inﬂuence of processing parameters on particulate dispersion in...

Available online at www.sciencedirect.com

www.elsevier.com/locate/IJRMHM

International Journal of Refractory Metals & Hard Materials 26 (2008) 411–422

Influence of processing parameters on particulate dispersion indirect laser sintered WC–Cop/Cu MMCs

Dongdong Gu *, Yifu Shen, Jun Xiao

College of Materials Science and Technology, Nanjing University of Aeronautics and Astronautics, 29 Yudao Street, 210016 Nanjing, PR China

Received 31 March 2007; accepted 20 September 2007

Abstract

Homogenizing the particulate dispersion in the matrix is an important consideration in obtaining high-quality particulate reinforcedmetal matrix composites (MMCs) using direct metal laser sintering (DMLS). In this paper, the effects of processing parameters in termsof laser power, scan speed, and powder layer thickness on the particulate dispersion in the DMLS-processed submicron WC–Cop/Cu1

MMCs were investigated. It shows that for a given scan speed larger than 0.04 m/s, a proper increase in the laser power between �650 Wand �750 W leads to a uniform dispersion of smooth and fine WC particulates, due to a sufficient liquid formation and an improvedwettability. Whereas, an excessive increase in the laser power results in a severe particulate aggregation, because of the balling effect.There exists a critical scan speed of 0.04 m/s, above which the particulates can be well engulfed by the advancing dendritic front, therebyhomogenizing the particulate dispersion in the matrix. A proper decrease in the powder layer thickness to 0.20 mm can alleviate the par-ticulate aggregation, due to the elevated Marangoni convection and liquid capillary forces.� 2007 Elsevier Ltd. All rights reserved.

Keywords: Metal matrix composites (MMCs); Particulate reinforcement; Rapid prototyping (RP); Direct metal laser sintering (DMLS); WC–Co

1. Introduction

Particulate reinforced metal matrix composites (MMCs)exhibit a favorable combination of metallic matrix and stif-fer and stronger reinforcements [1,2]. Depending on thenature of the final product desired, the selection of typesof metallic matrix and reinforcement is performed [3].WC–Co, as a well-known hard metal, has excellent proper-ties such as high strength, high hardness, wear resistance,and reasonable fracture toughness [4]. Copper is character-ized by excellent electrical and thermal conductivity andoutstanding resistance to fatigue and corrosion [5,6]. Inorder to combine their superior properties, the develop-

0263-4368/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.

doi:10.1016/j.ijrmhm.2007.09.005

* Corresponding author. Tel.: +86 25 52112904x80517; fax: +86 2552112626.

E-mail addresses: [email protected], [email protected] (D. Gu).

1 All the subscript ‘‘p’’ indicates powder form.

ment of WC–Co particulate reinforced Cu matrix compos-ites is of significance.

Direct metal laser sintering (DMLS), as a typical rapidprototyping (RP) technique, enables the quick productionof complex shaped three-dimensional (3D) parts directlyfrom metal powder [7–11]. The DMLS process createsparts in a layer-by-layer fashion by selectively fusing thinlayers of loose powder with a scanning laser beam. Eachscanned layer represents a cross-section of the object’smathematically sliced CAD model. After consolidation ofone cross-section, a fresh layer of powder is depositedand the process is repeated until a 3D part is finished. Thistechnique competes effectively with other conventionalmanufacturing processes when the part geometry is com-plex and the production run is not large [12]. Our recentresearch efforts reveal that DMLS process, due to its flexi-bility in materials and shapes, exhibits a great potential fornet-shape fabrication of complex shaped WC–Co particu-late reinforced Cu matrix composites that cannot be easilydeveloped by other conventional methods [13].

mailto:[email protected]



412 D. Gu et al. / International Journal of Refractory Metals & Hard Materials 26 (2008) 411–422

However, as is evident from our previous work [14], theaggregation of the reinforcing particulates and the resul-tant crack formation between the reinforcement and thematrix occurs frequently in the laser processed MMCs,due to the complex metallurgical nature of DMLS involv-ing multiple modes of heat, mass, and momentum transfer[15,16]. Thus, in order to obtain high-quality MMCs byDMLS, it is important to establish the process conditionsunder which laser fabrication is able to generate a uniformdistribution of particulates in such composites. These con-ditions include both powder characteristics (e.g., particleshape, particle size and its distribution, component ratio)and processing parameters (e.g., laser power, scan speed,powder layer thickness). Besides the optimization of pow-der characteristics [14], the capability of defining the pro-cessing window in improving the particulate dispersionhomogeneity is becoming another important consideration.

