Available online at www.sciencedirect.com GeochimicaetCosmochimicaActa73(2009)183–196 www.elsevier.com/locate/gca The influence of artificial radiation damage and thermal annealing on helium diffusion kinetics in apatite David L. Shustera,*, Kenneth A. Farleyb aBerkeleyGeochronologyCenter,2455RidgeRoad,Berkeley,CA94709,USA bDepartmentofGeologicalandPlanetarySciences,MS170-25,Caltech,Pasadena,CA91125,USA Received15July2008;acceptedinrevisedform10October2008;availableonline21October2008 Abstract Recentwork[ShusterD.L.,FlowersR.M.andFarleyK.A.(2006)Theinfluenceofnaturalradiationdamageonhelium diffusionkineticsinapatite.EarthPlanet.Sci. Lett.249(3–4),148–161]revealingacorrelationbetweenradiogenic4Hecon- centrationandHediffusivityinnaturalapatitessuggeststhatheliummigrationisretardedbyradiation-induceddamagetothe crystalstructure.Ifso,theHediffusionkineticsofanapatiteisanevolvingfunctionoftimeandtheeffectiveuraniumcon- centrationinacoolingsample,afactwhichmustbeconsideredwheninterpretingapatite(U–Th)/Heages.Herewereportthe resultsofexperimentsdesignedtoinvestigateandquantifythisphenomenonbydeterminingHediffusivitiesinapatitesafter systematically adding orremovingradiation damage. Radiationdamagewasaddedtoasuiteofsyntheticandnaturalapatitesbyexposuretobetween1and100hofneutron irradiationinanuclearreactor.Thesampleswerethenirradiatedwitha220MeVprotonbeamandtheresultingspallogenic 3Heusedasadiffusantinstep-heatingdiffusionexperiments.Ineverysample,irradiationincreasedtheactivationenergy(E ) a andthefrequencyfactor(D /a2)ofdiffusionandyieldedahigherHeclosuretemperature(T)thanthestartingmaterial.For o c example,100hinthereactorcausedtheHeclosuretemperaturetoincreasebyasmuchas36(cid:2)C.Foragivenneutronfluence themagnitudeofincreaseinclosuretemperaturescalesnegativelywiththeinitialclosuretemperature.Thisisconsistentwith alogarithmicresponseinwhichtheneutrondamageisadditivetotheinitialdamagepresent.Indetail,theirradiationsintro- ducecorrelatedincreasesinE andln(D /a2)thatlieonthesamearrayasfoundinnaturalapatites.Thisstronglysuggeststhat a o neutron-induced damagemimics the damageproduced byU andTh decayin natural apatites. To investigate the potential consequences of annealing of radiation damage, samples of Durango apatite were heated in vacuum to temperatures up to 550(cid:2)C for between 1 and 350h. After this treatment the samples were step-heated using theremainingnatural4Heasthediffusant.Attemperaturesabove290(cid:2)CasystematicchangeinT wasobserved,withvalues c becominglowerwithincreasingtemperatureandtime.Forexample,reductionofT fromthestartingvalueof71to(cid:2)52(cid:2)C c occurredin1hat375(cid:2)Cor10hat330(cid:2)C.TheobservedvariationsinT arestronglycorrelatedwiththefissiontracklength c reduction predicted from the initial holding time and temperature. Furthermore, like the neutron irradiated apatites, these samples plot on the same E (cid:3)ln(D /a2) array as natural samples, suggesting that damage annealing is simply undoing the a o consequences of damageaccumulation in terms ofHediffusivity. Takentogetherthesedataprovideunequivocalevidencethatattheselevels,radiationdamageactstoretardHediffusionin apatite,andthatthermalannealingreversestheprocess.Thedataprovidesupportforthepreviouslydescribedradiationdam- agetrappingkineticmodelofShusteretal.(2006)andcanbeusedtodefineamodelwhichfullyaccommodatesdamagepro- duction andannealing. (cid:3)2008Elsevier Ltd. Allrights reserved. * Correspondingauthor. E-mailaddress:[email protected](D.L.Shuster). 0016-7037/$-seefrontmatter(cid:3)2008ElsevierLtd.Allrightsreserved. doi:10.1016/j.gca.2008.10.013 184 D.L.Shuster,K.A.Farley/GeochimicaetCosmochimicaActa73(2009)183–196 1. INTRODUCTION log (kerma) (log(MeV/g)) 10 14 15 16 17 18 Application of (U–Th)/He cooling ages to the solution ofgeologicproblemsrequiresthatheliumdiffusionkinetics 90 hr 140 quantified in the laboratory be accurately extrapolated to 10 hr conditions of interest in nature, specifically to lower tem- 0 hr 1 hr peratures and over geologic timescales. The recent sugges- 120 tions that naturally occurring radiation damage modifies He diffusion in apatite (Farley, 2000; Crowley et al., 2002; 100 Green et al., 2006; Green and Duddy, 2006; Shuster C) 90-100hr etal.,2006)underscorethechallengeofthisextrapolation: o(C helium diffusivity is anevolving function of time governed T 80 00 hhrr by both temperature and effective uranium concentration (eU)(Shusteretal.,2006;Flowersetal.,2007).Inthisstudy 60 wepresentresultsofcontrolledexperimentsinwhichweex- ploretheresponseofHediffusioninapatitetoincreasesin 0 hr crystal damage caused by artificial irradiation, and to 40 0-1hr annealing of damage at elevated temperatures. In a subse- quent publication, we will investigate the implications of -2 -1 0 1 2 3 theseobservations forapatite Hethermochronometry. log ([4He] ) (log(nmol/g)) 10 eq The interactions between radiation and crystalline mat- ter involve ionizations, nuclear transmutations and direct Fig. 1. Neutron irradiation experiments: Helium closure temper- scattering displacements of constituent atoms (Ewing ature (Tc) versus the log of the concentration of equivalent 4He etal.,1995;Weberetal.,1998).Basedonthetotalnumber (log10([4He]eq) and the total kinetic energy released into matter of atomicdisplacements perunit time,the mostimportant (log10(kerma)), as described in the text. Values of Tc were calculated for a cooling rate of 10(cid:2)C/Ma using the formulation of these interactions for He diffusion in minerals is likely of Fechtig and Kalbitzer (1966) for diffusion