Contents lists available at ScienceDirect

Vacuum

journal homepage: www.elsevier.com/locate/vacuum

Microstructure characteristics and its formation mechanism of selectivelaser melting SiC reinforced Al-based composites

Xuan Zhaoa,b, Dongdong Gua,b,∗, Chenglong Maa,b, Lixia Xia,b, Han Zhanga,b

a College of Materials Science and Technology, Nanjing University of Aeronautics and Astronautics, Yudao Street 29, Nanjing, 210016, Jiangsu Province, PR Chinab Jiangsu Provincial Engineering Laboratory for Laser Additive Manufacturing of High-Performance Metallic Components, Nanjing University of Aeronautics andAstronautics, Yudao Street 29, Nanjing, 210016, Jiangsu Province, PR China

A R T I C L E I N F O

Keywords:Selective laser meltingMicrostructureIn-situ reactionNanohardness

A B S T R A C T

SiC ceramic reinforced Al-based composites were successfully synthetized by selective laser melting (SLM) of8.5 vol % SiC/AlSi10Mg powder mixtures. The mechanism of microstructure evolution in an individual moltenpool of the SLM processing was investigated. Three characteristic zones were well distinguished, includingcenter and boundary areas of the molten pool as well as the heat-affected zone. The results revealed that themicrostructure changing from columnar dendrite growth to cellular growth along the building direction could beattributed to the change of G/R (G is temperature gradient and R is solidification velocity). Phase identificationof the SLM-fabricated parts was performed by X-Ray Diffraction (XRD) and Energy Dispersive Spectroscopy(EDS) analysis, further demonstrating the in-situ formed phases in the molten pool. Al4SiC4 crystals were formedin situ around the partially melted SiC ceramic particles. The nanoindentation tests conducted on different areasof the SLM-fabricated Al-based composites exhibited the high nanohardness (2.27 GPa and 2.15 GPa) and theelastic modulus (78.94 GPa and 75.46 GPa) in the center and boundary of molten pool, respectively, which weremuch higher than the reported value of conventional materials.

1. Introduction

In recent years, demands of high-performance aluminum matrixcomposites (AMCs) in the automotive and aerospace industries aredrastically increasing due to their high specific strength and stiffness,excellent wear resistance and low coefficient of thermal expansion[1,2]. AMCs combined with favorable properties of aluminum matrixand reinforcing phases, have also received considerable research in-terest [3,4], in which a variety of ceramic reinforcements, such as SiC,Fe2O3, TiC, Al2O3 and TiB2 [5–9], have been often used due to theirhigh hardness and excellent corrosion resistance. Notably, largeamounts of functional components with complex geometries, such asairscrew, oil tube, etc., are involved in the automotive and aerospacefields. These complex configurations along with relatively low ambientductility of AMCs in comparison with aluminium alloys, remain chal-lenging for precisely shaping of the complex components through tra-ditional processing routes [10].

Selective laser melting (SLM), as one of rapidly developed additivemanufacturing techniques, has received much attention in terms of itsgreat potential to near net shaping of metallic components with

complex structures [11–16]. Based on complete melting philosophy,SLM process is characterized by selectively melting of the powder layerand subsequently rapid solidification of the mesoscale molten poollayer by layer [17,18]. The complex non-equilibrium physical andchemical metallurgical nature of SLM processing can cause a homo-genous dispersion of reinforcements in the matrix and induce complexinteractions at the interface between the Al matrix and the reinforce-ments [19,20], thereby tailoring microstructures and performances ofthe manufactured AMC components appropriately. Dadbakhsh et al.investigated the influence of materials and processing windows duringSLM on fabrication of Al/Fe2O3 composite parts from powder mixtures[6]. An in-situ reaction occurred by modifying the visual surface andaltering material characteristics in the SLM processing windows andimproved mechanical properties were obtained with addition of Fe2O3

[6]. Gu et al. fabricated novel TiC/AlSi10Mg nanocomposites partswith ring-structure nano-TiC reinforcements uniformly distributedalong the Al grain boundaries by SLM [7]. These nanocomposite partsdemonstrated a remarkable improvement in both tensile strength andmicrohardness. Obviously, addition of the reinforcements not onlycomplicates the physical/chemical reaction at the interface between the

https://doi.org/10.1016/j.vacuum.2018.11.022Received 6 September 2018; Received in revised form 7 November 2018; Accepted 13 November 2018

∗ Corresponding author. College of Materials Science and Technology, Nanjing University of Aeronautics and Astronautics, Yudao Street 29, Nanjing, 210016,Jiangsu Province, PR China.

E-mail address: dongdonggu@nuaa.edu.cn (D. Gu).

