ContentslistsavailableatScienceDirect
Materials Chemistry and Physics
j our na l h o me p a g e: w w w . e l s e v i e r . c o m / l o c a t e / m a t c h e m p h y s
Synthesis and characterization of xMgO–1.5Al 2 O 3 –5SiO 2 (x = 2.6–3.0) system using mainly talc and kaolin through the glass route
Johar Banjuraizah
a, Hasmaliza Mohamad
b, Zainal Arifin Ahmad
b,∗aSchoolofMaterialsEngineering,UniversitiMalaysiaPerlis,02600Kangar,Perlis,Malaysia
bSchoolofMaterialsandMineralResourcesEngineering,UniversitiSainsMalaysia,EngineeringCampus,14300NibongTebal,Penang,Malaysia
a r t i c l e i n f o
Articlehistory:
Received25October2010
Receivedinrevisedform23February2011 Accepted13May2011
Keywords:
A.Glasses B.Crystallization C.Powderdiffraction D.Thermalproperties D.Dielectricproperties
a b s t r a c t
Threenon-stoichiometriccordieritewithcompositionsofxMgO–1.5Al2O3–5SiO2(x=2.6–3.0mol)were synthesizedusingmainlytalcandkaolinthroughtheglassroute.Thedensificationandcrystallization behaviorsoftheseglasspowderswereinvestigatedusingDTA,dilatometer,andXRD.Sampleswerethen heattreatedat900◦Cfor2handfurtheranalyzedusingXRD,CTE,FESEM,densityandporositytests, andalsodielectrictesttofurtherinvestigatetheirproperties.The2.8MgO·1.5Al2O3·5SiO2glasscompo- sitionfullydensifiedandcrystallizedintohighpurity␣-cordieriteattheheattreatmenttemperature.
ItexhibitedlowerCTEanddielectricconstantcomparedtoothercomposition,makingitsuitablefor highfrequencyapplications.Otherformulationswhichcontainmulticrystallinephasesrequiremuch highertemperaturesfordensification.TheCTEvaluedependsonthetypeofcrystallinephasewhich existinthesample,whilethedielectricconstantdecreasedwithincreasingMgOamountsinthesample compositions.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
␣-Cordierite-basedglass–ceramicshaveattractedsomeatten- tioninrecentyears owingtotheirlowCTE, dielectricconstant, dielectriclossanddensitywhichmakethemsuitableforhighfre- quencyapplications.Thedifficultyofobtaininghighcrystallinity
␣-cordieritephaseatorbelow900◦Cwithashortsoakingtime usingsolidstate[1–5]andsol–gel[6–9]methodscanbereviewed intheliterature.Althoughithasbeenreportedthat␣-cordierite canbecrystallizedat900◦Cusingtheglassroute[10],thecrys- tallization of secondary crystalline phases, together with large amorphous phases, will significantly affect its properties. Fur- thermore,thecrystallizationthatoccursbeforethedensification processwouldcausetheviscosityoftheglasstoincrease,resulting inhigherporosity ofglass–ceramics.Thisproblemcanbeover- comebyreducingthesofteningtemperatureandviscosityofthe glass.Besidestheadditionofsinteringaidsand pulverizinginto veryfineparticles,theviscosityoftheglasscanbefurtherreduced bymodifyingthestoichiometriccompositions.
Previously, we reported the effects of excess MgO concen- tration from cordierite stoichiometric compositions on phase transformationwhichisintheseriesofxMgO·2Al2O3·5SiO2(where x=2–4mol) [11], and also the densification and crystallization
∗Correspondingauthor.Tel.:+6045996128;fax:+6045941011.
E-mailaddress:[email protected](Z.A.Ahmad).
