Characterization of Zinc Powder

Compactions: Factors Affecting

Mechanical Properties and Analytical

Powder Metallurgy Models

Ben J. RaelMechanical Engineering Department,

MSC01-1150 University of New Mexico,

Albuquerque , NM 87131

Casey L. DyckMechanical Engineering and Sciences Department,

University of Illinois at Urbana-Champaign,

Champaign, IL 61820

Tariq A. KhraishiMechanical Engineering Department,

MSC01-1150 University of New Mexico,

Albuquerque, NM 87131

Mehran Tehrani

Marwan S. Al Haik

Virginia Tech. Engineering Science and Mechanics,

MC 0219 223 Norris Hall,

Blacksburg, VA 24061

In this study, the effectiveness of analytical models which attemptto predict the density of unsintered powder metallurgy (PM) com-pacts as a function of consolidation pressure is investigated.These models do not incorporate the nonuniform densification ofpowder compacts and may insufficiently describe the pressure/densification process. Fabrication of uniform and nonuniformZinc (Zn) tablets is conducted to assess the validity of the pres-sure/density model developed by Quadrini et al. (Quadrini andSqueo, 2008, “Density Measurement of Powder MetallurgyCompacts by Means of Small Indentation,” J. Manuf. Sci. Eng.,130(3), pp. 0345031–0345034). Different tablet properties wereobtained by varying the compaction pressure and fabricationprotocol. Density gradients within Zn tablets result in a spatialdependence of Vickers microhardness (HV) throughout the fabri-cated specimen. As a result, micro-indentation testing is usedextensively in this study as a characterization tool to evaluate thedegree of nonuniformity in fabricated Zn tablets. Scanning elec-tron microscopy (SEM) is also employed to verify tablet densityby visual examination of surface porosity as compaction pressureis varied and sintering is applied. [DOI: 10.1115/1.4005404]

Introduction

Powder metallurgy is a very attractive forming and fabricationtechnique due to its relatively low costs and the ability to formcomplex near net shapes. The key to fabricating successful PMcomponents are selection of the right material appropriate for adesired application and processing it under optimal processing

conditions. Key processing variables include consolidation pres-sure, green density, and sintering conditions. Extensive work hasbeen carried out by researchers to characterize consolidationpressure to achieve a certain green density of PM components.However, theoretical models, which relate the consolidation pres-sure to achieved green density, assume spatially uniform densifi-cation throughout the powder compact. For these reasons, a morein-depth look, as achieved in this paper, is needed into the effec-tiveness of theoretical models which attempt density predictionsof statistical materials such as PM compacts.

A plethora of studies have been carried out to understand thedensification mechanism in the PM process as well as to deter-mine the mechanical properties of the compacts. Researchers haveprimarily focused on density measurements and calculations.Quadrini et al. used small flat-surface indentations and an analyti-cal model to try and predict the density of powder metal compacts[1]. Ludwig et al. proposed the use of electrical conductivitymeasurements as a means to measure the density of green statemetal compacts [2]. Dawson et al. used ultrasonic measurementsto track density changes during compaction [3]. While manyfocused on measuring the density of the compactions, Fleck et al.showed that the yield stress, ultimate tensile stress, resistance tofatigue crack propagation, and fracture toughness of a compactedmetal powder is directly related to the compaction density [4].Carnavas et al. explored the effects that particle shape and sizehave on the physical properties of PM compacts [5]. All the aboveproposed methods may be effective in finding the overall densityof the compacts but fail to address the issue that a completely uni-form compaction is nearly impossible to achieve.

The primary focus of this paper is to show that models whichdescribe the pressure/densification process, as in Ref. [1], aresensitive to fabrication procedures. If great care is not taken tohomogenize PM components to the furthest extent, possible poorcorrelation between data and theory can result. One method pro-posed by Quadrini et al. [1] to obtain compacted tablets with higherdegrees of homogeneity (by first compacting one side, flipping thetablet over, and then compacting the other side) is used in this study.The sensitivity of the Quadrini model to this uniforming procedure(flipping of the tablets) is characterized through hardness mappingand a least-squares analysis. Finally, visual confirmation of the den-sity and porosity is achieved with the use of a SEM to illuminate theeffects of compaction pressure magnitude and sintering.