Vacuum 160 (2019) 189–196

Available online 14 November 20180042-207X/ © 2018 Published by Elsevier Ltd.

T

Al matrix and the reinforcements, but also results in unfavorablebonding due to poor wettability of the reinforcements by molten Alalloys. Due to rapid melting/consolidation nature of SLM, the moltenmaterial tends to experience a distinctively physical and chemicalmetallurgical process during high-energy non-equilibrium processing,indicating complicated modes of heat and mass transfer and forming aunique microstructure in the AMCs [21]. Limited work has been per-formed on quantitatively thermodynamic investigation of AMCs duringSLM processing, which has an important influence on the micro-structure evolution. Owing to rapid melting/solidification of powdermaterial and resultant small molten pool dimension, it is difficult todirectly obtain transient temperature and velocity profiles through ex-periment methods [22,23]. Hence, numerical simulation is essential toanalyze temperature evolution within an individual molten pool duringSLM.

This work has investigated the microstructural evolution of theSLM-processed Al-based composite parts within an individual moltenpool. Based on a numerical simulation, the solidification nucleation andgrowth mechanism of the local molten pool were discussed. The formedphases and interaction mechanism of the SLM produced SiC/AlSi10 Mgduring SLM processing were analyzed in detail. The structure-depen-dent nanohardness measured within an individual molten pool werealso discussed.

2. Experimental procedures

2.1. Materials

Spherical AlSi10Mg powder (99.8% purity) with a particle sizeranging from 20 to 63 μm and poly-angular SiC ceramic powder (99.6%purity) with a mean particle size of 3.5 μm were used in this study. AQM series Planetary Mill (Nantai instrument co. LTD., NanJing) wasemployed to achieve a homogeneous powder mixture, which wascomposed of 91.5 vol % AlSi10Mg and 8.5 vol % SiC ceramic powder.The powder mixture was prepared in an argon atmosphere in a glovebox. Then the milling process was carried out under a condition of aball-to-powder weight ratio of 1:1 and a rotation speed of 200 rpm. Themilling time was performed in a mode containing a 15-min milling and5-min pause and was set to 2 h. Subsequently, mixed powder with aweight of 2 kg was prepared for the SLM process.

2.2. Selective laser melting process

An independently developed SLM system is equipped with a YLR-500-WC ytterbium fiber laser (IPG Laser GmbH, Germany) with a spotsize of 70 μm and a maximum power of 500W, an automatic powderspreading device, and a computer system for process control. Before theSLM processing, an aluminum substrate was fixed on a building plat-form (174.50 mm×174.50 mm×14.50mm) and leveled precisely.Then, the processing chamber was pumped to high vacuum conditionsand then filled with high-purity argon at a pressure of 200 Pa to care-fully control the O2 content below 20 ppm and minimize oxidationduring the overall SLM process. During SLM, the mixed powder wasspread homogeneously on the substrate by the powder spreading deviceand the laser beam scan the pre-spread powder selectively based on theslice profile of computer-aided design (CAD) model. An inter-layerstagger scanning strategy was applied in this work. According to theprevious experiments, the processing parameters were set as follows:laser power p=350W, scan speed v=2500mm/s, hatch spaceh=50 μm and layer thickness l=50 μm. Finally, the samples withdimensions of 11mm×11mm×5mm were fabricated accordingly.

2.3. Material characterization

The cross sections of the SLM samples along the building directionwere grounded and polished according to the standard procedures. For

microstructure observations, the polished surfaces were then etchedwith Keller's reagent containing HF (1.0mL), HCl (1.5 mL), HNO3

(2.5 mL), and distilled water (95mL) for 20 s. The microstructure of theSLM-processed part was characterized using an S-4800 field emissionscanning electron microscope (FE-SEM, Hitachi, Tokyo, Japan) at avoltage of 5.0 kV. The chemical compositions were determined by anEDAX energy dispersive X-ray spectroscope (EDX) (EDAX, Inc.,Mahwah, NJ), using a super-ultrathin window sapphire detector. Thephases of the specimens were identified by a Bruker D8 Advance X-raydiffractometer (XRD) (Bruker AXS GmbH, Karlsruhe, Germany) with CuKα radiation at 40 kV and 40mA, using a continuous scan mode of 4°/min. Nanohardness tests were performed at room temperature usingDUH-W201S nanoindentation tester (Shimazdu Corporation, Tokyo,Japan). Before the nanoindentation tests, the surface of the sample waspolished according to the standard procedures. A loading-unloadingtest mode was used, and the maximum load applied was 1000 μN with aloading/unloading rate of 250 μN/s, holding time for 5 s at the max-imum load. During the nanohardness measurements, the load and in-dentation depth were recorded, and then were used to constructloading-unloading plots. The uncertainty of nanohardness and elasticmodulus values is about 3% and 4%, respectively.

