Arpakkam, Tamil Nadu, India Characterization and chemometric analysis of ancient pot shards trenchedfrom Technology Journal of Applied Researchand

Availableonlineatwww.sciencedirect.com

Journal of Applied Research and Technology

www.jart.ccadet.unam.mx JournalofAppliedResearchandTechnology14(2016)345–353

Original

Characterization and chemometric analysis of ancient pot shards trenched from Arpakkam, Tamil Nadu, India

Duraisamy Seetha

, Gothandapani Velraj

aDepartmentofPhysics,PeriyarUniversity,Salem636011,TamilNadu,India

bDepartmentofPhysics,AnnaUniversity,Chennai600025,TamilNadu,India Received1February2016;accepted17August2016

Availableonline1October2016

Abstract

ThechemometricmethodbasedontheEDSanalysishasbeenutilizedtoclassifyarcheologicalceramicfragmentsfromArpakkamsite(eight samples)inKanchipuramdistrict.EDSwasusedtodeterminetheirelementalcompositionandtheresultsweretreatedstatisticallyusingtwo methods:factoranalysis(FA)andclusteranalysis(CA).Thistreatmentrevealedtwomaingroups;thefirstoneencompassesonlythetwosamples (ARMP4andARMP8)andthesecondonecomprisestheremainingsamples(ARMP1–ARMP7)(exceptARMP4).Groupingofselectedartifacts wascarriedoutusingtheratiosofSiO2 toAl2O3 concentrations becauseoftheirnon-volatilenature.Themineralogicalcompositionsofthe potteriesweredeterminedbymeansofX-raypowderdiffractionandFT-IRspectroscopy.Thefiringtemperatureandthemethodswereassessed bySEM,FT-IRandXRDspectroscopicanalysis.Thepresentstudiesevaluatethemineralogical,chemicalandmicrostructuralcharacteristicsand thegroupingofpotteries.Thesetechniquesmightbeusefulmethodsforthematerialanalysisofceramics.

©2016UniversidadNacionalAutónomadeMéxico,CentrodeCienciasAplicadasyDesarrolloTecnológico.Thisisanopenaccessarticleunder theCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords:FT-IR;XRD;SEM-EDS;Chemometricanalysis;SiO2/Al2O3

1. Introduction

Clays are generally termed as aluminosilicates and they have a definite crystalline structure. They contain aluminum oxideandsilicondioxideasuniversalmineralsandclayswere classified into phyllosilicates and layers silicates (Bailey &

Brindley,1979).Theyarecomposedprincipallyofsilica,alu- minaandwateroftenwithiron,alkalioralkalineearthmetals (Nguetnkam,Kamga,Villiéras,Ekodeck,&Yvon,2008).

Pottery has been the focus of research from a variety of disciplinaryperspectives,including archeology, fornumerous reasons.Pottery hasbeen producedby humans sinceat least 10,000BC,usuallybythoselivinginsedentarycommunities.

Because it is non-perishable,it tends to be preserved in the archeological record. Moreover, pottery was used by people

∗Correspondingauthor.

E-mailaddresses:[email protected](D.Seetha), [email protected](G.Velraj).

PeerReviewundertheresponsibilityofUniversidadNacionalAutónomade México.

throughoutsociety,notbeingrestrictedtotheupperstratum.It wasalsousedinvariedcontexts:fromadailyresidentialusetoa morespecializedone,suchasmortuarypurposes.Becauseofits plasticityasarawmaterial,claycouldbeformedintoanumber ofdifferentshapes,thusceramicswasusedforamyriadoffunc- tions.Ceramicsisrelatedtothemostubiquitousartifactclass, themajorityofwhichcomprisesvesselsandshards(fragments ofvessels).

Nowadays,amultidisciplinaryapproachisbeingpracticedin mostoftheareasofresearchtohavecomprehensiveinformation.

Itisparticularlyusefulinthearcheologicalinvestigationsasit isthecombinationofartsandscience.Infraredabsorptionbya mineralarisesfromthevibrationsofitsconstituentatoms,and thefrequenciesofthesevibrationsaredependentonthemassof theatoms,therestrainingforcesofthebonds,andthegeometry ofthestructure.Infrared(IR)spectroscopyhasalongandsuc- cessfulhistoryasananalyticaltechniqueandisusedextensively (McKelvyetal.,1996;Stuart,1996).Itismainlyacomplemen- tarymethodofX-raydiffraction(XRD)andothermethodsused toinvestigateclaysandclayminerals.X-raypowderdiffraction patterns(XRD)aretodefinehomogeneousgroupsofclaybased

http://dx.doi.org/10.1016/j.jart.2016.08.002

1665-6423/©2016UniversidadNacionalAutónomadeMéxico,CentrodeCienciasAplicadasyDesarrolloTecnológico.Thisisanopenaccessarticleunderthe CCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

standardinterpretationproceduresofXRD.Theexaminationof fresh-fracturedsurfacesofceramicsamplesunderthescanning electronmicroscope(SEM)providesinformationonthesurface morphologyoftheclaysdevelopedduringfiringandthedegree of vitrification (Maniatis & Tite, 1978,1981). Thisinforma- tioncombinedwithknownmorphologiesofclaysandpottery orbyrefiringsamplesfromtheterracottainthelaboratoryand re-examinationwiththeSEMleadstothedeterminationofthe firing temperatures used inantiquity.Provenance is normally assessedbyelementalanalysesusingEDS(Ravisankaretal., 2014).