As a further step in obtaining high performance MMCswith controllable microstructures, the present workemployed DMLS to prepare a series of WC–Cop/Cu MMCsunder different processing parameters. The effects of laserpower, scan speed, and powder layer thickness on micro-structural features of the laser processed MMCs, especiallythe dispersion states of the reinforcing particulates (e.g.,the degree of dispersion, particle shape, and particle size),were investigated.

2. Experimental

2.1. Materials

Electrolytic 99% purity Cu powder with an irregularshape and a mean particle size of 15 lm (Fig. 1a) andWC–10 wt.% Co composite powder with an irregular struc-ture and an average equivalent spherical diameter of0.6 lm (Fig. 1b) were used in this experiment. The WC–

Fig. 1. SEM images showing characteristic morphologies of the sta

Co composite powder was synthesized using a novel ‘‘spraydrying and fixed bed’’ technique, which involved spray dry-ing a precursor solution containing ammonium metatung-state (AMT) and Co(NO3)2, followed by roasting, ballmilling, reduction, and carbonization [17]. The two compo-nents (i.e., Cu and WC–Co) were mixed according to theCu:WC–Co weight ratio of 70:30 (the equivalent volumefraction of the WC–Co constituent of 20.4 vol.% and theWC constituent of 17.5 vol.%, respectively) in a vacuumball mill at a rotation speed of 150 rpm for 60 min, withballs to powders weight ratio of 5:1.

2.2. Processing

The DMLS apparatus mainly consisted of a continuouswave Gaussian CO2 laser with a maximum output power of2000 W, an automatic powder delivery system, and a com-puter system for the process control. Prior to the laser sin-tering process, a steel substrate was placed on the buildingplatform and leveled. Afterwards, a thin layer of the loosepowder was spread on the substrate by the roller. Subse-quently, a laser beam scanned the powder bed surface toform a layer-wise profile according to the CAD data ofthe part. The process was repeated and the part was pro-duced in a layer-by-layer fashion until completion. Theentire sintering process was performed in ambient atmo-sphere at room temperature. The following processingparameters were used: spot size 0.30 mm, scan line spacing0.15 mm, laser power 550–800 W, scan speed 0.01–0.07 m/s,and powder layer thickness 0.20–0.40 mm.

2.3. Characterization

Samples for metallographic examinations were cut,ground, and polished according to standard procedures.A solution consisting of FeCl3 (5 g), HCl (10 mL), and dis-

rting powder: (a) Cu powder and (b) WC–10 wt.% Co powder.

D. Gu et al. / International Journal of Refractory Metals & Hard Materials 26 (2008) 411–422 413

tilled water (100 mL) was taken as an etching agent, withan etching time being 30 s. Tensile strength tests were per-formed at room temperature in a digitally controlled CMT-5105 testing machine at a loading rate of 1.0 mm/min.Microstructure was characterized using a QUANTA 200scanning electron microscope (SEM) in both secondaryelectron (SE) and back-scattered electron (BSE) modes.Chemical composition was examined by EDAX energy dis-persive X-ray spectroscope (EDX).

3. Results

3.1. Process map

In the present study, a series of laser sintered sampleswere prepared under different processing parameters (indi-cated by different symbols in Fig. 2), in order to establishthe process window in which laser sintering is able to gen-erate a uniform distribution of particulates in such com-posites. The changes of the dispersion states of thereinforcing particulates in laser sintered samples under dif-ferent processing conditions are depicted in Fig. 2. Over theentire range of laser powers and scan speeds, the followingfour processing windows are defined.

(I) Severe aggregation caused by balling effect: at a rela-tively high laser power for all scan speeds, a consider-ably high energy input leads to a large amount ofliquid formation with a too low melt viscosity, result-ing in the rupture of liquid scan tracks into rows ofspheres due to significant surface tension effects

Fig. 2. Process windows for dispersion states of reinforcing particulatesover a wide range of laser powers and scan speeds. laser sintering usingprocessing parameters denoted by ‘‘$’’ results in severe aggregation ofparticulates caused by balling effect; laser sintering using processingparameters denoted by ‘‘s’’ leads to partial aggregation of particulates;laser sintering using processing parameters denoted by ‘‘ ’’ yieldsuniform dispersion of particulates; laser sintering using processingparameters denoted by ‘‘4’’ results in severe aggregation of particulatescaused by limited liquid formation.