2.4. Numerical simulation

In order to study the thermal evolution behavior and velocity fieldwithin the molten pool during the SLM processing of AlSi10Mg/SiCcomposites, the simulation was conducted using the commercial com-putational fluid dynamics software FLUENT. The model used in thisstudy has been described elsewhere [19,22]. The laser heat source witha scan speed of 2500mm/s followed a Gaussian distribution is mathe-matically defined as a heat flux in this model. Furthermore, the powderbed is regarded as the homogeneous and continuous media. The fluidmotion in the molten pool is considered to be Newtonian and in-compressible fluid. Flow in the molten pool is considered to be laminar[23]. Meanwhile, thermal conductivity (k) and specific heat (CP) of theused materials are temperature-dependent in the melting process. Theas-used material properties and SLM processing parameters are listed inTable 1. The variation of thermal-physical parameters of AlSi10Mg andSiC with temperature are shown in Fig. 1.

3. Results and discussion

3.1. Microstructural evolution in different regions of molten pool

Fig. 2 shows the microstructural features in different regions of amolten pool. The adjacent melting tracks after solidification were ob-vious and boned well without pores observed, thus indicating that theSiC/AlSi10Mg composites possessed good processability as well asgood wettability between the SiC ceramic particles and Al matrix.Fig. 2a shows the cross section of the molten pool with three typicalcharacteristic zones of SLM-processed AMCs [27]: center zone (zone I),boundary zone of a molten pool (zone II), and heat-affected zone (HAZ)(zone III). It is observed that the remaining SiC ceramic particles wererandomly distributed in the molten pool (as seen in Fig. 2a). Thecharacteristic morphologies of molten pool in three different regions

Table 1Properties of as-used material and SLM processing parameters.

Parameters Values

Laser absorptivity of the AlSi10Mg powder 0.09 [16]Laser absorptivity of the SiC powder 0.78 [24]

Ambient temperature, T0 300 KPowder layer thickness, l 50 μm

Hatching space, h 50 μmRadius of the laser beam, ω 70 μm

X. Zhao et al. Vacuum 160 (2019) 189–196

190

are shown in Fig. 2b–d. In the center of molten pool (zone I), finecellular microstructure dominated and the average cell size reached∼0.3 μm (Fig. 2b). A relatively clear interface between adjacent α-Algrains was observed (Fig. 2b), which could be attributed to the pre-cipitation of primary Si particles along the grain boundaries. But theinterface became vague and fiber-like microstructure with typical

eutectic microstructure formed along the α-Al grain boundaries close tothe zone II (Fig. 2b). The microstructure located at the boundary ofmolten pool (zone II) was transformed from cellular to dendritic mor-phology (Fig. 2c), simultaneously showing an apparent orientationwhich could be ascribed to great temperature gradients. This is in goodagreement with the results reported by Kimura et al. [28]. A smallamount of SiC ceramic particles and newly formed phases with a stripstructure were observed in the zone II (Fig. 2c). It is noted that the fiber-like microstructure extended largely in this zone, which is ascribed to arelatively small cooling rate and coarsening growth at the boundaries.For the heat-affected zone (zone III), a number of dispersed particleswere formed in the matrix, inferred to Si precipitates (Fig. 2d). Simi-larly, Thjis et al. found dispersed particles in the heat-affected zone ofthe SLM-prepared AlSi10Mg, demonstrating that these particles wereprecipitated eutectic Si phase [29].

It is found that the microstructure of molten pool was non-uniformand showed a significant change at different regions. This is expected tobe induced by various temperature gradients in the small-sized moltenpool during the high-energy laser irradiation. This is in good agreementwith the results reported by Liu et al. [21], where the cooling rate in-duced discrepancy in microstructural gradient and element distributionduring SLM. The temperature at the boundary was much lower thanthat inside the molten pool because of the heat flux based on a Gaussiandistribution. The temperature distribution plots of melt pool from topand cross-section view are shown in Fig. 3. The melt pool exhibited anelongated shape along the scanning direction (as marked by whitearrow). The appearance of the ellipse shape instead of circular shapecan be attributed to the effect of the movement of heat source. Fig. 3bexhibits the transient temperature distribution at the cross section ofthe molten pool as the laser interacts with the powder bed. The whitedash line corresponds to the isotherm of the melting temperature of theAlSi10Mg powder, the region inside which had higher temperaturethan the melting temperature of AlSi10Mg, resulting in a small moltenpool in this area. As the maximum temperature of molten pool obtainedfrom simulation was 1150 K, much lower than 2827 °C (the meltingpoint of SiC ceramic [30]), it is supposed that the SiC ceramic particleswere not melted completely under the experimental condition.

