Properties of closed-cell nickel-graphite composite

Properties of Closed-Cell Nickel-Graphite Composite

AI LIU, GERHARD E. WELSCH, ROBERT L. MULLEN, and DOV HAZONY

Nickel/graphite cells consisting of graphite cores and nickel cell walls were formed into formlightweight cell composite samples by hot-pressing in reducing atmosphere. In the composite sam-ples, the cells walls were diffusion-bonded to form a continuous metal-matrix. The composite densityranged from 2 to 4 g/cm3. Mechanical properties of the composite were measured in bending,compression and by ultrasound wave propagation, and were correlated with the properties of theconstituent elements. Tensile strength derived from bending tests, compression strength, ductility intension and compression, Young’s modulus, velocity of sound, and damping coefficient were meas-ured on samples with a density of 3.35 g/cm3. Young’s modulus values ranged from 4 to 35 GPadepending on the measurement methods. The flow stress in compression ranged from 35 to 120 MPa.Plastic deformations of over 15 pct were obtained in compression testing before shear failureoccurred. The composite has high damping capacity. Attenuation measurements of transmitted ultra-sound pulses yielded a value for the attenuation constant of 0.34 mm�1, more than five times than thatof gray iron.

I. INTRODUCTION

NATURAL or manmade cellular materials are oftenformed from hollow or fluid-filled cells, and tend to belightweight compared to solid monolithic materials. Theirstructure endeavors to give them favorable combinations ofphysical (e.g., density) and mechanical properties (e.g.,flexibility). The material can be used for structural or func-tional purposes.[1] Structurally, a cell composite is made ofa network interconnected of solid struts or plates, whichform the edges and faces of cells.[2] The present work dealswith a metal matrix cell composite of nickel/graphite cells.The cell cores consist of graphite, and the cell walls consistof nickel, i.e., they are filled closed cells. When consoli-dated by hot pressing in reducing atmosphere, a cell com-posite is obtained, which has a continuous matrix of nickel.Volumetrically, nickel is the minority phase, and graphite isthe majority phase. The nickel/graphite composite of thisstudy differs in significant ways from other cell materialssuch as honeycomb, open-cell foam, or closed-hollow cellstructures. The latter cell structures often contain hollow orgas-filled pockets, which makes for low density. However,they have low compression strength because of bending andbuckling of the cell walls. The nickel/graphite cells, on theother hand, use graphite as a pressure-resistant cell interiorfor support against buckling or bending of thin metal cellwalls. The graphite thus has a structural or functional pur-pose.

Nickel/graphite cells have the following features: Eachcell is a structural entity of graphite that is fully envelopedby a layer of nickel. When such cells are placed into a formand compacted by hot pressing, diffusion bonding takesplace between the cell walls. Both the cell walls and the

graphite cores are plastically formable. During compres-sion, they can flow and fill the interstitial spaces so thatdensification occurs. The consolidated composite consistsof an interconnected metal matrix with embedded graphiteparticles. Depending on the degree of densification, a cer-tain small interstitial volume fraction may remain in thecomposite.Research by others on foams and nonfilled cellular solids

have shown three types of compressive stress-strain curves,labeled ‘‘elastomeric,’’ ‘‘elastic-plastic,’’ and ‘‘elastic-brittle,’’and observations by optical microscopy and scanning elec-tron microscopy have shown their deformation to bestrongly controlled by bending and compressive bucklingof cell walls.[3] Hollow metal sphere structures were shownto have mechanical properties close to those of open cellfoams.[4] Their stress-strain response in compression can bedivided into three stages. At low compression stress, theyexhibit elastic bending of the cell walls. The thin struts orcell walls bend elastically and provide only a low strength.The end of useful elastic deformation is reached when thestruts or cell walls begin to bend plastically. The furthercompression proceeds by plastic bending, buckling, andcollapse of the cell wall. This plastic deformation stageexhibits a stress plateau with little strain hardening, whilethe empty volume is steadily diminished. Stage 3 occursafter severe collapse and results in densification and a rapidincrease in resistance against further compression.[1,2]

Composites of solid- or liquid-filled cells are expected tobe stronger in compression and shear. The mechanical behav-ior of such composites was evaluated by Ozgur et al.,[5,6]

using predictive finite element modeling. The model com-posites consisted of filled closed cells, with metal walls thatconfine a fluidlike substance in their interiors. Graphite canbe considered to be fluidlike, because it can plastically flowin shear. Composites made from cylindrical cells withsquare, hexagonal, and circular cross sections were eval-uated. The predicted compression stress-strain curvesshowed the beneficial effect of load-bearing substances inthe cell interiors. They were compared with the deforma-tion of nonfilled and with gas-pressure-filled cell structures.The calculations showed that the filled cell structures had

AI LIU, Project Design Engineer, is with Bemdorf Belt Systems, Elgin,IL, 60123. Contact e-mail: [email protected]. GERHARD E.WELSCH, Professor, Department of Materials Science and Engineering,ROBERT L. MULLEN, Professor and Chairman, Department of CivilEngineering, and DOV HAZONY, Professor (Emeritus), Department ofElectrical Engineering, are with Case Western Reserve University,Cleveland, OH 44106.

