art-3A10.1023-2FA-3A1013781927227

Embed Size (px)

Citation preview

  • 8/10/2019 art-3A10.1023-2FA-3A1013781927227

    1/11

    J O U R N A L O F M AT E R I A L S S C I E N C E 3 7 (2002) 603 613

    Effect of particle size and surface treatment onconstitutive properties of polyester-cenospherecomposites

    R. J . CARDOSO, A. SHUKLA Dynamic Photomechanics Laboratory, Department of Mechanical Engineering and Applied Mechanics, University of Rhode Island, Kingston, RI 02881, USAE-mail: [email protected]

    A. BOSEDepartment of Chemical Engineering, University of Rhode Island, Kingston, RI 02881, USA

    Cenospheres (hollow, aluminum silicate spheres ranging from 10 to 400 m in diameter)are used as ller in a homogeneous polyester composite. Particle size was varied to studyits effect on mechanical properties of the composite. The effect of particulate surfacemodication using a silane coupling agent was also studied. Properties of the compositeswere characterized using standard testing methods. When compared to the largestparticulate used, an increase in compression strength was achieved by particle sizereduction and use of coupling agent. The Elastic modulus increased by using ne particles,while Poissons ratio remained constant and independent of silane treatment or particlesize. Fracture toughness increased with particle size reduction and increased further withsilane surface modication. Dynamic compressive strength increased with particle sizereduction, while silane did not show improvement. The addition of cenospheres as well assilane treatment increased the glass transition temperature for polyester. A given massfraction of particulate, of a mean diameter D , will have the surface area between theparticulate and matrix scale as D 1 (specic surface area). The sensitivity of these

    properties to cenosphere size is a direct function of the interfacial surface contacts betweenthe polyester and the cenospheres and the specic surface area.C 2002 Kluwer Academic Publishers

    1. IntroductionAn inexpensive, readily available byproduct of y-ashfrom coal burning or heavy fuel oil combustion, calledcenospheres [13], are used as the ller material in apolyester resin matrix to produce a macroscopicallyhomogeneous particulate composite. Published ceno-

    sphere morphology and chemical composition indi-cates; spherical aluminum silicate micro spheres rang-ing from 10 to 400 microns in diameter, porous wallshaving a thickness of approximately 10% sphere diam-eter and bulk density of approximately 400 kg/m 3 . Ac-curacy of the reported morphology and chemical com-position was veried by theauthors. Using cenospheresas ller can reduce the density of composite materialsand structural members, while maintaining or increas-ing their strength, making these light-weight compos-ites very desirable to the automotive, aircraft/aerospaceindustries. Theuse ofcenospherescanalso impact other mechanical, thermal and acoustic properties. Ceno-spheres are currently considered waste material anddisposed of in landlls; therefore recycling is also a

    Author to whom all correspondence should be addressed.

    benet. Fig. 1 shows SEM images of the 7510 msize range cenospheres before casting in polyester andclearly show the porous walls.

    Cenospheres in aluminum matrix composites [4]andpolymer matrices other than polyester [5, 6] have beenstudied in the past. Parameswaran and Shukla [7, 8]

    studied polyester-cenosphere composites cast as func-tionally gradient materials (FGMs). This study is ex-clusively on homogeneous polyester-cenosphere com-posites and focuses on the role of cenosphere size andsurface properties on the mechanical properties of thecomposite.Thecompositeswerecharacterizedby statictensile loading experiments (ASTM D638) [9], com-pressive experiments (ASTM D695) [10], and frac-ture toughness experiments (ASTM D5045) [11]. Thedynamic mechanical behavior was characterized withthe Split Hopkinson Pressure bar technique [12] anda dilatational wave speed measurement. In addition tothe mechanical tests, a series of microscopic and ther-mal tests were conducted to analyze the microstruc-ture of the composite, the surface composition of the

    00222461 C 2002 Kluwer Academic Publishers 603

  • 8/10/2019 art-3A10.1023-2FA-3A1013781927227

    2/11

    Figure 1 SEM images of a cenosphere before casting in polyester, which clearly show the porous nature of the walls.

    cenospheres and bonding of the silane-coupling agent

    to the cenospheres. Differential Scanning Calorimetrywas used to measure changes in the glass transitiontemperature of the polyester. Scanning Electron Mi-croscopy (SEM) and optical microscopy were used toexamine the cenosphere s surface. Energy DispersiveSpectroscopy (EDS) and X-Ray Diffraction were usedto identify cenosphere surface composition [13, 14].