kinetics determined from alpha recoil of heavy nuclei in the 238U, 235U, and from stepwise release fractions of proton-induced 3He or natural 232Th decay chains (Weber et al., 1997; Trachenko et al., radiogenic 4He. We estimated uncertainty in T (1r; shown as c 2002),butmayalsoincludetheeffectsofthealphaparticles vertical lines) solely from the linear regression statistics from the themselves and spontaneous U fission. Regardless of Arrheniusplots.Theopen,light-graycirclescorrespondtonatural source, the resulting displacements (Ewing et al., 1995) apatites and the solid black curve is the ‘‘trapping model” from canmodifytheenergeticenvironmentalongdiffusionpath- Shuster et al. (2006). The gray circles plotting between T (cid:2)120 c waysandmayeitherimpedeorpromotediffusion.Inasuite and140(cid:2)Caretheresultsofneutronirradiationexperimentsona ofnaturalapatites,Hediffusivitiesbelow(cid:2)200(cid:2)Cdecrease synthetic fluorapatite sample shown in Fig. 2a. The dark gray points (0 and 1h irradiations) and black points (90 and 100h with increasing radiation damage (Shuster et al., 2006). irradiations)correspondtothefiveneutronirradiationexperiments Thisleadstotheideaofaradiationdamage‘‘trap”,alocal conductedonasuiteofnaturalapatitessummarizedinTable1and damaged region of the lattice in which He is preferentially showninFig.2bandcandFig.EA1.Uptriangles=CJ50,down sited. Diffusion through a radiation-damaged crystal in- triangles=00mr18, circles=DYJS5, diamonds=01mr59, volvestwoenergeticterms:onetoescapethedamagedsite, squares=Durango.Timescorrespondtonumberofhoursinthe and another to migrate through the undamaged lattice. CLICITfacilitywithinthenuclearreactor(seetext). This model predicts a log-linear relationship between the volume fraction of radiation damage in an apatite and theheliumclosuretemperature;(Shusteretal.,2006)docu- ogical composition. Limited work on apatite (Farley, mented such a correlation in a suite of natural apatites 2000) suggests that annealing enhances He diffusion, as using 4He as a proxy for radiation damage (Fig. 1). These wouldbeexpectediftheradiation damagetrapsare elimi- dataprovidedthecalibrationforakineticmodelforHedif- nated by this process. At present we have insufficient data fusion that includes both temperature and radiation dam- toquantifyhowannealinginfluencesHediffusioninapatite ageas independentvariables (blackcurve in Fig.1). oranyothermineral.Theradiationdamagetrappingmodel At high enough radiation doses many solids approach of Shuster et al.(2006) included this effectonlyin an indi- an aperiodic or glasslike state. Previous work shows that rect manner. zirconsandtitanitesthatreachsuchastatediffuseHevery For thermochronometry, we are most interested in the rapidly(Hurley,1952,1954;Nasdalaetal.,2004).Thefact extensivepropertiesofthedamagedmaterial,i.e.,theeffec- thatapatitetendstohavelowerconcentrationsofUandTh tivebulkdiffusivityofthematerialatagivenpointintime. and has a greater propensity to anneal radiation damage Ourgoalinthisworkistoconfirmandquantifytheeffects may prevent it from reaching this condition, although at of damage accumulation and annealing on the kinetics of presentthereare insufficient datato know. heliumdiffusion,ratherthanilluminatedetailsattheatom- Annealingorreversionofdamagedsitesbacktoacrys- icscale.Ourapproachistwofold.First,tosimulatetheef- tallinestateoccursinmanysolids.Therateofthisprocess fects of natural radiation damage we subjected apatites to dependsonmanyvariablesincludingtemperature,thevol- neutron irradiation, a technique by which to induce vari- ume fraction of damaged sites, and chemical and mineral- ableandcontrolledamountsoflatticedamagecomparable Heliumdiffusionkinetics 185 to that produced by alpha recoil. Second, we subjected a held within the cadmium lined in-core irradiation tube natural apatite to a suite of experimental conditions at (CLICIT) facility of the Oregon State University TRIGA which lattice annealing is known to occur to varying ex- nuclear reactor for 1, 10, 90, or 100h. The samples tents.Afterthesetreatmentswemaderoutinestepped-heat- were subjected to a 1MeV equivalent neutron fluence ing measurements onthe samples to quantifythe resulting (U ) equal to (cid:2)2.3(cid:4)1016, (cid:2)2.7(cid:4)1017, (cid:2)2.0(cid:4) eq, 1MeV, Si changes in He diffusion kinetics. Throughout this paper, 1018, and (cid:2)2.7(cid:4)1018n/cm2, respectively. We calculated we focus on the closure temperature (T) as a parameter these values of U from an external calibration c eq, 1MeV, Si whichencapsulatesthecombinedeffectsofmodifyingboth oftheCLICITfacility(referencedtosilicon)usingconven- E andD /a2inourexperiments.However,ultimatelyitis tional techniques (ASTM, 1994). The energy spectrum a o thesetwofundamentalparametersofdiffusion(ratherthan within the CLICIT is dominated by moderate energies; T) that are both directly observable in an Arrhenius plot roughly50%oftheincidentneutronshaveenergybetween c andmostrelevantforquantifyingHediffusivityonlabora- 1 and 4MeV (Figs. EA3 and EA4). We expect the higher tory andgeologic timescales. energy neutrons to induce a greater proportion of damage thanthoseatlowerenergies.Forthe90and100hirradia- 2. SAMPLES ANDMETHODS tions, the samples were not continuously exposed to the neutron flux. The 90h irradiation was conducted in (cid:2)33 2.1.Samples consecutive steps between January 23 and February 23, 2006,andthe100hirradiationwasconductedin21consec- The well-studied helium diffusivity (Zeitler et al., 1987; utive steps betweenNovember 29andJanuary 