Methods

Metal tablets were fabricated by “cold” compaction of commer-cially available zinc powder (97.5% pure, median 6–9 lm,Alfa-Aesar, Inc., Ward Hill, MA), Fig. 1 (inset). Particle size

Fig. 1 Particle size distribution for zinc powder (median6–9 lm) used in this study, (inset) SEM micrograph of zincpowder

Contributed by the Materials Division of ASME for publication in the JOURNAL OF

ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received April 6, 2010; finalmanuscript received October 26, 2011; published online August 24, 2012. Assoc.Editor: David Bahr.

Journal of Engineering Materials and Technology OCTOBER 2012, Vol. 134 / 044502-1Copyright VC 2012 by ASME

Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 09/29/2013 Terms of Use: http://asme.org/terms

distribution of the Zn powder was determined using a JEOL5800LV scanning electron microscope, where the diameters of800 Zn particles were measured (Fig. 1).

The die and discs used in the compaction process wereconstructed of high-strength (HS) steel and produced cylindricaltablets with diameters of approximately 12.72 mm. A half-inch(12.7 mm) nominal diameter steel plunger was used to transfer thecompaction force “F” from an Instron

VR

4400 R tensile/compres-sion testing machine to the upper disc and onto the powder duringthe compaction process (Fig. 2).

Approximately, 5 g of Zn powder was weighted to manufactureeach tablet. Before consolidation of Zn tablets could commence, itwas necessary to apply zinc stearate [6] to the inner die wall anddisks. This is done to reduce wear and friction between movingcomponents in Fig. 2. In order to adequately lubricate the innerdie surface, it is first swabbed with acetone, a mixture of shreddedcork and zinc stearate is then consolidated at a pressure of50 MPa, after which the mixture is ejected from the die. Afterthis lubrication protocol is carefully implemented, Zn powderis carefully loaded into the die with the compaction pressurevarying from 150 MPa to 450 MPa in 50 MPa increments. Theadvancement rate of the Instron machine was held constant at0.254 mm/min until the desired compaction pressure was achievedat which point the unloading phase would commence at0.254 mm/min. This advancement rate was chosen as such tofacilitate particle rearrangement/movement and to help with mini-mizing density gradients in the green compact. The tablets werethen ejected from the die, the density of the cylindrical tablets was

then determined using a micrometer and electronic weight bal-ance. These tablets were flipped and then reloaded into the diewith the same lubrication and compaction protocol performed asoutlined above. Upon completion of the second compaction cycle,the densities of the cylindrical tablets were determined. Multiplesamples at each compaction pressure were fabricated in order toensure accuracy and repeatability.

Half of the samples from each compaction pressure were sin-tered at 378 �C for 45 min with 60 standard cubic centimeters perminute (sccm) of nitrogen flowing to provide an inert environmentfor sintering. This is done to see the effect of sintering on theporosity of compacts compared to their green state.

Characterizing the material properties and microstructure of thefabricated tablets consisted of Vickers microhardness tests andSEM micrographs, in addition to the density measurements men-tioned above. The microhardness tests were performed using aBuehler Micromet II microhardness tester with 50–300 g loadswhich were held constant for 10 s. Hardness tests were performedon both green and sintered tablets, on both sides of the tablets, andacross tablet diameters (indents spaced 400 lm apart). Hardnesstests were also performed along the depth of the fabricated tablets,before and after the second compaction cycle (indents spaced500 lm apart). SEM micrographs (Fig. 3) of the Vickers indentsconfirm that no surface preparation of green tablets is necessarysince the indents are clearly visible and symmetric. If the surfacewas not completely flat, then the shape of the indents would notbe a perfect diamond. It should also be noted that the indent inFig. 3 encompasses a large number of Zn particles. This is indica-tive of adequate indentation loads used in this study. To performindentations along the specimen depth, a HARIG 618 Automaticprecision grinder was utilized to create flat surfaces. The hardnessvalues used in this study are the mean values from equally placedindentations on multiple samples.