The provenance studiesarecarried out intwo waysusing physicalandchemicalmethods.Physicalmethodsexplainthe size, shapeandsurface decoration of artifactsandreflect the ancient cultures (Shepard, 1980; Tykot, 2004). Similarly, the chemicalmethodsareusedfortheidentificationoftrace,minor andmajorelementalcompositionofartifacts.Elementalcom- position plays an important role to provide the provenance studies,usingarcheologicalartifacts.Alkali,alkalineearthele- mentsand rare earthelements (REEs) are used for grouping thearcheologicalartifacts(Beal&Olmez,1997;IAEA,2003).

Elementalconcentrationdatainconjunctionwithstatisticalanal- ysis methods like factor analysis and/or cluster analysis are usedextensivelyforgrouping/provenancestudyofclayceram- ics(Neff,2000;Tykot,2004).Thesechemometricanalysesare helpfultofindthesimilaritiesbetweenthesamples,whichgive meaningful interpretation toarcheological studies(Glascock, Braswell, & Cobean, 1998; Oliveira, Munita, & Hazenfratz, 2010).Thus,thestatisticalanalyseswereperformedforgroup- ing of artifacts by using the WINSTAT and SPSS software packagesrespectively.

2. Materialsandmethods

2.1. Sitedetails–Arpakkam(12^◦49^N79^◦42^E)

Arpakkam is avillage panchayat located inKanchipuram districtofTamilNadustate,India.TheArpakkamsitemapis showninFigure1andthephotographofsamplesisshownin Figure2.Thisplaceislocatedinbetweentworivers,namely, Cheyyar andPalar. The places around Arpakkam,Kanchipu- ram,MamallapuramandUttiramerurarearcheologicallyvery significant for theirPallava rockcuts,structural temples, and epigraphicalremains.ItisaknownfactthatKanchipuramand Mamallapuram were thepolitical capitaland portcity of the Pallavasrespectively,whiletheexistenceofawell-established electoralsystemisendorsedinlithicrecordatUttiramerur.Apart fromthese,theregionisstuddedwithmanyJainacenterswith naturalcavernsonceinhabitedbyJainamonks,containingstone bedsandBrahmiinscriptionsdatabletotheearlycenturiesofthe Christianera(Mahadevan,1994).Fromtheepigraphicalrecords, itisgleanedthatArpakkamwasformedpartofMagaralnaduof EyilKottaminTondaimandalam.ArpakkamwasahubofBud- dhism,Jainism,SaivismandVaishnavismby10thcenturyA.D.

inthismicroregion.

Fig.1.ArcheologicalsitemapofArpakkaminKanchipuramDistrict.

Duringthecourseoftheexplorationundertakeninthevillage, thepresenceofburialurnsinthenorthandnorthwesternparts of the present villageproper was observed. Very nearto the burialsite,inrecentyearsadwellingsitewasidentifiedbythe DepartmentofAncienthistoryandArcheologyduringitsfurther intensive survey. The archeologicalmoundrising toa height of nearlytwometerswastraced.Inthesurfaceofthemound, plenty of all black ware,black andred wareandred slipped wareshardsindifferentshapeswerecollected.Moreover,afew terracottaobjectsandbeadswerealsocollected.Thepresenceof potshardscoupledwithotherartifactswouldverywellendorse thatthearcheologicalmoundwouldcontainculturalproperties, whichwouldhelpintracingthematerialhistoryoftheancient Tamils.Furthermore,theyaretwoofthefewburialanddwelling siteslocatedsidebysideforacomparativestudy.

2.2. Fouriertransforminfraredspectroscopy(FT-IR)

FT-IR spectra were recorded in the mid infrared region (4000–400cm⁻¹)inanevacuatedchamberofaBrukerTensor 27spectrometerusingKBrdiscsasmatrices.Thespectralreso- lutionisof±2cm⁻¹andthespectrawereaccumulatedover16 scans.TheFT-IRspectroscopywasappliedforallthesamples.

ThesamplewasmixedwithKBrintheratioof1:20.Atfirst, thesamplewascrushedandgroundwithKBr.Themixtureof thegroundsampleandKBrwaspelletizedasadisk.Thefitting ofpeaksandsmoothingweredonewiththeOPUS6.5software intheinstrumentovertheworkingwindow.

2.3. X-raydiffraction(XRD)

Forthe X-raydiffractionanalysis,asmall fragmentof the specimen of about 0.2g of the sample wasground manually usinganagatemortar.Thirtymilligramsofthepowderedsam- ple were homogeneously dispersed on a cover glass which was served as a sample holder in the X-ray diffractometer.

ARMP1 ARMP2

ARMP3

ARMP4

ARMP8 ARMP7

ARMP6 ARMP5

Fig.2.ArcheologicalpotshardsofArpakkamsite.

