Dongdong Gu1

College of Materials Science and Technology,

Nanjing University of Aeronautics and

Astronautics (NUAA),

Yudao Street 29,

Nanjing 210016, China;

Institute of Additive Manufacturing (3D Printing),

Nanjing University of Aeronautics and

Astronautics (NUAA),

Yudao Street 29,

Nanjing 210016, China

e-mail: dongdonggu@nuaa.edu.cn

Donghua DaiCollege of Materials Science and Technology,

Nanjing University of Aeronautics and

Astronautics (NUAA),

Yudao Street 29,

Nanjing 210016, China;

Institute of Additive Manufacturing (3D Printing),

Nanjing University of Aeronautics and

Astronautics (NUAA),

Yudao Street 29,

Nanjing 210016, China

Wenhua ChenCollege of Materials Science and Technology,

Nanjing University of Aeronautics and

Astronautics (NUAA),

Yudao Street 29,

Nanjing 210016, China;

Institute of Additive Manufacturing (3D Printing),

Nanjing University of Aeronautics and

Astronautics (NUAA),

Yudao Street 29,

Nanjing 210016, China

Hongyu ChenCollege of Materials Science and Technology,

Nanjing University of Aeronautics and

Astronautics (NUAA),

Yudao Street 29,

Nanjing 210016, China;

Institute of Additive Manufacturing (3D Printing),

Nanjing University of Aeronautics and

Astronautics (NUAA),

Yudao Street 29,

Nanjing 210016, China

Selective Laser Melting AdditiveManufacturing of Hard-to-Process Tungsten-Based AlloyParts With Novel CrystallineGrowth Morphology andEnhanced PerformanceSelective laser melting (SLM) additive manufacturing (AM) of hard-to-process W-basedparts with the addition of 2.5 wt.% TiC was performed using a new metallurgical process-ing mechanism with the complete melting of the high-melting-point powder. The influenceof SLM processing parameters, especially laser scan speed and attendant laser fluence(LF), on densification behavior, microstructural development, and hardness/wear per-formance of SLM-processed W-based alloy parts was disclosed. The densificationresponse of SLM-processed W-based parts decreased both at a low LF of 10.7 J/mm2,caused by the limited SLM working temperature and wetting characteristics of the melt,and at an excessively high LF of 64 J/mm2, caused by the significant melt instability andresultant balling effect and microcracks formation. The laser-induced complete melting/solidification mechanism contributed to the solid solution alloying of Ti and C in Wmatrix and the development of unique microstructures of SLM-processed W-based alloyparts. As the applied LF increased by lowering laser scan speed, the morphologies ofW-based crystals in SLM-processed alloy parts experienced a successive change from thecellular crystal to the cellular dendritic crystal and, finally, to the equiaxed dendriticcrystal, due to an elevated constitutional undercooling and a decreased thermal under-cooling. The optimally prepared W-based alloy parts by SLM had a nearly full densifica-tion rate of 97.8% theoretical density (TD), a considerably high microhardness of809.9 HV0.3, and a superior wear/tribological performance with a decreased coefficientof friction (COF) of 0.41 and a low wear rate of 5.73� 10�7 m3/(N m), due to the com-bined effects of the sufficiently high densification and novel crystal microstructures ofSLM-processed W-based alloy parts. [DOI: 10.1115/1.4032192]

Keywords: additive manufacturing, selective laser melting, tungsten, crystal growth,wear

1 Introduction

Tungsten (W), due to the unique properties such as highmelting point, high tensile strength, and low thermal expansioncoefficient, is very promising for the applications as high heat fluxcomponents and high-power density structural materials in radia-tion environments [1–3]. Nevertheless, W, as a typical hard-to-process material, demonstrates serious inherent embrittlement [4],which limits its applications as the structural material. It hasbeen revealed that the significant embrittlement and resultant low

ductility of W material typically originate from the microstruc-tural factors including the heterogeneity in grain size and its distri-bution and the grain growth and coarsening by recrystallization[5]. The room-temperature ductility of W material is thereforestrongly dependent on grain structure and its size and can beenhanced considerably by decreasing grain size or homogenizinggrain structure [6]. On the other hand, it has been disclosed thatwhen heated above the recrystallization temperature, the micro-structure of W material tends to change considerably because ofthe grain growth effect, which further reduces strength and hard-ness and, meanwhile, causes embrittlement [4]. Fortunately, theprevious work has proved that the strengthening and/or toughen-ing of W material with the addition of ceramics, which are typi-cally added in the discontinuous form of particles dispersed

1Corresponding author.Manuscript received June 22, 2015; final manuscript received November 23,

2015; published online March 28, 2016. Assoc. Editor: Z. J. Pei.

Journal of Manufacturing Science and Engineering AUGUST 2016, Vol. 138 / 081003-1Copyright VC 2016 by ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 03/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

throughout the matrix, is a promising method to decrease the brit-tleness of W material [7]. In particular, as the ceramic reinforcingparticles are refined to a nanometer scale to prepare W-basednanocomposites, the obtainable mechanical performance isexpected to be enhanced significantly, due to the unique propertiesof nanoparticles including small size effect and surface/interfaceeffect [8–10]. A uniform microscopic distribution of ceramicnanoparticles along grain boundaries is effective in handicappinggrain growth and grain boundary migration at higher workingtemperatures, which in turn elevates the recrystallization tempera-ture and decreases the brittleness of W material [11].

Powder metallurgy (PM) methods are typically applied to pro-duce bulk-form W-based parts, e.g., spark plasma sintering [12],hot-press sintering [13], liquid-reactive sintering [14], etc. In thecase of PM methods with limited operative temperatures, one ofthe significant technological problems is the formation of residualporosity, due to the considerably high melting temperatures ofW-based materials and attendant limited densification activity.Furthermore, nearly all PM-processed pieces are required to bepreworked by some dedicated tools such as moulds or dies toobtain desired shapes. However, due to the limitation of availabletools, some complex-structured W-based components are nor-mally difficult to be produced. On the other hand, for bulk-formW-based parts condensed from ultrafine nanopowder by conven-tional PM methods, the uncontrolled agglomeration of nanoscaleparticles due to the considerably large van der Waals attractiveforces tends to cause the microstructural inhomogeneity and eventhe coarsening of original favorable nanostructures. In this situa-tion, differential stresses caused by heterogeneous microstructuresare directly related to the propagation of internal cracks inPM-consolidated solids. Therefore, novel processing methods arerequired to achieve complex shaped, fully dense W-based partswith homogeneous microstructures and resultant elevatedperformance.