Fig. 1. Variation of thermal conductivity and specific heat of (a)AlSi10Mg and(b) SiC with temperature [25,26].

Fig. 2. FE-SEM image showing the cross sections of characteristic melt pool of the SLM-processed parts (a) and magnified structure located in different zones: (b)Center zone I; (c) Molten pool boundary zone II; (d) Heat-affected zone III.

X. Zhao et al. Vacuum 160 (2019) 189–196

191

However, with a high-energy laser the SiC ceramic particles would insitu react with aluminum melt partially, thereby forming a strip-struc-tured phases close to the unmelted SiC. As reported by D. Gu et al. [31],reinforcing phases included unmelted micron-sized SiC particles, in-situformed micron-sized Al4SiC4 strips and in-situ produced submicronAl4SiC4 particles in the SLM of AlSi10Mg/SiC (20 wt.%). It can bepreliminarily inferred that the newly formed strip-structured phases areAl4SiC4.

Fig. 3c shows the Marangoni flow with a radially circulating flowpattern, transferring heat and mass within the molten pool. As the laser-induced molten pool exhibits large temperature gradients due to in-troduce of Gaussian laser heat source, this will make contributions tosurface tension gradients accordingly and cause a Marangoni flow fromlow surface tension to high surface tension area on the surface of themelt pool. During the SLM of SiC/AlSi10Mg composites, migration ofSiC ceramic reinforcements is affected by two different forces. The SiCceramic particles were dragged by the friction force resulting from theviscosity of the melt despite the thermal capillary force driving themigration of SiC ceramic particles. The SiC ceramic particles can cir-culate several times by the motion of fluid in the molten pool. The in-tensity of the fluid flow plays a significant role in the dispersion of theSiC ceramic reinforcements in the Al matrix. This explains why the SiCceramic particles were randomly distributed in the molten pool (seeFig. 2a).

The profiles of temperature distribution versus distance from the topsurface and cooling rate versus time calculated by finite element ana-lysis method are shown in Fig. 4 in order to reveal the solidificationfeatures of zones I, II and III in the molten pool. The cooling rate varied

from a negative value to a positive value when the laser beam scansapproach and leave away from the detected point, corresponding to therespective melting and solidification processes of the SLM-induced meltpool. The calculated maximum temperature of zones I, II and III was1149 K (Figs. 4b), 1060 K (Fig. 4c) and 879 K (Fig. 4d), respectively andthe corresponding maximum cooling rate was 2.55×106 K/s,1.80×106 K/s and 0.33×106 K/s, respectively. This indicates that themaximum cooling rates decrease with distance from the top surfacewithin an individual molten pool. As reported by Rosenthal et al. [32],the solidification of Al alloy in the SLM processing depends on thetemperature gradient(G) and on the solidification velocity (R) in themolten pool. A relative low value of R at a constant G contributed toformation of a stable planar consolidation front, whereas with an in-creasing solidification velocity the consolidation front changes from acellular to dendritic structure. Similarly, for the different morphologiesin the molten pool of the SLM-processed Al-based composite parts, G,solute concentration (C) and R (R=V/G, V is cooling rate) are the threefactors affecting the growth mode of structure.

According to the data obtained directly from the simulation results,the G/R in the center region (3.56× 106 K s/m2) was almost one ordermagnitude smaller than that at the boundary (6.72×107 K s/m2) of theindividual molten pool. As a result, the combined effects of relativelylow G/R and low solute concentration of Si lead to development of acellular structure in the center of the molten pool. On the contrary,columnar dendrites is well developed in the boundary region (Fig. 2).This explains that the microstructure of the SiC/AlSi10Mg compositeschanged from a cellular structure to a dendritic structure with an in-creasing distance from the center of molten pool.

Fig. 3. Temperature contour plots from top view (a), the simulated molt pool profile at cross section (b), and Velocity vector plots in the molten pool with the actionof Marangoni effect from cross-sectioned view (laser power 350W, scan speed 2500mm/s) (c).

X. Zhao et al. Vacuum 160 (2019) 189–196

192

Due to the layer processing strategy of SLM, remelting process isanother possible factor contributing to formation of non-uniform mi-crostructure, which occurs between two adjacent molten pools. Theregions at the boundary of the molten pools can melt twice because ofthe hatch overlap, and the grains developed along the positive tem-perature gradient. Si phases can precipitate from Al matrix sufficientlyand even globular Si particles grow coarsely. As a result, the remeltingeffect leads to the precipitation of large Si particles in the heat-affectedzone (the light-grey particles as seen in Fig. 2d).