Subsequently, theX-ray diffractionpatterns wereobtained at ambienttemperature withaRigakuminiflexIIdiffractometer usingCuK␣radiationofλ=1.5406 ˚A.Diffractionpatternswere recordedat2θanglesintherangeof20–70^◦.

2.4. Scanningelectronmicroscopyandenergydispersive spectroscopy(SEM-EDS)

ThesampleswerepreparedfortheSEManalysisbysimply crackingtheartifact(freshfractures)andthencoatedwithathin goldlayerinordertoobtainaconductivesurface.Averynarrow beamofelectronsisscannedrapidlybackandforthacrossthe sample.Aselectronshitthesample,somewerescatteredback fromthesurface,andlow-energyelectronswerealsoknocked outofthesample.Theseback-scatteredandsecondaryelectrons wereconvertedintoelectronicsignalswhichgiverisetoimages oftheareascannedonadisplaymonitor.

The morphological investigations were undertaken on a JEOL, JSM-6390 scanning electron microscope at an accel- eratingvoltageof20kVandabeam currentof 1–3nA.SEM images were obtained by either secondary electrons (SE) or back-scatteredelectrons(BSE).Theinstrumentwas equipped withanenergydispersivespectrometer(EDS) system,forthe analysisoftheX-raysemittedbythesampletodeterminethe elementalcompositionforelementalidentificationduringSEM observations.Pure element oxidesand natural minerals were usedasstandardsforthequantitativeanalysis.

2.5. Chemometricanalysis–softwaredetails

Thestatisticalmethodssuchasclusteranalysis(CA)andfac- toranalysis(FA)wereperformedwiththehelpoftheWINSTAT Exceland SPSS16.0 software packages,respectively. Factor analysisandtheclusteranalysisareacommonapproachusedas atooltoexaminegraphicallythegroupingpatternofthesamples intermsofchemicalcomposition.Theauto-scalingprocedure hadbeenperformedtocarryoutmultivariatestatisticalanalysis

4000 3500 1000 500

Transmittance (au)

Wavenumber cm^–1 ARMP 8

ARMP 7

ARMP 6

ARMP 5

ARMP 4

ARMP 3

ARMP 1 ARMP 2

Fig.3.FT-IRspectraofARMPpotshards.

onthedata.Intheclusteranalysis,simplelinkageandEuclidean distancewereusedtogroupingtheartifacts.

3. Resultsanddiscussion

3.1. Fouriertransforminfrared(FT-IR)Analysis

FT-IRspectraofthesamplescollectedfromArpakkamsite areshowninFigure3.Toprovideadequateidentificationofa claymineralbyinfraredspectroscopy,thespectrawererecorded.

Absorptionbandsofadsorbedwatermoleculesappearedat3400 and1630cm⁻¹andC–Habsorptionbandsat2850–2960cm⁻¹, whicharise from organic contaminants. The identificationof clay minerals by FT-IR spectroscopy is greatly promoted by referencespectra(Farmer,1968).

Inthe1100–400cm⁻¹regions,vibrationalbandscorrespond- ing to Si–O, Si–O–Si, Al–O and Al–Si–O were detected in allofthesamples.Thesevibrationalbandscouldbeattributed

FT-IRabsorptionbandsinwavenumber(cm⁻¹)withrelativeintensities Tentativevibrational assignments

ARMP1 ARMP2 ARMP3 ARMP4 ARMP5 ARMP6 ARMP7 ARMP8

3435M 3433M 3433M 3433S 3434M 3433W 3433S 3432S O–Hstr.ofadsorbed

water

– – – – – 2928W C–Hstr.oforganic

contaminants

1880VW 1881VW 1881VW 1880VW 1880VW 1880VW 1880VW 1880VW C–Ostr.ofcalcite

1785VW – – – – – – – C–Ostr.ofcalcite

1627W 1632W 1633W 1634W 1625W 1644W 1633W 1632W H–O–Hbendingof

adsorbedwater

– – – – 1084VS 1080VS – – Al–Si–Ostr.of

amorphous aluminosilicates

1051VS 1052VS 1045VS 1041VS – – – 1043VS Si–Ostr.ofclayminerals

– – – – – – 1036VS – Si–Ostr.ofclayminerals

778M 778M 777M 778M 779M 776S 778M 778S Si–Osymmetricalstr.of

quartz

693W 694VW 693W 692VW 692VW 694VW 692W 694VW Si–Osymmetricalstr.of

quartz

641VW 642W 646VW 640W 642VW 638VW 642W 643VW Al–O–Sistr.offeldspar

584W 584W 581W 580W 584W 584W – 584M Fe–Obendofmagnetite

535W 535W 533M 534M 536W 537W 531M 536M Fe–Obendofhematite

463S 469S 470S 466S 464S 461S 466S 465S Si–O–Sibendingof

silicates VS–verystrong,S–strong,M–medium,W–weak,VW–veryweak.

Table2

EstimatedfiringtemperatureandprevailedatmosphereconditionsofARMPpotshards.