SLM, as a newly developed AM/3D printing technology[15–17], enables the rapid production of 3D parts having anycomplex structures directly from powder materials [18–20]. SLMbuilds components in a layer-by-layer fashion by selective fusionand consolidation of thin layers of loose powder, using a high-energy laser beam. SLM possesses a number of advantagesincluding net-shape production without requirements of molds ordies, a high level of processing flexibility, and a wide range ofapplicable powder materials [21]. Typically, SLM process demon-strates a unique nonequilibrium physical metallurgical nature,including multiple modes of heat, mass, and momentum transfer[22]. During SLM, the high-energy laser beam focused on thepowder layer surface can reach a power density up to1010–1012 W/cm2 and can heat a material surface speedily to atemperature up to 105 K, followed by a superfast cooling at arate up to 106–107 K/s [23]. In this situation, even the high-melting-point W material tends to melt completely and, subse-quently, experience laser-induced rapid solidification process. Itaccordingly provides a high potential for obtaining very finemicrostructures with superior mechanical properties. Neverthe-less, little previous work has been reported on laser processing ofW-based parts through the complete melting/precipitation mecha-nism. A large degree of shrinkage occurs during liquid–solidtransformation, which tends to produce considerable stresses andeven microcracks in SLM-processed W-based parts. Furthermore,the uncontrollability of the crystallization and microstructuraldevelopment of W alloy in an SLM route remains a major chal-lenge. Recent work from Zhou et al. [24] on SLM of pure Wrevealed that the “balling” effect of melted droplets at the laserfocal points and entrapped cavities hindered the preparation offully dense parts. An analysis of the balling mechanism indicatedthat such a metallurgical defect was determined by the intrinsic Wproperties and laser processing parameters.

In the present work, bulk-form W-based alloy parts havingnovel microstructural features were successfully produced bySLM of a W–Ti–C system. The melting behavior and

microstructural development of W-based alloy parts at variousSLM parameters were studied and the mechanical propertiesincluding densification rate, microhardness, and tribological/wearproperties were assessed, with an aim to establish a relationship ofprocess, microstructure, and mechanical performance of W-basedalloy parts by SLM.

2 Experimental Procedures

2.1 Powder Materials. The starting powder materials were99.9% purity, equiaxed W powder with an average particle size of7 lm and 99.0% purity TiC nanopowder with a near sphericalshape and a mean particle size of 50 nm. The TiC/W powder sys-tem consisting of 2.5 wt.% TiC was milled in a Pulverisette 6planetary high-energy ball mill (Fritsch GmbH, Germany) at arotation speed of 250 rpm for 5 hrs, using a ball-to-powder weightratio of 10:1. The as-prepared TiC/W nanocomposite powder hada sufficiently high flowability and a sound spreading ability onpowder bed, which was particularly important for a successfulSLM production. For the current nano-TiC/macro-W compositesystem, the nano-TiC particles were dispersed and mixed uni-formly during ball milling treatment. After milling process, theas-prepared composite powder was immediately used for SLMproduction, which demonstrated a high flowability on the powderbed and a sound laser processing ability. With the aid of ball mill-ing process to obtain the homogeneous W-based nanocompositepowder, the agglomeration of powder after the milling processwas insignificant and the powder could be transferred to the pow-der bed smoothly for SLM.

2.2 SLM Processing. A schematic of the entire SLM equip-ment used in the present study is depicted in Fig. 1(a), and adetailed SLM processing route for the fabrication of bulk-formTiC/W nanocomposite parts is illustrated in Fig. 1(b). Theself-developed SLM equipment in NUAA consisted mainly of aYLR-500 Ytterbium fiber laser with a power of �500 W and aspot size of 70 lm (IPG Laser GmbH, Burbach, Germany), ahurrySCAN

VR

30 laser scanner (SCANLAB AG, Puchheim, Ger-many), an automatic powder layering apparatus, an inert argongas recycling and protection system, and a computer system forprocess control. When samples were to be prepared, a steel sub-strate was fixed on the building platform and leveled. The argongas with an outlet pressure of 30 mbar was fed into the sealedbuilding chamber and the resultant oxygen content decreasedbelow 10 ppm. A thin layer of TiC/W nanocomposite powder witha layer thickness of 25 lm was then deposited on the substrate bythe powder spreading system. For the spreading of the powder oneach layer, a reciprocating mode was applied to achieve a uniformspreading of the powder in every position on the layer. Eventhough there was some slight agglomeration of the powder in thefirst place of the layer, the powder in this place would be recoatedhomogeneously during the reciprocation process of the powderspreading mechanism. With the aid of the reciprocating mode forpowder spreading, the as-spread powder layer was uniform andcould well meet the requirements of SLM production. Afterward,the laser beam scanned the powder bed surface to form a layer-wise profile according to computer-aided design data of the sam-ples. A laser power of 160 W, which was optimized by thepreliminary SLM experiments, was fixed during one batch ofexperiment. Meanwhile, the scan speeds were settled periodicallyat 50, 100, 200, and 300 mm/s by SLM control program, in orderto change the applied processing conditions. Basically, the laserscan speeds with a uniform scan speed increase of 100 mm/s wereapplied. Furthermore, in order to investigate the effects of laserscan speed and resultant laser energy density on SLM processingability of W-based parts, a significantly low scan speed of 50 mm/s was also applied for SLM. Laser scanning of each layer wasbased on a linear raster scan pattern with a scan vector length of4 mm and a hatch spacing of 50 lm between neighboring lines. In

081003-2 / Vol. 138, AUGUST 2016 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 03/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

order to assess the laser energy input to the powder layer beingmelted, an integrated parameter LF was defined [25]

LF ¼ P

vh(1)

where P is the laser power, v is the scan speeds, and h is thehatch spacing. Four different LFs of 10.7, 16, 32, and 64 J/mm2

were applied to build bulk-form parts with dimensions of8 mm� 8 mm� 5 mm, and the amount of layers that had beendeposited on the substrate was 200.

2.3 Microstructural Characterization and MechanicalProperties Tests. Phase identification of SLM-processed partswas performed by a D8 Advance X-ray diffractometer (XRD)(Bruker AXS GmbH, Germany) with Cu Ka radiation(k¼ 0.154 nm) at 40 kV and 40 mA, using a continuous scanmode. A quick scan at 4 deg/min was first performed over a widerange of 2h¼ 30–120 deg to give a general overview of the dif-fraction peaks. A slower scan rate of 1 deg/min was further usedin the vicinity of the first and second strong peaks to give a moreaccurate determination of 2h location and intensity of the peaks.Specimens for metallographic examinations were cut, ground, andpolished according to the standard procedures and etched with asolution comprising HF (10 mL), HNO3 (30 mL), and distilledwater (70 mL) for 15 s. High-resolution microstructures werecharacterized using a LEO 1550 field emission scanning electron

microscope (FE-SEM) (Carl Zeiss NTS GmbH, Germany) in asecondary electron mode at 5.0 kV. Chemical compositions wereanalyzed by an EDAX energy dispersive X-ray (EDX) spectro-scope (EDAX Inc., Mahwah, NJ), using a super-ultra thin windowsapphire detector.