behaviorofthesecompositions[12].Althoughcompositionswith highMgOconcentrationshadenhancedthecrystallizationofthe␣- cordieritephase,smallamountsofspinel(MgAl2O4)andforsterite alsotendedtocrystallize[11].Additionally,ithasbeenprovenfrom thesestudiesthatexcessMgOinastoichiometriccordieritecom- positionwould notonlycause thedensification temperatureto decrease,butalsothecrystallizationtemperature[12].Infact,some ofthestudiedcompositionstendtocrystallizebeforedensification hascompleted.Crystallizationthatoccursbeforedensificationmay havebeeninitiatedfrominsufficientmeltingtemperatureusedto melttheglasscompoundoralowquenchingrate.Thiswillcause smallamountsof glasstocrystallize duringmeltingand conse- quentlyretardstheviscoussinteringprocessduringheattreatment oftheglass.McMillan[13]inhisbookproposedthatthemelting temperatureoftheglasscompoundcanbereducedbyreducing theintermediate oxideoftheglass,which is Al2O3 in thissys- tem,andthiswillconsequentlydecreasetheviscosityoftheglass ifthesamemeltingtemperaturewasemployed.Inadditiontothat, areductioninAl2O3moleratioisexpectedtoretardtheforma- tionofspinelasexcessMgOwillreactwithAl2O3toformspinel (MgAl2O4),whilevariationsinMgOconcentrationswillretardthe formationofforsterite.Therefore,bysystematicallymodifyingthe compositionsofstudy,thecrystallization oftheintendedphase canbecontrolledbyimpedingthereactionandcrystallizationof thesecondaryphase.Althoughithasalsobeenreportedinthelit- erature[14–18]thattheMAScompositionusedwithexcessMgO andlessAl2O3wouldcontributetobetterdensificationandcrys- 0254-0584/$–seefrontmatter© 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.matchemphys.2011.05.026
Table1
ElementalanalysisofmineralsandoxidecompoundbyXRF.
Kaolin Talc SiO2 Al2O3 MgO
wt%
MgO 0.88 49 – – 99
SiO2 59 47 99.5 0.2 –
Al2O3 35 0.16 0.04 99.46 –
K2O 3 – – – –
CaO 0.014 3 – 0.075 0.13
TiO2 0.84 – – – –
Fe2O3 0.6 0.45 0.04 0.15 0.0036
Cr2O3 0.032 – – – –
NiO 0.014 0.028 – 0.022 –
P2O5 0.073 0.095 – – –
ZrO2 0.05 – – – –
SO3 – 0.06 – – 0.07
CuO – 0.016 – 0.014 –
Table2
Compositionofoxidesinthemixtures.
Sample MASratio MgO(g) Al2O3(g) SiO2(g)
X1 3:1.5:5 21.0548 26.6322 52.3131
X2 2.8:1.5:5 19.9309 27.0113 53.0578
X3 2.6:1.5:5 18.7745 27.4014 53.8241
tallizationbehavior,itmustbenotedthatpureoxideswereused astheirinitialrawmaterialsandthecompositionsweredifferent fromtheoneusedinthepresentstudy.
Therefore,besidestheuseof compositionswithexcess MgO concentrationinthispresentstudy,moleofintermediateoxidewas alsoreduced.Al2O3wasreducedto1.5moleratio,whileSiO2mole ratiowaskeptconstantat5molesastheexactcordieritestoichio- metriccompositionforallinvestigatedcompositions.Theeffectof MgOconcentrationinaseriesofxMgO·1.5Al2O3·5SiO2sampleswas examined.Thecompositionsofthepresentresearchweresystem- aticallychosenbasedonpreviousstudiessoastoincreasethepurity ofthe␣-cordieritephaseaswellitsdensificationbehavior.
2. Materialsandmethods
MgO–Al2O3–SiO2glass–ceramicswithdifferentmoleratiosweresynthesized usingkaolin(KaolinIndustry,TapahPerak,Malaysia)andtalc(IpohCeramic,Sdn Bhd,IpohMalaysia).Silica(IpohCeramic,SdnBhd.Ipoh,Malaysia),alumina(Metco, Westbury,USA)andmagnesia(Merck,WhitehouseStation,NJ)wereaddedtocom- pensatetheformulationofthecompositions.Theamountoftalcandkaolinused ineachcompositionstudiedwasweightedtotheoptimumamountsinorderto minimizetheuseofpuremetaloxides.Theelementalcompositionoftheinitial rawmaterialswasdeterminedbyX-rayflorescencespectroscopy(RigakuX-Ray SpectrometermodelRIX3000)andthedetailsaregiveninTable1.Tables2and3 summarizethedetailedcompositionsofoxideandtheweightpercentofinitialraw materials.Thehomogenizedmixturesofrawingredientsweremeltedinalumina cruciblesat1500◦Cfor4hinair.Fritswereproducedbyquenchingthemeltsin distilledwater,andweredrymilledusingPulverized6toobtainfineglasspowders withaverageofparticlesizesintherangeof1–3m.
TwosetsofDTAanalyseswerecarriedout.Thefirstdatawascollectedfromthe glasspowderheatedat5◦Cmin−1fromroomtemperatureupto900◦Candwaskept isothermallyat900◦Cfor1h(ModelLinseis).ThesecondDTAwascollectedfromthe samplenon-isothermallyheatedfromroomtemperatureto1000◦Cat5◦Cmin−1 heatingrate.Rectangularspecimens(5mm×3mm×20mm)forthedilatometry testwerepreparedbycompactionoftheglasspowderusingaHydrotexuniax- ialpressingmachineat120MPa.Sinteringreactionsatincreasingtemperatures weretestedusingadilatometrytestwithahightemperatureverticaldilatome- ter(ModelLinseis)fromroomtemperatureto1100◦C.TheX-raydiffraction(XRD)
Table3
Compositionofinitialsamplebasedminerals.