    2. Composite preparation, testingand characterization

    2.1. Sieving

    All cenospheres used in this study were purchased fromSphere Services Inc. TN, USA. All of the different par-ticle size distributions in this study with the exceptionof the 75 10 range were sieved from the 300 10 gradewith a standard mesh sieve column. The sieving wasdone on a mechanical shaker following standard soiltesting procedures. The 300 10 grade and the 75 10grade were purchased in that condition.

    2.2. Composite castingThe composite was prepared using cenospheres in apolymeric matrix. The cenospheres have a low spe-ci c gravity(0.67)compared to that of polyester (1.18),therefore will migrate against the direction of gravity.This would result in a heterogeneous distribution of cenospheres in the matrix. In order to achieve a ho-mogeneous composite, all castings were rotated abouta horizontal axis at two revolutions per minute, dras-tically reducing the effective gravitational eld insidethe mold.

    For a given mass fraction of cenospheres in a com-posite, the overall interfacialareabetween the inorganicsilica-alumina surface and the polyester scales as D1(speci c surface area), where D is the diameter of

    the cenospheres. Since a large proportion of failurein these materials is the result of cenosphere delami-nation, the particle size can be expected to in uencea range of mechanical properties. In addition, modi- cation of the interface to enhance the coupling be-

    tween the polyester and cenosphere surface should also

    have signi cant effects on these mechanical properties.The experiments described in this paper are geared to-wards determining the effects of particle size on sev-eral mechanical properties of the composite. In addi-tion to ller sizes, ller surface modi cations weretested to study their effects on the properties of thecomposite. A silane coupling agent [Silquest A-174,Gamma-Methacryloxypropyltrimethoxy Silane, WitcoChemical Company] is coated onto the cenospheres toincrease bonding between the polyester and the ceno-spheres. The silane group hydrolyses and bonds di-rectly to the aluminum silicate cenosphere surface.

    The organic chain penetrates into the polyester ma-trix. This arrangement increases the bonding energybetween the aluminum silicate and the surface of thepolyester matrix. Ideally, only a monomolecular layer should be deposited, allowing the correct concentrationof the organic chain.

    The mold in which the sample was prepared in con-sisted of two sheets of acrylic having dimensions of 250 mm 250 mm 12 mm, a rib section was madeusing three pieces of aluminum bar stock screwed to-gether to form a U shaped frame and a lid. Mylar sheets having thickness of 0.18 mm were cut to thesame size as the acrylic sheets (250 mm

    250 mm) a

    small amount of glycerin was poured on the acrylic,the Mylar placed on top and the two were pressed to-gether with a roller to remove all air bubbles. Siliconespray release agent was sprayed on the Mylar to aid indemoldinganda thin bead of sensor safe silicone caulk-ing was applied to both sides of the aluminum ribs toseal the mold. Mylar was used to aid in casting removalas well improve surface nish.

    Polyester resin (MR-17090), manufactured by theAshland Chemical Company is used as the matrix.In order for the matrix material to cure properly acatalyst (Methyl Ethyl Ketone Peroxide, 0.85% w/w)

    and an accelerator (Cobalt Octoate, 0.03% w/w) wereused. The polyester resin was mixed thoroughly withthe proper amounts of both catalyst and accelerator inthat order. Cenospheres, 25% by weight of resin wereslowly added to insure complete wetting as well as

    604

  • 8/10/2019 art-3A10.1023-2FA-3A1013781927227

    3/11

    avoid clumping. Once thoroughly mixed, the samplewas placed in a 600 mm of Hg vacuum for 20 minutesto remove any trapped air bubbles. The mixture wasthen slowly poured into the mold and then sealed withthe lid. Once sealed, the casting was cured rotating at2 revolutions per minute for 48 hours at room tempera-ture. After the 48-hour cure cycle, the mold was openedandthe casting waspost-cured in anair-circulatingoven

    on a

    at pane of glass, which was sprayed with mold re-lease and coated with cenospheres to prevent sticking.The post-curing cycle was 4 hours at 52 C followed by5 hours at 63 C. The post-curing at elevated tempera-tures ensures complete cross-linking of thepolymerandyields the maximum possible stiffness and strength of the casting [7].