The experimentally measurable density of SLM-processed partswas defined as mass divided by volume. The mass of the speci-mens was measured by the analytical balance, and the volume ofthe specimens was determined using the Archimedes’ method.Then, the densification rate of SLM-processed parts was deter-mined by the ratio of the experimentally measurable density to theTD of the corresponding material. The Vickers hardness wasmeasured using a MicroMet 5101 microhardness tester (BuehlerGmbH, Germany) at a load of 0.3 kg and an indentation time of20 s. Based on the ASTM G99 standard, the wear/tribologicalproperties of the specimens were estimated by the dry slidingwear tests conducted in a HT-500 ball-on-disk tribometer (Lanz-hou ZhongKe KaiHua Sci. & Technol. Co., Ltd., Lanzhou, China)in air at room temperature. Surfaces of samples to be tested wereground and polished prior to wear tests. A bearing steel GCr15ball having a diameter of 3 mm and an average hardness ofHRC60 was taken as the counterface material, using a test load of5.2 N. The friction unit was rotated at a speed of 560 rpm for15 min, and the rotation radius was 2 mm. The COF was recordedautomatically during wear tests. The wear volume (V) was deter-mined gravimetrically using

V ¼ Mloss=q (2)

Fig. 1 Schematic of SLM equipment used in the present study (a) and SLM proceduresfor producing bulk-form parts (b)

Journal of Manufacturing Science and Engineering AUGUST 2016, Vol. 138 / 081003-3

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 03/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

where Mloss is the weight loss of the specimens after wear tests,and q is the density. The wear rate (x) was calculated by

x ¼ V=ðWLÞ (3)

where W is the contact load, and L is the sliding distance duringtests.

3 Results and Discussion

3.1 Phase Evolution. Figure 2 depicts the typical XRD spec-tra of SLM-processed W-based parts obtained within a wide rangeof 2h¼ 30–120 deg. The strong diffraction peaks corresponding toW (JCPDS Card #04-0806) were generally detected in SLM-processed parts at all given LFs. The XRD characterization withinsmall 2h angles in the vicinity of the first and second strong dif-fraction peaks of W, as depicted in Table 1, revealed that the exact2h locations and intensities of W peaks showed an apparentchange due to the addition of TiC and the variation of SLMparameters. Here, the standard diffraction peaks for W, which is

located at 2h¼ 40.264 deg and 2h¼ 73.195 deg, were taken for acomparison. It revealed that as the applied LF increased by lower-ing laser scan speed, the 2h locations of the W diffraction peaksgenerally shifted to the higher 2h angles. The shift tendency of 2hangles to larger values became more significant at a higher LF(Table 1). Meanwhile, the intensity of the W diffractionpeaks showed a significant decrease when a higher LF was applied(Table 1), which implied the formation of considerably refinedcrystalline microstructures in SLM-processed W-based part in thissituation.

Based on the Bragg’s law for XRD analysis [26]

2d sin h ¼ nkðn ¼ 1; 2; 3;…Þ (4)

the detected increase of 2h at a higher LF (Table 1) indicates adecrease in the lattice plane distance (d), which is believed to becaused by the alloying of Ti and C atoms with W matrix. DuringSLM process, the high-energy fiber laser beam focused onthe powder bed can concentrate a power density up to1010–1012 W/cm2 and can heat the powder layer surface within25 ms to a working temperature up to 105 K [27]. SLM of the pres-ent TiC/W system is accordingly processed via a metallurgicalmechanism with the complete melting of the powder. Both the Wmatrix powder with a melting point of 3695 K and the TiC addi-tive powder with a melting point of 3433 K tend to become meltedfully during SLM process, thereby releasing the mobile Ti and Catoms in laser-induced molten pool. During the subsequent rapidsolidification process after the laser beam moves away, the solidsolution alloying of Ti and C atoms in W matrix predominatesduring this course. The solid solution of C in W is interstitial andthe rate of expansion of W lattice caused by the solid solution ofC atoms amounts to 7� 10�5 nm per at. % C [28]. On the otherhand, the W–Ti tends to form substitution solid solution. Accord-ing to the Hume–Rothery rule [29], the extensive substitutionalsolid solution occurs only if the atomic size factor, i.e., the relativedifference between the atomic radii of the two elements, is lessthan 15%. If the difference is larger than 15%, the solubility islimited. The atomic radii of Ti and W are 144.8 pm and 137.1 pm,respectively, and the corresponding atomic size factor is 5.6%,which indicates an extensive solid solubility of Ti in W. Nor-mally, the occurrence of solid solution phenomena, both the inter-stitial solution of C in W and the extensive substitutional solutionof Ti in W, is accompanied by the microscopic volume expansion,which in turn creates stresses at grain boundaries that influencethe lattice parameters of W matrix in SLM-processed W-basedalloy parts.

3.2 Melting and Densification Mechanisms. Figure 3 showsthe typical surface morphologies of SLM-processed W-basedalloy parts at various laser scan speeds and resultant LFs. High-magnification micrographs of the corresponding surfaces, asrevealed in Fig. 4, illustrate the melting and solidification mecha-nisms of the liquid front under different SLM conditions. Theresultant densification levels of SLM-processed W-based alloyparts are depicted in Fig. 5. At a relatively low LF of 10.7 J/mm2

induced by a high v of 300 mm/s, a heterogeneous, porous surface

Fig. 2 XRD spectra over a wide range of 2h 5 30–120 degshowing constitutional phases of SLM-processed W-basedalloy parts using different SLM processing parameters: (a)LF 5 10.7 J/mm2, v 5 300 mm/s, (b) LF 5 16 J/mm2, v 5 200 mm/s,(c) LF 5 32 J/mm2, v 5 100 mm/s, and (d) LF 5 64 J/mm2,v 5 50 mm/s

Table 1 XRD data showing variations of displacement and intensity of W diffraction peaks in SLM-processed W-based alloy partsusing different SLM parameters

First strong peak Second strong peak

Samples and SLM processing parameters 2h (deg) Intensity (CPS) 2h (deg) Intensity (CPS)

Standard W sample (PDF#04-0806) 40.264 — 73.195 —LF¼ 10.7 J/mm2, v¼ 300 mm/s 40.28 7068.9 73.22 2064.7LF¼ 16 J/mm2, v¼ 200 mm/s 40.30 5134.2 73.22 1694.7LF¼ 32 J/mm2, v¼ 100 mm/s 40.32 2359.6 73.28 1189.8LF¼ 64 J/mm2, v¼ 50 mm/s 40.36 2189.7 73.32 1024.8