Manuscript submitted September 14, 2005.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 37A, SEPTEMBER 2006—2849

JOBNAME: MTA 37A#9 2006 PAGE: 1 OUTPUT: Thursday August 10 09:14:04 2006

tms/MTA/123238/ETP05446A12

significantly increased stiffness and delayed the onset ofplastic yielding. The filling of cells enables the cell compo-sites to better resist buckling and crushing of the cell walls.

The cell filler material, graphite, has a relatively lowshear modulus value and high bulk modulus value. A lowshear flow stress enables it to flow plastically. When thegraphite-filled cells are formed into a dense composite, thecomposite is expected to sustain a significant compressivestress before plastic shearing causes it to yield. Since graph-ite has a low density (1.7 to 2.2 g/cm3), the composite isexpected to also have a relatively low density. Furthermore,based on the internal friction research data of Weller andWert,[7] the graphite is expected to impart a high energydissipation capacity to the composite.

Microstructure: A schematic structure of a noncompactedspherical cell composite is shown in Figure 1. The graph isidealized. In reality, the cells have an irregular shape, asshown in the micrograph in Figure 2.

The objectives of the present research are to evaluate themicrostructure of a compacted nickel/graphite closed-cellcomposite and to evaluate the physical and mechanical prop-erties of density, Young’s modulus, velocity of sound, energydissipation, yield strength, and ductility and to correlate themwith the properties of nickel, graphite, and pore constituents.

II. EXPERIMENTAL PROCEDURE

A. Material and Specimen Preparation

The fabrication of the composite material involved twoprocessing steps: (1) production of cells and (2) forming ofthe composite by hot pressing. The cell particles wereobtained from INCO-NOVAMET Inc. (INCO-Clydach,Swansea, South Wales, U.K.) They had been manufacturedby chemical vapor deposition of nickel on graphite particlesin a fluidized bed by the reaction[8]

NiðCOÞ4 1Graphite ! Ni coated Graphite1CO[

The coated graphite powder had a size distribution rangingfrom 50 to 200 mm. The nickel weight fraction can bevaried from 25 to 85 wt pct. The powder for the presentwork had a nickel weight fraction of 60 wt pct. Metallo-graphically prepared cross sections of nickel-coated powderparticles or ‘‘nickel/graphite cells’’ are shown in Figure 2.

Consolidation: Nickel/graphite cell particles were pouredinto cylindrical graphite dies and then uniaxially compressedin vacuum between 900 °C and 1100 °C at pressures rangingfrom 5 to 20 MPa.[9] The material for the current study wascompacted at 1100 °C and 20 MPa for about 30 minutes.During compaction, the nickel cell walls bonded by diffu-sion. Bulk cylindrical specimens with 38.5-mm diameterwere made by this method. Specimens for mechanical andultrasound propagation tests were cut from a cylinder pieceof 38.5-mm diameter and 28.1-mm height. From the volumeand the weight of the cylinder, a density of 3.35 g/cm3 wasdetermined. The finished composite appeared as a relativelydense bulk material of a metallic gray color.

B. Structure and Density of Nickel-Graphite CellComposite Specimens

The hot-pressed composite cylinder was sliced into discsand rectangular pieces; then, it was mechanically polished

with sandpaper. Micrographs of the cell particles (Figure 2)and the compacted sample were taken. The microstructureof the composite cross section is shown in Figure 3. Themicrograph is from a polished cross section taken with anoptical microscope. It shows that the cell size ranges from50 to 150 mm. The thickness of the nickel cell walls rangesfrom less than 5 to over 20 mm. The composite contains asmall volume fraction of pores. Due to the fact that theoriginal nickel/graphite cells had irregular shapes, the poresize ranges from 10 to 50 mm. Nickel forms a continuousmatrix that envelopes the graphite particles. The porosity isestimated from area fraction measurements to have a vol-ume fraction of less than 5 pct.

The density values of nickel and graphite are given inTable I. The densities of Ni-G cell composites of variouspore volume fractions are shown for comparison.

Fig. 1—Schematic structure of an aggregate of spherical nickel/graphitecells before it is compacted into a composite.

Fig. 2—Optical micrograph of cross sections through Ni-coated graphiteparticles. They are filled closed cells with graphite forming the cell interiorand nickel forming the cell wall. Preparation: Particles were embedded inepoxy mount, and then polished with the help of Struers MetallographicalLab (Westlake, OH).