Live images were taken using a JEOL 5800LV scanning elec-tron microscope which provided insight into surface porosity andmorphology of zinc tablets consolidated at different pressures.SEM micrographs also give valuable insight into the increasedbonding between zinc particles that is a result of the sinteringprocess.

An analytical model for density prediction was proposed byQuadrini et al. [1] which relate the relative density “R” and com-paction pressure “p” by

p ¼ rs

g�2� R2� �

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiR� R0

1� R0

� �n=R�

1� R2ð Þ 4� R4ð Þ

vuuut(1)

To assess the density prediction capability of Eq. (1), a MATLABVR

nonlinear least-squares fitting routine was applied to the experi-mental data set [Runiform, p(Runiform)] to fit the Quadrini model(determine the g* and n parameters). The parameter, rs, is consid-ered deterministic and has the value of 124 MPa for zinc [1]. Thisoptimization procedure minimizes the sum of the squares of theresiduals [7]

Sr ¼Xm

i¼1

pi;measured � pi;model

� �2(2)

where “m” is the total number of points. The MATLABVR

“lsqcurvefit” routine (algorithm: “large-scale, trust-region reflec-tive Newton”) performed repetitive fittings until the sum of theresiduals squared (Sr)uniform is minimized. This provides usthen with the correlation coefficient (r2)uniform. With the parame-ters g and n determined, Sr and r2 are then determined for non-uniform tablets (single-sided compactions) using the [Rnonuniform,p(Rnonuniform)] data set. Correlation coefficients r2 and Sr valuesfor uniform and nonuniform tablets are then compared.

Fig. 2 Experimental setup

Fig. 3 Scanning electron micrograph of Vickers indent of sam-ple consolidated at 300 MPa

044502-2 / Vol. 134, OCTOBER 2012 Transactions of the ASME

Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 09/29/2013 Terms of Use: http://asme.org/terms

Results and Discussion

In the fabrication of each tablet, a loading and unloading curvewas created for both the first and second compaction cycles.Figure 4 shows the typical loading and unloading curves for thefirst and second compaction cycles.

During fabrication of the tablets, it can be seen from Fig. 4 thatflipping the tablet over after the initial compaction cycle doesallow for further consolidation during the second loading andunloading phase. The idea set forth by Quadrini et al. was that thisprocedure would create tablets with a more uniform density [1].However, Ref. [1] did not demonstrate this idea as herein usingexperimental load- and unload-displacement curves. The large netdisplacement of the first loading and unloading curves in Fig. 4shows majority of the consolidation occurring during this part ofthe procedure. The small net displacement seen in the secondloading and unloading cycle shows that flipping the tablet overdoes indeed allow for further consolidation, though not nearly asmuch as the first loading and unloading cycle. Furthermore, thenearly collapsed loading and unloading curves for the secondcycle in Fig. 4 is indication that the specimens were compacted tothe furthest extent possible.

Further indication that Zn tablets are attaining a higher degreeof uniformity due to the fabrication procedure adopted in this

study is exemplified in Fig. 5. For the uniform tablet, HV as afunction of depth remains constant throughout the depth of thetablet. For the nonuniform tablet, HV is highest on the compactedsurface and overall is less than for the uniform tablet. This can beexplained by deformation strengthening due to high compressivestresses at the upper disc/tablet interface (compacted surface). Forthe lower disc/tablet interface (noncompacted surface), compres-sive stresses due to the plunger are lower and grain size reductiondrops. As a result, HV values are lowest for the noncompactedsurface. This can be seen in Fig. 5 where the double-compactedtablets were more densified throughout than the single-compactedones resulting in a harder material response, i.e., less overallporosity. Note also in Fig. 5 how at one depth point for the non-uniform tablets the HV values rapidly decline toward the noncom-pacted side. This indicates a loss of compaction efficiency thereand should be guarded against in the design of PM-manufacturedparts intended to be load-bearing in service.