    X-Ray diffraction of the polyester composite wasused to learn more about the crystal structure of thepolyester and the chemical identity of the cenospheres.The Bragg diffraction peaks for the cenospheres mostclosely matched aluminum silicate. The X-ray diffrac-tionplot forpolyestershown in Fig. 2, clearly illustratesthe amorphous nature of the polyester.

    Figure 2 X-Ray diffraction trace plotting showing peak vs. angle for polyester sample which shows the amorphous nature of the polyester.

    Figure 3 EDS trace showing elemental composition information of a cenosphere sample.

    2.3. Cenosphere surface treatmentEnergy Dispersive Spectroscopy (EDS) yielded chem-ical composition values that matched Sphere Service spublished literature, and is shown in Fig. 3. All EDStesting was done on a JEOL 1200 EX instrument inSEM mode, using a Noran Instruments detector, whichused a Germanium crystal detector element and a lightelement (novar) window. A total of three tests were

    done on a 100-micron square at 3000X magni

    cationand 20 kV in a low pulse processor rate. The resultsshow 52% Oxygen, 23% Aluminum, 23% Silica andtrace amounts of other elements.

    The procedure for applying the silane coupling agentto the cenospheres was outlined in manufacturer liter-ature. Special attention was needed in the solution mixdue to the silane reacting with Silica, present in thebeakers used. The Silane reacts with Silica only in thepresence of water therefore two beakers were used toensure separation of water and silane until the desiredtime. Also for optimum performance the pH of the so-lution had to be closely maintained between 4 and 5and monitored continuously. The 290-gram sample of

    605

  • 8/10/2019 art-3A10.1023-2FA-3A1013781927227

    4/11

    Figure 4 Density of casting at various locations in g/cc.

    cenospheres required 0.029 grams of silane to produceone of the concentrations used (0.01%).

    The sample was thoroughly mixed for 15 minutes.The still moist cenospheres were air dried for two hours

    at room temperature followed by drying in an air circu-lation oven at 80 C for an additional four hours. After the drying process clumping was observed, so sievingwas done to ensure better surface wetting when mixedwith polyester.

    2.4. Physical propertiesThe density of the castings was measured at variouslocations as shown in Fig. 4. This was done to ensurethat density did not vary signi cantly with position,which would indicate a non-homogeneous casting. Bythis method the casting technique was proven effectivein consistently producing homogeneous sheets with nointernal voids due to air being trapped at time of moldsealing.

    2.5. Quasi-static elastic properties2.5.1. Tensile experiments Following ASTM standard D 638 [9] for rigid plas-tics, the initial portion of the tensile stress-strain curvewas measured for the various ller grades tested. Thespecimen dimensions were chosen according to theASTM speci cations. Two 120 Ohm resistance straingages from Micro-Measurements company type EA-06-250BG-120 were bonded to each specimen tested,one axially and one transversely. The testing was doneon an Instron model 1125 tensile test machine using a90kN( 20,000 lb ) load cell. A total of ve repetitionswere done for each experiment. The average values of Young s moduliandPoisson sratiosforallthe ller sizedistributions used can be seen in Table I. The table lists ller grades in a range of their ller size (in microns)and is in order of decreasing mean diameter. (Note.S1 and S2 are different silane concentrations, 0.003%and 0.01% respectively). The bands were calcu-lated assuming normal distribution using the standardstudent s t [15] method with a 95% con dence. This

    con dence was done for all subsequent testing.The Poisson s ratios remained essentially constant

    and independent of ller size used. As the data shows,addition of cenospheres reduces the Poisson s ra-tio when compared to virgin polyester. The relation

    TA BL E I Table displaying theresultsof elastic modulus andPoisson sratio for the studied composites