2850—VOLUME 37A, SEPTEMBER 2006 METALLURGICAL AND MATERIALS TRANSACTIONS A


tms/MTA/123238/ETP05446A12

Volume fraction estimations can be made based on theknowledge that the starting powder had 60 wt pct nickel.Assuming the ratio of graphite and nickel was unchangedduring compression, one obtains the volume fractions ofnickel and graphite of a fully dense composite.

Use rG 5 1.8 (g/cm3) and rNi 5 8.9 (g/cm3) for thecalculation of volume fraction,

VG ¼ WG=rGWG=rG 1WNi=rNi

[1]

and VG 5 0.767 and VNi 5 0.233 for fully dense composite,

Density5 0:767 3 1:81 0:233 3 8:95 3:453 ðg=cm3Þ [2]The measured density of the composite was 3.35 g/cm3,which is 97.1 pct of the fully dense composite. Porosityaccounts for the balance.

The volume fractions of the components are shown inTable II.

C. Mechanical Testing

Bending, compression, and hardness indentation testswere performed on the composite.

1. Bend testsIn the three-point bend test shown in Figure 4, a rectan-

gular beam-shaped specimen was loaded in the center whilethe ends were supported by two pins. The test length is L.The specimens were rectangular beams cut from a hot-pressed cylinder. An ISOMET2000 wafering machine was

used. Two test samples had dimensions of 3.2 3 4.0 3 35.0mm and 3.0 3 4.1 3 35.0 mm, respectively. The distance Lbetween the supporting spans was 20.3 mm. A photo of aspecimen is shown in Figure 4. During elastic bending, themodulus of elasticity was obtained from load-deflectionmeasurements. The testing procedure of ASTM Standard3.01 was followed.[10]

From the load-deflection curve, the Young’s modulus andthe maximum bending strength can be calculated using anelastic beam bending formula:

E 5PL3

4bh3d5

L3

4bh3P

d[3]

smax ¼ 3PmaxL

2h2b[4]

where E5 modulus of elasticity in bending, L5 span lengthbetween supports, b 5 specimen width, h 5 specimen

Fig. 3—Hot-pressed nickel/graphite powders form a composite with acontinuous nickel matrix. The white regions are nickel, gray regions aregraphite, and black regions are porosity.

Table I. Density of Nickel, Graphite, and of Ni/G Composites

Item Number Material Density (g/cm3) Reference

1 nickel 8.88 Budinski2 graphite 1.6 to 2.2 Budinski3 Ni-graphite 3.35 present work4 fully dense Ni-G 3.45* calculated5 95 pct dense Ni-G 3.28* calculated6 90 pct dense Ni-G 3.11* calculated

*Calculated based on a graphite density of 1.8 g/cm3.

Table II. Volume Fractions of the Sample Composite ofDensity of 3.35 g/cm3

Graphite Nickel Porosity

Volume fraction 74.4 pct 22.6 pct 3 pct

Fig. 4—(a) Illustration of specimen for three-point bending. (b) Illustra-tion of loading arrangement for three-point bending.



tms/MTA/123238/ETP05446A12

thickness, P 5 load increment as measured from preload,and d 5 elastic deflection at measured midspan location.

2. Compression testsIn uniaxial compression testing, the specimen was sub-

jected to an axial compressive load; the load was graduallyincreased and the change in specimen height was moni-tored. Specimen sizes were 6.5 3 6.2 3 13.3 mm and6.6 3 6.3 3 13.4 mm, respectively. Stress/strain curvescan be obtained from this experiment showing both elasticand plastic deformation regimes. The Young’s modulus canbe obtained from the linear elastic portion of the stress-strain curve.

Some of the compression tests were carried out withunloading/reloading cycles in order to observe the defor-mation process during unloading and reloading, and also toobtain the value of Young’s modulus from the unloadingand reloading portions of the stress/strain curve. The setupfor this test was the same as for uniaxial compression test-ing. The dimensions of the specimens were 6.6 3 6.6 36.35 mm and 6.8 3 6.6 3 5.7 mm, respectively.

3. HardnessA hardness indentation test was done with a hard steel

ball indenter of 1/16-in. diameter (D). A Rockwell 15-Tscale superficial hardness tester with a 15-kg load was usedin the manner of a Brinell hardness test. Hardness values inunits of N/m2 were obtained by dividing the applied load bythe area of indentation. The diameter of indentation, d, wasmeasured with an optical microscope after removal of theindenter. Hardness values were calculated according to Eq.[5], in analogy to a Brinell hardness value and convertedinto units of N/m2.

H 5P

p

2DðD�

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiD2 � d2

pÞ

[5]

where H is the hardness value, d is the diameter of theindentation in the test specimen, and D is the diameter ofthe indenting sphere.