3.2. Phase identification

Fig. 5 shows the XRD spectra of the SiC/AlSi10 Mg compositepowder (black curve) and SLM-processed SiC/AlSi10 Mg parts (redcurve). Strong diffraction peaks corresponding to Al were detected.Besides, weak SiC diffraction peaks were identified, indicating that asmall amount of unmelted SiC particles remained in the matrix aftersolidification. Meanwhile, additional diffraction peaks from Al4SiC4

were observed in the XRD spectra of the SLM-processed SiC/AlSi10 Mgparts, which could be ascribed to in-situ reaction between the Al meltand the SiC ceramic particles in laser-induced molten pool. At hightemperature of SLM processing, Si and C from SiC reinforcementstended to dissolve in molten aluminum alloys and the in-situ reactionbetween the aluminum melt and SiC occurred forming new Al4SiC4

phase. A small quantity of Si phases was also found, consistent with themicrostructure observation in microstructure section. Although alu-minum alloys have a high affinity to oxygen element, the characteristicpeaks of aluminum oxide were not detected in the SLM-processed Al-based composites in the limit of XRD detection. This implies that theatmosphere for preparation and SLM processing of the AlSi10Mg/SiC

specimens could avoid the oxidation behaviors of powder and as-fab-ricated samples.

The high-magnification FE-SEM images, as shown in Fig. 6 andFig. 7a, present characteristic morphologies of the remaining SiCceramic particles and the newly formed strip-structured Al4SiC4 in themolten pool. The morphology of the SiC ceramic particles significantlydepends on their locations in the molten pool. As the heat is accumu-lated greatly in the center of the molten pool at a high-energy irra-diation of laser beam, it is reasonable to expect that the SiC ceramicparticles located in the central region were melted partially and the in-situ reaction of SiC ceramic particles with Al melt occurred at the

Fig. 4. Temperature distribution profile (a) and cooling rate versus time at different locations of the melt pool: (b) Center zone; (c) Boundary of molten pool; (d)Heat-affected zone.

Fig. 5. XRD spectra of the SiC/AlSi10Mg composite powder and SLM-processedSiC/AlSi10Mg parts.

X. Zhao et al. Vacuum 160 (2019) 189–196

193

boundary of SiC particles. The SiC particles became very smooth in-stead of the poly-angular shape used as starting powder. While at theboundary of the molten pool, the energy absorbed by the powder issomewhat lower than that in the center of the molten pool, the SiCceramic particles have more tendency to remain sharp and rough (asseen in Fig. 6).

Fig. 7 illustrates the EDS results of the chemical compositions col-lected from the SLM-processed composites in Fig. 7a. Three differentphases are measured, including the large-sized SiC ceramic particles,strip-structured Al4SiC4, and eutectic Al-Si. As shown in Fig. 7b, EDSpoint detected in the grey area (point 1) revealed that the fibrous phaseshowed a proportion of Al (88.03 at. %) and Si (11.97 at. %), corre-sponding to eutectic Al-Si. The Al matrix with a cell microstructure(point 2) exhibited a small amount of Si element (8.85 at. %), inferredto α-Al matrix. As revealed in Fig. 7(d), the Al, Si, and C elements withan approximate Al:Si:C atom ratio of 4:1:4 was detected in the strip-structured reinforcements (point 3). It demonstrated that Al4SiC4 phasewas in situ formed during SLM via the reaction between molten alu-minum and SiC, in good agreement with the XRD results.

Due to the relatively high mass fraction of SiC (8.5 vol %) in thestarting powder and incomplete in-situ reaction during the laser-in-duced rapid melting/solidification process, it is reasonable to expectthat there existed primary SiC phase after SLM. The in-situ formation ofAl4SiC4 can be explained based on the Al-Si-C ternary alloy phasediagram [33], as given in Fig. 8. The starting component used in thisstudy is the Al:Si:C mole ratio of 78:15:7, which is far below the contentfor formation of Al4SiC4. In such situation, the SiC firstly reacts withmolten aluminum to form Al4C3 in the temperature range of940–1620 K according to the previous study [31]. With dissolution ofpartial SiC ceramic particles in the melt, the reactions P and M willthermodynamically occur accompanied with reaction Q in the moltenpool [33]. Therefore, Al4SiC4 phase are formed close to the SiC ceramicparticles. This finding is in good agreement with the results reported byWalter et al. [34].

P: L + C + Al4C3→Al8SiC7

Fig. 6. The morphologies of the SiC particles in the SLM-processed SiC/AlSi10Mg composite.

Fig. 7. (a) The SiC particles in the molten pool center of SLM-processed SiC/AlSi10Mg composite and EDS results showing the chemical compositions collected fromthe SLM-processed composites: (b) fibrous area (point 1); (c) cellular area (point 2); (d) strip-structured reinforcement (point 3).