081003-4 / Vol. 138, AUGUST 2016 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 03/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

consisting of irregular-shaped, large-sized open pores with a meansize of �150 lm was produced after SLM (Fig. 3(a)), resulting ina limited densification rate of 87.5% TD (Fig. 5). The solidifiedtracks were considerably crumpled, having the disordered liquidsolidification front. A number of small-sized unmelted W particleswith an average size below 10 lm appeared in the solidified liquidfront (Fig. 4(a)). On increasing the applied LF to 16 J/mm2 bylowering v to 200 mm/s, the SLM-processed surface was highlydense and free of any apparent pores or cracks (Fig. 3(b)), and thedensification response of SLM-processed part was enhancedmarkedly to 93.4% TD. The solidified scan tracks had the orderly

distributed liquid solidification front, exhibiting an improvedmicrostructural homogeneity of scan tracks (Fig. 4(b)). Interest-ingly, at a suitable LF of 32 J/mm2, the laser-processed surface ofW-based alloy parts was completely dense, consisting of parallel-distributed and coherently bonded scan tracks (Fig. 3(c)). Theordered and homogeneous liquid solidification front was producedin the solidified scan tracks (Fig. 4(c)). In this situation, a nearfully dense W-based part with 97.8% TD was obtained after SLM(Fig. 5). However, at an even higher LF of 64 J/mm2 caused by aconsiderably low v of 50 mm/s, several typical metallurgicaldefects associated with SLM occurred on the surface of

Fig. 3 Typical surface morphologies (FE-SEM) of SLM-processed W-based alloy parts atdifferent SLM processing parameters, revealing the variation of densification behaviors:(a) LF 5 10.7 J/mm2, v 5 300 mm/s, (b) LF 5 16 J/mm2, v 5 200 mm/s, (c) LF 5 32 J/mm2,v 5 100 mm/s, and (d) LF 5 64 J/mm2, v 5 50 mm/s

Fig. 4 High-magnification surface morphologies (FE-SEM) of SLM-processed W-based alloyparts under various SLM processing conditions, showing typical features of the solidified liq-uid front in laser scanned tracks: (a) LF 5 10.7 J/mm2, v 5 300 mm/s, (b) LF 5 16 J/mm2,v 5 200 mm/s, (c) LF 5 32 J/mm2, v 5 100 mm/s, and (d) LF 5 64 J/mm2, v 5 50 mm/s

Journal of Manufacturing Science and Engineering AUGUST 2016, Vol. 138 / 081003-5

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 03/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

laser-processed W-based part. First, the narrow microcracks wereformed on the surface, typically in the middle of the scan trackswhere the SLM working temperature is highest. Second, a numberof spherical-shaped inclusions were present on the surface(Fig. 3(b)). High-magnification characterization revealed thatthese inclusions were the solidified liquid splashes caused by thesignificant liquid instability and resultant balling effect duringSLM process with a high laser energy input (Fig. 4(d)). The occur-rence of defects including microcracks formation and ballingphenomenon lowered the densification level of SLM-processedpart to 92.4% TD in this case (Fig. 5).

Typically, SLM is processed based on a complete melting met-allurgical mechanism of the powder, and the dynamic viscosity(l) of the molten pool consisting of the complete liquid formationis controlled by SLM working temperature (T) and is definedby [30]

l ¼ 16

15

ffiffiffiffiffiffiffiffim

kBT

rc (5)

where m is the atomic mass, and c is the surface tension of the liq-uid. Using a higher laser scan speed (v) results in a shorter dwell-ing time of laser beam on the surface of molten pool and attendantlower LF acted on the pool, thereby decreasing the operative Twithin the pool. The dynamic viscosity of the W–Ti–C meltwithin the pool accordingly increases (Eq. (5)). The combinedeffects of a shorter liquid lifetime and an elevated viscosity of themelt result in a poor rheological property of the melt and, accord-ingly, a limited wettability of the melt with the neighboringscanned track and the previously processed layer. Furthermore,due to the limited SLM working temperature, the appearance ofthe unmelted W particles ahead the liquid solidification front(Fig. 4(a)) further decreases the wetting characteristics of themelt. Consequently, a large amount of intertrack and interlayer re-sidual porosity tends to be remained during laser rapid solidifica-tion process (Fig. 3(a)), thereby lowering the densificationresponse of SLM-processed W-based part (Fig. 5).

The present study reveals that as a reasonably high LF of32 J/mm2 is applied, the W-based powder system can be meltedfully, which is demonstrated by the continuous and ordered liquidmelting/solidification front (Figs. 3(c) and 4(c)). The completemelting of W-based powder and the subsequent stable solidifica-tion behavior of the melt play a crucial role in achieving a suffi-ciently high densification of SLM-processed W-based alloyparts. Nevertheless, care should be taken to control the input LF;the application of an excessive LF of 64 J/mm2, although it pro-vides a high potential for melting the W-based powder com-pletely, exerts a detrimental influence on the densification

behavior of SLM-processed part (Fig. 5), due to the presence ofmicrocracks and balling effect in this situation (Figs. 3(d) and4(d)). In laser-induced molten pool containing the complete liquidformation, a significant temperature gradient will be presentbetween the center and edge of the pool. The temperature gradientin the pool gives rise to surface tension gradient and resultantMarangoni convection. As a connective stream, the formation ofMarangoni flow within the pool increases the magnitude of thethermocapillary force and resultant instability of the melt [31]. Ithas been disclosed that a higher LF and attendant elevated SLMworking temperature result in the intensified Marangoni convec-tion and liquid capillary instability effect [18,19]. Consequently,the liquid surface flow occurs from a region of low surface tensionto a region of high surface tension, changing the direction of liq-uid flow from a radially outward to an inward [32]. The radiallyinward flow causes the liquid to spheroidize toward the laserbeam center where the SLM working temperature is highest.Under this condition, a number of small-sized liquid droplets tendto splash from the surface of the molten track being solidified, dueto the reduction in the surface energy of liquid at short lengthscales. After solidification, a large amount of micrometer-scaledspherical splashes are formed and included on laser-processed sur-face (Fig. 4(d)), which is known as balling effect [33]. As a typicalmetallurgical defect associated with SLM process, the occurrenceof balling effect tends to disturb the advancing liquid solidifica-tion front significantly, hence producing interrupted scan tracksand, finally, lowering the densification rate of SLM-processed partbased on multitrack and multilayer laser scanning (Fig. 5).