Samplecodes Octahedral sheetstructure (Bandat1100 and915cm⁻¹)

Positionofthe silicateband (cm⁻¹)

Estimated firing temperature (^◦C)

Existenceof iron compounds

Firing atmosphere prevailed

ARMP1 Disappeared 1051 >800 Hematite/magnetite Oxidizing

ARMP2 Disappeared 1052 >800 Hematite/magnetite Oxidizing

ARMP3 Disappeared 1045 <800 Hematite/magnetite Reduction

ARMP4 Disappeared 1041 <800 Hematite/magnetite Oxidizing

ARMP5 Disappeared 1084 >900 Hematite/magnetite Oxidizing

ARMP6 Disappeared 1080 >900 Hematite/magnetite Oxidizing

ARMP7 Disappeared 1036 <800 Hematite/magnetite Oxidizing

ARMP8 Disappeared 1043 <800 Hematite/magnetite Reduction

tothe feldsparsandquartz.Quartzwasobservedthrough the appearanceofthebandsataround776and694cm⁻¹whereas theweakAl–O–Si stretchingoffeldspar inall thepotshards wasindicatedthepresenceoftheabsorptionpeaksintherange of641–646cm⁻¹.Veryweakcalcitebands(around 1880and 1785cm⁻¹)werepresentinallthe samples(Seetha&Velraj, 2015).ThemaximumoftheSi–Obandchangedsystematically withthefiringtemperatureoftheclayandthemainSi–Ostretch- ing bands were located at about 1035–1084cm⁻¹ (Table 1).

AccordingtoShoval(1994),asinglepeakwiththemaximum of 1042cm⁻¹ was foundat700^◦C.In 800^◦C,theband split andtwomaximaappearedat1050cm⁻¹and1078cm⁻¹respec- tively. A single band was observed at1082cm⁻¹ at 900^◦C.

Hence, samples ARMP3, ARMP4, ARMP7 and ARMP8 exhibiting a ν(Si–O) band around 1036–1045cm⁻¹ suggest that the firing temperature was in the range of 700–800^◦C

(<800^◦C)(Damjanovi´cetal.,2011).Conversely,theabsorption spectraofSi–Obandoftheremainingantiquerelicsobserved at 1051, 1052, 1080 and 1084cm⁻¹ for ARMP1, ARMP2, ARMP5andARMP6respectively,indicatesthefiringtemper- atureat900^◦Corevenhigher(>800^◦C).Theestimatedfiring temperatureandprevailedatmosphericconditionsaregivenin Table2.

The FT-IR spectra showed absorptionbands of 466cm⁻¹, 535cm⁻¹ and 584cm⁻¹ (Fig. 3) are due to Si–O–Si bend- ing of silicates, Fe–O bending of hematite and magnetite respectively. The higher amount of magnetite represents a reducing atmosphereandthe higheramountofhematite indi- catesanoxidizingatmosphere.Theconcentrationratioofiron oxides (Fe3O4/Fe2O3) determined the prevailed firing atmo- sphereduringtheproductionofrelics andwasdeterminedby the I/I0 method(Seetha&Velraj, 2015).Iftheintensity ratio

20 30 40 50 60 70

Intensity (au)

2θ in degree

ARMP 8

ARMP 7

ARMP 6

ARMP 5 ARMP 4

ARMP 3

ARMP 2

ARMP 1

Fig.4.XRDpatternsofARMPpotshards.

of magnetite andhematite is less thanone (I/I0<1), thereby containsahighamountofhematiteorotherwise(I/I0>1)itcon- tainsahighamountofmagnetite.Thissuggeststhattheserelics mighthavebeenfiredintheopenair(oxidizingatmosphere)and closedkiln(reducingatmosphere).Fromtheseresults(Table2), itcanbedeterminedthatthe artisansof Arpakkamwerewell awareof boththe techniquesof firingthe potteriesinoxidiz- ingandreducingatmospheresandthatthesamplesmighthave been fired tothe temperature around andbelow 800^◦C.The identificationof clayminerals by FT-IRwere complemented withtheXRDanalysisandwereconfirmed.AccordingtoTite (1995),thefiringtemperatureintheopenaircanchangerapidly fromreducingtooxidizing atmospheresubjecttofluctuations inwindflowandweather.Andtheartifactsareveryrarelyfully oxidized dueto the insufficientfiring time for organic mate- rial within the clay to be burnt out. As discussed by Pavía (2006),coloristemperaturedependent:increasingthetemper- ature turns the clay a shade of red and latterly darkens the color.At 700^◦C and 800^◦C, the clay displays an orange to light red color, whereas above 900^◦C, it turns a deeper red color.

3.2. X-raydiffraction(XRD)

Inordertoidentifyclayminerals,theX-raydiffractograms wererecordedin2θrangeof20–70^◦.Identificationofcrystalline minerals was carried out using the Joint Committee of Pow- der Diffraction Standards (JCPDS, 2003) data bank. Table 3 includesasummaryof themineralassemblageof thepottery samplesandthecorrespondingdiffractionpatternsareshownin Figure4.Quartzandalbitearemajorconstituentsintheselected samples,whereasironoxidessuchashematiteandmagnetiteare alsopresentinallthesamples.Calcite,bytownite,orthoclase, illite, muscovite,high-temperature silicaprobably microcline andmullitewereoccasionallyrecorded.

Themineralogyofthesamplesrevealsthatclayispredomi- nantlynon-calcareousanditisinveteratedbytheEDSanalysis.