3, 2007. Wolf et al., 1996; Farley, 2000) and annealing behavior Following the neutron irradiations, these samples were (Green, 1988; Carlson et al., 1999; Ravenhurst et al., subjectedtoprotonirradiationatTheFrancisH.BurrPro- 2003)ofthefluorapatitefromCerrodeMercado,Durango, ton Therapy Center at Massachusetts General Hospital to Mexico (Young et al., 1969) make it a logical choice to produce a uniform spatial distribution of 3He using previ- investigatehere.OurspecimenofDurangoconsistsoffrag- ouslypublishedprocedures(Shusteretal.,2004).Forthese mentsproducedbycrushingaslabcutfromtheinteriorofa experimentsweuseda220MeVprotonbeamandafluence large, gem-quality megacryst. Two different grain sizes of of(cid:2)5(cid:4)1016p/cm2.Theirradiationstookplaceoveracon- Durango were analyzed. Most experiments were done on tinuous(cid:2)8hperiodoneitherApril30,2006orOctober21, shards of a standard we call CIT Dur-B, sieved to a mean 2007. cross section of (cid:2)170lm. This fraction is from the same Note that the Durango aliquots used in the annealing megacrystandissimilarinsizetotheCITDur-Amaterial experiments (see below) were neither neutron irradiated studiedpreviously atCaltech and currently usedas anage nor protonirradiated. standard (Farley, 2000). Dur-A shards have a helium clo- sure temperature, T =71.7±1.9(cid:2)C (here and below 2.3. Initial treatment: thermalannealing c assuming dT/dt=10(cid:2)C/Myr). For a few experiments we analyzed aliquots that consisted of a single large (few Asuiteofexperimentswasundertakentoassesswhether mm) chipfrom the sameDurango megacryst. 4Hediffusionismodifiedwhenapatiteisheatedtotemper- OnlyDurangowasinvestigatedintheannealingexperi- atures at which radiation damage annealing is expected ments, but the irradiation experiments were performed on based on fission track annealing kinetics (e.g., Green, Dur-Baswellasfournaturalapatitesfromgranitoidrocks 1988). For these experiments (cid:2)5–30mg aliquots of Dur-B and a synthetic fluorapatite. The synthetic fluorapatite or Durango chips were loaded in copper foil pouches and (WSAp) was synthesized in a flux of CaF (Prener, 1967; heatedinvacuumusingaprojectorlampapparatus(Farley 2 Cherniak,2005).Theanalyzedcrystalshadameancrosssec- etal.,1999).Thisdeviceiscapableofrapidtemperaturecy- tionof(cid:2)300lm,andweassumethatthismaterialinitially cling and has an estimated temperature uncertainty of less contains no radiation damage. Helium diffusion kinetics than2(cid:2)C. andcompositionaldataforthefourgranitoidapatiteswere Annealing durations were between 1 and 350h at tem- reported earlier (Shuster et al., 2005, 2006). They were se- peraturesrangingfrom20to500(cid:2)C.Duringtheannealing lectedforneutronirradiationexperimentsbecausetheyhave step theevolvedHewasaccumulated,thenanalyzed using significantlylowerheliumretentivitythanDurangoapatite isotopedilutionquadrupolemassspectrometryasdescribed (T for 01mr59=49.2±6.6(cid:2)C, DYJS5=49.9±5.1(cid:2)C, elsewhere (Farley, 2000). Following the annealing step the c CJ50=49.4±3.7(cid:2)C(CJ50),and00mr18=47.2±9.6(cid:2)C). sample was subjected to a routine diffusion coefficient ThelowerHeretentivityinthesesamplesisconsistentwith step-heatexperimentwithoutbreakingvacuum(seeSection arelativelylowvolumefractionofradiationdamageandsug- 2.4). geststhesesampleswillbereadilyaffectedbytheadditionof damagefromneutronirradiation.Meancrosssectionsofthe 2.4. Diffusion experiments analyzedgrainsrangefrom(cid:2)95to120lm. For the neutron irradiation experiments, we quantified 2.2.Initial treatment: neutron irradiation–proton irradiation heliumdiffusionkineticsusingstepped-releaseofproton-in- duced 3He (Shuster et al., 2004), while for the annealing Toinduceradiationdamagethesixsampleswereloaded experiments we used the residual natural 4He as the diffu- into an Al disk (routinely used for 40Ar/39Ar dating), and sant (Wolf et al., 1996; Farley et al., 1999; Shuster et al., 186 D.L.Shuster,K.A.Farley/GeochimicaetCosmochimicaActa73(2009)183–196 2004).Aliquotsranginginsizefromasinglecrystaltomany et al., 2006). Below 600(cid:2)C, the samples were heated using mgwereheldatacontrolledtemperatureforaknowntime aprojectorlampheatingapparatusinacontrolledfeedback in a chamber under static vacuum (Farley et al., 1999). loop with a thermocouple, as described by Farley et al. After heating, evolved helium was purified and cryogeni- (1999).Fortemperaturesbetween600and1200(cid:2)C,weused callyseparatedfromothernoblegasesusingactivatedchar- a70Wdiodelaser(k=810nm)defocusedontothesample coal held at 32K and analyzed using either a sector-field in the same thermocouple apparatus, which ultimately en- mass spectrometer at BGC (3He) or a quadrupole mass abledcompleteheliumextraction.Bothheatingprocedures spectrometer at Caltech (4He). Using the fraction of 3He weretunedforrapidresponseandtoavoidovershootofthe or4Hereleasedandthedurationofeachstep,wecalculated set-point temperature. Temperature control across this en- thediffusioncoefficient(D)normalizedtothecharacteristic tire rangewastypically better than±2(cid:2)C. diffusive length scale a, (i.e., D/a2) using published equa- We used previously described criteria (Shuster et al., tions and the assumptions therein (Fechtig and Kalbitzer, 2006) to establish data subsets for Arrhenius regression 1966). modelsforthe3Hebasedexperiments:(i)weusedtheentire We report D/a2 values (rather than D values) for all of setofmeasured3HereleasefractionstocalculateD/a2val- our samples. The only exception is for the Durango large