D. Ultrasound Pulse Propagation Tests

Young’s modulus was obtained from velocity of soundand material density, Eq. [7]. Ultrasound wave propagationtests were performed using a setup described by Hazonyet al.[11] For this purpose, disc specimens with parallel faceswere prepared. Two discs had thicknesses of 3.2 and3.1 mm, respectively. The experimental setup is shown inFigure 5.

A transducer T sends out short ultrasound pulses througha plexiglass medium and through the test specimen. Thepulse is reflected at the boundaries. From the time differ-ence between the pulse launch and the arrivals of thereflected signals back at the transducer, the times of flightthrough the medium and through the specimen wererecorded and the velocity of sound was obtained. Fromthe time difference between the return arrival times of thesignals from reflection at the plexiglass/specimen interfaceand reflection at the specimen/air interface, the soundvelocity in the specimen was obtained, Eq. [6].

v 5 2l=t [6]

where l 5 thickness of specimen, and t 5 time of flight ofthe sound pulse in the specimen as measured from the timeinterval between the respective return signals.

From the time interval between the first and secondreflected signals (a1 and b1), and knowing that the distanceof travel is twice the sample thickness, one can calculatethe speed of sound in the sample. A dynamic Young’smodulus value is obtained from this experiment. TheYoung’s modulus is obtained from Eq. [7], which statesthe relationship with density r of the specimen and thevelocity v of sound traveling in the specimen:

E 5 v2r [7]

Fig. 5—Schematic illustration of sound pulse propagation through amedium and specimen. (a) Piezoelectric transducer plus plexiglassmedium assembly. Ultrasound pulses of intensity a0 emanating from thetransducer transverse the medium and after reflection return to the trans-ducer. (b) Assembly with attached specimen. Multiply reflected soundpulses (b1, b2, . . .) return to the transducer.



tms/MTA/123238/ETP05446A12

The damping coefficient of the material can be obtainedfrom the attenuation of the pulses as measured from thedecreased pulse amplitudes of the reflected signals. Thetimes of flight and the attenuation of ultrasound plusesare first measured on the ‘‘transducer plus plexiglass’’ as-sembly. Then, the specimen was placed on the plexiglass andthe measurements were repeated. From the comparison ofthe amplitudes of the respective reflected wave signals, thereflection coefficient of the specimen was obtained (Eq. [8]).

The reflection coefficient of the plexiglass/specimenassembly can be expressed as the ratio of the amplitudesof the first reflected signals (Eq. [8]), without and with thespecimen attached:

G ¼ a1a0

[8]

A brief description is given here of the ultrasound pulsepropagation and reflection through the array of the mediumand the sample, as shown in Figure 5. For a detaileddescription of the method, refer to the article by Hazony.[11]

Consider that the flat thin specimen is being irradiated bya short pulse with pressure p. As the pulse is reflected at theinterfaces, the pulse bounces back and forth inside thespecimen. A periodic set of pulses is formed and returnsto the transducer. The rays of the pulses and their corre-sponding amplitudes can be written as

a1 ¼ Gp [9]

�b1 ¼ �ð1� G2ÞpD2 [10]

�b2 ¼ �ð1� G2ÞGpD2 [11]

where P is the input pulse pressure from the transducer; anda1, b1, and b2 are the amplitudes of the returning pulse rays.The minus signs indicate a reversal of phase between a and b.The term G is the reflection coefficient, and D is attenua-tion. Attenuation can be expressed as[12]

D ¼ e�al [12]

where a is the attenuation constant, and l is the path length.With the thickness of the specimen being denoted as t, thepath length is 2t. The attenuation constant is obtained fromEq. [13]:

a ¼ � 1

2tln

G

1� G2

b1a1

� �[13]

The ultrasound pulse propagation tests yield data on thedynamic Young’s modulus and on the ultrasound attenua-tion. Attenuation coefficient values were measured on discsof the Ni-G cell composite and also on discs made fromlow-carbon steel and gray cast iron. Disc samples of eachmaterial were tested in identical fashion in order to obtaincomparison of their Young’s moduli and sound attenuation.

III. RESULTS AND DISCUSSION

A. Mechanical Test Results

1. Three-point bend test resultsFigure 6(a) shows the load-displacement curves for two

samples subjected to three-point bend tests. The values of theYoung’s modulus were calculated from the linear elastic partof the load-displacement curves. For sample A, the moduluswas calculated from the slope of A1 to A2, while for sampleB, modulus values were calculated from slopes B1 to B2 andB2 to B3. The maximum stress values were taken to be equalto the fracture stresses. They were obtained from the peakloads using Eq. [3]. The values are listed in Table III.During loading, the specimens bent slightly, and then

broke at loads between 20 and 25 N. The onset of fracturedid not result in an immediate load drop to zero. The tear-ing cell walls offered some resistance beyond the onset ofthe fracture, as is evident from the gradual down-slope ofthe load-displacement curves.Scanning electron micrographs were taken from the frac-

ture surfaces (Figure 6(c). The fracture morphology is con-trolled by the size and shape of the cells. In the SEMmicrographs, the white areas are nickel, and the dark areaswith cleavage fracture striations are graphite. Cleavage orshear fracture of the layer structure of the graphite cellcores shows up as lathlike features.The nickel cell walls failed by tensile necking, as inferred

from the elongation and slenderizing of many of the torn cellwalls. In a few instances, the fracture path proceeded alongthe cell boundaries. They appear as white areas. However,the vast majority of the fracture surface was transcellular.This suggests that the bonding between the cells is relativelystrong. The failure mechanisms are by cleavage and shear of