However,the pottery occasionally containstraces of primary calciteonlyinfewofthesamples.Thepresenceofsomeofthe

mineralsisaclearindicationofthefiringtemperature.Accord- ing to Peters and Iberg (1978) and Maggetti (1982), calcite decomposesatapproximately600–850^◦Cinnaturalclay,and calcium silicates (diopside–wollastonite) appear in the range 850–900^◦C. But in the present work, calcite was present in fewofthesamplesandtherewasanabsenceofsecondarycal- ciumsilicateswhichindicatesthatthepotteryclaysampleswere non-calcareousinnature.

Thepresenceofmineralssuchasquartz(PDFNo.891961) at 3.35, 1.82, 2.46, 4.28, 1.38, 1.53 ˚A, hematite (PDF No.

898104) at1.69, 1.83 ˚A andmagnetite (PDF No.893854) at 2.95,2.54,1.45 ˚Awereverifiedwiththecorrespondingstandard dataset.Thehightemperaturemineralphasemicrocline(3.72, 3.45, 4.22 ˚A) (PDF No. 831604) and mullite (2.21 ˚A) (PDF No.831881)wereobtainedfromshardsARMP5andARMP6.

Conversely, the remainingrelics ARMP1,ARMP2,ARMP3, ARMP4,ARMP7andARMP8containedlowfiringtemperature mineral,suchascalcitewithalessintensepeak.Theexistence of Illite(PDFNo. 310968)istypically representedbyapeak appearingat2.21 ˚A.Thepresenceofillite/muscoviteindicates that the firingtemperature waslowerthan 900^◦C(Franquelo etal.,2008),whereasthepresenceofhematiteindicatesafir- ingtemperature ofabout850^◦C(Cardianoetal., 2004).This agrees with the firing experiments, which demonstrated that increasingthetemperatureturnstheclayadarkershadeofred (above800^◦C).Thecalcitepeaks2.28,1.98,3.01 ˚Awerealmost observedinallthesamplesexceptinARMP6.Nevertheles,from theFT-IRresults,theweakcalcitebandwaspresentinallthe samples.Thismaybe duetothe poor crystallinityof calcite.

Thefeldspargroupmineralhassimilarcharacteristicsbecause ofasimilarstructure.Thealkalifeldsparsgroupmineralsortho- clase (2.13 ˚A) andanorthoclase(3.25 ˚A)were presentin few ofthesamples.Theplagioclasefeldspargroupmineralsalbite (3.18,4.04,3.21 ˚A)(PDFNo.898575)wasabundantinallsam- plesandbytownite(3.77 ˚A)waspresentinsomeofthesamples (Table3).

3.3. Scanningelectronmicroscopyandenergydispersive spectroscopy(SEM-EDS)Analysis

Inordertoestimatethefiringtemperature,scanningelectron microscopy haslong been used to characterize the morphol- ogyanddegreeofvitrificationofarcheologicalceramics(Tite, Freestone, Meeks, & Bimson, 1982). The degree of particle interconnectionisimportant,whichsuggestsanincreaseinthe firing temperature. Firing inareducing atmosphere, as com- pared toan oxidizingatmosphere,resultedintheloweringof the temperatures at which the various vitrification structures formedatabout50^◦C.ThemicromorphologyandEDSspectra of the selected potshards are showninFigure 5.Concentra- tions of major elements andtrace elements are displayed in Table4.

Inthepresentwork,theisolatedsmooth-surfaceareasorfila- mentsofglassinsomeofthesamples(ARMP1–ARMP4)were appeared.AccordingtoManiatisandTite(1981),thisstructure, whichisreferredtoastheinitialvitrificationstage,wassimi- larforboththenon-calcareousandcalcareousclaysanditwas

Minerals ARMP1 ARMP2 ARMP3 ARMP4 ARMP5 ARMP6 ARMP7 ARMP8

Quartz + + + + + + + +

Albite + + + + + + + +

Microcline − − − − + − − −

Calcite + + + + + − + +

Magnetite + + + + + + + +

Hematite + − + + − + + −

Illite − − − − − + + −

Orthoclase − + + + + + + −

Mullite − − − − + − − −

Bytownite + + − + − − − +

Muscovite − − − − − + − −

Dolomite + − + + + + − +

Dickite − − + − − + − −

+=present;−=absent.

Fig.5.SEMimageswithEDSspectraofARMPpotshards.

Table4

Elementalconcentrations(EDS)ofARMPpotshards.

Nameoftheelements Elementalcomposition–wt(%)

ARMP1 ARMP2 ARMP3 ARMP4 ARMP5 ARMP6 ARMP7 ARMP8

O 51.00 50.39 52.17 54.33 53.60 50.31 57.28 52.17

Na 1.60 1.31 – 1.31 – 1.59 1.34 1.63

Al 9.55 10.02 9.09 7.51 8.47 9.00 8.91 7.55

Si 26.70 27.24 26.11 26.85 27.02 26.58 25.36 27.42

Cl – – – – – – – 0.69

K 1.95 3.03 3.10 1.62 1.75 2.24 1.58 2.86

Ca 2.17 1.40 1.67 1.57 1.66 1.94 1.06 1.73

Fe 7.04 6.61 7.86 6.81 7.48 8.35 4.46 5.95

developed typically at firing temperatures in the 800–850^◦C range in an oxidizing atmosphere. The isolated glassy areas ofthestageincreaseuntiltheycoalescetoformanessentially continuousvitrified layer,which is known as the continuous vitrificationstage appearing in ARMP5.In orderfor contin- uousvitrification(CV)tooccur, thefiring temperatureof the samplemaybebelow1000^◦C.Somefracturesdisplaycontin- uous glass andbloating pores, indicating strong vitrification.