(‘‘balling’’ effect). The reinforcing particulates aggre-gate severely during balling process, resulting in anonuniform dispersion of the reinforcement in thematrix.

(II) Partial aggregation: at a relatively low laser powercombined with a lower scan speed, the deliveredenergy density of the laser beam is moderate, therebyforming a suitable amount of liquid phase to avoidballing effect. However, the used lower scan speedlimits the sufficient trapping of the particulates byan advancing liquid-solid interface, resulting in a par-tial aggregation of the reinforcing particulates in thefinally solidified metal matrix.

(III) Uniform dispersion: laser sintering at a suitable laserpower (�650 W to �750 W) generates a sufficientamount of liquid phase; and, meanwhile, a higherscan speed (>0.04 m/s) favors the particulate trap-ping effect, leading to a homogeneous distributionof the reinforcement in the matrix.

(IV) Severe aggregation caused by limited liquid forma-tion: at a lower laser power combined with a higherscan speed, a significantly low energy input resultsin an insufficient liquid formation with a high meltviscosity, hence decreasing the capillary forces forparticle rearrangement and, ultimately, resulting ina serious aggregation of the reinforcing particulatesafter sintering.

3.2. Microstructure

Fig. 3 shows the characteristic microstructures of thepolished samples prepared at different laser powers.The typical morphologies of the reinforcing particulatesin the corresponding samples are provided in Fig. 4. TheEDX spot analysis revealed that the dispersed white partic-ulates in Fig. 3 were generally composed of the W and Celements with a near iso-atomic proportion, while the sur-rounding matrix was the Cu and Co (�5 wt.%) dissolvedwith the W (�8 wt.%) and C (�6 wt.%) atoms. Further-more, the phase identification by X-ray diffraction(XRD), which was performed in our previous work [13],revealed that the laser processed materials were mainlycomposed of WC and Cu phases. Thus, it is reasonableto conclude that the WC particulate reinforced Cu matrixcomposites can be successfully prepared. However, the dis-persion states and microstructural features of the reinforc-ing particulates are significantly influenced by the laserpower. At a low laser power of 625 W (point 1 in regionIV), the WC particulates exhibited a serious agglomeration(Fig. 3a) and had an irregular polygonal shape with a par-ticle size larger than 1 lm (Fig. 4a). Laser sintering at ahigher laser power of 700 W (point 2 in region III) favoredthe deagglomeration of clusters of particulates, leading to apretty homogeneous distribution of particulates possessingsignificantly smoothened surfaces (Figs. 3b, 4b). However,at an even higher laser power of 775 W (point 3 in region

Fig. 3. BSE images showing characteristic microstructures of laser sintered samples at different laser powers: (a) 625 W (point 1 in Fig. 2); (b) 700 W (point2) and (c) 775 W (point 3). Fixed processing parameters are scan speed 0.06 m/s and layer thickness 0.20 mm.


I), although the WC particulates were partially smooth-ened (Fig. 4c), large clusters of the reinforcing particulateswere visible in the matrix (Fig. 3c).

Fig. 5 shows the characteristic microstructures on thepolished sections of the laser sintered samples prepared atdifferent scan speeds. The etched microstructures on thecorresponding sections are exhibited in Fig. 6. It is clearthat the distribution states of the reinforcing particulatesand the dendritic growth morphologies of the matrix metalvary with the used scan speeds. At a low scan speed of0.02 m/s (point 4 in region I), significantly coarsened clus-ters of the WC reinforcing particulates were visible(Fig. 5a). Such particulate clusters were pushed by thegrowing dendrites and segregated towards the tips of den-drites (Fig. 6a). With increasing the scan speed to 0.03 m/s(point 5 in region II), the dispersion homogeneity of the

WC particulates was improved, although some small-scaled aggregates were still visible (Fig. 5b). In this case,the reinforcing particulates ceased to be pushed, but startedto be incorporated into the growing dendrites (Fig. 6b).Interestingly, laser sintering at a higher scan speed of0.05 m/s (point 6 in region III) led to an almost uniformdispersion of the fine WC particulates (Fig. 5c). The den-drites of the matrix metal were well developed and the rein-forcing particulates were completely trapped in thedendrites (Fig. 6c).