Sample MASratio Talc(g) Kaolinite(g) MgO(g) Al2O3(g) SiO2(g) X1 3:1.5:5 30.0000 64.0000 6.3550 4.2320 0.4531 X2 2.8:1.5:5 31.0000 65.0000 4.7400 4.2600 0.1400 X3 2.6:1.5:5 31.0000 65.0000 3.5845 4.6514 0.9040
Fig.1. XRDpatternofglasspowderforsamplesX1,X2andX3.
patternsoftheglasspowdersandsinteredproductswereobtainedusingaBruker D8AdvancedoperatedinBragg–Brentanogeometry,withCuK␣radiation.Count- ingtimewasfixedat71.5sforeach0.03◦2step.TheX-raytubewasoperatedat 40kVand30mA.AllXRDinstrumentparameterssetupwerefixedtostandardvalue duringthemeasurements.Bulkdensityandapparentporosityweremeasuredby Archimedesimmersiontechnique.Microstructuresofthefracturedsampleswere observedusingafieldemissionscanningelectronmicroscopy(VPFESEM-Supra 35VP)at30kV.SpecimensfortheCTEanddielectrictestwerepreparedbycom- pactionoftheglasspowderusingaHydrotexuniaxialpressingmachineat120MPa.
Thesamplesweresinteredinairunderisothermalconditionsfor2hat900◦Cwith a5◦Cmin−1heatingratebeforeCTEanddielectrictesting.TheCTEtestwascar- riedoutusinghightemperatureverticaldilatometertests(ModelLinseis)inair fromroomtemperatureto1200◦C,whilethedielectricmeasurementsweremade usinganImpedanceAnalyzer(HewletPackardmodelHP4291).Thespecimenswere groundandpolisheddowntoa1mdiamondpaste,washedinwaterandacetone andheatedat100◦Cfor2htoremovemoisturepriortothedielectricmeasurement.
Theresultsofdielectricconstantanddielectriclossasafunctionoffrequencycan beobtaineddirectlywhenthesamplethicknessisfedintothemachine.
3. Resultsanddiscussion 3.1. XRDpatternofglasspowder
Fig.1showsalargeamorphoushumpandnocrystallinepeaksin thediffractionpatternsfortheallthreeglasspowdersamples.This indicatesthatallcompoundsinthesampleshavetransformedinto amorphousorglassyphaseafterbeingmeltedat1500◦Cfor4h.
Reductionofintermediateoxide(Al2O3)intheglasscompositions wasproventobeabletoreducethemeltingtemperatureaswell astheviscosityoftheglassat1500◦Ccausingnocrystallinephase toappearinthediffractionpatternofmilledglasspowder.
3.2. Non-isothermalandisothermalDTAanalysis
Fig.2demonstratestheDTAscanoftheglasssampleinanon- isothermalcondition.Onlyasingleexothermicpeakwaspresent ineachsample.TheweakexothermicpeakbelongstosampleX3 withcomposition2.6MgO·1.5Al2O3·5SiO2whichhastheleastMgO concentrationamongthethreestudiedsamples.Thepositionof thecrystallizationpeakalsovarieswithcompositions.Theincrease of MgOconcentrationcaused a shiftof thecrystallization peak maximatolowertemperatures. Thecrystallizationtemperature decreasedfrom925◦Cto876◦Cforsampleswith2.6–3.0MgOcon- centrationasgivenin Fig.3.Thisresultwasinagreementwith experimentsconducted using a seriesof sampleswith thefor- mulationxMgO·2Al2O3·5SiO2 [11].DTAcurvesalsoindicatethat thecrystallizationpeakbecomesmoreintense,andtheareaunder theDTApeakbecomeslargerwithincreasingamountsofMgO,as demonstratedinFig.4.
Fig.2. NonisothermalDTAcurvesofsampleswithvariousMASratios.
Fig.3. CrystallizationtemperatureforsampleswithincreasingMgOconcentration forseriesxMgO·1.5Al2O3·5SiO2.
Althoughthe nonisothermalDTA curvein Fig.2 shows that the crystallization range of samples X1 and X2 ended above 900◦C,prolongedsoakingfor2hwouldbebeneficialtoimprove the crystallization process at intended heat treatmenttemper- atures,as demonstrated inFig. 5. Asshown in Fig.5 complete exothermicpeakswereobservedfor samplesX1andX2inthe isothermalDTAcurveduringheatingat900◦Cfor2h.However, noexothermicpeak was observed for sample X3withcompo-
Fig.4.IntegratedareaundernonisothermalDTApeaks.