PotterysampleARMP6includesglassyareasassociated with mediumporesCV(MB),whichareformedthroughoutthepot- terybodyatabout1050–1100^◦C(Maniatis&Tite,1981).In somecases,itwaspossibletorecognizeaslightlyearlierstage inthedevelopmentofvitrificationinwhichtherewasnodefi- nitesmooth-surfaceareasofglassbutsomeslightbucklingand roundingoftheedgesoftheclayplates.Thisstructureisrep- resentedbynovitrification(NV),whichappearedinARMP7 and ARMP8. The incomplete dehydroxylation exhibited by thesesamplesintheFT-IRstudyfurthersupportthatthefiring temperature was below 750^◦C.The clay typesand vitrifica- tionstagewithcorrespondingfiringtemperaturearedepictedin Table5.

Calciumisanessentialnutrientthatplaysanimportantrole in determining the physical and chemical characteristics of soil. Clays canbe divided into two main typessuch as non- calcareous(CaO<6%)and calcareous(CaO>6%).From the EDSdata,itshouldbenotedthatallthesamplesanalyzedwere non-calcareous(concentrationofCaOlessthan6%)(Musthafa, Janaki,&Velraj,2010;Velraj,Janaki,Musthafa,&Palanivel, 2009).

Table5

VitrificationstagesandfiringtemperaturerangesofARMPpotshards.

Shardscode CaO% Claytype Vitrificationstage Firingtemperature (^◦C)

ARMP1 3.04 NC IV <750

ARMP2 1.96 NC IV 800–850

ARMP3 2.34 NC IV 800–850

ARMP4 2.10 NC IV 800–850

ARMP5 2.32 NC CV <1000

ARMP6 2.71 NC CV(CB) 1050–1100

ARMP7 1.48 NC NV 800–850

ARMP8 2.42 NC NV <750

3.4. Chemometricanalysis

Thestatisticalanalysissuchasclusteranalysis(CA)andfac- toranalysis(FA)wereperformedforthegroupingofartifacts withthewinSTATandSPSSsoftwarepackages.ElementsO, Na,Al,Si,K,Ca,ClandFepresentintheselectedpotshards wereusedforthechemometricanalysis.

Afactoranalysis(FA)asadatareductiontechniquewasper- formedintwostepswhichincludedtheextractionandrotation offactors.TheFAcanbeexpressedas(Zarei&Bilondi,2013), z_ji=a_f1f1i+a_f2f2i+···+a_fmf_mi+e_fi (1) where,zisthecomponentscore,aisthecomponentloading,f isthefactorscore,eistheresidualtermaccountingforerrorsor othersourcesofvariation,iisthesamplenumberandmisthe totalnumberofvariables.

Therotatedcomponentmatrixvaluesof7elementspresented intheselectedsamplesaregiveninTable6.Asshowninthis table,factor1explains37.91%ofthetotalvarianceinthedataset duetohighpositiveloadingsofelementsSi,CaandFe,whereas factor2explainsthehighpositiveloadingsofelementsAland Kandthetotalvarianceis23.41%.Finally,factor3explainsthe totalvarianceisabout18.04%andthehighpositiveelementis Na.Therearethreefactorsextractedfromthedatasetandthe totalvarianceis79.45%.Accordingtothis, thetotalvariance ismorethan50% whichrevealsthattheprinciplecomponent (PC)fittothedatasetisgood.Inaddition,thefactorscorewas obtainedforthegroupingofartifacts.FromtheFAanalysis,two groupsofsampleswereidentified:groupIconsistingofelements Si,CaandFe,andgroupIIcontainingAlandK.Figure6and7 showthebivariateplotofthreeprincipalcomponentscores.

Table6

FactoranalysisofARMPpotshards.

Variables Factor1 Factor2 Factor3

O −0.804 −0.587 −0.067

Si 0.787 −0.051 0.235

Ca 0.850 0.049 −0.057

Fe 0.735 0.201 −0.555

Al −0.147 0.886 −0.067

K 0.307 0.682 −0.010

Na 0.098 −0.007 0.942

%ofvarianceexplained 37.907 23.407 18.041

–0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6

1 0.5

0 –0.5

–1

Factor 2 (23.41)

Factor 1 (37.91)

Na Si

Ca

O

Fe K

Group I

Fig.6.BivariateplotofFactor1andFactor2ofARMPpotshards.

1.2 1 0.8 0.6 0.4 0.2 0 –0.2

–1 –0.5 0 0.5 1

–0.4 –0.6

Factor 3 (18.04)

Factor 1 (37.91%) Group II

Group K

Ca Si

Fe K

Na

Al O

–0.8

Fig.7.BivariateplotofFactor1andFactor3ofARMPpotshards.