Fig. 6—(a) Load-displacement curves of two three-point-bend test samples, A and B. (b) Three-point bent specimen after fracture. The arrows indicate theloading direction in bend test (also Fig. 4). (c) Scanning electron microscope images of fracture cross sections of bend samples. On the left: crack starting; onthe right: crack termination. The graphite cell interiors have cracked by cleavage and shear. The nickel cell walls failed by ductile tearing.



tms/MTA/123238/ETP05446A12

graphite within the cells and by tensile and necking andrupture of the nickel cell walls, but not separation of cells.

2. Compression test resultCompression of rectangular samples of hot-pressed Ni/G

cell composite yielded load vs displacement curves, whichwere converted into true stress–true strain curves using Eqs.[14] and [15].

s ¼ sðe1 1Þ [14]

e ¼ ln ðe1 1Þ [15]

where s is true stress, s is engineering stress, e is engineer-ing strain, and e is true strain.[13]

The curves for two specimens are shown in Figure 7(a).The Young’s modulus values were calculated from the lin-ear deformation regimes at the beginning of the compres-sion curves. After this linear region, one observes yielding,and the specimens exhibit plastic deformation and strainhardening. The load increased with strain with a parabolic

dependence until facture occurred. Results of the tests aregiven in Table IV.

The data show the material to be relatively soft elasti-cally and to have significant compression strength and com-pression plasticity. The stress-strain curves do not suffer thelow-strength plateau of nonfilled cell materials. Plasticstrains of 11 to 12 pct and strain hardening from around38 to 105 MPa true stress was recorded. In the end, shearfracture occurred on a macroscopic shear plane inclined atapproximately 45 deg to the compression axis. A photo-graph of a fractured specimen is shown in Figure 7(b).

An SEM micrograph of the shear fracture surface is pro-vided in Figure 7(c). It shows sliding traces in the directionof the shear motion. Other steplike features in the fracturesurface are often associated with sheared nickel cell wallsor tensile-ruptured ledges.

A second set of compression tests was conducted withunloading-reloading stages. It showed elastic, anelastic, andplastic deformation behavior (Figure 8). The unloadingstress-strain slopes are steeper than the loading ones andgive a Young’s modulus that is believed to be more repre-sentative than the modulus obtained from the up-loadingcurve. The hysteresis loop areas between the unloading/reloading curves are measures of dissipated strain energy.They increase in magnitude with stress and strain. Theyindicate that a significant fraction of the cycles’ elasticstrain energies is dissipated as anelastic energy; i.e., thematerial is expected to have high damping capacity. Theratio DW/W is a measure of the damping capacity, whereDW is the dissipated strain energy and W is the elastic

Table III. Three Point Bend Test Results: Young’s Modulusand the Maximum Tensile Stress at Which Fracture Started

Bend Sample Modulus (GPa)Maximum Stress � Tensile

Rupture Stress (MPa)

A 5.9 (A1 to A2) 17B 21.0 (B1 to B2) 18

3.0 (B2 to B3)

Fig. 7—(a) True stress–true strain curves of compression-tested parallelepiped specimens. Note the noninstantaneous load drop after the onset of fracture. (b)Compression specimen. The arrow indicates the loading direction during the compression test. Note the bulging of the cylindrical sides of the sample causedby plastic deformation. (c) SEM micrograph of the shear fracture surface inclined at 45 deg to the compression axis. The sliding direction upon shear fractureis shown. (d) Illustration showing the shear fracture including tensile rupture at steps in the specimen.



tms/MTA/123238/ETP05446A12

strain energy during loading or unloading.[14] Values forDW/WLoading and DW/WUnloading are given in Table V.

The compression specimens in these tests failed aftersome 15 pct plastic strain by shear fracture.

3. Hardness test resultUsing a modified Rockwell 15-T superficial hardness

testing method, a 1/16-in.-diameter steel sphere waspressed into the surface of the Ni/G cell composite. A hard-ness value of 370 MPa was obtained from the measuredindentation diameter using Eq. [5]. An approximate (empir-ical) correlation between hardness and uniaxial flow stress is

H 5 3suniaxial [16]

Based on this empirical rule, the hardness value predicts auniaxial compression flow stress of around 123 MPa,slightly higher than the values from the uniaxial compres-sion tests.