Fig.5.IsothermalDTAplotforsamplesX1,X2andX3.
sition 2.6MgO·1.5Al2O3·5SiO2. Although the XRD pattern from compactedglassspecimenof sampleX3demonstratedthat the crystallinepeakappearedafterprolongedsinteringfor2hat900◦C, theintensityofthepeaksarelowercomparedtotheothertwosam- ples.Itwasexpectedforsurfacecrystallizationtohaveoccurredin thiscompactedandsinteredsamplesincenoexothermicpeakwas presentintheDTAcurveofthesameglasspowdersampleheated atthesameheattreatmentparameters.Thecrystallizationof- cordieriteormagnesiumdialuminumsilicatephaseinsampleX3 afterheattreatmentstronglyindicatesthatthecrystallizationpro- cessisrestrictedtotheglasssurface[19,20].Theabsenceofthe exothermicpeakintheDTAcurvewasexpectedtohaveresulted fromthelowerdrivingforceforsurfacecrystallizationoftheloose glasspowdersample(forDTAtest)comparedtocompactedglass powdersample(forXRDtest).Therefore,thenumbersofcrystals nucleatedstronglydependonthechemicalcompositionofthepar- entglassanddonotdependontemperatureandtimeoftheheat treatmentprocess[21].
3.3. Dilatometrycurve
Thedilatometrytestwascarriedoutonun-sinteredsamplesto determinethesinteringreactioncurveasafunctionoftemperature.
Therectangulargreensampleswerenon-isothermallyheatedfrom roomtemperatureto1000◦C.Fig.6showsatypicaldilatometric
Fig.6. Dilatometrycurveofunsinteredcompactedglasspowder.
Fig. 7.Activation energy for densification of glass sample with composition xMgO·1.5Al2O3·5SiO2.
curveforun-sinteredsampleswithdifferentMgOconcentrations.
Itcanbeseenthattheexpansionvaluesforallsampleswithdiffer- entMgOconcentrationsareveryclosetoeachotherattheinitial stagewhenheatedfromroomtemperatureupto775◦Candaccel- eratedat825◦C. Atthistemperaturerange,theviscosityofthe glassisdramaticallydecreased.Allsamplesstopobviousshrink- ingapproximatelyat875◦Cwithasimilarpercentageofshrinkage (12%).Becauseofthesurfacetensioneffectthatvariesasafunction ofsurfacecurvature,materialflowsaredriventowardtheparticle necks,thusdensifyingaglasspowdercompact.
Althoughthesofteningandfinalshrinkagetemperatureswere thesamefor all threesamples, theshrinkagerate ofthe three sampleswasdifferentasafunctionoftemperature.Thiscanbe observedthroughtheslopeattheshrinkagepart.Basedontheslope ofthecurves,sampleX1hasahighershrinkageratefollowedby samplesX2andX3.Thedifferenceinshrinkagerateofsamplesis greatlyrelatedtotheviscosityofglass.Glasswithlowviscosity easilyflowed,causingsampleswithhigherMgOconcentrationsto havehighershrinkagerates.Thus,varyingtheamountofMgOcon- centrationsinnon-stoichiometriccordieritecompositionwithless intermediateoxideswouldalsoaffecttheviscosityoftheglass.
3.4. Activationenergyfordensification
Activationenergy for densification of glass powder samples with different compositions is shown in Fig. 7. The activation energyfordensificationQwascalculatedwhenthesampleexperi- encesshrinkage(between825and875◦C)fromthedilatometry curve using the Arrhenius equation [12]. From the figure, the activationenergyfordensificationtendedtodecreasewithincreas- ing MgO concentrations in the compositions. This result (in xMgO·1.5Al2O3·5SiO2 series)wasingoodagreementwithresults obtainedfrompreviousstudiesin non-stoichiometriccordierite samples in xMgO·2Al2O3·5SiO2 series where Q decreased with increasingMgO[12].However,theactivationenergyfordensifi- cationtendstoplateauabove2.8molMgO.