To further affirm that the fabrication technique adopted fromQuadrini et al. [1] did indeed create a more uniformly dense tab-let (i.e., with less porosity), Vickers microhardness tests wereperformed radially on both sides of uniform and nonuniformtablets. Figure 6 indicates higher overall HV values for double-compacted tablets compared to single-compacted ones. Further-more, the single-compacted (nonuniform) tablets see an increasein HV near the tablet edges on the compacted side, and adecrease in HV near tablet edges on the noncompacted side. Thereason for this is due to machining tolerances between the twodiscs and the inner die cavity. Because of this tolerance, particlesduring pressurization are squeezed out the tolerance of the sta-tionary disc forming a severely deformed lip (an elephant footshape) due to frictional and shearing effects. This is at theexpense of losing particles from right below that region, i.e., adeficiency of particles at the other tablet-disc interface. Thisexplains the disparate HV behavior of uniform and nonuniformtablets. These results for microhardness as a function of tablet ra-dial distance agree in trend with results reported and explainedelsewhere [8].

Fig. 5 Microhardness as a function of specimen/tablet depth.Indent load 5 300 g. Each data point represents the average ofthree specimens tested. Eight grams of Zn powder was used foreach tablet.

Fig. 6 Hardness as a function of radial distance. Tablets con-solidated at 400 MPa in the green state, 300 g load. Two gramsZinc powder was used for each tablet here.

Table 1 Least-squares curve fitting of the parameters used inminimizing Eq. (2)

rs (MPa) g* qf (g/cm3) Ro n

124 0.7141 7.133 0.477 0.971

Fig. 4 Superposed loading and unloading curves for a speci-men consolidated at 450 MPa. Two grams of Zn powder wasused for the tablet here.

Journal of Engineering Materials and Technology OCTOBER 2012, Vol. 134 / 044502-3

Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 09/29/2013 Terms of Use: http://asme.org/terms

The fitting parameters n and g* from minimizing Eq. (2) arelisted in Table 1. Figure 7 shows optimized Eq. (1) superposedon experimental data. It is seen in this figure that computed R ofboth uniform and nonuniform tablets exhibit a greater sensitivityto compaction pressure in the lower pressure range. Also, asthe compaction pressure increases, the rate of increase of densifi-cation falls off. At later stages of the compaction process Znparticles are work hardened, which consequently impedes densifi-cation [9]. This work hardening is related to dislocation densitysaturation [10].

Although Eq. (1), the Quadrini model, captures the trend of theexperimental data it does not do as good a job capturing the val-ues. There are multiple reasons for why this may have happened.First, the Quadrini model does not take into account particlerearrangement, rotation, and sliding [11]. This is active especiallyduring the early phases of compaction, i.e., for the lower pressureregion. Later in the compaction process, i.e., at higher pressurevalues, particle sliding, rearrangement, and rotation become lesslikely with plastic deformation being the responsible agent fordensification. In fact, the Quadrini model, and similar such models[9,10,12–18], only take into account the densification of powdercompacts as a result of plastic deformation and nothing else. Sec-ond, rs used in Eq. (1) was considered a deterministic parameterwith a value (provided in Table 1) that was assumed constant with

deformation. However, according to Tszeng et al. [19] it wasshown that the apparent yield stress of a powder compactincreases with particle morphological index “M” (M¼ 0 for spher-ical particles, M¼ 1 for highly irregular particles). What thismeans is that rs is actually a function of M. Furthermore, M is afunction of compaction pressure p as the more the compaction

Fig. 7 Numerical optimization of Eq. (1) superposed on experi-mental data. Each data point represents the average of threespecimens tested. Five grams of Zn powder was used for eachtablet.

Fig. 8 Hardness comparison for green and sintered tablets. Allhardness indents were performed close to the center of the tab-let using a 50 g load. Each data point represents the average offive hardness measurements. Two grams of Zinc powder wasused for each tablet here.

Fig. 9 SEM image of zinc powder compacted at 200 MPa ingreen state

Fig. 10 SEM image of zinc powder compacted at 400 MPa ingreen state

Fig. 11 SEM image of zinc powder compacted at 400 MPa aftersintering at 378 �C for 45 min with 60 sccm nitrogen flow

044502-4 / Vol. 134, OCTOBER 2012 Transactions of the ASME

Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 09/29/2013 Terms of Use: http://asme.org/terms

the more M increases, especially starting out with spherical par-ticles as in our case (Fig. 1, inset). Furthermore, for this fit(r2)uniform¼ 0.9757, which indicates a goodness of fit between theQuadrini model and [Runiform, p(Runiform)] data. However, for(r2)nonuniform¼ 0.8543, the Quadrini model does not describewell Zn tablets that have not undergone a homogenization/uniforming procedure. This is further evident in the higher valuefor (Sr)nonuniform. This clearly shows that applying models such asthe model in Eq. (1) to double-sided compactions would givefitting parameters for the model that are not suitable for single-sided compactions of the same powder. In other words, the fittingparameters in models such as the Quadrini model are very sensi-tive to the degree of densification of produced PM parts.