On the other hand, the formation of microcracks on the surfaceat an excessive LF (Fig. 3(d)) is believed to be thermal crackscaused by residual thermal stresses. During SLM densificationprocess, shrinkage occurs speedily once the powder materialbecomes fully molten. A nominal flat surface of powder bedbecomes a curved surface or a recessed region due to the effect ofdensification caused by conversion of the loose powder to denseliquid [34]. Based on the previous studies by Dai and Shaw [35]and Xiao and Zhang [36], a dominant mechanism that inducesresidual stresses in laser additive manufactured part is that thecool-down phase of the molten top layer tends to shrink due to thethermal contraction; however, simultaneously, such a deformationis inhibited by the underlying previously solidified layers. Thethermal shrinkage for a given metallic material increases as theinput laser energy density increases [37], thereby accumulatingconsiderable thermal stresses in SLM part being solidified. Themicrocracks in SLM-processed W-based alloy part, as noted inFig. 3(d), typically belong to the hot cracking, caused by the inter-ruption of liquid film at grain boundaries in the solidificationtemperature range, due to the action of the complex tensilestresses [38].

3.3 Microstructures and Chemical Compositions. Thecharacteristic microstructures of W-based crystals on the etchedcross sections of SLM-processed W-based alloy parts using vari-ous LFs are provided in Figs. 6(a)–6(d). The chemical composi-tions of the corresponding W-based crystals are detected by EDXmethod, as revealed in Table 2. The underlying crystallizationmechanisms of W crystals during SLM are schematically depictedin Fig. 7. Generally, the microstructural features of W-based crys-tals in SLM-processed W-based alloy parts differed entirely fromthe grain-structured crystals in the conventionally PM-processedW-based parts based on solid-state sintering or liquid-phase sin-tering mechanisms [12–14]. The complete melting/solidificationprocessing mechanism contributed to the formation of uniquemicrostructures of SLM-processed W-based alloy parts. EDXelement analysis revealed that the crystals in SLM-processedstructures were generally identified as the W crystals dissolvedwith Ti and C elements. As the applied LF increased, more Ti andC elements were found to be dissolved in the W matrix (Table 2).Meanwhile, it was found that the applied LF played a key role in

Fig. 5 Effects of laser scan speed and resultant LF on densifi-cation levels of SLM-processed W-based alloy parts

081003-6 / Vol. 138, AUGUST 2016 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 03/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

determining the microstructural development of W crystals duringSLM. At a relatively low LF of 10.7 J/mm2 combined with a highv of 300 mm/s, the directionally solidified W crystals with a cellu-lar structure were present on the cross section of SLM-processedpart. The cellular crystals were significantly refined, with the aver-age length and diameter of cellular crystals less than 15 lm and1 lm, respectively (Fig. 6(a)). As the LF increased to 16 J/mm2,although the as-developed W-based crystals almost demonstrateda directionally solidification feature, the mixed crystal structuresconsisting of cellular crystals and cellular dendritic crystals werepresent. A fraction of W-based crystals developed into a cellulardendritic morphology, and the refined secondary dendrites start toform along the cellular crystals (Fig. 6(b)). At a sufficiently highLF of 32 J/mm2, the homogeneous cellular dendritic W-basedcrystals were produced after SLM, having a reduced crystal lengthof �5 lm and a slightly coarsened cellular width of �1.5 lm(Fig. 6(c)). The microstructural features of SLM-processed

W-based alloy part changed completely at an even higher LF of64 J/mm2; the equiaxed dendritic W-based crystals were formedafter SLM, with the average crystalline size in the order of2–4 lm (Fig. 6(d)). For comparison, the typical microstructure ofSLM-processed pure W part without the addition of TiC is pro-vided in Fig. 6(e). A relatively coarsened granular crystal featurewas observed, which was different from the cellular/ dendriticgrowth features of SLM-processed W-based alloy parts(Figs. 6(a)–6(d)).

It was accordingly reasonable to conclude that as the appliedLF increased by lowering laser scan speed v, the morphologies ofW-based crystals in SLM-processed W-based alloy parts experi-enced a successive change from the cellular crystal to the cellulardendritic crystal and, finally, to the equiaxed dendritic crystal, asdepicted schematically in Fig. 7. During laser rapid solidificationprocess, the velocity of the solidification front is determined bythe temperature field. As the velocity of the solidification front

Fig. 6 Characteristic microstructures (FE-SEM) on the etched cross sections ofSLM-processed W-based alloy parts using different SLM processing parameters,showing laser-controlled crystallization features of W-based crystals: (a) LF 5 10.7 J/mm2,v 5 300 mm/s, (b) LF 5 16 J/mm2, v 5 200 mm/s, (c) LF 5 32 J/mm2, v 5 100 mm/s, and (d)LF 5 64 J/mm2, v 5 50 mm/s. Typical microstructure of SLM-processed pure W part at LF of32 J/mm2 is provided for a comparison purpose (e).

Table 2 EDX elemental analysis in point #1, #2, #3, and #4 in Figs. 6(a)–6(d) showing the chemi-cal compositions of W-based crystals in SLM-processed W-based alloy parts

Chemical compositions (wt.%)

Test position W C Ti

Point 1, Fig. 6(a) 95.71 6 4.3 2.99 6 1.2 1.30 6 0.1Point 2, Fig. 6(b) 92.04 6 3.6 6.19 6 1.5 1.77 6 0.1Point 3, Fig. 6(c) 91.53 6 3.3 6.88 6 1.6 1.58 6 0.1Point 4, Fig. 6(d) 93.40 6 3.9 4.96 6 1.4 1.65 6 0.1

Journal of Manufacturing Science and Engineering AUGUST 2016, Vol. 138 / 081003-7

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 03/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

(vs) is along the maximum temperature gradient (i.e., along thenormal to the liquid–solid interface), vs can be expressed by [39]

jvsj ¼ v � n ¼ jvj cos h (6)

where n is the unit direction normal to the solid–liquid interface(i.e., parallel to vs), v is the laser scan speed, and h is the anglebetween the scan direction and normal to the liquid–solid inter-face. The crystallization and growth of crystals along with theadvancing solidification front are further controlled by the under-cooling of melt being solidified. The thermal undercooling DTt

and the constitutional undercooling DTc within the molten poolare expressed as follows [40]:

DTt ¼DHf

clp

FIv Ptð Þ (7)

DTc ¼ mc0 1� m�=m

1� 1� k vsð Þ� �

FIv Pcð Þ

!(8)