X. Zhao et al. Vacuum 160 (2019) 189–196

194

M: L + C + Al8SiC7→Al4SiC4

Q: L + C→Al4SiC4+SiC

3.3. Nanohardness

The nanoindentation tests were performed on three different areas:reinforcing structure, center of molten pool and boundary of moltenpool. The hardness is defined as a ratio of the maximum indentationload (Fmax) to the projected area of hardness impression (A). The na-nohardness (H) can be calculated as follows [35]:

=H FA

max(1)

where Fmax is the maximum load and A is the projected area of thehardness impression.

According to Oliver and Pharr method [35], the reduced elasticmodulus ER can be determined by the data of a complete loading/un-loading cycle. The reduced elastic modulus is expressed as:

=ER π SA2 (2)

where ER is the reduced elastic modulus, S is the stiffness and A is theprojected area of the hardness impression.

Load-depth curves and the corresponding nanohardness and elasticmodulus values of the SLM processed SiC/AlSi10 Mg composites areshown in Fig. 9. The nanoindentation depth and the deformation re-cover behavior varied with different zones of molten pool in the SLM-processed SiC/AlSi10 Mg composites. The nanohardness significantlydepends on the microstructure and phases developed in the molten pool[36]. The respective nanohardness and elastic modulus values calcu-lated from measurements are 2.27 GPa and 78.94 GPa in the center,2.15 GPa and 75.46 GPa at the boundary of molten pool. Moreover,unmelted SiC particles and newly formed Al4SiC4 phases exhibitedmuch higher nanohardness (7.21 GPa) and elastic modulus(197.13 GPa) than that of the molten pool. It has to be mentioned thatthe measured nanohardness of the composites is coincident with themeasured microhardness across the molten pool along the buildingdirection. Notably, the nanohardness values obtained from SLM-fabri-cated AlSi10Mg/SiC composite is more than twice than that measuredfrom as-cast AlSi10Mg alloy (∼0.97 GPa), while it is comparable to thenanohardness of laser melted AlSi10Mg alloy (∼2.2 GPa) [37]. Al-though 8.5 vol % SiC particles were added in the matrix, it seems that

they imposed negligible effects on the enhancement of average nano-hardness of matrix. The nanohardness value in the center is somewhatlarger when compared to that measured at the boundary of moltenpool, appearing in contrast to the reported results of AlSi10Mg pro-cessed by SLM [37]. This might be slightly influenced by the addition ofSiC reinforcements as well as in-situ formed phase Al4SiC4, which madecontributions to the structure development in different regions ofmicro-sized molten pool.

4. Conclusions

This work has investigated the microstructure evolution and nano-hardness of the SiC/AlSi10Mg composite fabricated by SLM tech-nology. The following conclusions are drawn:

(1) The microstructure in the center of the molten pool was dominatedby fine cellular structure and the average size of cell reached∼0.3 μm, while the main microstructure at the boundary of moltenpool was columnar dendrites. In the upper part of central zone, arelatively clear interface between adjacent α-Al grains was ob-served, while in the lower part, the interface became vague andfiber-like microstructure is developed along the α-Al grainboundary. The solidification in the SLM processing of SiC/AlSi10Mg alloy depends on three factors of the temperature gra-dient (G), the solidification velocity (R), and the concentration ofsolute Si in the micro-sized molten pool, suggesting that a variationof G/R and the concentration of Si contributes to a structuralchange from columnar dendrite growth to cellular growth.

(2) The in-situ reaction between aluminum melt and SiC ceramic

Fig. 8. Al-Si-C ternary alloy phase diagram [33].

Fig. 9. (a) The load-depth curves and (b) calculated nanohardness and elasticmodulus of the SLM-processed SiC/AlSi10Mg samples in the different regions:A: Reinforcing structure; B: Center of molten pool; C: Boundary of molten pool.

X. Zhao et al. Vacuum 160 (2019) 189–196

195

particles led to the formation of Al4SiC4 reinforcements close to thepartially melted SiC ceramic particles. Three different phases in-cluding the unmelted micron-sized SiC ceramic particles, the in situformed Al4SiC4 phase, and the eutectic Si phase, are observed in theas-fabricated SiC/AlSi10 Mg composites.

(3) The hardness and elastic modulus of the SLM-produced SiC/AlSi10Mg parts are 2.27 GPa and 78.94 GPa in the center, 2.15 GPaand 75.46 GPa at the boundary of molten pool. The nanohardnessvalues of SLM-prepared composites is more than twice than that ofas-cast AlSi10Mg alloy, while it is comparable to the reported na-nohardness of laser melted AlSi10Mg counterparts. The SiCceramic particles and in-situ formed phase Al4SiC4 play a significantrole in the structure development and nanohardness change indifferent regions of micro-sized molten pool.