Microhardness results for the sintered specimens are shown inFig. 8. The hardness of the sintered samples was approximately12–14% greater than that of the green tablets fabricated underthe same compaction pressure. Solid-state diffusion bonding,facilitated by sintering, between adjacent particles at or near themelting point of zinc [20] accounts for this increase in hardness.Figure 8 also indicates that sintering plays a more meaningful rolein the hardening of the compact under higher pressures. Ascompaction pressure increases, the total contact area between Zngranules increases, which increases the amount of diffusing regionavailable.

SEM images (Figs. 9–11) show the physical organization of thezinc particles after compaction and sintering. The higher densityand lower porosity, due to the higher compaction pressures, canbe clearly seen when comparing Figs. 9 and 10. The increasedbonding between zinc particles due to the sintering process canbe visually observed from the SEM micrographs presented inFigs. 10 and 11.

Conclusions

The Quadrini model (Eq. (1)) developed for uniformly densetablets is not an accurate description of the pressure–density rela-tionship for Zn tablets that have not undergone a density-uniforming procedure. Furthermore, the Quadrini and such models[9,10,12–18] are based on densification stemming from plastic de-formation only and thus do not take into account factors affectingthe density such as particle sliding, movement, re-arrangement[11]. Also, such models [9,10,12–18] do not take into account theeffect of particle morphological changes on the apparent yieldstress of the powder material [19].

To improve the density prediction capabilities of Eq. (1), theauthors suggest including particle morphology in the constitutivemodel which this analytical model was derived from. This maybe accomplished by introducing an appropriate state variableK¼K(M) into the modified yield criteria from which Eq. (1)is based off, where a morphology index, M 2 (0,1), is used todescribe particle morphology.

Another lesson learned from this study is that sintering is moreeffective on mechanical properties at higher compaction pressurevalues compared to lower values. In addition, this study clearlyshows the positive effect that double-sided compactions have onuniforming the material hardness throughout the PM product(both radially and axially) and increasing its value.

Nomenclature

F ¼ compaction forcero ¼ tablet radiusr ¼ radial distance along tablet

H ¼ tablet heightz ¼ distance along height of tablet

p, p(Runiform), p(Rnonuniform) ¼ compaction pressure for uniformand nonuniform zinc tablets

R, Runiform, Rnonuniform, ¼ relative density (compactedpowder density/max theoreticaldensity qf) for uniform andnonuniform zinc tablets

R0 ¼ relative density of zinc tablets atzero compaction

Sr, (Sr)uniform, (Sr)nonuniform ¼ sum of the squared residuals foruniform and nonuniform test data

pi-measured, pi-model ¼ measured and calculated pressuresfor the ith specimen

r2,(r2)uniform, (r2)nonuniform ¼ correlation coefficients for uniformand nonuniform test data

Greek Symbols

rs ¼ maximum yielding stress of thecompacted powder (at R¼ 1)

g* ¼ efficiency coefficient of a compac-tion test due to the die friction

n ¼ hardening coefficient of the metalpowder

References[1] Quadrini, F., and Squeo, E., 2008, “Density Measurement of Powder Metal-

lurgy Compacts by Means of Small Indentation,” J. Manuf. Sci. Eng., 130(3),pp. 0345031–0345034.

[2] Ludwig, R., Apelian, D., and Leuenberger, G., 2005, “An NDE Methodology toPredict Density in Green-State Powder Metallurgy Compacts,” J. Nondestruct.Eval., 24(3), pp. 109–119.