Fig. 7 shows the characteristic dispersion states of thereinforcing particulates in the fracture surfaces (i.e. cross-sections) of the laser sintered samples at different layerthicknesses. At a relatively high layer thickness of0.30 mm, the fracture surface was consisted of large andcoarsened clusters of the reinforcing particulates, showing

Fig. 4. BSE images showing characteristic morphologies of reinforcing particulates at different laser powers: (a) 625 W (point 1 in Fig. 2); (b) 700 W (point2) and (c) 775 W (point 3). Fixed processing parameters are scan speed 0.06 m/s and layer thickness 0.20 mm.


a heterogeneous microstructure (Fig. 7a). With decreasingthe layer thickness to 0.25 mm, the reinforcing particulatesstarted to be refined and dispersed (Fig. 7b). At a low layerthickness of 0.20 mm, the reinforcing particulates were uni-formly inserted in the matrix, exhibiting a homogenous dis-tribution state (Fig. 7c).

4. Discussion

The present experimental results, as listed in Section 3,reveal that the used processing parameters, e.g., laserpower, scan speed, and layer thickness, significantly influ-ence the particulate dispersion in the DMLS-processedMMCs. The observed different microstructures are relatedto the variation in the thermodynamic and kinetic charac-teristics, such as solid-liquid wettability, solid-interface

interaction, and particle rearrangement, during laser sinter-ing. Therefore, it is necessary to rule out the exact reasonswhy the particulates aggregate or disperse under certainprocessing conditions.

4.1. Influence of laser power on solid-liquid wettability

When the laser beam scans over the powder bed, the laserenergy is directly absorbed by the solid particles throughboth bulk coupling and powder coupling mechanisms[18]. Firstly, the laser energy is absorbed in a narrow layerof individual powder particles determined by the bulk prop-erties of the material, leading to a high temperature of thesurface of the particles during the interaction. The heatflows mainly towards the center of the particles until a localsteady state of the temperature within the powder particle is

Fig. 5. BSE images showing characteristic microstructures of polished samples with variation of scan speeds: (a) 0.02 m/s (point 4 in Fig. 2); (b) 0.03 m/s(point 5); (c) 0.05 m/s (point 6). Fixed processing parameters are laser power 700 W and layer thickness 0.20 mm.


obtained. Afterwards, the surrounding powder propertiesare responsible for the further thermal development [18].Due to the size effect of the fine grained submicron WC–Co powder, the spreading of the binder phase Co on theWC grains occurs at a relatively low temperature of�1000 �C [4], so as to wet the WC phase firstly. A furtherincrease in the sintering temperature above the meltingpoint of the Cu powder (�1083 �C) leads to a larger amountof liquid formation, leading to a second wetting of the WC/Co system by the liquid. Since the molten Co has beenspread around the WC particulates in the initial stage of sin-tering, the liquid Cu directly contacts the metallic phase Co,rather than the ceramic phase WC. For a metal/metal sys-tem, the wettability is generally better than a metal/ceramicsystem because of the lower surface energy [6]. Thus, suchCu/Co (metal/metal) interface can efficiently improve

liquid-solid wetting characteristics, in comparison to theCu/WC (metal/ceramic) interface. The reinforcing particu-lates, consequently, experience a rapid rearrangement underthe influence of capillary forces exerted on them by the wet-ting liquid. In other words, the excellent wetting character-istics of the solid phase by the liquid phase are crucial inobtaining a uniform dispersion of the reinforcement in thefinally solidified matrix. The obtained results, as shown inFigs. 2–4, reveal that for a given scan speed, a properincrease in the laser power can homogenize the dispersionof particulates. Based on Tolochko et al.’s results [19], theviscosity of a solid-liquid mixture l during DMLS can beestimated by the following equation:

l ¼ l0 1� 1� ul

um

� ��2

ð1Þ

Fig. 6. SE images showing characteristic microstructures of etched samples with variation of scan speeds: (a) 0.02 m/s (point 4 in Fig. 2); (b) 0.03 m/s(point 5); (c) 0.05 m/s (point 6). Fixed processing parameters are laser power 700 W and layer thickness 0.20 mm.