    Particulate sizerange ( m) E (GPa) Poisson s ratio

    Polyester 3.98 0.35300 180 4.16 +/ 0.10 0.29 +/ 0.01150 105 4.31 +/ 0.03 0.30 +/ 0.01105 10 4.33 +/ 0.02 0.30 +/ 0.01

    75

    10 4.45 +/ 0.08 0.29 +/ 0.0175 10 S1 4.31 +/ 0.07 0.28 +/ 0.0175 10 S2 4.37 +/ 0.10 0.28 +/ 0.01

    between the lowering of Poisson s ratio by the addi-tion of particulate closely follows the standard rule of mixtures. Using Poisson s ratio and volume fractionof cenospheres, 40% and 0.22 respectively, and thePoisson s ratio of polyester, 0.35; the calculated com-positePoisson s ratio of0.30 agreeswell with measuredvalues.

    When compared to virgin polyester, Young s mod-ulus increases with the addition of cenospheres. TheYoung s modulus increased as the mean diameter de-creased, indicating possibly better packing, or moreuniform packing. Figs 5 and 6 shows data plots for a75 10 composite for both engineering stress-strain andtransverse strain versus axial strain respectively. TheYoung s modulus of the composite is a function of theYoung s modulus of its constituents and their percentvolume fraction by the Halpin Tsai relations, which hasvolume fraction as a variable. Since in this case volumefraction is held constant, the particle size becomes amodi er to the Halpin Tsai relations.

    2.5.2. Compressive experiments The quasi-static compressive strength of the compositewas measured following ASTM standard D 695 [10].The specimen chosen was a rectangular prismatic col-umn having width and depth equal to one another andto half the height.

    All compression specimens failed in shear, along aplane at 45 to the loading direction. The maximumcompressive strength of the composites tested is givenin Table II, which shows the ller grades used (size dis-tribution in m) and in order of decreasing numericalmean diameter.

    The size effect trend is clearly seen where the com-pressive strength increases as the mean ller diame-ter decreases. The study of the crack surface indicatesbreakage of weak large cenospheres and pop out of ner ones. In the ner grade composites, the larger,weaker cenospheres are not present, so the compres-sive strength increases. Another possible reason maybe that the projected cross-sectional area of the smaller cenospheres is larger than the large cenospheres. Hav-ing a larger area and being stronger, the compressivestrength increases for the ne particulate composite.

    2.6. Fracture toughnessThe quasi-static fracture toughness of the compositewas measured following ASTM standard 5045 [11] us-ing the single edge notched specimen. A notch was cutinto the specimen to simulate a crack. The crack was

    606

  • 8/10/2019 art-3A10.1023-2FA-3A1013781927227

    5/11

    Figure 5 Plot of axial stress versus axial strain data with a curve t to nd the elastic modulus.

    Figure 6 Plot of transverse strain versus axial strain data with a curve t to nd the Poisson s ratio.

    TA BL E II Table of compressive strengths of composites studiedfound under quasi-static loading condition

    Particle size range ( m) Compressive strength (MPa)

    Polyester 155300 180 104.0 +/ 1.4150 105 114.9 +/ 1.3105 10 115.8

    +/

    0.7

    75 10 120.3 +/ 0.975 10 S1 127.6 +/ 0.575 10 S2 125.7 +/ 1.0

    305 m wide and was subsequently sharpened with asharp razor blade, Fig. 7. The fracture toughness wascalculated from Equation 1 using the failure load ( F ).The equation, which indicates the critical stress inten-sityfactor fora plane-straincondition allows predictionof failure stress when a maximum aw size in the mate-rial is known, or to determine maximum allowable awsize when the stress is set.

    K I = F S

    BW 3/ 23 x {1.99 x (1 x )[2 .15 3.93 x +2.7 x 2]}

    2(1 +2 x )(1 x )3/ 2 (1)

    TA BL E II I Table displaying results of fracture toughness experi-ments found with single edge notched 3 point bend test

    Particle size range ( m) K IC (MPa m)Polyester 0.51300 180 1.10 +/ 0.02150 105 1.35 +/ 0.06105 10 1.38

    +/

    0.09

    75 10 1.45 +/ 0.0275 10 S1 1.52 +/ 0.0575 10 S2 1.38 +/ 0.08

    where a is the crack length, W is specimen thickness, B is specimenheight, S is specimen lengthand x =a/ W .All fracture testing was done on an Instron model1125 test machine with the exure test attachment anda 90 kN (20,000 lb). load cell. The only parameter recorded was the failure load.