Here, FIvðxÞ � x exp ðxÞE1ðxÞ denotes the Ivantsov function (withE1 as the first exponential integral function), Pt ¼ vsR=2DT thethermal P�eclet number (with R as the curvature radius of the crys-tal tip and DT as the thermal diffusivity), DHf the heat of fusion,cl

p the specific heat of the liquid, m� the velocity dependent liqui-dus slope, m the liquidus slope, c0 the nominal concentration ofthe metal, Pc ¼ vsR=2DL the chemical P�eclet number (with DL asthe diffusion coefficient), and kðvsÞ ¼ ½ðke þ vs=vDÞ=ð1þ vs=vDÞ�is the velocity dependence of the partition coefficient (with ke asthe equilibrium partition coefficient and vD as the diffusive speedthat is treated as a free parameter). At a lower v and resultanthigher LF, the interfacial velocity of the advancing solidificationfront (vs) decreases (Eq. (6)). In this situation, the thermal under-cooling DTt shows a decrease, whereas the constitutional under-cooling DTc is elevated, as revealed by Eqs. (7) and (8). With theenhancement of constitutional undercooling DTc, the smoothparabolic-shaped interfaces of cellular crystals (Fig. 7(a)) becomesignificantly unstable and, therefore, the primary dendrites tend tobe bulged along the cellular crystals, leading to the formation ofcellular dendritic crystal structure (Fig. 7(b)). Furthermore, due toa decrease of thermal undercooling DTt and a significant thermalaccumulation within the pool caused by the combination of a lowv and a high LF, the effect of quenching that occurs by conductionof heat through the previously processed layers is not significant

in this situation. A large amount of heat is accordingly accumu-lated around the growing secondary dendrites, which are formeddue to a significant interfacial instability caused by a high DTc,hence resulting in the formation of equiaxed dendritic crystals atan excessively high LF (Fig. 7(c)).

3.4 Microhardness and Tribological/Wear Performance.Figure 8(a) depicts the microhardness and its distribution meas-ured along the SLM building direction on the cross sections ofSLM-processed W-based alloy parts. With an increase of theapplied LF from 10.7 J/mm2 to 16 J/mm2, the average microhard-ness increased from 700.4 HV0.3 to 757.6 HV0.3 and, meanwhile,the fluctuation of the microhardness distribution was weakenedsignificantly. At a suitable LF of 32 J/mm2, a uniform distributionof the microhardness with a considerably enhanced mean value of

Fig. 7 Morphological change of W-based crystals during SLMwith the increase of LF

Fig. 8 Microhardness (a), COF (b), and wear rate (c) of SLM-processed W-based alloy parts at different LFs

081003-8 / Vol. 138, AUGUST 2016 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 03/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

809.9 HV0.3 was achieved, due to the combined effects of thenearly full densification of SLM-processed part (Fig. 5) and theformation of homogeneous, refined cellular dendritic microstruc-tures of W-based crystals (Fig. 6(c)). A further increase in theapplied LF to 64 J/mm2, however, decreased the obtained micro-hardness to 778.3 HV0.3 and, meanwhile, intensified the fluctua-tion of hardness values. In this instance, the limited densificationlevel of SLM part caused by balling effect and microcracks for-mation (Figs. 3(d) and 4(d)) and the presence of relatively coars-ened, heterogeneous equiaxed dendritic crystals (Fig. 6(d)) wereresponsible for the decrease in hardness performance. Here,Arshad et al.’s work [13] on hot-press sintering of W–5 wt. %VW-based alloy was taken for a comparison, in which the obtainedmaximum microhardness was 589.8 HV0.2. The microhardness ofSLM-processed W-based alloy parts with the novel crystallinemicrostructures, accordingly, showed a maximum 27% increaseupon the conventionally PM-processed W-based materials.

The variations of COFs and resultant wear rates of SLM-processed W-based alloy parts are depicted in Figs. 8(b) and 8(c),respectively. It revealed that the applied LF exerted a significanteffect upon the wear/tribological properties of W-based parts. Anenhancement of LF from 10.7 J/mm2 to 16 J/mm2 lowered the av-erage COF slightly from 0.58 to 0.52 and the resultant wear ratedecreased accordingly from 8.36� 10�7 m3/(N m) to 7.02� 10�7

m3/(N m). The W-based alloy part processed at LF of 32 J/mm2

demonstrated the superior wear performance. An apparentlydecreased COF of 0.41 was obtained, leading to the lowestwear rate of 5.73� 10�7 m3/(N m). A further increase of LF to64 J/mm2, however, worsened the wear properties; the mean COFand attendant wear rate of W-based alloy part increased markedlyto 0.49 and 6.85� 10�7 m3/(N m), respectively.

In order to disclose the underlying wear mechanisms responsi-ble for the variations of wear/tribological properties, the typicalmorphologies of worn surfaces are provided in Fig. 9. At a rela-tively low LF of 10.7 J/mm2, the worn surface was considerablyrough and wavy, consisting of parallel grooves (Fig. 9(a)). Suchwear morphology revealed that the sample suffered a heavyabrasive wear, thereby producing a relatively high wear rate

(Fig. 8(c)). Both the low densification level of SLM-processedpart caused by the residual porosity (Figs. 3(a) and 5) and theformation of directionally solidified cellular W crystals with inter-cellular shrinkage porosity (Fig. 6(a)) were responsible for thedecrease of wear performance in this instance. As the applied LFincreased to 16 J/mm2, the grooves on the worn surface becameshallow (Fig. 9(b)). When a sufficiently high LF of 32 J/mm2 wasapplied, the worn surface was covered completely with the denseand smooth adhesion tribolayer, without the local plowing or frac-turing of the worn surface (Fig. 9(c)). During dry sliding tests, thecounterface ball slid against the surface continuously. The wornsurface experienced the sufficient plastic deformation at a temper-ature below its recrystallization temperature. The materialstrengthening was expected to occur, due to the movement andpinning of dislocations within the crystalline structure of the ma-terial, which was known as strain-hardened tribolayer [41,42].The formation of significantly refined cellular dendritic crystals ofW-based alloy part (Fig. 6(c)) contributed to the formation ofstrain-hardened tribolayer to enhance wear resistance of the sur-face. However, at an excessive LF of 64 J/mm2, although thestrain-hardened, smooth worn surface was still visible, the tribo-layer underwent a severe fragment and the interconnected, thin,and deep cracks were present on the worn surface (Fig. 9(d)),resulting in a significant increase of COF and wear rate (Figs. 8(b)and 8(c)). In this situation, the limited densification caused byboth balling effect and thermal cracking (Figs. 3(d) and 4(d)) andthe coarsening of equiaxed dendritic crystals of W-based alloypart (Fig. 6(d)) were responsible for the spalling of tribolayer and,accordingly, the decrease in wear performance.

4 Conclusions

SLM AM of W-based alloy parts was performed in this studyand the following conclusions were drawn:

(1) The applied LF played a significant role in determiningdensification behavior of SLM-processed W-based alloyparts. The densification response of SLM-processed parts

Fig. 9 Typical morphologies (FE-SEM) of worn surfaces of SLM-processed W-basedalloy parts at various SLM processing parameters: (a) LF 5 10.7 J/mm2, v 5 300 mm/s, (b)LF 5 16 J/mm2, v 5 200 mm/s, (c) LF 5 32 J/mm2, v 5 100 mm/s, and (d) LF 5 64 J/mm2,v 5 50 mm/s

Journal of Manufacturing Science and Engineering AUGUST 2016, Vol. 138 / 081003-9

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 03/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

decreased both at a low LF of 10.7 J/mm2, caused by thelimited SLM working temperature and wetting characteris-tics of the melt, and at an excessively high LF of 64 J/mm2,caused by the significant melt instability and resultant ball-ing effect and microcracks formation.