where l0 is a strong temperature-dependent viscosity whichdecreases with increasing the working temperature [5], ul

the volume fraction of liquid phase, and um is a critical vol-ume fraction of solids above which the mixture has essen-tially infinite viscosity. Increasing the laser power at aconstant scan speed leads to a higher sintering temperatureand, accordingly, a lower l0. Also, an increase in the oper-ating temperature results in a larger amount of liquid for-mation, i.e., a higher ul. According to Eq. (1), both alower l0 and a higher ul lead to a lower mixture viscosityl, thereby facilitating a sufficient removal of the molten li-quid in conjunction with solid particles. Under this condi-tion, the liquid can successfully surround the solidparticles, hence increasing the liquid-solid wettability andthe amount of particle rearrangement. The clusters of rein-forcing particulates, therefore, undergo a significant deag-

glomeration (Fig. 3a and b), hence changing from severeaggregation to uniform dispersion in the finally solidifiedmetal matrix (Fig. 2). Meanwhile, the edges of irregularlyshaped particles are expected to dissolve in the wetting li-quid, leading to the smoothening of particulate surfaces(Fig. 4a and b).

On the other hand, since laser scanning is carried outline by line, the laser energy causes melting along a rowof powder particles, hence forming a continuous liquidscan track in a cylindrical shape. The liquid track is likelyto break up into a row of spheres in order to diminish thesurface tension (so called ‘‘balling’’ effect) [20]. Accordingto Agarwala et al.’s results [21], the balling effect is con-trolled by the mixture viscosity l. This viscosity shouldbe high enough to prevent balling during DMLS. However,using an even higher laser power at a fixed scan speed tends

Fig. 7. SE images showing dispersions of reinforcing particulates in fracture surfaces of laser sintered samples at different layer thicknesses: (a) 0.30 mm;(b) 0.25 mm and (c) 0.20 mm. Fixed processing parameters are laser powder 700 W and scan speed 0.06 m/s.


to induce an excessive amount of liquid formation, sincethere is an effect of remelting that may take place on previ-ously sintered layer when a higher energy input is used. Inthis instance, although the particle surfaces become smoothdue to the favorable solid phase wetting behavior (Fig. 4c),the mixture viscosity l decreases overmuch, hence causingthe significant balling. The reinforcing particulates tend toaggregate severely during balling process, resulting in a het-erogeneous distribution of the particulates in the matrix(Fig. 3c).

4.2. Influence of scan speed on particle-interface interaction

In general, the laser-induced melting of such a compos-ite system is followed by a non-unidirectional solidificationprocess which is characterized by the rapid growth of den-

drites [22]. This is because (i) a local destabilization of aninitially planar solid-liquid interface caused by the differentthermal conductivities and specific heats of the solidifyingmatrix and the particulate material is inherent to the solid-ification process of a composite system; (ii) the laser gener-ated turbulence in the melt pool will easily destroy such aplanar solidification interface. For the two matrix metals(Cu and Co) in the solidifying system, the weight fractionof Co is �4.1 wt.%. According to the Co–Cu phase dia-gram (Fig. 8a), the high-temperature phase (a-Co) precipi-tates primarily in a dendritic morphology duringsolidification (Fig. 8b). Afterwards, the Cu phase presentsand surrounds the primary phase (a-Co) by means of theperitectic reaction (Fig. 8b). However, due to the laserinduced rapid non-equilibrium solidification process, thesolute has insufficient time to transfer and penetrate the

Fig. 8. Binary phase diagram of Co–Cu system (a) [23] and schematic ofmicrostructural development during Co-Cu peritectic reaction (b) [24].


new-developed second-phase (Cu), hence handicapping afurther proceeding of peritectic reaction [24]. When theoperating temperature decreases below the peritectic tem-perature, the residual liquid phase crystallizes to form Cuphase.

The dendritic interface morphology during solidificationintroduces a new mode of particle-interface interaction inthe melt pool, i.e., the interaction between the particlesand the growing dendrites. It is known that during unidi-rectional solidification process, the particle-interface inter-action proceeds with a planar solidification front, since theparticles have only one degree of freedom [25]. In thisinstance, the particles are either pushed or engulfed bythe advancing solidification front, due to the negligible set-tling velocity of small particles [26]. It is obvious that thepushing results in pile-ups and local aggregation of partic-ulates at regions finally solidified, while the engulfingbrings about a uniform dispersion of the particulates inthe matrix. In the case of dendritic solidification, however,the mechanisms of particle-interface interaction change sig-nificantly, because the lateral migration of the particles intothe inter-dendritic regions becomes possible. Therefore, the

velocity of the dendrite tip and the lateral growth velocityof the dendrite trunk should be taken into account in theanalysis of the particle-interface interaction during den-dritic solidification.