Fig.8demonstrates plots of densificationand crystallization temperatureas a function of MgO. The range of crystallization temperatureslightlydecreased withincreasingMgOconcentra- tionratio,whilethetemperaturerangefordensificationissimilar for all three samples which is below the range of crystalliza- tiontemperature.Thisindicatesthatdensificationoccurredbefore crystallization.Thegapbetweendensificationandcrystallization temperaturesishighinsampleX3followed bysamplesX2and X1,respectively.AlthoughsampleX1hasthelowestcrystallization temperatureamongthethreesampleswherecrystallizationmay startbelow900◦C,thecrystallizationprocessisdifficulttocontrol
Fig.8.DensificationandcrystallizationtemperatureofsampleswithvariousMAS ratios.
sincethegapbetweendensificationandcrystallizationtempera- tureisveryclose.
3.5. XRDanalysisofsinteredpellet
As shown in Fig. 9, four crystalline phases that were observedinthese3samplesare␣-cordierite(ICSD98-005-8888), forsterite (ICSD 98-008-5876), spinel (ICSD 98-005-8457), and
-cordieriteormagnesiumdialuminiumtrisilicate(ICSD98-000- 770).Althoughnoexothermalpeakwasobservedintheisothermal DTAcurveofsampleX3withcomposition2.6MgO·1.5Al2O3·5SiO2, resultsofXRDpatternonthesinteredcompactedpelletrevealed theappearanceofcrystallinepeaksinthediffractionpattern.Both glasspowder(DTAtest)andcompactedglasspowder(XRD)sam- plesofthesamecompositionexperiencedthesameheattreatment profile,butpressedpellets,incontrasttoloosepowdersamples, tendtocrystallizeduringprolongedholdingatthesameheattreat- menttemperature.Thediffusionandthearrangementofatomsfor crystallizationwouldbeenhancediftheparticlesareverycloseas incompactedpowders.
Asdemonstratedinthediffractionpattern,themaincrystalline phasein sinteredpelletsforallthreesampleswas␣-cordierite.
Itcanbeclearlyseenthattheintensityofthe␣-cordieritepeak
Fig.9.X-raydiffractionpatternsofsampleswithdifferentMASratiossinteredat 900◦Cfor2h(␣:␣-cordierite,f:forsterite,:-cordierite,sp:spinel).
Table4
ResultsofRietveldrefinement.
Sample Phases Quantity(wt%) RBragg Rwp S Rexp Rp
X1 Amorphous 2.30
␣-Cordierite 89.20 12.35 11.31 2.78 6.78 9.19
Forsterite 8.50 33.21
X2 ␣-Cordierite 100.00 11.52 9.55 1.55 6.17 7.50
X3 Amorphous 46.20
␣-Cordierite 45.78 17.54 11.50 3.21 6.41 9.00
-Cordierite 7.59 2.59
Spinel 0.38
RBragg:agreementindicesonBraggvalue;Rwp:agreementindicesontheweighted value;S:goodnessoffit;Rexp:agreementindicesonexpectedvalue;Rp:agreement indicesontheprofilevalue.
becamehigherastheamountofMgOusedwasincreased,andsam- pleX3withchemicalformulation2.6MgO·1.5Al2O3·5SiO2hadthe lowestintensitycomparedtotheother2samples.Smallamounts of secondary phases were observed in samplesX1 and X3. - Cordierite and spinel peaks appeared in sample X3, while the forsteritepeakappearedinsampleX1.Thesescrystallinephasesare normallypresentincrystallizationofMASglass[18,22].Theexis- tenceofthesetypesofsecondaryphaseinsampleswith2.6moland 3molMgOissimilarwiththeseriesofxMgO·2Al2O3·5SiO2samples [11].ResultsofphasequantificationaretabulatedinTable4.
Decreasing the amount of MgO retarded the formation of theforsteritephase,whiledecreasingtheamountofAl2O3from stoichiometric2.8MgO·2Al2O3·5SiO2successfullyretardedthefor- mationofthespinelphase.Singlephase␣-cordieritewasobserved insampleX2withcomposition2.8MgO·1.5Al2O3·5SiO2.Acompar- isonof theintensitycountsof ␣-cordieriteat(022)planealso indicatesthattheintensityofthe␣-cordieritephaseishigherin thesamplewith2.8molofMgO,andtheintensitycountsforthe threesamplesareparallelwiththeamountof␣-cordieritephase presentinthesamplesasshowninFig.10.
Fig.11showsoneoftheRietveldplotsoftheXRDdataforsample X3whichwasheattreatedat900◦Cfor2h.TheRietveldplotsshow agoodqualityofrefinement.Theresultsarenormalizedto100%of crystallinefraction,sothehyphotheticalamorphouscontentofthe samplewasassumedtobenegligible.
Percentageoftheamorphouscontentwascalculatedusingthe Fullprofilepatternmethod.Theresultsofthequantitativephase analysis,includingtheamorphousphase,aretabulatedinTable4.
