 D10 38 mm
Zr
Mn
Fe [12]
Others [12]
0.49
0.068 D50 6.2 mm D50 58 mm
0.05 Bal. D90 49.9 mm D90 83 mm
Al
2.3. Measurement techniques The density (r) of consolidated material was determined by the Archimedes method, using a theoretical maximum density of 2.67 g/cm3, as described by Spierings et al. [15]. Density measurements were corroborated by optical measurements showing minor residual porosity in the builds. The microstructures of as-processed samples were analyzed using a FEI Helios dual beam FIB instrument equipped with an EDAX-EBSD system. Electro-polished surfaces were prepared using a Struers Tenupol 3 twin jet electro-polisher, a 5 vol.% perchloric acid in 95 vol.% methanol electrolyte, cooled to 50 8C, voltage of 25 V. High resolution EBSD maps were collected using a step size of 80 nm in areas of fine grained material (10K magnification), and 250 nm steps were used to map large areas of coarse grained material at lower 750 magnification. All EBSD maps were recorded using beam conditions of 20 kV, 5.5 nA and a 12 mm working distance. The cutting plane orientations for SEM measurements are given in the respective figure captions, using the coordinate system shown in Fig. 1.
A.B. Spierings et al. / CIRP Annals - Manufacturing Technology 65 (2016) 213–216
215
2.4. Static mechanical testing Static tensile specimens according to DIN-500 125-A525 were manufactured in the horizontal (908) and vertical (08) direction. The samples were turned to a surface finish N5 in the as-built condition (without heat treatment), and mechanically tested in accordance with EN-10002/ISO-6892. From the measured ‘‘engineering’’ stress– strain curves, the true stress–strain curves and true ultimate strength (Rm, T) is calculated according to Hosford [5] by
s T ¼ s ð1 þ eÞ;
Rm;T ¼ Rm ð1 þ eÞ;
2 T ¼ lnð1 þ eÞ
(2)
3. Results 3.1. SLM-processing window The processing window shows good attainable material density >99% in a range of EV between 75 J/mm3 and 240 J/mm3 and for d between 135 mm and 165 mm, as shown in Fig. 2. The high thermal conductivity of the Al-alloy leads to scan tracks widths of 260 20 mm, allowing hatch distances d larger than the laser spot diameter. Smaller d are leading to overheating effects resulting in uneven scanning surfaces, hence to irregular next powder layers and consequently to a bad build morphology. Higher d reduces EV to values too small to reach sufficient density. The general reproducibility is acceptable with a mean standard deviation of 0.09% for the considered d values and corresponding densities >99%. The size and type of defects per mm2, taken from xy-micrographs of cube samples, is presented in Fig. 3 showing acceptable values for EV > 120 J/mm3. Only for very high EV > 230 J/mm3 the number of defects can further be minimized, but on the costs of scanning productivity (Fig. 3). Two different EV-levels were selected for the production of tensile test specimens: EV1 = 238 J/mm3 and EV2 = 135 J/mm3, to analyze the impact of varying EV on mechanical properties.
Fig. 4. Static mechanical properties for as-processed Scalmalloy.
These values indicate a comparably very low anisotropic mechanical behaviour of the SLM-processed alloy, which is correlated to the material microstructure. 3.3. Material microstructure The microstructure of the consolidated material was analyzed for two EV-levels EV1 and a second level EV3 = 115 J/mm3, which is roughly half of EV1, and close to level EV2. Fig. 5 shows SEM microstructure processed using EV1, showing a very fine grained (FG) microstructure, next to regions of coarser grains (CG). In Fig. 5, a high amount of bright particles can be found in the FG material, most of them being located at grain boundaries. Additional EDS analysis on TEM foils reviled that these are Mg–O particles with varying composition, probably also MgAl2O3 spinel. Fig. 6 presents the microstructure of material processed with ½EV1 showing a similar structure with regard to FG and CG material. Due to the weld-lines depth of 100 mm and the corresponding high amount of re-melting of previous layers, the perpendicular and longitudinal weld-line cross-sections cannot be identified very clearly (Fig. 6a). But it shows that solidification takes place either in a columnar structure with elongated grains along the build (z-) direction, or a very fine grained structure without any preferential grain orientation, as the corresponding discrete pole figure (Fig. 6c) does not indicate any preferential grain orientation. In contrast, Fig. 6d shows that the CG material displayed a preferential crystallographic h100i orientation approx. parallel to the build (z-) axis of the samples, which is typical for AM-processed materials.