    The trend observed was that the fracture toughnessincreased 116% compared to virgin polyester with the

    addition of the large 300 180 m cenospheres. Thefracture toughness increased up to 185% as the mean ller diameter was decreased to the 75 10 m range.The full results can be seen in Table III with the ller grade posted in decreasing mean diameter. A small

    607

  • 8/10/2019 art-3A10.1023-2FA-3A1013781927227

    6/11

    Figure 7 Image of notch cut into single edge notched specimen used in the three point bend experiments to determine fracture toughness. Spacebetween scale marks is 100 m.

    increase of 5% was also observed between theuntreatedversus silane treated specimens. A possible explanationis that theaddition of cenospheres increases the fracturetoughnessof the composite by blunting the crack.Also,since the fracture toughness of cenosphere material ishigher than the matrix the fracture process zone devel-ops through the matrix and around the cenospheres andthis dissipates more energy, thereby increasing the frac-ture toughness. In the case of smaller ller, the area be-tween matrix and ller is substantially increased whilemaintaining the weight fraction constant.

    Fractographic analysis was done on the fracture sur-faces of these specimens using a Nikon SMZ-U opticalmicroscope. In the composites with the higher meandiameter ller, many broken large cenospheres wereobserved. This may be because larger cenospheres aremore easily broken than the smaller ones. Figs 8 and 9show the fracture surface of a fracture specimen. InFig. 8 broken large cenospheres at the crack initiationregion can be seen and are believed to be due to thecenosphere volume fraction being high enough that thecenospheres as well as the matrix share the load. Inthe case of the ner grade ller as well as the larger grades, step marks perpendicular to the crack surfacewere seen, indicating that thecrack front is splitting into

    different planes and then rejoins to form a planar crackfront. This was noted especially downstream of crackgrowth. Cenosphere pop out was much more evidentthan breaking in the ner ller composites.

    TAB LE IV Table displaying measured dilatational wave velocitiesfor the various composites studied as well as comparing the calculateddynamic elastic modulus to the static elastic modulus for each

    Particle range Velocity Dynamic E Static E( m) (m/s) (GPa) (GPa) % Increase

    300 180 2175 +/ 20 4.33 4.16 4150 105 2250 +/ 30 4.61 4.31 7105 10 2300 +/ 20 4.81 4.33 11

    75 10 2400 +/ 25 5.28 4.45 1975 10 S1 2420 +/ 25 5.36 4.31 2375 10 S2 2420 +/ 30 5.36 4.37 25

    2.7. Dynamic propertiesThe dilatational wave speed of the composite was cal-culated by measuring the arrival times of a compres-sive wave, which was initiated by a low velocity im-pact on a free edge of the composite plate. The waveswere witnessed by two accelerometers a known dis-tance apart. Fig. 10 shows the arrival times of twopulses. The voltage output accelerometers used werePCB model 303M37, mounted with wax50.8 mmapart.A Tektronix model TDS-3014 oscilloscope was used torecord the pulses. The striker was swung by hand witha small piece of plastic attached to the free surface of

    the casting to protect it from impact damage.The collected dilatational wave speeds can be seenin Table IV. The dynamic modulus is calculatedfrom Equation 2 using the given Poisson s ratio and

    608

  • 8/10/2019 art-3A10.1023-2FA-3A1013781927227

    7/11

    Figure 8 Fractographof fracturesurfaceof a fracture toughnessexperimenttakennearthe crack initiation zone. Space between scale marks is 100 m.

    density for each of the grades used and the measureddilatational wave speeds ( CL). The dynamic modulusappears to closely follow a trend similar to that of thestatic modulus except that it is much higher which isnormally the case for all materials.