SampleX3stillcontainsahighamountofamorphousphase.As obtainedinthenon-isothermalDTAtest,thecrystallizationtem-
Fig.10.Intensityof␣-cordieritepeakcountat(022)planeasafunctionofMgO concentration.
Fig.11.RietveldplotofsampleX3.
Fig.12. BulkdensityandporosityforseriesofxMgO·1.5Al2O3·5SiO2samples.
peraturerangeofsampleX3isabove900◦C.Therefore,prolonged holdingofthecompactedpelletupto2hwould noteffectively transformtheglassintoafullycrystallinephase.Thiscomposition requiresamuchhigherheattreatmenttemperaturetoenhancethe crystallizationprocess.Smallamountsofamorphousphasewere
Fig.13. Proportionlineforcalculatingcoefficientofthermalexpansionofsintered rectangularpelletsofsamplesX1,X2,andX3.
Fig.14.CoefficientofthermalexpansionofsamplesX1,X2andX3.
stillpresentinsampleX1.Theexistenceofmulticrystallinephase withdifferentcrystalstructuresinthesamplecausedpartofthe spacebetweenthecrystaltobearrangeddisorderly.
Itisobviousthatwithoutanysinteringaids,excessMgOupto 2.6molwasnotsufficienttoretardtheformationof-cordierite,
Fig.15. LatticechangesasafunctionofMgOconcentrationratio.(a)Latticeaand (b)latticec.
Fig.16.Dielectricconstantofsamplesasafunctionoffrequency.
whileatand/orabove3molMgOwillresultintheformationof aforsteritephase.Therefore,samplesshouldbeformulatedwith approximately2.8mol MgO withless Al2O3 (1.5mol) fromthe exactcordieritestoichiometriccompositioninordertoobtainsin- glephase␣-cordierite.
3.6. Densityandporositytest
Fig.12illustratesbulkdensityandpercentofporosityofsintered pelletasafunctionofMgOconcentrationratio.Thetrendofdensity plotsasafunctionofMgOconcentrationratioistheoppositetothe trendofporosity.SampleX2hasthehighestdensityandlowest percentageofporosityamongthethreesamples.Theformationof asecondaryphasewithdifferentcrystalstructureslimitstheden- sificationprocessandconsequentlyproducessampleswithhigher porosityandlessbulkdensity.
Theformationofmorethanonecrystallinephaseduringthe sinteringprocessinsamplesX1andX3causedthekineticprocess fortheviscousflowtobequiteslowandrequiredmuchenergy tocompletetheprocess.Therefore,samplesX1andX3with8.7 and 14.0wt% secondaryphases, respectively,have lowerdensi- fication thansample X2which hashighest wt% of ␣-cordierite phase.
Fig.17.Dielectriclossofsamplesasafunctionoffrequency.
Fig.18.(a)Microstructureoffracturesurfaceat10,000×magnificationand(b)microstructureoffracturesurfaceat1000×magnification.
3.7. Coefficientofthermalexpansion(CTE)
Thermalexpansionwasmeasuredusingsinteredrectangular pelletsinadilatometrytestmachine.TheCTEwascalculatedin theproportionregionwhichisattemperaturerange100–700◦C.
Therefore, thecurvesof dL/Lo(%)vs T wasreplottedin ascat- terdiagram,asshowninFig.13,toobtainalinearequation.The slopevalueofthecurveswasdividedby100togivetheresults ofCTE.ResultsofCTEforsampleswithincreasinglevelsofMgO arepresentedinFig.14.MostofthesamplesshowlowCTE,i.e.
2–4×10−6K−1.Thelowthermal expansioncoefficientofthe␣-
cordieritesamplesistheresultofitsrelatively‘rigid’tetrahedral framework,andtheanisotropicexpansionoftheoctahedralframe- workofthe␣-cordieritestructure.AsshowninFig.14,CTEplots droppedfromsampleX1toX2,andincreasedforsampleX3.The decrease in CTE values for samplesX1–X2 is associated to the increasesof␣-cordieriteweightpercentageasafunctionofMgO concentration.Theexistenceof-cordierite,spinelandamorphous phase(Table4)wasresponsibleforthehigherCTEinsampleX3, whiletheincreaseinCTEvaluesforsampleX1with3.0molMgO mayhaveresultedfromtheformationofaforsteritephasewhich hashigherCTE.Thecrystallizationprocessgreatlyalteredthether-
malexpansionoftheglasses.Therefore,theglass-ceramicmaterials mayhavehighorlowcoefficientsofexpansiondependingonthe expansioncoefficientsandelasticpropertiesofthecrystalphases formed[23].