B. Young’s Modulus from Ultrasound Propagation Test

1. Young’s modulusThe first, second, and third (a1, b1, and b2) reflected

ultrasound wave signals (Figure 5) were used for evalua-tion. Their amplitudes and times of flight were recorded.The signals correspond to ultrasound pulses having propa-gated through a plexiglass layer, through plexiglass plusonce through the specimen, and through the plexiglassand twice through the specimen. The first signal (a1) in aplot of signal amplitudes vs time (Figure 9(a)) correspondsto a travel path of two times the thickness of the plexiglasslayer and the reflection at the plexiglass/specimen interface.The second signal (b1) is the reflection from the back faceof the specimen plate. The abscissa is a time scale and hasunits of microseconds: The ordinate shows the pulse ampli-

tude in arbitrary units. It is equivalent to sound pressure,which has units of energy.Two Ni/G composite specimens were evaluated by ultra-

sound propagation. Because the specimens were cut fromdifferent parts of the hot-pressed Ni/G composite cylinder,they had a slightly different density. The density variation ina hot-pressed sample comes from a pressure gradient in thepressing die that is due to friction of the powder particlesagainst the wall of a cylindrical container. Slices taken fromdifferent axial positions of the compacted cylinder have dif-ferent density. This is a rationale for explaining the differentsound velocities of different specimens in Table VI.

2. Attenuation of transmitted ultrasound pulsesA measure of attenuation or a coefficient of damping can

be obtained from comparison of amplitudes of signals fromsound pulses that have traveled different path lengthsthrough the sample. The amplitude changes come fromreflections and from attenuation in the material. A set ofpulses returning to the transducer and their amplitude equa-tions are shown in Figure 5. Rays indicating the travel pathsare shown at an angle for clarity. In reality, the paths areperpendicular to the interfaces of transducer/medium/speci-men. In the illustration, the attenuation factor D ¼ e�al anda is the attenuation constant. On an oscilloscope, succes-sive returned pulses show up as signals with diminishingheights, namely, a1, -b1, -b2,. . ., etc, in Figure 9(a).To find out the attenuation constant of the material, the

reflection coefficient of specimen G was determined firstusing Eq. [8]. From the ratio of the signal amplitudes a1/a0,we obtain the reflection coefficient between the specimen andthe plexiglass.

G ¼ a1a0

51:7

3:625 0:47

Identical comparison experiments were carried out withspecimens of plain carbon steel and of gray cast iron.The reflection coefficients from these two specimen mate-rials were measured, as described previously. The results ofthe ultrasound propagation experiments are shown in Fig-ures 9(a) through (c). The a0 signals are from ultrasoundreflection without specimen. The signals (a1, b1, etc.) wereobtained with Ni/G (Figure 9(a)), carbon steel (Figure9(b)), and gray cast iron (Figure 9(c)) specimens attachedto the plexiglass medium.The attenuation constants of the specimens were calcu-

lated from the amplitudes of successive reflected pulseamplitudes (a1, b1, b2) using Eq. [17] with the previously

Fig. 8—Engineering stress–engineering strain curves of compression testswith intermediate unloading/reloading cycles.

Table V. Mechanical Properties Obtained from CompressionTests with Intermediate Unloading and Reloading Cycles

Sample 1 2 Average of Two

Unloading Young’s modulus(GPa)

10 18 14

Yield point (MPa) 39 35 37Maximum stress (MPa) 95 105 100Strain to fracture 0.15 0.16 0.155DW/W loading 0.27 0.29 0.28DW/W unloading 0.50 0.45 0.47

Table IV. Mechanical Properties Obtained fromCompression Tests

Sample 1 2

Young’s modulus (GPa) 2.94 3.49Yield stress (MPa) 36 38Maximum stress (MPa) 121 115Strain to fracture 0.110 0.113



tms/MTA/123238/ETP05446A12

determined reflection coefficients and measured pulseamplitude ratios (b1/a1):

a ¼ � 1

2lln

G

1� G2

b1a1

� �[17]

The ultrasound pulse propagation data for the nickel/graph-ite cell composite and for low-carbon steel and gray cast-iron specimens are listed in Table VII.

IV. DISCUSSION

A. Microstructure

The Ni/G cell composites consist of an interconnectednickel matrix formed of diffusion-bonded nickel cell walls.

In a fully dense composite, the nickel phase occupies22.6 vol pct (60 wt pct). The thickness of the nickel wallsis approximately twice that of the starting cell walls, i.e., 5to 15 mm. The second phase, graphite, was present withsome 74.4 vol pct (40 wt pct) in the form of 50- to 150-mmparticles enveloped by nickel. The graphite had a density of;1.8 g/cm3, as inferred from the density of a cell compo-site that consisted of ;97 pct graphite and nickel and ;3pct porosity. The porosity is a component in the composite.