ues;(ii)weexcludedD/a2valuesfromregressionmodelsfor chip experiments in which we scaled from the D/a2 values stepswhenRF3He60.5%(again,tominimizetheinfluence actuallymeasuredtoa=85lm(equivalenttoDur-B)using of small dust fragments which may be present as well as the measured grain dimensions and an equivalent sphere largeuncertaintiesatsmallyields),and(iii)excludedvalues model(MeestersandDunai,2002).Giventhe complicated fortemperaturesP280(cid:2)C(toavoidtheinfluenceofanneal- geometry of the analyzed chips this computation is fairly ing; see below). The only exception was for the synthetic uncertain, so these data, especially the D/a2 and closure apatites,whichdidnotyieldappreciableamountsofhelium temperaturevalues,aredeemedlessreliablethanthoseob- until well above 300(cid:2)C. Step numbers included in these tainedonDur-B. regressions are indicatedin boldin Table EA1. We quantifiedthe temperature dependenceof the diffu- sioncoefficientfromaseriesofstepsoneachsample.Such 2.4.2.Heatingscheduleandregressioncriteriaforannealing datapermitlinearregressionofln(D/a2)against1/Tassum- experiments ing the Arrhenius relationship D(T)/a2=D /a2exp((cid:3)E / Foraliquotsannealedattemperaturesabove250(cid:2)C,the o a RT),whereE istheactivationenergy,D isthediffusivity heatingschedulestartedattemperaturesatleast10degrees a o atinfinitetemperature,andRisthegasconstant.Theheat- lower thanthe initialannealingtemperaturesoastomini- ing schedule and data selection criteria used in the regres- mize further annealing. It then included a series of retro- sionsare described in the nextsection. grade steps, followed by prograde steps typically up to Incalculatingandinterpretingourdataweareexplicitly 400(cid:2)C. Starting with a retrograde series was adopted to assuming that 3He and 4He diffusion kinetics are inter- minimize the transient artifact known to occur when tem- changeable despite a substantial mass difference. Previous peratures are dropped by a large amount between subse- works suggests this conclusion isvalid for apatite (Shuster quent diffusion steps (Farley, 2000). For the experiments et al., 2004). Similarly, other than for the coarse Durango where the initial temperature of annealing was <250(cid:2)C, chipsforwhichwearecertainwecan(andmust)scaledif- only the prograde schedule was used. All of these experi- fusivity from one grain size to another (Farley, 2000), we ments were performed with the projector lamp apparatus are making no adjustments for the small grain size differ- described in Section 2.4.1. These experiments usually did ences among our samples. This follows the approach of not degas the aliquot entirely, so the final yield necessary our previous work in which the small range of grain sizes for diffusion coefficient calculations was obtained either routinely analyzed for He dating yields variations in diffu- by atotal fusionstep in aresistance furnace (chips),orby sivity far smaller than variations caused by the factors of calculation based on the mass of the sample and several interestinthepresentstudy(Shusteretal.,2006).Theissue determinations of the concentration of He in Dur-B of grain size will be considered further in a subsequent (8.2nmol/g). publication. Arrheniusparameters weredeterminedfor eachaliquot usingalldatapointsmeasuredattemperatures<280(cid:2)C(at 2.4.1.Heating scheduleandregression criteria forneutron whichpointfurtherannealingseemslikelybasedonfission irradiatedsamples track experiments (e.g., Ketcham et al., 2007). In some The heating schedule for the synthetic apatites began cases the first steps after annealing were isothermal, in withasetofisothermalstepsateither150or250(cid:2)Ccontin- which case only the last of the isothermal measurements uing with sets of isothermal steps up to 600(cid:2)C. All of the wasincludedintheregression(itwouldmakenodifference natural samples were analyzed using a heating schedule if allisothermal pointswere included). startingwithatleast3isothermalextractionsat200(cid:2)C,fol- lowed by a series of decreasing temperature (retrograde) 3. RESULTS steps, followed by increasing temperature (prograde) steps up to 600(cid:2)C. Beginning our experiments at moderate 3.1. Neutron irradiation diffusionexperiments (ratherthanminimum)temperaturesmoreefficientlymini- mizes a potential artifact introduced by small dust frag- Results of the neutron irradiation and helium diffusion ments containing 3He adhered to crystal surfaces (Shuster experimentsaresummarizedinTable1andthecorrespond- Heliumdiffusionkinetics 187 ing Arrhenius plots for the six sets of experiments are today.While[4He]isaneasilymeasurableproxyforradia- showninFig.2 andFig. EA1in theelectronic annex.Pri- tiondamage,itisnotidealforidentifyingthefundamental mary datafor allexperiments are tabulatedin TableEA1. controls on helium diffusion. This is because correlations For all samples, He diffusion coefficients define highly lin- betweenHediffusivityand4Hemightarisefromconcentra- ear Arrhenius arrays at temperatures <280(cid:2)C (e.g., tion-dependentdiffusionratherthanradiationdamagecon- Fig. 2). Exposure to a 1MeV equivalent neutron fluence trol, and because radiation damage accumulation (and of 2.3(cid:4)1016n/cm2 causedno noticeablechange in helium annealing) in general will not correlate perfectly with the diffusionkineticsinanyofthesixsamples.However,expos- accumulation of alpha particles. Here we improve on the ing each to a fluence of 2.0–2.7(cid:4)1018n/cm2 caused in- [4He]proxybydeveloping analternativeindicator ofradi- creases in both the activation energy (E ) and