    CL = E(1 2) (2)The dynamic stress-strain behavior of the mate-

    rial was also investigated using the Split Hopkinson

    609

  • 8/10/2019 art-3A10.1023-2FA-3A1013781927227

    8/11

    Figure 9 Fractographshowingsurfaceof a fracture toughnessexperimenttakenhalfway between thecrackinitiationzoneand thetop of thespecimen.It is clearly seen that large cenospheres break while small ones either do not, or pop out. Space between scale marks is 100 m.

    Pressure Bar (SHPB) technique in compression [12].Thecompressive SHPB setup consists ofanincident bar and transmitterbar both ofwhich are 12.7 mm( 1/ 2 inch)diameter steel and instrumented with strain gages, seeFig. 11. The specimen is sandwiched between the inci-

    dent bar and transmitter bar. The impact of a striker bar generates a compressive stress pulse of a nite lengthinto the incident bar. Upon reaching the specimen someof the stress pulse gets re ected back as a tensile pulseand some gets transmitted through the specimen into

    610

  • 8/10/2019 art-3A10.1023-2FA-3A1013781927227

    9/11

    Figure 10 Plot showing output pulses from accelerometers used to nd the arrival times at each and subsequently used to determine dilational wavevelocity.

    Figure 11 Schematic representation of the compressive SHPB setup used to conduct the dynamic compressive loading experiments.

    the transmitter bar. The strain histories, which are timeresolved arerecordedandused foranalysis by a LeCroymodel 6810 data acquisition module.

    The dynamic stress-strain response of the specimencan be obtained from the recorded strain histories usingone dimensional wave propagation theory. Assuming ahomogeneous specimen

    s(t ) = 2cbls

    t

    0 r (t ) d t

    cb = Ebb s(t ) = Eb

    Ab As

    t(t ) (3)

    deformation the stress and strain ( s , s) in the speci-men can be generated as a function of time from themeasured re ected and transmitted strains ( r , t) us-ing the relations shown in Equation 3. Where Ab and As are the cross-sectional areas of the bar and specimen

    respectively, ls is the specimen length, cb is the wavespeed in the bar material and Eb and b are the Young smodulus and density of the bar material respectively.Cylindrical specimens with a diameter of 11 mm andthickness of 4 mm were used to obtain the dynamicstress strain pro le of the composite.

    The observed results were that the dynamic peakstress increased as mean ller diameter decreased verysimilar to thequasi-staticcompressivestrength reportedin a previous section. The values ranged from 155 MPafor the 300 180 composite to 178 MPa for the 75 10composites. No signi cant difference was observed be-tween the two silane treated batches and the untreatedcomposite of the same grade. A typical result fromthe analysis of the SHPB data is shown in Fig. 12.Good equilibrium was observed for all experiments,(see lower right-hand equilibrium , where the curvehovers at 1 and is the stress ratio on both faces of thespecimen).

    TheSHPBtesting was done with a 203.2 mm(8-inch)projectile red at 207 kPa (30 psi). All specimens shat-tered into very small pieces. Upon closer examination

    611

  • 8/10/2019 art-3A10.1023-2FA-3A1013781927227

    10/11

    Figure 12 Results of analysis of SHPB strain histories used to nd dynamic compressive strength of the composites.

    TA BL E V Table ofSHPBexperimentalresultsshowingthemaximumdynamic peak stress for each of the studied composites

    Dynamic compressiveParticle size range ( m) strength (MPa)

    Polyester 265300 180 154.0 +/ 1.3150 105 167.5 +/ 1.4105

    10 174.5 +/ 0.675 10 178.0 +/ 1.0

    75 10 S1 179.5 +/ 0.675 10 S2 176.0 +/ 0.8

    it was observed that there was delamination betweencenosphere and matrix and some, but very few bro-ken cenospheres in the ner grade composites whichwere much more evident in the larger grade compos-ites. Some matrix and cenosphere crushing was alsoobserved in the ner debris collected from the larger grade specimens. Table V shows the measured valuesof dynamic compressive strength. The dynamic peakstresses are on average 50 MPa higher than the staticpeak stresses. The strain is applied dynamically; thestress builds faster than failure can relieve it, henceallowing higher dynamic stress.