Fig. 2. Material density in dependence of EV and hatch distances d. Dotted lines represent too small d values.
Fig. 5. CBS-images taken in a xz-/yz-plane for EV1. Fine (FG) and coarse (CG) grains are indicated.
Fig. 3. Number, size and type of defects per mm2 for different EV-levels.
3.2. Static mechanical properties High static mechanical properties, overcoming the values reported for AlSi10Mg (Table 1), are reached in the as-processed condition (Fig. 4), with true Rm,T > 400 MPa and RP0.2 > 277 MPa for all build orientations and EV. The E-modulus is 68.0 0.8 GPa for EV1 and 73.6 1.2 GPa for EV2. The anisotropy between horizontal (908) and vertical (08) build orientation in both EV-levels is <2% for true Rm and <3.4% for RP0.2.
Fig. 6. Microstructure processed with EV3 showing FG and CG regions. (a) Edched and EBSD (b) microstructure of the same region. (c) EBSD of FG-material and [001] Pole figures for FG (left) and CG region (right).
216
A.B. Spierings et al. / CIRP Annals - Manufacturing Technology 65 (2016) 213–216
An image analysis revealed that the FG, as-built structure accounts for approx. 52–56% of the total cross-section. Similar to this processing window shown in Fig. 6, Fig. 7(left) presents the microstructure of material processed with EV1, showing a comparable amount of 49–56% of the total cross-sections as FG. The distribution of the misorientation angles between single grains (Fig. 7(right)) in the FG-region supports that the fine grains are rather randomly oriented, as the distribution follows nicely the ideal distribution for random orientation [8]. With regard to grain sizes Fig. 8 presents its distributions for FG and CG regions of samples processed with the two energy levels EV1 and EV3. Only small differences in FG-regions from both processing energy levels EV can be observed. Assuming a normal distribution of grain sizes, with mean xEV1 ¼ 0:6 mm and xEV2 ¼ 1:05 mm. The ¯ ¯ CG-material thereby shows coarser grains, with sizes up to 13 mm. Even such grain sizes are considered as ‘‘fine’’ compared to other manufacturing processes.
layer from the powder raw material, which is broken up during processing. In FG material, such particles are embedded at the grain boundaries, whereas in CG material they are expected to be pushed at the solidification front, leading to a much lower concentration of such particles. 5. Outlook Work is proceeding with regard to the full understanding of the formation mechanisms for the observed microstructure, different types of particles and precipitates. Appropriate heat treatments will be developed, and related static and dynamic mechanical properties have to be measured. Full understanding of the ‘‘working principle’’ of this alloy system might pave the way to the development of other AM-designed alloys. Acknowledgments The authors gratefully acknowledge Prof. Dr. K. Wegener, Mr. Demont, Mr. Frauchiger and Mr. Vincenz for their support of this project. Part of this project is co-financed by the Swiss Commission for Technology and Innovation, CTI, No. 17365.1.
Fig. 7. EBSD in selected area (10K magnification) in FG as-built material processed by EV1 (left), grain orientation angle distribution in this FG area (right).
Fig. 8. Grain size distribution for FG material (EV1 and EV3) and CG material (EV1).