in ln(D / ation damage. This transformation requires a relationship a o a2),byasmuchas(cid:2)25kJ/moland6.4ln(s(cid:3)1),respectively. between the crystal damage caused by neutron irradiation For example, E ranges from 139kJ/mol in an untreated andanequivalentamountofdamagecausedbynaturalal- a aliquot of Durango apatite to >161kJ/mol in those ex- phadecay.Tomakethetransformation,westartbyassum- posedto90hofneutronirradiation.Ineffect,thedataar- ing that the amount of crystal damage caused by a raysintheArrheniusplotsrotateclockwiseattemperatures particulareventisproportionaltothetotalamountofion- below 550(cid:2)C. izing kinetic energy released into the matter of interest (or Because the two diffusion parameters are so strongly kerma(ASTM,1994)).Withascalingrelationshipbetween correlated (Fig. 3), for discussion of our observations it thekermaduetoneutronirradiationandthekermadueto makessensetocombinethemintoasingleparameter,spe- alpha decay, we can inter-relate the amount of damage cifically the closure temperature. As shown in Fig. 3, the caused by each process. This approach assumes that the observedarrayofdiffusionparameterscutsobliquelyacross atomic displacements caused by neutrons are similar to contours of closure temperature, with higher values of E those caused by alpha decay. In addition, we assume that a andln(D /a2)associatedwithhigherT.Thisindicatesthat damageassociatedwithinducedfissionof235Uisinsignifi- o c thepositivechangesinE duetoradiationexposurearethe cant,particularlybecauseweusedCdshieldingwhichmin- a dominantcontrolonlow-temperatureretentivity.Theradi- imizes the thermal neutron flux. We will focus our ation resulted in lower values of D/a2 at lowest tempera- discussiononkerma,sinceitisaquantitythatismoreeas- tures and diffusion kinetics with stronger temperature ilyrelatedtoradiationdamagethanis[4He],andallowsus sensitivity. The resulting closure temperatures increased to compare damagecaused bydifferent types ofradiation. by16to27(cid:2)Cafterirradiation.ChangesinT (DT)nega- Thekermaforaparticularneutronirradiationisafunc- c c tivelyscalewiththeT priortoirradiation,indicating that tionoftheenergyspectrumandtheamountoftimeasam- c apatites with higher initial T were less perturbed by the ple resided within a specific nuclear reactor. The energy c neutron irradiation thansamples withlower initial T. spectrumwithintheCLICITfacilityatOSUiswell-known c at the location of our samples, shown in Fig. EA3. Using 3.2.Thermal annealing diffusionexperiments the American Society for Testing and Materials (ASTM) standardpracticeforcharacterizingneutronenergyfluence Completestepped-heatingdatafortheannealingexper- spectra for radiation hardness testing (ASTM, 1994), the iments are compiled in Table EA2 and the segments used 1MeV equivalent monoenergetic damage function for sili- forcomputationofArrheniusparametersplottedinSupple- con has been calibrated for the OSU CLICIT facility by mentaryFig.EA2.TypicalresultsareshowninFig.2c.All S. Reese (personal communication), equaling 6.3(cid:4)109n/ aliquots yielded highly linear Arrhenius arrays including kWscm2. From this calibration, we can estimate the both the retrograde and the prograde sequences. Both the 1MeV equivalent neutron fluence (U ) for each eq, 1MeV, Si slope and the intercept of the He diffusion arrays vary experimentalcondition(notetheOSUreactorwasoperated among the aliquots (Table 2). For example, the activation at 1MW for ourexperiments). Since the kerma for silicon energy(E )rangesfrom139kJ/molinanuntreatedaliquot hasalsobeenquantified(Namesonetal.,1972),wecansim- a (identicaltopreviousresults(Farley,2000))to<100kJ/mol plyrelateU totheSikermabyusingtheASTM eq,1MeV,Si in thoseexposed to the most intense annealing conditions. standard 1MeV neutron displacement kerma factor (K D, For a given annealing duration, both activation energy ) of 95±4MeVmb (ASTM, 1994) and the mean 1MeV, Si andD /a2generallydecreasewithtemperatureand,aswith atomic mass of Si asshown inFig. 4a. o theneutronirradiationexperiments,thereisastrongcorre- The kerma for a-decays along the U- and Th-decay lation between the two Arrhenius parameters (Fig. 3). Ta- chainsisgivenbythemeanQ-valueforthesetofa-decays ble 2 shows that T of the variably annealed Durango ((cid:2)5.7MeV/a-decay).Byassumingquantitativeretentionof c aliquots rangesfrom (cid:3)2 to78(cid:2)C. 4He, which will be true at temperatures below (cid:2)40(cid:2)C, we can relate the kerma resulting from a-decay to an 4. DISCUSSION equivalentradiogenic4Heconcentrationinapatite(‘‘alpha equivalent units”, or [4He] ) (Fig. 4a). Note that the eq 4.1.Relating neutrondamageto naturalalpha damage reaction Q-value carries all available energy for causing crystal damage, including both the kinetic energy of the Inourpreviouswork,weconsideredtheradiogenic4He a-particleandtherecoiledparentnucleus.Here,weignore concentration([4He])tobeaproxyforthevolumefraction thesmallamountoftotalkermaduetonaturallyoccurring ofnaturallyoccurringradiationdamagewithineachsample butrelatively infrequentspontaneous 238U fissionevents. 