    2.8. Glass transition temperatureDifferential Scanning Calorimetry was used to studythe glass transition temperature to see the effects of the presence of cenospheres as well as silane surfacetreatments. This transition from solid to the glassy stateis endothermic andappears as a minimum on the typical

    DSC plot. Our results show an increase in the glasstransition temperature by 3 C for coated batch number 1 and 4 C for silane batch number 2.

    The rise in the glass transition temperature is relatedto loss of mobility of the polymer chains located near

    the solid cenosphere surfaces. The coupling betweenthe organic groups in the silane with the organic ma-trix further restrict the mobility of the polymer chains.This arrangement of polymer chains affects the lo-cal microstructure and thus the interfacial mechanicalproperties.

    3. ConclusionsAddition of cenosphere particulate as llers into apolyester matrix creates internal cenosphere-polyester interfaces. A range of mechanical properties of cenosphere-polyester composite are affected by parti-clesize rangeused in thecomposite.These include elas-tic modulus increase of 7%, static compressive strengthincrease of 16%, dynamic compressive strength in-crease 16%, and fracture toughness increase of 32%compared to the largest particulate range used. Whencompared to virgin polyester, a 116% increase in frac-ture toughness was obtained by adding 300 180 m ller, but a 185% increase was obtained by using un-treated75 10 ller. When silanetreated75 10 m ller was used, a 200% increase in fracture toughness wasachieved.

    Coating of a silane coupling agent onto the ceno-spheres prior to their embedding within the polyester increases the interfacial strength, and allows several of the mechanical properties to improve. The incorpora-tion of a controlled size, surface modi ed cenospheresinto a polyester matrix is therefore a feasible routeto producing lightweight, high strength polyester materials.

    AcknowledgementsThe authors extend acknowledgement to the NationalScience Foundation for their nancial support under

    612

  • 8/10/2019 art-3A10.1023-2FA-3A1013781927227

    11/11

    grant number CMS-9900138. The authors also ac-knowledge the Ashland Chemical Company who verygenerously furnished all the polyester resin used in thisstudy.

    References1. T . WA N D E L L , The American Ceramic Bulletin 75 (6) (1996) 79.

    2. S . P E RV E Z , G . S . PA N D E Y and V . K . J A I N , Research and Industry 38 (1993) 99.3. R . M . C L AY TO N and L . H . B A C K , Journal of Engineering

    for Gas Turbines and Power 111 (1989) 679.4. P . K . R O H AT G I , R . Q . G U O , B . N . K E S H AVA R A M and

    D . G O L D E N , Transactionsof the AmericanFoundrymens Society(103) (1995) 575.

    5. E . E . B E R RY , R . T . H E M M I N G S and J . L E I D N E R , PlasticsCompounding 9 (7) (1986) 12.

    6. N . C H A N D , J. Mater. Sci. Lett . 7 (1) (1988) 36.7. V . PA R A M E S WA R A N and A . S H U K L A , J. Mater. Sci . 35

    (2000) 1.8. Idem. , ibid. 33 (1998) 3303.

    9. American Society for Testing and Materials, Standard test methodfortensile properties of plastics, Annual Bookof ASTMStandards,Designation D638-98 (1998).

    10. American Society for Testing and Materials, Standard test methodforcompressiveproperties of rigidplastics, AnnualBookof ASTMStandards, Designation D695-96 (1996).

    11. AmericanSociety forTesting andMaterials, Standard testmethodsfor plane-strain fracture toughness and strain energy release rate of plastic materials, Annual Book of ASTM Standards, DesignationD5045-99 (1999).

    12. H . K O L S K Y , Stress Waves in Solids, Dover Publications, Inc.(New York, 1963).

    13. S . P . K I M and S . C . K I M , Polymer Engineering and Science31 (2) (1991) 110.

    14. P E R E Z - C A R D E N A S , C . F E R N A N D O , D E L C A S T I L LO ,F E L I P E and V E R A - G R A Z I A N O , R I C A R D O , J. Appl. Polym.Sci. 43 (1991) 779.

    15. W . H . B E Y E R , CRC Standard Mathematical Tablesand Formu-lae (CRC Press, Florida, 1991) p. 514.

    Received 18 September 2000and accepted 27 June 2001

    613