[3] Dawson, A. L., and Bussiere, J. F., 1998, “Ultrasonic Characterization of IronPowder Metallurgy Compacts During and After Compaction,” Adv. Perform.Mater., 5, pp. 97–115.

[4] Fleck, N., and Smith, R., 1981, “Effect of Density on Tensile Strength, FractureToughness, and Fatigue Crack Propagation Behavior of Sintered Steel,” PowderMetall., 24(3), pp. 121–125.

[5] Carnavas, P., and Page, N., 1998, “Elastic Properties of Compacted MetalPowders,” J. Mater. Sci., 33, pp. 4647–4655.

[6] Kumar, P., and Kumaran, S., 2010, “Comparison Study of Fly Ash ReinforcedAA6061 Composites Using Press Sinter Extrusion and Press Extrusion,”Powder Metall., 53, pp 163–168.

[7] Chapra, S., and Canale, R., 1998, Numerical Methods for Engineers, 3rd ed.,pp. 438–443, WCB/McGraw-Hill, Boston.

[8] Kandeil, A., Malherbe, M. C., Critchley, S., and Dokainish, M., 1977, “The Useof Hardness in the Study of Compaction Behavior and Die Loading,” PowderTechnol., 17, pp. 253–257.

[9] Al-Qureshi, H. A., Soares, M. R., Hotza, D., Alves, M. C., and Klein, A. N.,2008, “Analyses of the Fundamental Parameters of Cold Die Compaction ofPowder Metallurgy,” J. Mater. Process. Technol., 199, pp. 417–424.

[10] Kim, H., Estrin, Y., Gutmanas, E. Y., and Rhee, C., 2001, “A ConstitutiveModel for Densification of Metal Compacts: The Case of Copper,” Mater. Sci.Eng., A307, pp. 67–73.

[11] Martin, C. L., Bouvard, D., and Shima, S., 2003, “Study of Particle Rearrange-ment During Powder Compaction by the Discrete Element Method,” J. Mech.Phys. Solids, 51, pp. 667–693.

[12] Kim, H. S., Won, C. W., and Chun, B. S., 1998, “Plastic Deformation of PorousMetal With an Initial Inhomogeneous Density Distribution,” J. Mater. Process.Technol., 74, pp. 213–217.

[13] Lee, D., and Kim, H., 1992, “Plastic Yield Behavior of Porous Metals,” PowderMetall., 35(4), pp. 275–279.

[14] Yoon, S. C., Kwak, E. J., Kim, T., Cheon, B. S., and Kim, H. S., 2008, “Yieldand Densification Behavior of Rapidly Solidified Magnesium Powders,” Mater.Trans., 49(5), pp. 967–971.

[15] Alves, M. M., Martins, A. F., and Rodrigues, M. C., 2006, “A New Yield Func-tion for Porous Materials,” J. Mater. Process. Technol., 179, pp. 36–43.

[16] Narayanasamy, R., Ponalagusamy, R., and Subramanian, K. R., 2001,“Generalized Yield Criteria of Porous Sintered Powder Metallurgy Metals,”J. Mater. Process. Technol., 110, pp. 182–185.

[17] Ponalagusamy, R., Narayanasamy, R., and Subramanian, K. R., 2005,“Prediction of Limit Strains in Sheet Metals by Using New Generalized YieldCriteria,” Mater. Des., 28, pp. 913–920.

[18] Narayan, S., and Rajeshkannan, A., 2011, “Densification Behavior in Formingof Sintered Iron–0.35% Carbon Powder Metallurgy Preform During ColdUpsetting,” Mater. Des., 32, pp. 1006–1013.

[19] Tszeng, C. T., 1998, “Considering Particle Morphology in a ConstitutiveModel for Metal Powders Compaction,” Metall. Mater. Trans. A, 30, pp.1159–1162.

[20] Kalpakjian, S., and Schmid, S., 2006, Manufacturing Engineering and Technol-ogy, 5th ed., Pearson–Prentice-Hall, Upper Saddle River, NJ, pp. 483–512.

Journal of Engineering Materials and Technology OCTOBER 2012, Vol. 134 / 044502-5

Downloaded From: http://materialstechnology.asmedigitalcollection.asme.org/ on 09/29/2013 Terms of Use: http://asme.org/terms