B. Young’s Modulus

The Ni/G composite having Ni/G cells in a randomlyaggregated form is expected to have an isotropic Young’smodulus. It is further expected to be a function of theYoung’s moduli and volume fractions of the constituent

Fig. 9—(a) Plot of ultrasound signal amplitudes vs time of arrival. The pulse signals are from a series setup of transducer/plexiglass/Ni-G specimen, cf.Fig. 6. The specimen was a 3.2-mm-thick plate of polished nickel/graphite composite. The a0 reflection is not shown in this graph. It was measured separately.The reflection coefficient between plexiglass and specimen was a1/a0 5 0.47. (b) Reflected ultrasound pulse signals in a series setup of transducer/plexiglass/steel plate, cf. Fig. 6. The specimen was a low-carbon-steel plate of 6.1-mm thickness. The reflection coefficient between plexiglass and the steel specimenwas 0.87. (c) Reflected ultrasound pulse signals in a series setup of transducer/plexiglass/cast iron assembly. The specimen was a gray cast-iron plate of6.1-mm thickness. The reflection coefficient was 0.86.



tms/MTA/123238/ETP05446A12

phases. The Young’s modulus of natural graphite consistingof random graphite grains that is not fully dense (r 5 1.8g/cm3 vs 2.2 g/cm3 theoretical density) was taken as 3 GPa.The Young’s modulus of nickel was taken to be 180 GPa.The packing of the cells is such that the components appearin series and to a degree in parallel to a uniaxial stresspulse. The composite modulus for nickel and graphite inseries and in parallel is plotted in Figure 10. Experimentaldata are shown for a composite that consists of 22.6 vol pctnickel, the balance being graphite. Therefore, porosity isnot taken into account.

The measured Young’s modulus values ranged from 4to 35 GPa, as measured in hardness, compression, andultrasonic propagation tests. The ultrasonically measured

value is higher than the modulus value from mechanicaltesting. In mechanical tests, when load testing a compo-site material that has some of its components oriented inseries along the load direction, the weakest part in thespecimen will yield first, and it is hard to obtain a truelinear-elastic stress-strain curve. During elastic unload-ing, the stress-strain curves had steeper slopes, and theYoung’s modulus values measured from the slopes of theunloading stress-strain curves are closer to the ultrasonicvalues.Also, because of the limited amount of samples we have,

two samples were tested in bending, compression, com-pression with reloading tests, and the ultrasound test,respectively. The scatter in the data of yield stress andYoung’s modulus may then be due to changes in the densityof two specimens.

C. Tensile Strength and Ductility

The compression strength of the Ni-G composite was inthe range of 95 to 121 MPa. The yield strength rangedfrom 35 to 39 MPa. Ductility in compression ranged from11 to 16 pct. In tension or in bending, the ductility wasless than 1 pct. The high ductility in compression is atleast in part due to the strain-hardening ability of the filledcell structure. Because the cells are filled, buckling orcrushing of cell walls is not a failure mechanism. Eventualfailure appears to occur by fracture of shear bands, atapproximately 45 deg to the compression axis, but notuntil some 10 to 16 pct of plastic deformation hasoccurred.

1. Energy dissipation propertyThe nickel/graphite cell composite has a sound pulse

attenuation constant for sound pulses of ;0.35/mm. It isseveral times larger than those of gray cast iron (;0.06/mm)and carbon steel (;0.02/mm). It is a highly dampingmaterial. Figure 11 shows a comparison of the three mate-rials. The large anelastic energy dissipation ability is also

Table VII. Comparison of Ultrasound Pulse Propagation Data for Ni/G Cell Composite, Low-Carbon Steel, and Gray Cast-IronSpecimens Including Reflection Coefficient and Attenuation Constants (Experiments were conducted by Dr. D. Hazony)

SpecimenSpecimen

Thickness l (mm)Reflection

Coefficient G*Ratio of Pulse

Amplitudes b1/a1

Attenuation Constanta (mm�1)

Ni/graphite composite 3.2 0.4696 0.1818 0.3456Gray cast-iron 6.1 0.8554 0.1458 0.0628low-carbon steel 6.1 0.8675 0.2308 0.0173

*Reflection coefficient at the interface between the plexiglass medium and specimen.

Fig. 10—Composite modulus varies with volume fraction and Young’smodulus of components. The Ni/G cell composite has 22.6 vol pct nickel,74.4 vol pct graphite, and 3 vol pct porosity (refer to Eqs. [1] and [2], andTable II). The lower bar of experimental data shows modulus values from3.49 to 21 GPa determined from mechanical tests, and the upper bar showsmodulus values from 25 to 35 GPa determined from ultrasound propaga-tion tests.