Linkage distance ARMP8

0

10

20

30

40

50

60

ARMP4 ARMP2 ARMP7

Sample Code

ARMP6

Group II

ARMP5

Group I

ARMP3 ARMP1

Fig.8.ClusteranalysisdendrogramfortheARMPpotshards.

Intheclusteranalysis,simplelinkagemethodsandEuclidean distance were used and two groups were obtained: (i) ARMP1–ARMP3, ARMP5–ARMP7 (ii) ARMP4, ARMP8.

Thedendrogramof Arpakkampotshards,showninFigure8, showsthatthetwogroupsofsamplesintheselectedsitearein goodagreementwithFA.

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

ARMP1 ARMP2 ARMP3 ARMP4 Sample code

ARMP5 ARMP6 ARMP7 ARMP8 SiO2 /Al2 O3

Fig.9.SiO2/Al2O3ofARMPpotshards.

The Si and Al concentration ratio values are expected to remain the same in fired and raw clay source because of their non-volatile nature. Most of the clays can be describedashydrousaluminumsilicateswithgeneralformulaof Al2O3·2SiO2·2H2O.Thus,forobtaininginformationonprove- nance/groupingstudiesofpotteries,theconcentrationratiosof SiO2toAl2O3 wereplottedfor allpotterysamples asshown inFigure 9.It wasalso confirmedthat theselected sitesam- pleswereseparatedastwogroups,whichmayduetodifferent sources.

4. Conclusion

Spectroscopic techniques coupled with chemometric tools FAandCAwereusedtocharacterizetheselectedclaypotter- ies.Accordingtothevibrationalassignmentsandthemineralogy performedbyFT-IRandXRD,thefiringtechniquesoftherelics during manufacturewere obtained. Both of these techniques were wellconcurringwitheachother.Furthermore, theSEM analysiswascarriedouttoestimatetherangeoffiringtemper- aturesisalsoingoodagreementwiththe resultsof theabove techniques.Fromalltheaboveresults,lowfiring(<800^◦C)and high firing (>800^◦C)temperatures were obtained inthe oxi- dizingandreducingatmosphere.Hence,theresultsoftheEDS analysis indicate that some of the majorand minorelements suchasO,Na,Al,Si,Cl,K,FeandSi,Al,Nawereinvestigated.

ElementsSi,AlandNawerenon-volatileinnatureandwere usedforthepreliminarygroupingofartifacts.Thechemometric tool,i.e.,CAshowedthat thereare twodistinctgroupsinthe selectedartifacts.Thisindicatesthatthepotteriesaremadeout ofclaysofdifferentcomposition;theseresultsareinagreement withtheresultsofthefactoranalysis.

Conflictofinterest

Theauthorshavenoconflictofinteresttodeclare.

Acknowledgements

TheauthorsaredeeplythankfultotheDepartmentofAncient HistoryandArcheology,UniversityofMadras,Chennaiforpro- vidingthearcheologicalsamplesandtheinstrumentationCenter, KarunyaUniversityfortheinstrumentationfacilityofSEMwith EDSanalysis.Theoneoftheauthorsarealsoacknowledgewith thankstothePeriyarUniversityforthefinancialassistancepro- videdthroughUniversityresearchfellowship(URF)andforthe financialsupportprovidedthroughmajorresearchprojectinthis fieldbyDAE-BRNS,NewDelhi.

References

Bailey,S.W.,&Brindley,G.W.(1979).Structureofbasicclaymineralogy groups.ClayMineral,27,238–239.

Beal,J.W.,&Olmez,I.(1997).Provenancestudiesofpotteryfragmentsfrom medievalCairo,Egypt.JournalofRadioanalyticalandNuclearChemistry, 221(1–2),9–17.

Cardiano,P.,Ioppolo,S.,DeStefano,C.,Pettignano,A.,Sergi,S.,&Piraino, P.(2004).Studyandcharacterizationoftheancientbricksofmonastery of“SanFilippodiFragalà”inFrazzanò(Sicily).AnalyticaChimicaActa, 519(1),103–111.

Damjanovi´c,L.,Holclajtner-Antunovi´c,I.,Mioˇc,U.B.,Biki´c,V.,Milovanovi´c, D.,&Evans,I.R.(2011).Archaeometricstudyofmedievalpotteryexca- vatedatStari(Old)Ras,Serbia.JournalofArchaeologicalScience,38(4), 818–828.

Farmer,V.C.(1968).Infraredspectroscopyinclaymineralstudies.ClayMin- erals,7(4),373–387.

Franquelo,M.,Robador,M.,Ramírez-Valle,V.,Durán,A.,JiménezdeHaro, M.,&Pérez-Rodríguez,J.(2008).Romanceramicsofhydraulicmortars usedtobuildtheMithraeumhouseofMérida(Spain).JournalofThermal AnalysisandCalorimetry,92(1),331–335.

Glascock, M.D.,Braswell, G.E.,& Cobean,R.H.(1998). Asystematic approachto obsidiansource characterization. InM. S.Shackley(Ed.), Archaeologicalobsidianstudies(pp.15–65).NewYork:PlenumPress.