According to Zhang et al.’s results [27], in the laserinduced solidification process, the local growth rate of den-drite tip V is given by the following equation:

V ¼ V b cos h ð2Þ

where Vb is the laser beam scan speed and h is the angle be-tween the vectors V and Vb. A close look at the Eq. (2) andthe obtained experimental results, as shown in Figs. 2, 5cand 6c, reveals that the scan speed plays a significant rolein determining the growth velocity of dendrites and, thus,the crystalline morphologies of dendrites and the attendantparticulate dispersion states. The mechanisms of the parti-cle-dendrite interaction at different scan speeds can be clas-sified into the following three categories, as schematicallyshown in Fig. 9.

(i) Pushing: at a low laser scan speed, i.e. a small den-dritic growth velocity, the reinforcing particulatesare rejected by a moving dendritic interface; that is,they are pushed ahead into the liquid, traveling alongwith the interface as it advances. These particles inev-itably impinge upon other particles in the liquid,resulting in pile-ups which extend a number of parti-cles in the direction normal to the dendritic interface(Fig. 6a). Consequently, the significantly coarsenedparticulate aggregates are segregated along grainboundaries after solidification (Figs. 5a, 9a).

(ii) Semi-engulfment: with increasing the scan speed,some reinforcing particulates start to be entrappedby the tip of the dendrite trunk and the first sidebranches near the trunk (Fig. 6b), because of a largerdegree of dendrite thickening and side branch coars-ening caused by a higher dendritic growth velocity.Such entrapped particulates can eventually get incor-porated in the metal matrix, while other particulatesthat are pushed by the advancing dendritic interfacedwell near the interior of grains after complete solid-ification (Figs. 5b, 9b).

(iii) Engulfment: as the scan speed increases above a crit-ical one of �0.04 m/s (Fig. 2), the reinforcing partic-ulates cease to be pushed by the advancing dendriticinterface, but are well incorporated in the growingdendrite trunk and the side branches (Fig. 6c),thereby leading to a uniform dispersion of the rein-forcing particulates in the solidified metal matrix oncooling (Figs. 5c, 9c).

Therefore, it is reasonable to conclude that a successivetransition from pushing to semi-engulfment and, subse-quently, to engulfment of the particulates by the dendriticsolidification front occurs with increasing the scan speed,leading to a change of the particulate dispersion states fromsevere aggregation to partial aggregation and, afterwards,

Fig. 9. Mechanisms of particle-dendrite interactions and the resultant dispersion states of reinforcing particulates: (a) pushing, aggregation; (b) semi-engulfment, partial aggregation and (c) engulfment, dispersed.


to uniform dispersion. For a given laser power between�650 W and �750 W, there exists a critical scan speed,0.04 m/s, above which the particulates are much easy tobe engulfed by the mobile dendritic interface (Figs. 2, 5cand 6c).

4.3. Influence of powder layer thickness on metallurgical

behavior of melt pool

In DMLS process, the incident laser energy causes aselective melting of the matrix metal, forming a melt poolcontaining both liquid (Co and Cu) and solid (WC) phases.Since a Gaussian laser beam is used in the sintering, a sig-

nificant temperature gradient tends to form between thecenter and edge of the melt pool, giving rise to the surfacetension gradient and the resultant Marangoni convection[28,29]. The formation of Marangoni convection inducescapillary forces for liquid flow, facilitating an efficient rear-rangement of the reinforcing particulates by the wettingliquid. Our previous work [30] reveals that in order toobtain a high sintered densification with a coherent inter-layer bonding ability, a suitable powder layer thickness(<0.40 mm), which is less than the thermal penetrationdepth of the laser beam, is required to allow the input laserenergy reaching the bottom of the powder layer. In thepresent study, although the used powder layer thicknesses


are generally less than 0.40 mm, i.e., within the thermalpenetration depth of the laser beam, the particulate disper-sion states in the sintered structures vary considerably withthe used layer thickness (Fig. 7). Here, it should be pointedout that although the shape and size of the melt pools cannot be observed directly in the solidified structures, thecharacteristic morphologies of the reinforcing particulates(Fig. 7) can well reflect the metallurgical behaviors of theparticulates in the different melt pools during sintering.