Table VI. Properties Obtained from Ultrasonic Propagation Test

SampleTimet (ms)

Thickness of Ni/Gspecimen l (mm)

UltrasoundVelocityv (m/s)

MeasuredDensity of Ni/G

Composite r (g/cm3)

Young’s Modulusof Ni/G Composite

E (GPa)

1 2.4 3.2 2670 3.55 252 1.9 3.1 3250 3.35 35



tms/MTA/123238/ETP05446A12

evident from the unloading/reloading stress-strain hystere-sis loops in Figure 8.

2. ApplicationBecause of the lack of plateau behavior, high damping

coefficient, and relatively poor mechanical property of thiscomposite, it is not suggested that it be used as a conven-tional structural material by itself. However, an applica-tion such as brake disc is suggested because of theproperty of absorbing a large amount of energy with asmall volume of material. The vibration energy will bedissipated into heat and this property makes the nickel-graphite cell composite a potential material for this kindof application. Also, if used as filling material insidethe steel or other metal tube, the sandwich structure willprovide good mechanical property as well as energyabsorption.

V. CONCLUSIONS

Nickel/graphite cell composite is a lightweight material.Composite with 60 wt pct Ni and 40 wt pct graphite can bemade with a density from 2.53 to 3.55 g/cm3, depending on

the processing conditions under which the specimens aremade. The microstructure is one of well-bonded cells.Nickel forms a continuous matrix in the composite andgraphite is a plastically formable fill material inside thecells. Stress-strain curves show elastic, anelastic, and plas-tic deformation; strain hardening; and a relatively high duc-tility (up to 16 pct) in compression. A modest tensilestrength (;18 MPa) is believed to be provided by the con-tinuous nickel cell-wall matrix. Filling of the cells withgraphite prevents buckling and crushing of the cell walls.Mechanical tests show that Young’s modulus ranges from4 to 20 GPa, depending on the characterization method.Ultrasound propagation yielded Young’s modulus valuesfrom 25 to 35 GPa, depending on the density of the speci-men. The Ni/G cell composite has a high damping capacityand forms stress-strain hysteresis loops with significantenergy dissipation during unloading and reloading cyclesin compression. Ultrasound pulses are rapidly attenuated.The attenuation constant of 0.34/mm is six times that ofgray cast iron.

REFERENCES

1. H. Bart-Smith, A.-F. Bastawros, D.R. Mumm, A.G. Evans, D.J.Sypeck, and H.N.G. Wadley: Acta Mater., 1998, vol. 46 (10), pp.3583-92.

2. L.J. Gibson and M.F. Ashby: Cellular Solids: Structure & Proper-ties, 1st ed., Pergamon Press, New York, NY, 1988, pp. 11-17 and121-31.

3. E. Andrews, W. Sanders, and L.J. Gibson: Mater. Sci. Eng., 1999, vol.A270, pp. 113-24.

4. T.-J. Lim, B. Smith, and D.L. McDowell: Acta Mater., 2002, vol. 50,pp. 2867-79.

5. M. Ozgur, R.L. Mullen, and G. Welsch: Acta Mater., 1996, vol. 44 (5),pp. 2115-26.

6. M. Ozgur, R.L. Mullen, and G. Welsch: Int. J. Num. Methods Eng.,1996, vol. 39, pp. 3715-30.

7. M. Weller and C.A. Wert: J. Phys., 1983, vol. 44 (12).8. Rees, E.L. LL., Heck F.W., Dibari, G.A.: INCO Europe Ltd./Novamet

Specialty Products (Clydach, Swansea, South Wales, U.K.) unpub-lished research.

9. P.T. Szozdowski and G. Welsch: Case Western Reserve University,Cleveland, OH, unpublished research, 1999.

10. Annual Book of ASTM Standards, ASTM, 1994, Sect. 3, vol. 3.01,ASM International, West Conshohocken, PA, pp. 101-08, 118-30,and 650-56.

11. D. Hazony, L.J. Katz, and N. Kharin: Case Western Reserve Univer-sity, Cleveland, OH, unpublished research, 2001.

12. J. Krautkramer and H. Krautkramer: Ultrasonic Testing of Materials,2nd ed., Springer-Verlag, New York, NY, 1977, pp. 23-25 and 107-09.

13. G.E. Dieter: Mechanical Metallurgy, 3rd ed., McGraw-Hill, Inc., NewYork, NY, 1986, pp. 284-85.

14. K.G. Budinski and M.K. Budinski: Engineering Materials, Pearson,Prentice Hall, Englewood Cliffs, NJ, 2005, pp. 826-27.

Fig. 11—Attenuation constants of steel, gray cast-iron, and nickel/graphitecell composite.



tms/MTA/123238/ETP05446A12

Date post:	14-Nov-2023
Category:	Documents
Upload:	sc
View:	1 times
Download:	0 times

Properties of closed-cell nickel-graphite composite

Documents