(2) SLM of W–2.5 wt. % TiC powder system was processedusing a novel metallurgical mechanism with the completemelting of the powder. The laser-induced complete melt-ing/solidification mechanism contributed to the solid solu-tion alloying of Ti and C in W and the development ofunique microstructures of SLM-processed W-based alloyparts. As the LF increased by lowering laser scan speed, themorphologies of W-based crystals experienced a successivechange from the cellular crystal to the cellular dendriticcrystal and, finally, to the equiaxed dendritic crystal, due tothe elevated constitutional undercooling and the decreasedthermal undercooling.

(3) The optimally prepared W-based alloy parts by SLM at aproper LF of 32 J/mm2 had a nearly full densificationrate of 97.8% TD, a considerably high microhardness of809.9 HV0.3, and a superior wear/ tribological performancewith a decreased COF of 0.41 and a low wear rate of5.73� 10�7 m3/(N m), due to the combined effects of thesufficiently high densification and novel crystal microstruc-tures of SLM-processed W-based alloy parts. However,the microhardness and wear resistance decreased at anexcessive LF of 64 J/mm2 due to the formation of thermalmicrocracks, significant balling effect, and heterogeneousequiaxed dendritic crystals.

Acknowledgment

The authors gratefully appreciate the financial support from theNational Natural Science Foundation of China (Nos. 51322509 and51575267), the Top-Notch Young Talents Program of China theOutstanding Youth Foundation of Jiangsu Province of China (No.BK20130035), the Program for New Century Excellent Talents inUniversity (No. NCET-13-0854), the Science and Technology Sup-port Program (The Industrial Part), Jiangsu Provincial Departmentof Science and Technology of China (No. BE2014009-2), the 333Project (No. BRA2015368), the Science and Technology Founda-tion for Selected Overseas Chinese Scholar, Ministry of HumanResources and Social Security of China, the Aeronautical ScienceFoundation of China (No. 2015ZE52051), the Shanghai AerospaceScience and Technology Innovation Fund (No. SAST2015053),and the Fundamental Research Funds for the Central Universities(Nos. NE2013103 and NP2015206).

References[1] Jahan, M. P., Saleh, T., Rahman, M., and Wong, Y. S., 2010, “Development,

Modeling, and Experimental Investigation of Low Frequency WorkpieceVibration-Assisted Micro-EDM of Tungsten Carbide,” ASME J. Manuf. Sci.Eng., 132(5), p. 054503.

[2] Arif, M., Rahman, M., and San, W. Y., 2012, “A Model to Determine the Effectof Tool Diameter on the Critical Feed Rate for Ductile–Brittle Transition inMilling Process of Brittle Material,” ASME J. Manuf. Sci. Eng., 134(5),p. 051012.

[3] Alemohammad, H., and Toyserkani, E., 2011, “Metal Embedded Optical FiberSensors: Laser-Based Layered Manufacturing Procedures,” ASME J. Manuf.Sci. Eng., 133(3), p. 031015.

[4] Smid, I., Akiba, M., Vieider, G., and Pl€ochl, L., 1998, “Development ofTungsten Armor and Bonding to Copper for Plasma-Interactive Components,”J. Nucl. Mater., 258–263, pp. 160–172.

[5] Zhang, G. Q., and Gu, D. D., 2013, “Synthesis of Nanocrystalline TiCReinforced W Nanocomposites by High-Energy Mechanical Alloying: Micro-structural Evolution and Its Mechanism,” Appl. Surf. Sci., 273, pp. 364–371.

[6] Qiao, Z. H., Wang, H. J., Ma, X. F., Zhao, W., and Tang, H. G., 2008,“Nanostructured Bulk Novel Hard Material With “Rounded” Grains Obtainedby Nanocrystalline “Rounded” (W0.5Al0.5)C Powders,” Mater. Sci. Eng. A,496, pp. 507–511.

[7] Qiao, Z. H., Ma, X. F., Zhao, W., and Tang, H. G., 2008, “Fabrication, ThermalStability and Mechanical Properties of Novel Composite Hardmetals Obtainedby Solid-State Reaction and Hot-Pressing Sintering,” Mater. Chem. Phys., 108,pp. 364–368.

[8] Wu, J. G., Zhou, S. Y., and Li, X. C., 2015, “Ultrasonic Attenuation BasedInspection Method for Scale-Up Production of A206–Al2O3 Metal MatrixNanocomposites,” ASME J. Manuf. Sci. Eng., 137(1), p. 011013.

[9] Ye, C., and Cheng, G. J., 2010, “Laser Shock Peening of Nanoparticles Inte-grated Alloys: Numerical Simulation and Experiments,” ASME J. Manuf. Sci.Eng., 132(6), p. 061017.

[10] Gu, D. D., Wang, H. Q., Dai, D. H., Yuan, P. P., Meiners, W., and Poprawe, R.,2015, “Rapid Fabrication of Al-Based Bulk-Form Nanocomposites With NovelReinforcement and Enhanced Performance by Selective Laser Melting,” Scr.Mater., 96, pp. 25–28.

[11] Attar, H., Bonisch, M., Calin, M., Zhang, L. C., Scudino, S., and Eckert, J.,2014, “Selective Laser Melting of In Situ Titanium–Titanium Boride Compo-sites: Processing, Microstructure and Mechanical Properties,” Acta Mater., 76,pp. 13–22.

[12] Rodriguez-Suarez, T., D�ıaz, L. A., Torrecillas, R., Lopez-Esteban, S., Tuan,W. H., Nygren, M., and Moya, J. S., 2009, “Alumina/Tungsten NanocompositesObtained by Spark Plasma Sintering,” Compos. Sci. Technol., 69(14),pp. 2467–2473.

[13] Arshad, K., Wang, J., Yuan, Y., Zhang, Y., Zhou, Z. J., and Lu, G. H., 2015,“Development of Tungsten-Based Materials by Different SinteringTechniques,” Int. J. Refract. Met. Hard Mater., 50, pp. 253–257.

[14] Park, J. W., Suh, J. Y., Kang, S. W., Shin, S. E., and Bae, D. H., 2014, “TheEffect of the Size and Volume Fraction of Zr2Cu on the Sintering Behavior ofTungsten Matrix Composites During Liquid-Reactive Sintering,” Int. J. Refract.Met. Hard Mater., 43, pp. 157–163.

[15] Tapia, G., and Elwany, A., 2014, “A Review on Process Monitoring and Con-trol in Metal-Based Additive Manufacturing,” ASME J. Manuf. Sci. Eng.,136(6), p. 060801.