1 8 8 Table1 Neutronirradiationexperiments. Sample [4He]o Reactor J 1-MeVeq.neutron [4He]eq [3He] [4He]tot Ea(kJ/ (+/(cid:3)) ln(Do/a2) (+/(cid:3)) Tc (+/(cid:3)) D.L (nmol/g) time(h) fluence(1016n/cm2) (nmol/g) (nmol/g) (nmol/g) mol) (kJ/mol) ln(s(cid:3)1) ln(s(cid:3)1) ((cid:2)C) ((cid:2)C) . S h WSAp_A 0.000 0 0.00000 0.0 0.00 0.0037 0.037 132 0.8 5.66 0.14 117.2 0.9 u s WSAp_B 0.000 1 0.00027 2.3 0.13 0.0017 0.038 133 1.1 5.82 0.21 118.4 1.3 ter , WSAp_D 0.000 10 0.00260 26.8 1.56 0.0042 0.410 132 1.8 4.77 0.32 126.5 2.2 K WSAp_C 0.000 90 0.02343 204.1 11.86 0.0036 2.150 149 0.6 8.04 0.10 143.8 0.6 .A . F DAp_A 8.200 0 0.00000 0.0 8.20 0.0023 8.200 139 2.2 13.54 0.55 73.1 1.5 a r DAp_B 8.200 1 0.00027 2.3 8.33 0.0022 7.500 138 0.9 13.38 0.21 72.1 1.0 le y DAp_C 8.200 90 0.02343 204.1 20.06 0.0022 8.633 161 2.0 19.25 0.50 86.7 1.1 /G DAp_D 8.200 90 0.02343 204.1 20.06 0.0022 8.824 164 2.1 19.92 0.51 87.8 1.1 eo c 01mr59a 0.500 0 0.00000 0.0 0.50 0.0025 0.500 121 1.5 10.50 0.40 49.2 6.6 him 01mr59_A 0.500 1 0.00027 2.3 0.63 0.0019 0.793 123 1.2 11.73 0.29 44.5 1.2 ic a 01mr59_B 0.500 90 0.02343 204.1 12.36 0.0022 3.016 134 2.1 11.55 0.53 74.5 1.5 e t 01mr59_C 0.500 90 0.02343 204.1 12.36 0.0022 2.926 135 1.0 12.79 0.25 69.6 0.7 C o s DYJS5b 0.300 0 0.00000 0.0 0.30 0.0020 0.300 122 1.2 10.50 0.30 49.9 5.1 m o DYJS5_A 0.300 90 0.02343 204.1 12.16 0.0022 2.545 128 2.1 10.32 0.52 67.5 1.6 ch im CJ50b 0.010 0 0.00000 0.0 0.01 0.0020 0.010 121 0.9 10.20 0.20 49.4 3.7 ic a CJ50_A 0.010 100 0.02645 267.8 15.56 0.0026 1.668 136 0.8 10.86 0.16 85.0 0.8 A c 00mr18b 0.040 0 0.00000 0.0 0.04 0.0020 0.040 149 2.7 21.40 0.70 47.2 9.6 ta 7 00mr18_A 0.040 100 0.02645 267.8 15.60 0.0024 1.443 165 4.7 21.15 1.19 81.9 2.4 3 (2 [4He] istheinitialradiogenic4Heconcentration;Jistheneutronirradiationparameter(GrastyandMitchell,1966);the1-MeVequivalentneutronfluenceisestimatedfromempiricalcalibration 0 o 0 asdescribedinthetext;[4He]eqisthealphaequivalentconcentrationforeachirradiationestimatedfromFig.4and[4He]o;[3He]and[4He]totarethemeasuredheliumisotopeconcentrations;Eais 9) theactivationenergyforheliumdiffusion;ln(Do/a2)isthenaturallogofthefrequencyfactor,andTcistheheliumclosuretemperaturecalculatedfordT/dt=10(cid:2)C/Ma.nmol=10(cid:3)9mol. 183 a Shusteretal.(2005). –1 b Shusteretal.(2006). 96 Heliumdiffusionkinetics 189 T (oC) T (oC) 600 500 400 300 400 300 250 200 150 -12 -10 (a) Synthetic Fluorapatite (b) 01mr59 -12 -14 -14 -16 1)) 1)) -16 -s -18 -s n( n( 2a) (l -20 2a) (l -18 D/ D/ -20 n( n( l -22 l -22 01mr59; no neutrons WSAp_A; no neutrons (Shuster et al., 2005) -24 WSAp_B; 1 hr irradiation -24 01mr59_A; 1 hr irradiation WSAp_C; 90 hr irradiation 01mr59_C; 90 hr irradiation -26 -26 12 14 16 18 14 16 18 20 22 24 4 4 -1 10 /T (1/K) 10 /T(K ) T (oC) 400 300 250 200 150 -10 (c) Durango Apatite -12 -14 1)) -16 -s n( 2) (l -18 a D/ -20 n( l -22 DUR-B; 414oC 1 Hr anneal -24 DUR-A untreated DUR-B; 90 hr irradiation -26 14 16 18 20 22 24 4 10 /T (1/K) Fig.2. ExampleHediffusionArrheniusplotsfor(a)syntheticapatiteWSApbeforeandafterneutronirradiation,(b)01MR59beforeand afterneutronirradiation,and(c)DurangoapatiteshowingresultsfromuntreatedDUR-A(Farley,2000),andDUR-Bsubjectedto90hof neutronirradiationandDUR-Bsubjectedto1hofannealingat414(cid:2)C(graysquares).Linesaretheheliumdiffusionkineticsdeterminedby linearregressiontosubsetarraysselectedusingthecriteriadiscussedinthemaintextandindicatedinTablesEA1andEA2.Opensquaresare the control experiments without any time in the nuclear reactor or thermal annealing, circles are 1h irradiations and diamonds are 90h irradiations.InallcasesneutronirradiationcausesArrheniusarraystorotateclockwise,yieldinghigherclosuretemperatures.Incontrast, annealingdoesjusttheopposite.Arrheniusplotsforallsamplesareincludingintheelectronicannex. By assuming that the kerma for silicon during the time clearlyanoversimplification,wediscussbelowanddemon- in the OSU reactor is approximately equal to the kerma strateinFig.1thattherelationshipinFig.4bappearstobe for apatite for the same irradiation conditions, we can set valid. The relationship also allows us to express the previ- thetwofunctionsinFig.4aequaltooneanothertoderive ous results ofShuster et al. (2006)in terms of kerma(e.g., arelationshipbetweenU and[4He] .Thisallows Fig. 7). eq,1MeV,Si eq ustosemi-quantitativelyexpresstheexpectedcrystaldam- agecausedbyneutronirradiation(mostlikelyduetonucle- 4.2. Experimental complications ar scattering reactions) in terms of alpha equivalent units andtocomparetheseresultswiththenaturalvariationsob- Therelativelysmallamountofkermaintroducedbyour servedbyShusteretal.(2006).Althoughthisrelationshipis proton irradiation (Summers et al., 1993), as well as 190 D.L.Shuster,K.A.Farley/GeochimicaetCosmochimicaActa73(2009)183–196 25 temperature ofa samplewhilein the reactor. Asdiscussed natural samples intheelectronicannex,wecanestimatethe4Heproduction 1 hr rateinoursamplesfromknownreactioncrosssectionson 10 hr 20 100 hr 40Ca,16O,and19F.Sinceweknowthetotalamountoftime 350 hr spent in the reactor, we can use the measured diffusion -1n(s)) 15 1n ehurt rboign cirhraipds. kmineaentictsembpoethrabtuerfeoroefathnedsaafmteprleirurasidnigat4ioHne/t3oHseotlhveermfoorcthhre- 2) (l otrnoollmedetreyxp(SerhiumsetenrtainndvoFlvairnlegy,s2im00u4lt)a.nIneoeuffsecitn,-tghrioswisthacaonnd- a D/o 10 diffusionof4He.Fig.EA5showstheresultofthisanalysis, n( clearlyindicatingthatthesamplesreachedameantemper- l ature of (cid:2)260(cid:2)C. This temperature corroborates a previ- ous, independent measurement of maximum temperature 5 (270(cid:2)C) made in the same facility using a thermocouple 20o C 40 oC 60 oC80 oC100 0C eimmpboerdtdaendt iimnpalincaatilounmsinfourm40bAlor/c3k9A. rThgiesocthemropneorlaotguyr,ephaars- 0 80 100 120 140 160 ticularlywhenconductedonKbearingglasswhichislikely Ea (kJ/mol) to diffuse Ar at (cid:2)260(cid:2)C during irradiation (Shuster et al., 2005). Fig. 3. He diffusion parameters ln(D/a2) and E for natural Althoughthesmallamountofconcurrentannealingex- o a apatites (Shuster et al., 2006), for neutron irradiated natural pectedatthistemperatureinthereactorwillcausethesam- apatites, and for the annealed Durango apatite aliquots. The ples to plot at higher kerma in Figs. 1 and 7 than is specified number of hours corresponds to the Durango apatite appropriate, the shift is relatively small given the logarith- annealing experiments conducted at different temperatures, as mic x-axis. shown in Figs. 4 and 5. All samples define a roughly linear trajectory on this figure, which we associate with the effects of 4.3.Theinfluenceofneutronirradiationonsyntheticapatite radiation damage. Increasing damage moves samples to the northeast, annealing moves them to the southwest. Also shown ThetrappingmodelofShusteretal.