IAEA-416.(2003).Nuclearanalyticaltechniquesinarchaeologicalinvestiga- tions.TechnicalReportsSeriesNo.416,Vienna,Austria.pp.3–188.

MineralPowderDiffractionFile.JointCommitteeonPowderDiffractionStan- dards(JCPDS).2003.

Maggetti,M.(1982).Phaseanalysisanditssignificancefortechnologyand origin.InJ.S.Olin,&A.D.Franklin(Eds.),Archaeologicalceramics(pp.

121–133).Washington,DC:SmithsonianInstitutionPress.

Mahadevan,I.(1994).RecentdiscoveriesofJainacaveinscriptionsinTamil Nadu.pp.116–134.NewDelhi:RishabhSaurabh.

Maniatis,Y.,&Tite,M.S.(1978).CeramictechnologyintheAegeanworld duringtheBronzeAge.TheraandtheAegeanWorld,1,483–492.

Maniatis,Y.,&Tite,M.S.(1981).TechnologicalexaminationofNeolithic- BronzeAgepotteryfromcentralandsoutheastEuropeandfromtheNear East.JournalofArchaeologicalScience,8(1),59–76.

McKelvy,M.L.,Britt,T.R.,Davis,B.L.,Gillie,J.K.,Lentz,L.A.,Leugers, A.,etal.(1996).Infraredspectroscopy.AnalyticalChemistry,68,93–160.

Musthafa, A.M., Janaki,K.,& Velraj,G. (2010).Microscopy, porosime- try and chemical analysis to estimate the firing temperature of some archaeologicalpotteryshredsfromIndia.MicrochemicalJournal,95(2), 311–314.

Neff,H.(2000).Neutronactivationanalysisforprovenancedeterminationin archaeology.InE.Ciliberto,&G.Spoto(Eds.),Modernanalyticalmethods inartandarchaeology(pp.81–134).NewYork:JohnWiley.

Nguetnkam,J.P.,Kamga,R.,Villiéras,F.,Ekodeck,G.E.,&Yvon,J.(2008).

AssessingthebleachingcapacityofsomeCameroonianclaysonvegetable oils.AppliedClayScience,39(3),113–121.

Oliveira,P.M.S.,Munita,C.S.,&Hazenfratz,R.(2010).Comparativestudy betweenthreemethodsofoutlyingdetectiononexperimentalresults.Journal ofRadioanalyticalandNuclearChemistry,283(2),433–437.

Pavía,S.(2006).Thedeterminationofbrickprovenanceandtechnologyusing analytical techniques from thephysical sciences. Archaeometry, 48(2), 201–218.

Peters,T.,&Iberg,R.(1978).Mineralogicalchangesduringfiringofcalcium- richbrickclays.CeramicBulletin,57(5),503–509.

Ravisankar,R.,Naseerutheen,A.,Chandrasekaran,A.,Bramha,S.N.,Kana- gasabapathy,K.V.,Prasad,M.V.R.,etal.(2014).EnergydispersiveX-ray fluorescenceanalysisofancientpotteriesfromVelloreDistrictTamilnadu, Indiawithstatisticalapproach.JournalofRadiationResearchandApplied Sciences,7(1),44–54.

Seetha, D.,& Velraj,G.(2015). Spectroscopic and statisticalapproach of archaeologicalartifactsrecentlyexcavatedfromTamilnadu,SouthIndia.

SpectrochimicaActaPartA:MolecularandBiomolecularSpectroscopy, 149,59–68.

Shepard,A.O.(1980).Ceramicsforthearchaeologist(5thed.).Washington, DC:CarnegieInstituteofWashington.

Shoval,S.(1994).ThefiringtemperatureofaPersian-periodpotterykilnat TelMichal,Israel,estimatedfromthecompositionofitspottery.Journalof ThermalAnalysisandCalorimetry,42(1),175–185.

Stuart,B.(1996).Moderninfraredspectroscopy.NewYorkandChichester,UK:

JohnWiley&Sons.,180pp.

Tite,M.S.(1995)..pp.37–42.Firingtemperature—Howandwhy?(34)Stock- holm:KVHAAKonferenser.

Tite,M.S.,Freestone,I.C.,Meeks,N.D.,&Bimson,M.(1982).Theuseof scanningelectronmicroscopyinthetechnologicalexaminationofancient ceramics.InArchaeologicalceramics.pp.109–120.SmithsonianInstitution Press.

Tykot,R.H.(2004).Scientificmethodsandapplications.InM.Martini,M.

Milazzo,&M.Piacentini(Eds.),InternationalSchoolofPhysics“Enrico Fermi”CourseCLIV.IOSPress.

Velraj, G.,Janaki,K., Musthafa,A. M.,& Palanivel, R. (2009).Spectro- scopicandporosimetrystudiestoestimatethefiringtemperatureofsome archaeological potteryshredsfrom India. AppliedClay Science, 43(3), 303–307.

Zarei, H.,& Bilondi, M.P. (2013). Factor analysisofchemical composi- tionintheKaroonRiverbasin,southwestofIran.Appl.WaterSci.,3(4), 753–761.