[16] Ghariblu, H., and Rahmati, S., 2014, “New Process and Machine for LayeredManufacturing of Metal Parts,” ASME J. Manuf. Sci. Eng., 136(4), p. 041004.

[17] Paul, R., Anand, S., and Gerner, F., 2014, “Effect of Thermal Deformation onPart Errors in Metal Powder Based Additive Manufacturing Processes,” ASMEJ. Manuf. Sci. Eng., 136(3), p. 031009.

[18] Gu, D. D., 2015, Laser Additive Manufacturing of High-Performance Materi-als, Springer-Verlag, Berlin.

[19] Gu, D. D., Meiners, W., Wissenbach, K., and Poprawe, R., 2012, “Laser Addi-tive Manufacturing of Metallic Components: Materials, Processes and Mecha-nisms,” Int. Mater. Rev., 57(3), pp. 133–164.

[20] Mertens, R., Clijsters, S., Kempen, K., and Kruth, J. P., 2014, “Optimization ofScan Strategies in Selective Laser Melting of Aluminum Parts With Downfac-ing Areas,” ASME J. Manuf. Sci. Eng., 136(6), p. 061012.

[21] Kempen, K., Vrancken, B., Buls, S., Thijs, L., Van Humbeeck, J., and Kruth,J. P., 2014, “Selective Laser Melting of Crack-Free High Density M2 HighSpeed Steel Parts by Baseplate Preheating,” ASME J. Manuf. Sci. Eng., 136(6),p. 061026.

[22] Fu, C. H., and Guo, Y. B., 2014, “Three-Dimensional Temperature GradientMechanism in Selective Laser Melting of Ti–6Al–4V,” ASME J. Manuf. Sci.Eng., 136(6), p. 061004.

[23] Das, M., Balla, V. K., Basu, D., Bose, S., and Bandyopadhyay, A., 2010, “LaserProcessing of SiC-Particle-Reinforced Coating on Titanium,” Scr. Mater.,63(4), pp. 438–441.

[24] Zhou, X., Liu, X. H., Zhang, D. D., Shen, Z. J., and Liu, W., 2015, “BallingPhenomena in Selective Laser Melted Tungsten,” J. Mater. Process Technol.,222, pp. 33–42.

[25] Gu, D. D., and Shen, Y. F., 2009, “Effects of Processing Parameters on Consoli-dation and Microstructure of W–Cu Components by DMLS,” J. Alloys Compd.473, pp. 107–115.

[26] Gu, D. D., Hagedorn, Y. C., Meiners, W., Meng, G. B., Batista, R. J. S.,Wissenbach, K., and Poprawe, R., 2012, “Densification Behavior, Microstruc-ture Evolution, and Wear Performance of Selective Laser Melting ProcessedCommercially Pure Titanium,” Acta Mater., 60(9), pp. 3849–3860.

[27] Agarwala, M., Bourell, D., Beaman, J., Marcus, H., and Barlow, J., 1995,“Direct Selective Laser Sintering of Metals,” Rapid Prototyping J., 1(1),pp. 26–36.

[28] Goldschmidt, H. J., and Brand, J. A., 1963, “The Tungsten-Rich Region of theSystem Tungsten–Carbon,” J. Less Common Met., 5(2), pp. 181–194.

[29] Pan, J. S., 1998, Materials Science Fundamentals, 1st ed., Tsinghua UniversityPress, Beijing, China.

[30] Takamichi, I., and Roderick, I. L. G., 1993, The Physical Properties of LiquidMetals, Clarendon Press, Oxford, UK.

[31] Kruth, J. P., Levy, G., Klocke, F., and Childs, T. H. C., 2007, “ConsolidationPhenomena in Laser and Powder-Bed Based Layered Manufacturing,” CIRPAnn. Manuf. Technol., 56(2), pp. 730–759.

[32] Simchi, A., 2006, “Direct Laser Sintering of Metal Powders: Mechanism,Kinetics and Microstructural Features,” Mater. Sci. Eng. A, 428(1–2),pp. 148–158.

[33] Tolochko, N. K., Mozzharov, S. E., Yadroitsev, I. A., Laoui, T., Froyen, L.,Titov, V. I., and Ignatiev, M. B., 2004, “Balling Processes During SelectiveLaser Treatment of Powders,” Rapid Prototyping J., 10(2), pp. 78–87.

[34] Dai, K., and Shaw, L., 2005, “Finite Element Analysis of the Effect ofVolume Shrinkage During Laser Densification,” Acta Mater., 53(18),pp. 4743–4754.

[35] Dai, K., and Shaw, L., 2001, “Thermal and Stress Modeling of Multi-MaterialLaser Processing,” Acta Mater., 49(20), pp. 4171–4181.

[36] Xiao, B., and Zhang, Y. W., 2007, “Analysis of Melting of AlloyPowder Bed With Constant Heat Flux,” Int. J. Heat Mass Transfer, 50,pp. 2161–2169.

081003-10 / Vol. 138, AUGUST 2016 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 03/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

[37] Zhu, H. H., Lu, L., and Fuh, J. Y. H., 2006, “Study on Shrinkage Behaviour ofDirect Laser Sintering Metallic Powder,” Proc. Inst. Mech. Eng., Part B,220(2), pp. 183–190.

[38] Zhong, M. L., Sun, H. Q., Liu, W. J., Zhu, X. F., and He, J. J., 2005, “BoundaryLiquation and Interface Cracking Characterization in Laser Deposition ofInconel 738 on Directionally Solidified Ni-Based Superalloy,” Scr. Mater.,53(2), pp. 159–164.

[39] Yang, S., Huang, W. D., Liu, W. J., Zhong, M. L., and Zhou, Y. H., 2002,“Development of Microstructures in Laser Surface Remelting of DD2 SingleCrystal,” Acta Mater., 50(2), pp. 315–325.

[40] Schwarz, M., Arnold, C. B., Aziz, M. J., and Herlach, D. M., 1997, “DendriticGrowth Velocity and Diffusive Speed in Solidification Undercooled DiluteNi–Zr Melts,” Mater. Sci. Eng. A, 226–228, pp. 420–424.

[41] Gu, D. D., Chang, F., and Dai, D. H., 2015, “Selective LaserMelting Additive Manufacturing of Novel Aluminum Based CompositesWith Multiple Reinforcing Phases,” ASME J. Manuf. Sci. Eng., 137(2),p. 021010.

[42] Jain, A., Basu, B., Kumar, B. V. M., Harshavardhan, and Sarkar, J., 2010,“Grain Size–Wear Rate Relationship for Titanium in Liquid Nitrogen Environ-ment,” Acta Mater., 58(7), pp. 2313–2323.

Journal of Manufacturing Science and Engineering AUGUST 2016, Vol. 138 / 081003-11

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 03/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use