(2006)predictsthat andlabeledarecontoursofclosuretemperature(calculatedfordT/ dt=10(cid:2)C/Ma). Note that the sample array crosses closure proton-irradiated apatite otherwise devoid of radiation temperaturecontoursatanobliqueangle,showinghowincreasing damage should have diffusion kinetics defined by the two or decreasing amounts of radiation damage propagates into parameters: E =120kJ/mol and D /a2=1.58(cid:4)104s(cid:3)1, a o changesinclosuretemperature. correspondingtoT (cid:2)52(cid:2)C.Ourexperimentsonsynthetic c fluorapatite were designed, in part, to test this prediction. thermal annealing during the neutron irradiations may Assuming that the synthetic apatite has exactly the same complicate our ability to relate neutron-induced damage propertiesasnaturalapatite,weexpectedT tobeatorbe- c to the observations of Shuster et al. (2006). However, due low the least retentive apatites observed in Shuster et al. to the log-linear relationship between kerma and T, each (2006). However, this is not observed. Instead, even in the c introduces only a small bias in Fig. 1. The kerma intro- untreated specimen we found helium diffusion kinetics duced during the 8h of proton irradiation is only approximately equal to the most retentive natural apatite, (cid:2)1014MeV/g,comparedwith(cid:2)1016ormoreinmostnatu- or (cid:2)117(cid:2)C (Figs. 1 and 2a). We do not understand why ralsamples.Wethereforeexpectproton-induceddamageto the synthetic apatite is so He retentive, but we suspect it be insignificant relative to a sample’s initial or neutron-in- is related to large differences in defect density between the duced damage except for those samples containing very synthetic and natural specimens. low amounts of initial kerma (particularly the synthetic Another striking feature of the synthetic apatite Arrhe- apatite, CJ50 and 00mr18). Comparison of Fig. 1 with nius plot is a lack of curvature between 270 and 900(cid:2)C, a Fig.7(whichincludeskermaduetoprotonirradiation),re- linearity that is almost never observed in natural apatites veals only a slight difference in the location of these sam- (See Fig. EA1 and Farley, 2000; Shuster et al., 2004, ples. However, we now believe that the trapping model as 2005, 2006). While these discrepancies make the synthetic calibratedinShusteretal.(2006)(Fig.1)maybebiasedto- apatite an imperfect analog system, the sample’s response wardshighT forkermavaluesbelow(cid:2)1014MeV/g.Since to neutron irradiation is remarkably similar to natural c T is expected to rapidly increase with relatively small samples. c amountsofkerma,undermostcircumstancesthisisamin- The 10 and 90h neutron irradiations of the synthetic orartifact (see below). apatite caused E and D /a2 to increase dramatically a o Inadditiontocausingcrystaldamage,theneutronirra- (Table1).TheresultingchangeinT from90hofneutron c diations caused two bi-products during irradiation: 4He irradiation (DT (cid:2)+27(cid:2)C) spans most of the observed c (primarily from (n,a) reactions while in the reactor) and range in T for natural apatites (Fig. 1). When plotted c heat. Although the latter introduces the potential for a against alpha equivalent units ([4He] ) or kerma, we find eq smallamountofbiasinFig.1duetoconcurrentannealing the relative increasesin T are in good agreement with the c during neutron irradiations, we show that these complica- trapping model of Shuster et al. (2006) when shifted up- tionsalsoprovideaninterestingwaytomonitortheaverage wards to have a base T (cid:2)117(cid:2)C (gray curve in Fig. 1). c Table2 Thermalannealingexperiments. Aliquot AnnealingT((cid:2)C) Annealingtime(h) E (kJ/mol) (+/(cid:3))(kJ/mol) ln(D/a2)ln(s(cid:3)1) (+/(cid:3))ln(s(cid:3)1) T ((cid:2)C) (+/(cid:3))((cid:2)C) Fractionlostinannealing a o c step Shards Nopretreatment 141 1.7 14.23 0.40 72.5 1.3 H1-T75 75 1 140 2.5 14.21 0.60 71.2 1.9 1.00E(cid:3)04 H1-T300 300 1 137 1.7 12.98 0.40 72.0 1.3 7.30E(cid:3)02 H1-T325 325 1 133 1.6 12.90 0.39 63.9 1.3 1.14E(cid:3)01 H1-T350 350 1 127 1.5 10.75 0.37 62.4 1.3 9.50E(cid:3)02 H1-T354 354 1 123 2.3 10.17 0.55 57.4 2.1 1.50E(cid:3)01 H1-T365 365 1 121 1.2 9.05 0.38 59.4 0.4 1.80E(cid:3)01 H1-T375 375 1 117 1.5 8.54 0.35 52.4 1.2 1.99E(cid:3)01 H1-T397 397 1 112 1.9 7.21 0.46 46.6 1.7 2.14E(cid:3)01 H1-T414 414 1 108 1.3 6.78 0.32 39.2 1.1 3.13E(cid:3)01 H1-T450 450 1 94 1.3 4.57 0.32 14.1 1.1 8.70E(cid:3)01 H1-T500 500 1 79 1.3 0.50 0.32 -2.1 1.2 9.40E(cid:3)01 H e liu H10-T250 250 10 142 1.7 13.97 0.42 76.9 1.4 5.60E(cid:3)02 m H10-T275 275 10 135 1.7 12.67 0.40 68.8 1.3 1.14E(cid:3)01 d iff H10-T300 300 10 125 2.0 10.82 0.37 58.0 2.3 2.27E(cid:3)01 u s H10-T315 315 10 125 1.5 10.41 0.37 59.2 1.3 2.75E(cid:3)01 io n H10-T330 330 10 119 1.5 8.66 0.35 55.1 1.2 2.59E(cid:3)01 k H10-T345 345 10 109 