Since the delivered laser energy can well penetrate andmelt the powder layer, the depth of the formed melt poolincreases with increasing the powder layer thickness.Meanwhile, the energy density of the laser beam is con-stant, because of the same laser power and scan speed used.For a given energy input, an increase in the depth of themelt pool weakens the laser induced Marangoni convectionand the attendant liquid capillary forces, thereby slowingdown liquid flow and particle arrangement. Consequently,the reinforcing particulates are much likely to settle to thebottom of the melt pool under the action of the gravityforce, resulting in a severe aggregation and a local segrega-tion of the reinforcing particulates (Fig. 10). In otherwords, a proper decrease in the powder layer thickness

Fig. 10. Schematic for describing sintering behaviors of reinforcingparticulates in melt pools at different layer thicknesses.

facilitates an improvement in the thermodynamic andkinetic characteristics of both molten liquid and solid par-ticles in the melt pool during sintering, so as to obtain amore homogeneous dispersion of the reinforcing particu-lates in the matrix metal on cooling (Fig. 7).

5. Conclusions

We have performed a detailed investigation into theeffects of processing parameters on the particulate disper-sion in the WC–Cop/Cu MMCs prepared by DMLS, andthe following conclusions can be drawn:

(i) For a given scan speed larger than 0.04 m/s, a properincrease in the laser power between �650 W and�750 W can homogenize the dispersion of the rein-forcing particulates possessing a smoothened andrefined morphology, due to a sufficient liquid forma-tion and an improved wettability. However, an exces-sive increase in the laser power results in a severeaggregation of particulates, because of the ballingeffect caused by an excessive liquid formation witha too low viscosity.

(ii) A successive transition from pushing to semi-engulf-ment and, subsequently, to engulfment of the partic-ulates by the growing dendrites of the matrix metaloccurs with increasing the laser scan speed. Thereexists a critical scan speed of 0.04 m/s, above whichthe particulates can be well engulfed by the dendriticfront, leading to a homogeneous dispersion of thereinforcement in the matrix.

(iii) For a suitable laser energy input, a proper decrease inthe powder layer thickness to 0.20 mm can well alle-viate the particulate aggregation in the laser sinteredstructure, due to the elevated Marangoni convectionand liquid capillary forces.

Acknowledgements

The authors thank the financial supports from the Na-tional Natural Science Foundation of China (50775113,10276017), the Aeronautical Science Foundation of China(04H52061), and the Scientific Research Innovations Foun-dation of Nanjing University of Aeronautics and Astronau-tics (S0403-061). One of the authors (Dongdong Gu)gratefully appreciates the support from the GraduateSchool, Nanjing University of Aeronautics and Astro-nautics.

References

[1] Velasco F, Gordo E, Isabel R, Ruiz-Navas EM, Bautista A, TorralbaJM. Mechanical and wear behaviour of high-speed steels reinforcedwith TiCN particles. Int J Refract Met Hard Mater 2001;19:319–23.

[2] Gaard A, Krakhmalev P, Bergstrom J. Microstructural characteriza-tion and wear behavior of (Fe,Ni)-TiC MMC prepared by DMLS. JAlloys Compd 2006;421:166–71.


[3] Wong WLE, Karthik S, Gupta M. Development of high performanceMg–Al2O3 composites containing Al2O3 in submicron length scaleusing microwave assisted rapid sintering. Mater Sci Technol2005;21:1063–70.

[4] Breval E, Cheng JP, Agrawal DK, Gigl P, Dennis A, Roy R, et al.Comparison between microwave and conventional sintering of WC/Co composites. Mater Sci Eng A 2005;391:285–95.

[5] Gu DD, Shen YF. Effects of dispersion technique and componentratio on densification and microstructure of multi-component Cu-based metal powder in direct laser sintering. J Mater Process Technol2007;182:564–73.

[6] Gu DD, Shen YF. Influence of phosphorus element on direct lasersintering of multicomponent Cu-based metal powder. Metall MaterTrans B 2006;37:967–77.

