230P.K.CHEOAND C.H.WANG00=2~is about500W/cm.ThemeasuredIpandIsfor80= 0.46~are26and6W/cm,respectively.C.K.N.PatelandR.E.Slusher,Phys.Rev.Letters20,1087(1968).R.L.AbramsandA.Dienes,Appl.Phys.Letters~14237(1969).0=k/n(v),where&istheslopeof the linearplotand(e)theaveragethermalvelocityoftheSF&gas.In Bef.14,&wasdefineddifferentlyfrom oursbyafactor of2~. The cross sectionobtained inRef.14is1.2&10 cmPHYSICALBEVIEWAVOLUME1,NUMBER2IEBBUABY1970Ni LSelf-AbsorptionSpectrum*~D. ChopraPhysicsDepartment,EastTexasState University,Commerce,Texas75428(Received18July1969)0The shapeandpositionoftheNiL e x-rayemissionline(~=14.53A)has beenstudiedas afunction of bombardingelectronenergyrangingfrom2to30 keV.The bombardingelectronsstruck theanodeatnormal incidence,andxrayswereviewedbythespectrometerat a90'takeoffangle.The NiLnlinesrecordedundertheseconditionsexhibitan increasingatten-uation of intensityonthe high-energysideanda shiftingofthepeakpositiontowardslowerenergywith increasingbombardingelectronenergy.These changesare interpretedas due todifferentialself-absorptionof theemission linein theanode.Theemissionmeasurementshave beenusedtoobtain the"I.self-absorptionspectrum" ofNi,whichagreesfavorablywiththe absorptionspectrumreportedin theliterature.INTRODUCTIONEXPERIMENTALTheNiLax-rayemissionlinerepresentsthetransitionsofvalenceband(Mqvv)electronstovacanciesintheLzzzshellsofNiatoms.Itisbe-lievedbysomeauthorsthatifthecombinedwidthofthe¹iLzzzstatesandthespectral-windowfunc-tionoftheinstrument whichrecordsthe spectrumissmallcomparedwiththewidthofthe Nivalenceband,onecanlearnfromtheshapeoftheobservedNiLzlinethedensityof filledvalencestatesin Ni.Forpurposesofgainingsuchinformation,variousauthors'"haverecordedthe¹iL+line.It is asignificantfactthatalmostnoagreementexistsre-gardingthedetailedfeaturesofthe line andof thelocationof itspeakposition.The differingexper-imentalresults andthe attemptedexplanationsofvariousauthorssuggestedthatasystematicexper-imentalinvestigationof theNiLoemissionline asafunction of bombardingelectronenergyrangingfrom2to 30keVmightclarifytheunsettledpointsin theinterpretationofx-rayspectra.Asaresultofthe presentstudy,it hasbeenfound thatthechangesintheshapeandpositionof theNi Lnline arerelatedto theself-absorptionofradia-tionbythematerial of theanode.Thedata on theNiLex-ray emissionlinespectrawere recordedwith the two-crystal vacuum x-rayspectrometerat NewMexico StateUniversity.Potassium acid phthalate(KAP)crystals(2d=26.6kxu)were used asdiffracting elements,which weregrownfrom asupersaturatedwatersolutioninthislaboratory.The excellent spectrometricprop-erties of KAPcrystalsfor thewavelengthrangeofthepresentinvestigationhavealreadybeenestab-lishedThespectral window, impliedbythe width ofthe(l,—l)curveattheNiLo,wavelength,is=0.6eV(resolving power Xjd&-3000),whichisgoodenoughtoreproducesatisfactorilythe structure on thehigh-energysideof theNiLalinepreviouslyre-ported.The data wereobtainedbytheoperationof the two-crystalspectrometerinadouble-rota-tiontrackingmode."Theterm"double"impliesthatbothcrystalsarerotated forscanningthespectrum,ascomparedto the rotationof the first(central)crystalonlyin the caseofthesingle-ro-tationmode.In the methodof doublerotation, xrayswhichpassthroughtheinstrumentoriginatefrom