Acknowledgement

The authors gratefully acknowledge the financial support from theNational Natural Science Foundation of China (No. 51735005), theNational Key Research and Development Program “AdditiveManufacturing and Laser Manufacturing” (Nos. 2016YFB1100101,2018YFB1106302), the Priority Academic Program Development ofJiangsu Higher Education Institutions.

References

[1] S. Tjong, Z. Ma, Microstructural and mechanical characteristics of in situ metalmatrix composites, Mater. Sci. Eng. R. 29 (3) (2000) 49–113.

[2] K. Shorowordi, T. Laoui, A. Haseeb, J. Celis, L. Froyen, Microstructure and interfacecharacteristics of B4C, SiC and Al2O3, reinforced Al matrix composites: a com-parative study, J. Mater. Process. Technol. 142 (3) (2003) 738–743.

[3] A. Moghadam, E. Omrani, P. Menezes, P. Rohatgi, Mechanical and tribologicalproperties of self-lubricating metal matrix nanocomposites reinforced by carbonnanotubes (CNTs) and graphene– A review, Composites Part B 77 (2015) 402–420.

[4] S. Bakshi, D. Lahiri, A. Agarwal, Carbon nanotube reinforced metal matrix com-posites - a review, Int. Mater. Rev. 55 (1) (2010) 41–64.

[5] F. Chang, D. Gu, D. Dai, P. Yuan, Selective laser melting of in-situ Al4SiC4 + SiChybrid reinforced Al matrix composites: influence of starting SiC particle size, Surf.Coating. Technol. 272 (2015) 15–24.

[6] S. Dadbakhsh, L. Hao, P. Jerrard, D. Zhang, Experimental investigation on selectivelaser melting behaviour and processing windows of in situ reacted Al/Fe2O3,powder mixture, Powder Technol. 231 (2012) 112–121.

[7] D. Gu, H. Wang, D. Dai, P. Yuan, M. Wilhelm, R. Poprawe, Rapid fabrication of Al-based bulk-form nanocomposites with novel reinforcement and enhanced perfor-mance by selective laser melting, Scripta Mater. 96 (2015) 25–28.

[8] Q. Han, R. Setchi, F. Lacan, D. Gu, S. Evans, Selective laser melting of advanced Al-Al2O3, nanocomposites: simulation, microstructure and mechanical properties,Mater. Sci. Eng. 698 (2017) 162–173.

[9] X. Li, G. Ji, Z. Chen, A. Addad, Y. Wu, H. Wang, J. Vleugels, J. Humbeeck, J. Kruth,Selective laser melting of nano-TiB2, decorated AlSi10Mg alloy with high fracturestrength and ductility, Acta Mater. 129 (2017) 183–193.

[10] H. Ye, An overview of the development of Al-Si-Alloy based material for engineapplications, J. Mater. Eng. Perform. 12 (3) (2003) 288–297.

[11] J. Kruth, P. Mercelis, J. Vaerenbergh, L. Froyen, M. Rombouts, Binding mechanismsin selective laser sintering and selective laser melting, Rapid Prototyp. J. 11 (1)(2005) 26–36.

[12] D. Gu, Laser Additive Manufacturing of High-performance Materials, SpringerBerlin Heidelberg, 2015.

[13] B. Zhang, H. Liao, C. Coddet, Selective laser melting commercially pure Ti undervacuum, Vacuum 95 (2) (2013) 25–29.

[14] P. Mercelis, J. Kruth, Residual stresses in selective laser sintering and selective lasermelting, Rapid Prototyp. J. 12 (5) (2006) 254–265.

[15] L. Zhang, D. Klemm, J. Eckert, Y. Haod, T. Sercombe, Manufacture by selective lasermelting and mechanical behavior of a biomedical Ti–24Nb–4Zr–8Sn alloy, ScriptaMater. 65 (1) (2011) 21–24.

[16] Y. Liu, S. Li, H. Wang, W. Hou, Y. Hao, R. Yang, T. Sercombe, L. Zhang,Microstructure, defects and mechanical behavior of beta-type titanium porousstructures manufactured by electron beam melting and selective laser melting, ActaMater. 113 (2016) 56–67.

[17] E. Olakanmi, R. Cochrane, K. Dalgarno, A review on selective laser sintering/melting (SLS/SLM) of aluminium alloy powders: processing, microstructure, andproperties, Prog. Mater. Sci. 74 (2015) 401–477.

[18] E. Louvis, P. Fox, C. Sutcliffe, Selective laser melting of aluminium components, J.Mater. Process. Technol. 211 (2) (2011) 275–284.