[7] Pintsuk G, Brunings SE, Doring J-E, Linke J, Smid I, Xue L.Development of W/Cu-functionally graded materials. Fusion EngDes 2003;66–68:237–40.

[8] Wang Y, Bergstrom J, Burman C. Characterization of an iron-basedlaser sintered material. J Mater Process Technol 2006;172:77–87.

[9] Kruth JP, Kumar S, Van Vaerenbergh J. Study of laser-sinterabilityof ferro-based powders. Rapid Prototyping J 2005;11:287–92.

[10] Uzunsoy D, Chang ITH. The effect of infiltrant choice on themicrostructure and mechanical properties of Rapidsteel2.0. MaterLett 2005;59:2812–7.

[11] Gu DD, Shen YF, Yang JL, Wang Y. Effects of processingparameters on direct laser sintering of multicomponent Cu basedmetal powder. Mater Sci Technol 2006;22:1449–55.

[12] Xie JW, Fox P, O’Neill W, Sutcliffe CJ. Effect of direct laser re-melting processing parameters and scanning strategies on thedensification of tool steels. J Mater Process Technol 2005;170:516–23.

[13] Gu DD, Shen YF. WC–Co particulate reinforcing Cu matrixcomposites produced by direct laser sintering. Mater Lett2006;60:3664–8.

[14] Gu DD, Shen YF. Influence of reinforcement weight fraction onmicrostructure and properties of submicron WC–Cop/Cu bulkMMCs prepared by direct laser sintering. J Alloys Compd2007;431:112–20.

[15] Das S. Physical aspects of process control in selective laser sinteringof metals. Adv Eng Mater 2003;5:701–11.

[16] Simchi A, Pohl H. Direct laser sintering of iron-graphite powdermixture. Mater Sci Eng A 2004;383:191–200.

[17] Yang MC, Xu J, Hu ZQ. Synthesis of WC–TiC35–Co10 nanocom-posite powder by a novel method. Int J Refract Met Hard Mater2004;22:1–7.

[18] Fischer P, Romano V, Weber HP, Karapatis NP, Boillat E, GlardonR. Sintering of commercially pure titanium powder with a Nd: YAGlaser source. Acta Mater 2003;51:1651–62.

[19] Tolochko N, Mozzharov S, Laoui T, Froyen L. Selective lasersintering of single- and two-component metal powders. RapidPrototyping J 2003;9:68–78.

[20] Gu DD, Shen YF. Balling phenomena during direct laser sintering ofmulti-component Cu-based metal powder. J Alloys Compd2007;432:163–6.

[21] Agarwala M, Bourell D, Beaman J, Marcus H, Barlow J. Directselective laser sintering of metals. Rapid Prototyping J 1995;1:26–36.

[22] Wilde G, Perepezko JH. Experimental study of particle incorporationduring dendritic solidification. Mater Sci Eng A 2000;283:25–37.

[23] Yu JQ, Yi WZ, Chen BD, Chen HJ. Binary alloy phase diagrams. 1sted. Shanghai: Shanghai Sientific & Technical Publishers; 1987.

[24] Wang JX, Huang JR, Lin JS. Solidification of metals and its control.1st ed. Beijing: China Machine Press; 1983.

[25] Han Q, Hunt JD. Redistribution of particles during solidification.ISIJ Int 1995;35:693–9.

[26] Youssef YM, Dashwood RJ, Lee PD. Effect of clustering on particlepushing and solidification behaviour in TiB2 reinforced aluminiumPMMCs. Compos Part A 2005;36:747–63.

[27] Zhang K, Chen GN. Effect of SiC particles on crystal growth of Al–Sialloy during laser rapid solidification. Mater Sci Eng A 2000;292:229–31.

[28] Simchi A, Pohl H. Effects of laser sintering processing parameters onthe microstructure and densification of iron powder. Mater Sci Eng A2003;359:119–28.

[29] Gu DD, Shen YF. Microstructure of the WC–10%Co particulatereinforced Cu matrix composites fabricated by selective laser sinter-ing. Rare Metal Mater Eng 2006;35:276–9.

[30] Gu DD, Shen YF. Processing and microstructure of submicron WC–Co particulate reinforced Cu matrix composites prepared by directlaser sintering. Mater Sci Eng A 2006;435–436:54–61.

inﬂuence of processing parameters on particulate dispersion in...

Documents