InordertoseetheeffectofincreasingMgOinthecrystalstruc- tureof␣-cordierite,preciselatticeparametersweredetermined fromstructurerefinementusingLebail,PawleyandRietveldmeth- odsusingXRDsoftware.Aplotshowingtheeffectofa-andc-lattice parameterswithrespecttoincreasingMgOconcentrationratiois giveninFig.15.
SampleX3withcomposition2.6MgO·1.5Al2O3·5SiO2 hasthe highest lattice a and lattice c values. The expansion of lat- tice a and lattice c was expected due to the existence of higheramounts of secondary phase inside ␣-cordieritecrystals and finally expand the lattice a and lattice c. Volumes of unit cellsforthesesecondaryphases(forsterite289.93×106pm3,- cordierite=124.65×106pm3,spinel=524.39×106pm3)arelower thanthoseofthe␣-cordieritephase(771×106pm3).Therefore, someofthesecondaryphasescouldresideinterstitiallyinbetween the␣-cordieritecrystals,producingahighervolume.
3.8. Dielectricproperties 3.8.1. Dielectricconstant
ThedielectricconstantsofsampleswithdifferentMgOconcen- trationratiosarepresentedinFig.16asafunctionoffrequencyfrom 1MHzto1GHz.Thedielectricconstantdecreased withincreas- ingfrequencyfrom1MHzto1GHzforallsamples.Anincrease infrequencyinducesweakeningordisappearingofionicpolariza- tion,and consequentlydecreasesdielectricconstant values.The dielectricconstantofthesampleswithdifferentMgOconcentra- tionratiosisintherangeof5.3–5.5,fromafrequencyof1MHz to1GHz.Thevaluesareincloseagreementwithotherpublished values.DielectricconstantdecreaseswithincreasingMgO.Reduc- tionindielectricconstant asa functionofMgOin theseriesof xMgO·1.5Al2O3·5SiO2samplesresultedfromphasetransformation andtheamountofporosityinthesample.Althoughthepercentage ofporosityinsampleX3with2.6molMgOisthehighestamong thethreesamples,itpossesseshighdielectricconstant.Thehigh dielectricconstantinsampleX3mayhaveresultedfromhighresid- ualglasscontentinthis sample.Sampleswithhigher glassyor amorphousphasecontainmobileions.Thesemobileionscanjump betweentwositesintheopennetworkstructureofglasses,thus increasingthedielectric constantofthesample. SampleX1has higherdielectricconstantthansampleX2duetoitshigherporosity content.However,thedifferenceistoosmall,andstillwithinthe standardrange.
3.8.2. Dielectricloss
Thedielectriclossasafunctionof frequencyispresented in Fig.17.SampleX1withahigherpercentageofporosityhasthe highestdielectriclossfollowedbysamplesX3andX2.Thedielectric lossofallsamplesis within10−2 whichis slightlyhighercom- paredto reporteddata which used pure oxides asa precursor incordieritesynthesis.␣-Cordieritesynthesizedfrompureoxide generallyhasvery low dielectric loss <10−3 at10MHz[14,24].
Higher concentrations of ␣-cordieritephase do not necessarily havelowloss.Thehighvaluesofdielectriclossinthisstudy,com- paredto␣-cordieritesynthesizedusingpureoxideprecursors,are causedbyincreasedpolarizationdue todefectsinthestructure andimpuritiesormetaloxideatgrainboundaries.Someoxides, especiallythosewhichhaveexcessiveelectricalconductivityatele- vatedtemperatures,causedielectriclossesinceramicmaterials.
Dielectriclossesmayalsobecausedbythedipolemovementof polarizedmoleculesintheelectricalfieldornon-homogeneityof thestructure.
3.9. FESEM/EDXanalysis
MicrostructureoffracturesamplesasdemonstratedinFig.18(a) and(b)clearlyshowsthatsampleX2isfullydensifiedwiththe lowestporositycontent,followedbysamplesX3andX1.Theobser- vation of porosity from a microstructure analysis was in good agreementwith porosity contentmeasured by theArchimedes principle.Percentageofporositysignificantlyinfluencedthebulk densityanddielectricpropertyvaluesofthesamples.
4. Conclusions
All three non-stoichiometric cordierite samples with excess modifyingoxide and less intermediate oxidecan be melted at 1500◦Ctoobtainfullyamorphousglasspowder.Thecrystalliza- tiontemperaturedecreasedasa functionofMgO. Althoughthe softeningtemperatureandtemperatureatwhichthesamplesstop shrinkingarethesameforallthreesamples,samplesshrunkat differentratesduetothevariationin theviscosityof theglass.