thesameregionof theanode,strikethesameNi.LSELF-ABSORPTIONSPECT RUMtwo areasof thecrystals,and enter thedetectoratthe samepositionforeverysettingof thespec-trometer.Thus,distortionsresultingfrom pos-sible nonuniformities of the sourceorinhomoge-neities of the crystalsareeliminated.The effectsof beamwalkingwerenoted in the tailregionsoftheNiLnline wheretheintensityinthe single-rotationmode fallsbelowthatobtained inthedou-ble-rotationmode.Thedetectionsystemusedinthe investigationconsistedofaflowing-gas(P-10) proportionalcounterfollowedbyalinearamplifier,asingle-channelpulse-height analyzer,andanelectronicsealerandtimer.Thetargetself-absorptionplaysa significantrole in determiningthespectrumofvalence-bandelectrontoinner-shell holetransitionsinmetals.Thiseffectiscrucialfor the Ni Lemissionspec-trum wherethe absorptioncoefficientin thevicin-ityof theLz~qedgeundergoeslargefluctuationsbecauseoftherelativelynarrow andpartlyemptySdband.'4Forthisreason,one ofthemajorcon-cerns indesigningthe experimentwastheexplicitdeterminationof the effectsoftargetself-absorp-tion.This wasaccomplishedbyemployingnormalincidenceforthe bombardingelectronsandapprox-imatelya90'takeoffanglefor thexrays.Ade-mountablecontinuouslypumpedx-raytubewasused.Thecondensationofpumpoilvaporsandfil-ament evaporantson thesurfaceof theanodepro-ducesserioustargetcontamination.Toavoidthis,the filaments werescreenedfrom the anodebywater-cooled semicylindricalshieldsto avoiddi-rectcontamination ofthe anodebyfilamentevap-orants.At low bombardingelectronenergies(=2 keV)thecontamination ofthe anode provedtobedisastrous,evenat such lowoperatingpres-sures as1&&10-'Torr. Since thecountingrateswerequitelow for low bombardingelectronener-gies,the additionalintensityreduction due totar-getcontaminationpreventedsuccessfulexperi-ments. Anattemptwas made to avoid thiscon-taminationbythe useofahotanode. Thex-raytube wasoperatedso that the thinsheet(20milthick)ofelectropolishedgradeAnickelappearedred hot(T=800'C).However,thetargetwasnotso hot that the x-ray tubewalls showedanynotice-abletraces ofevaporatedNi. Thehottargetwasusedsuccessfullyas astable sourceofradiationfor bothlow andhighbombardingelectronenergies.In orderto transmitas much radiationaspossible,thinvinylplastic"windowswithan x-raytrans-missionofabout80%atthe NiLn wavelengthwereusedforthex-ray tube.and the detector.Separatevoltageregulatedpowersupplieswereemployedfor the x-ray tubeand detector.RESULTSTheresultsofthissystematic studyof theNi Lnlineshapeandpositionasafunction ofbombardingelectronenergyare presentedinFig.1.Thelinewasrecorded atdifferent bombardingelectronenergiesin therangeof 2 to 15 keV. Theordinatevaluesrepresenttheintensityof the¹iLeradia-tionwhichwasdeterminedto19~orbetter atthelinepeakin allthe measurements. Theback-groundcountingratefortheproportionalcount-er was 10 countspermin. Thedata havebeennormalizedtothe samepeakintensity.Thezeroofenergyisarbitrarilychosentobe atthepeakofthe2-keVcurve.Sincetheshapeof the line wastheprimeinterest of thisinvestigation,nowave-lengthcalibrationwasattempted.t IIII I II I IIII.O—& 04—WOCFIG.1.Ni Inlaneprof&lesforseveralbombardingelectronenergies.0.2II-8-6-4 -2024RELATIVEENERGY,ELECTRONVOLTS232D. CHOPRAIt isevident