[19] D. Dai, D. Gu, Influence of thermodynamics within molten pool on migration anddistribution state of reinforcement during selective laser melting of AlN/AlSi10Mgcomposites, Int. J. Mach. Tool Manufact. 100 (2016) 14–24.

[20] L. Xi, H. Zhang, P. Wang, H. Li, K. Prashanth, K. Lin, I. Kaban, D. Gu, Comparativeinvestigation of microstructure, mechanical properties and strengthening mechan-isms of Al-12Si/TiB2 fabricated by selective laser melting and hot pressing, Ceram.Int. 44 (2018) 17635–17642.

[21] Y. Liu, Z. Liu, Y. Jiang, G. Wang, Y. Yang, L. Zhang, Gradient in microstructure andmechanical property of selective laser melted AlSi10Mg, J. Alloy. Comp. 735 (2018)1414–1421.

[22] P. Yuan, D. Gu, D. Dai, Particulate migration behavior and its mechanism duringselective laser melting of TiC reinforced Al matrix nanocomposites, Mater. Des. 82(2015) 46–55.

[23] Q. Huan, M. Jyotirmoy, K. Hyungson, Numerical simulation of heat transfer andfluid flow in coaxial laser cladding process for direct metal deposition, J. Appl.Phys. 100 (2006) 024903.

[24] N. Tolochko, Y. Khlopkov, S. Mozzharov, V. Titov, M. Ignatiev, Absorptance ofpowder materials suitable for laser sintering, Rapid Prototyp. J. 6 (3) (2000)155–161.

[25] P. Yuan, D. Gu, Molten pool behaviour and its physical mechanism during selectivelaser melting of TiC/AlSi10Mg nanocomposites: simulation and experiments, J.Phys. D Appl. Phys. 48 (3) (2015) 035303.

[26] L. Snead, T. Nozawa, Y. Katoh, T. Byun, S. Kondo, D. Petti, Handbook of SiCproperties for fuel performance modeling, J. Nucl. Mater. 371 (1) (2007) 329–377.

[27] X. Li, X. Wang, M. Saunders, A. Suvorova, L. Zhang, Y. Liu, M. Fang, Z. Huang,T. Sercombe, A selective laser melting and solution heat treatment refined Al–12Sialloy with a controllable ultrafine eutectic microstructure and 25% tensile ductility,Acta Mater. 95 (2015) 74–82.

[28] T. Kimura, T. Nakamoto, M. Mizuno, H. Araki, Effect of silicon content on densi-fication, mechanical and thermal properties of Al-x Si binary alloys fabricated usingselective laser melting, Mater. Sci. Eng., A 682 (2017) 593–602.

[29] L. Thijs, K. Kempen, J. Kruth, J. Humbeeck, Fine-structured aluminium productswith controllable texture by selective laser melting of pre-alloyed AlSi10Mgpowder, Acta Mater. 61 (5) (2013) 1809–1819.

[30] X. Zhou, C. Tang, Current status and future development of coated fuel particles forhigh temperature gas-cooled reactors, Prog. Nucl. Energy 53 (2) (2011) 182–188.

[31] D. Gu, F. Chang, D. Dai, Selective laser melting additive manufacturing of novelaluminum based composites with multiple reinforcing phases, J. Manuf. Sci. Eng.137 (2) (2015) 021010.

[32] I. Rosenthal, A. Stern, N. Frage, Microstructure and mechanical properties ofAlSi10Mg parts produced by the laser beam additive manufacturing (AM) tech-nology, Metallography Microstructure & Analysis 3 (6) (2014) 448–453.

[33] J. Gröbner, H. Lukas, F. Aldinger, Thermodynamic calculation of the ternary systemAl-Si-C, Calphad 20 (2) (1996) 247–254.

[34] D. Walter, I. Karyasa, Solid state reactions in the Al‐Si‐C system, J. Chin. Chem. Soc.52 (5) (2005) 873–876.

[35] W. Oliver, G. Pharr, An improved technique for determining hardness and elasticmodulus using load and displacement sensing indentation experiments, J. Mater.Res. 7 (6) (1992) 1564–1583.

[36] S. Haghighi, G. Cao, L. Zhang, Nanoindentation study of mechanical properties of Tibased alloys with Fe and Ta additions, J. Alloy. Comp. 692 (2017) 892–897.

[37] N. Aboulkhair, I. Maskery, C. Tuck, I. Ashcroft, N. Everitt, Nano-hardness and mi-crostructure of selective laser melted AlSi10Mg scan tracks, Proc. SPIE (2015) 9567.

X. Zhao et al. Vacuum 160 (2019) 189–196

196