The sample withthehighest MgO concentration hasthe high- estshrinkagerate,andviceversa.AlthoughsampleX1withthe highest modifying oxide(MgO) wasexpected to have a better degreeofdensification,thecrystallizationofmulticrystallinephase withdifferentcrystalstructuresinthesamplecausedpercentage ofporositytoincrease.Thepresenceofothercrystallinephases causedintensitycountsof␣-cordieritepeakofsampleX1tobe lowerthanX2eventhoughsampleX1hasthehighestdegreeof crystallinity(measuredfromDTAcurves).Coefficientofthermal expansionof samplesdepends onthetype of crystallinephase whichexistsinthesample.Thesamplewiththechemicalformula- tion2.8MgO·1.5Al2O3·5SiO2possessesthelowestCTEcomparedto theothersamples.Thedielectricconstantdecreasedwithincreas- ingMgO.However,itsvalueswerestillwithinthestandardrange.
Thedielectriclosswasslightlyhigherthan␣-cordieritesynthesized frompureoxideduetotheexistenceofimpuritiesfromthestarting rawmaterials.
Acknowledgements
Theauthorsgratefullyacknowledgethefinancialsupportfrom IslamicDevelopmentBankSaudiArabiaandFundamentalResearch GrantScheme(9003-00171)UniversitiMalaysiaPerlis,andtechni- calassistantsfromUniversitiSainsMalaysia(USM).
References
[1]S.Kurama,H.Kurama,Ceram.Int.34(2008)269–272.
[2]E.Yalamac,S.Akkurt,Ceram.Int.32(2006)825–832.
[3]R.Goren,H.Gocmez,C.Ozgur,Ceram.Int.32(2006)407–409.
[4]R.Goren,C.Ozgur,H.Gocmez,Ceram.Int.32(2006)53–56.
[5]C.Ghitulica,E.Andronescu,O.Nicola,A.Dicea,M.Birsan,J.Eur.Ceram.Soc.27 (2007)711–713.
[6]I.Jankovic-Castvan,S.Lazarevic,D.Tanaskovic,A.Orlovic,R.Petrovic,D.Janack- ovic,Ceram.Int.33(2007)1263–1268.
[7]I. Jankovic-Castvan,S.Lazarevic,B.Jordovic,R.Petrovic,D.Tanaskovic,D.
Janackovic,J.Eur.Ceram.Soc.27(2007)3659–3661.
[8]M.K.Naskar,M.Chatterjee,J.Eur.Ceram.Soc.24(2004)3499–3508.
[9] A.M.Menchi,A.N.Scian,Mater.Lett.59(2005)2664–2667.
[10]Y.Hu,H.T.Tsai,Mater.Chem.Phys.52(1998)184–188.
[11]J.Banjuraizah,H.Mohamad,Z.A.Ahmad,Int.J.Appl.Ceram.Technol.(2010), doi:10.1111/j.1744-7402.2010.02486.x.
[12]J. Banjuraizah, H. Mohamad, Z.A. Ahmad, J. Am. Ceram. Soc. (2010), doi:10.1111/j.1551-2916.2010.04144.x.
[13]P.W.McMillan,in:J.P.Roberts(Ed.),Glass–Ceramics.CrystallizationandDevit- rification,vol.1,AcademicPressInc.,London,1969.
[14]B.H.Kim,K.H.Lee,J.Mater.Sci.29(1994)6592–6598.
[15]T.Rudolph,W.Pannhorst,G.Petzow,J.Non-Cryst.Solids144(1993)273–281.
[16]A.Goel,D.U.Tulyaganov,S.Agathopoulos,M.J.Ribeiro,J.M.F.Ferreira,Ceram.
Int.33(2007)1481–1487.
[17] G.-H.Chen,X.-Y.Liu,J.AlloysCompd.431(2007)282–286.
[18]G.-H.Chen,J.AlloysCompd.455(2008)298–302.
[19]J.Schmelzer,J.Möller,I.Gutzow,R.Pascova,R.Müller,W.Pannhorst,J.Non- Cryst.Solids183(1995)215–233.
[20]S.Reinsch,M.L.F.Nascimento,R.Müller,E.D.Zanotto,J.Non-Cryst.Solids354 (2008)5386–5394.
[21]E.D.Zanotto,J.Non-Cryst.Solids129(1991)183–190.
[22]M.Dittmer,M.Müller,C.Rüssel,Mater.Chem.Phys.124(2010)1083–1088.
[23]S.M.Salman,H.Darwish,E.A.Mahdy,Mater.Chem.Phys.112(2008)945–953.
[24]S.Wang,H.Zhou,L.Luo,Mater.Res.Bull.38(2003)1367–1374.