4. Discussion and conclusion The Sc-/Zr-modified Al-alloy (Scalmalloy1) is well processable with SLM, reaching densities >99%. Suitable processing windows require a laser energy input significantly higher compared to other Al-, Fe- or Ni-based alloys. The processing window is robust with regard to changes in EV, as long as the EV > 120 J/mm3. Both material density and general microstructure is not significantly affected by variations of EV, although lower energy input leads to a further refinement of the fine grains. This ensures stable mechanical properties, overcoming the properties of other typical SLM Al-alloys, e.g. AlSi10Mg. Thereby a higher EV reduces the number of defects, but with almost no effect on mechanical performance except for E-modulus, which is reduced for a slightly higher defect level. The mechanical performance is related to some very interesting microstructural features, making Scalmalloy1 an ideal candidate as a structural alloy for AM. Due to the Hall–Petch relationship, the high content of a very fine-grained microstructure without any preferential grain orientation supports high material strength, with low anisotropic properties. The reasons for the ‘‘bimodal’’ grain size distribution are not yet fully understood. However, it is assumed that a high number density of Al–Sc seed crystals are formed during consolidation, favouring the formation of FG-material, whereas for the CG-region, the density of such seed crystals is reduced. This results in a grain growth along the cooling gradient, and leading to coarser, elongated grains, as typical for AM-processed materials. Next to the formation of the microstructure, fine dispersed particles are found, basically Al–Mg-oxides with varying composition. Their origin has to be found in the oxide
References [1] Buchbinder D, et al (2011) High Power Selective Laser Melting (HP SLM) of Aluminum Parts. Physics Procedia 12(Part A):271–278. [2] Buchbinder D, Meiners W, Wissenbach K, Propawe R (2014) Selective Laser Melting of Aluminium Die-Cast Alloy. in Demmer A, (Ed.) Fraunhofer Direct Digital Manufacturing Conference 2014, 6. [3] Cloots M, Uggowitzer PJ, Wegener K (2016) Investigations on the Microstructure and Crack Formation of IN738LC Samples Processed by Selective Laser Melting Using Gaussian and Doughnut Profiles. Materials & Design 89:770– 784. [4] Davydov VG, Elagin VI, Zakharov VV, Rostoval D (1996) Alloying Aluminum Alloys With Scandium and Zirconium Additives. Metal Science and Heat Treatment 38(8):347–352. [5] Hosford WF (2005) Mechanical Behavior of Materials, Cambridge University Press, New York447. [6] Jerrard PGE, Hao L, Dadbakhsh S, Evans KE (2010) Consolidation Behaviour and Microstructure Characteristics of Pure Aluminium and Alloy Powders Following Selective Laser Melting Processing. 36th International MATADOR Conference, Springer London, Manchester UK487–490. [7] Kolobnev NI (2002) Aluminum–Lithium Alloys with Scandium. Metal Science and Heat Treatment 44(7–8):297–299. [8] Morawiec A, Szpunar JA, Hinz DC (1993) Texture Influence on the Frequency of Occurrence of CSL-Boundaries in Polycrystalline Materials. Acta Metallurgica et Materialia 41(10):2825–2832. [9] Norman AF, Prangnell PB, McEwen RS (1998) The Solidification Behaviour of Dilute Aluminium–Scandium Alloys. Acta Materialia 46(16):5715–5732. [10] Roder O, Schauerte O, Lu¨tjering G, Gysler A (1996) Correlation Between Microstructure and Mechanical Properties of Al–Mg Alloys Without and With Scandium. Materials Science Forum 217–222(3):1835–1840. [11] Røyset J, Ryum N (2005) Scandium in Aluminium Alloys. International Materials Reviews 50(1):19–44. [12] Schmidtke K, Palm F, Hawkins A, Emmelmann C (2011) Process and Mechanical Properties: Applicability of a Scandium modified Al-alloy for Laser Additive Manufacturing. Physics Procedia 12(Part 1):369–374. [13] Siddique S, et al (2015) Influence of Process-Induced Microstructure and Imperfections on Mechanical Properties of AlSi12 Processed by Selective Laser Melting. Journal of Materials Processing Technology 221:205–213. [14] Spierings AB, Herres N, Levy G (2011) Influence of the Particle Size Distribution on Surface Quality and Mechanical Properties in Additive Manufactured Stainless Steel Parts. Rapid Prototyping Journal 17(3):195–202. [15] Spierings AB, Schneider M, Eggenberger R (2011) Comparison of Density Measurement Techniques for Additive Manufactured Metallic Parts. Rapid Prototyping Journal 17(5):380–386. [16] Spierings AB, et al (2015) Processing ODS Modified IN625 Using Selective Laser Melting. in Bourell DL, (Ed.) Proceedings of the Annual International Solid Freeform Fabrication Symposium, . [17] Thijs L, Kempen K, Kruth J-P, Van Humbeeck J (2013) Fine-Structured Aluminium Products with Controllable Texture by Selective Laser Melting of PreAlloyed AlSi10Mg Powder. Acta Materialia 61(5):1809–1819. [18] Vilaro T, Abed S, Knapp W (2008) Direct Manufacturing of Technical Parts Using Selective laser Melting: Example of Automotive Application. Assises Europe´ennes de Prototypage Rapide. [19] Vilaro T, et al (2012) Microstructural and Mechanical Approaches of the Selective Laser Melting Process Applied to a Nickel-base Superalloy. Materials Science and Engineering: A 534(0):446–451. [20] Zakharov VV (2003) Effect of Scandium on the Structure and Properties of Aluminum Alloys. Metal Science and Heat Treatment 45(7):246–253.