1.3 6.16 0.32 46.3 1.1 2.79E(cid:3)01 ine tic H100-T250 250 100 135 1.7 12.33 0.40 72.8 1.4 8.90E(cid:3)02 s H100-T275 275 100 129 1.6 11.35 0.38 62.7 1.3 3.27E(cid:3)01 H100-T300 300 100 118 1.5 7.35 0.35 62.0 1.3 2.88E(cid:3)01 H100-T325 325 100 118 1.9 7.29 0.46 63.7 1.7 7.70E(cid:3)01 H350-T250 250 350 133 1.8 12.23 0.44 68.1 1.5 2.95E(cid:3)01 H350-T270 270 350 126 1.7 9.93 0.41 65.4 1.5 3.59E(cid:3)01 Chips(diffusivitiesrecomputedtor=85lm) Nopretreatment 141 2.7 14.00 0.60 74.1 2.3 BH1-T150 150 1 145 3.5 16.24 0.85 70.6 2.7 1.67E(cid:3)05 BH1-T250 250 1 143 2.0 14.74 0.47 73.7 1.5 1.20E(cid:3)03 BH1-T325 325 1 135 1.9 12.80 0.45 67.7 1.6 1.00E(cid:3)02 BH1-T350 350 1 133 1.6 12.65 0.39 63.4 1.3 2.00E(cid:3)02 BH1-T375 375 1 127 1.5 11.89 0.37 53.9 1.2 3.20E(cid:3)02 BH1-T391 391 1 115 1.4 9.97 0.33 35.7 1.1 5.40E(cid:3)02 BH1-T415 415 1 115 1.4 9.57 0.33 39.2 1.1 5.90E(cid:3)02 1 9 1 192 D.L.Shuster,K.A.Farley/GeochimicaetCosmochimicaActa73(2009)183–196 1e+18 initial Tc indicates that the neutron-induced damage was (a) kerma functions additivetothenaturaldamage,butinalogarithmicsense. Thisagreeswiththelog-linearrelationshipobservedinnat- 1e+17 uralsamplesandpredictedbythetrappingmodel(Shuster et al., 2006). This is most clearly observed in Fig. 1. For 1e+16 example,samplesCJ50and00mr18,whichcontainasmall g) amount of initial kerma and a T (cid:2)50(cid:2)C, end up with V/ c e T (cid:2)82 and 85(cid:2)C, (or differences of +32 and +35(cid:2)C). M c a ( 1e+15 On the other hand, Durango, which initially contains a m far larger amount of kerma and initial T =72(cid:2)C is only ker 1e+14 perturbedby+14(cid:2)C.Whenplottedagaincst[4He]eqorker- ma,allfivesamplesplotveryclosetotheclosuretempera- tures observed for natural apatites with an equivalent Φ 1e+13 eq,1Mev,Si amountofdamageandclosetothatpredictedbythetrap- [4He] ping model. eq ThemainobservationofShusteretal.(2006)wasacor- 1e+12 0.01 0.1 1 10 100 1000 relationbetweenTcandlog([4He])inasuiteofnaturalapa- (1016 n/cm2) or (nmol/g) tites. While radiation damage trapping of 4He is a reasonable explanation for this correlation, the possibility of concentration-dependent-diffusion could not be ruled 2m) 10000 (b) scaling relationship out. The neutron irradiation experiments described here n/c provideatestbetweenthesepossibilities,asdotheanneal- 160 ing experiments (see Section 4.5). Only a small fraction of 1 1000 neutrons produce 4He (see electronic annex), but almost ce ( all participate in damage production. In contrast, in the n e case of a-decay, the production of radiation damage is u nt fl 100 genetically tied one-to-one with 4He production. Thus, e the neutron irradiated samples accumulate radiation dam- al v age faster than 4He. For example, the measured 4He con- ui q 10 centrations following 90 and 100h neutron irradiations e V ([4He] in Table 1) are typically (cid:2)1 order of magnitude e tot M Q-value relationship lowerthanthealphaequivalentunitsofradiationdamage. 1 Mev,Si 1 9190 hh hrr rnp neroeututortonron irn rir ariradradiadiatiioationtionn irraFdiiga.ti1onshioswcsotrhraeltattheedrwesipthondsaemoafgHeeadcicffuumsiuolnattioonneruattrhoenr Φeq,1 0.1 than 4He accumulation. For example, the 100h experi- 0.01 0.1 1 10 100 1000 mentsforsamplesCJ50and00mr18plotaspredictedfrom their increased kerma (Fig. 1), yet if they were plotted [4He] (nmol/g) eq against [4He] , they would plot significantly higher in T tot c than the main population. The distinction between radia- Fig. 4. Kerma scaling. (a) The relationships between the total tion damage and [4He] is perhaps most clearly seen in the kinetic energy released into matter (kerma) and (i) the 1MeV Durango apatite results. The in-reactor 4He production equivalent neutron fluence (U ) and (ii) the alpha eq, 1 MeV, Si equivalent units due to U- and Th-decay series assuming quanti- and diffusive loss were fortuitously balanced such that the tative4Heretention([4He]eq).(b)Thescalingrelationshipbetween final 4He concentration after neutron irradiation ([4He]tot U and[4He] derivedby settingthe twolinesequal to (cid:2)8.7nmol/g) was approximately equal to the initial 4He eq,1MeV, Si eq one another in (a). Shown are the equivalent 4He concentrations concentration ((cid:2)8.2nmol/g). Despite nearly equivalent for90and1hintheOSUnuclearreactor,and9hofirradiationin [4He], the T after 90h of neutron irradiation was higher c the200MeVprotonbeamattheFrancisH.BurrProtonTherapy by(cid:2)15(cid:2)C.Thisclearlyindicatesthatavariableotherthan Center. [4He]iscontrolling heliumretentivity in thesample. When plottedagainst[4He] orkerma(Fig.1),theincreaseinT eq c is close to that predicted by the trapping model, again This agreement corroborates the parameters E =29kJ/ t mol and w=1.26(cid:4)10(cid:3)4g/nmol of Shuster et al. (2006) stronglyindicatingthatradiationdamageisthecontrolling variable. and also indicates that the kerma scaling relationship in Fig. 4 isnotgrossly inaccurate. 4.5. Kermaas a predictive variable 4.4.The influenceof neutron irradiation onnaturalapatite Whenscaledtothetotalkineticenergyavailablefordis- Justlikeinthesyntheticapatite,inallfivenatural apa- placing atoms, the neutron irradiations caused changes in tites,exposuretoincreasingamountsofneutronirradiation helium diffusion kinetics that are in good agreement with caused E , D /a2 and the He closure temperature to in- the observed range in natural apatites and the model of a o crease (Table 1). The fact that DT negatively scales with Shusteretal.(2006).Becausewedidnotinducesignificant c
Description: