isotopes in individual chondrules Early evolution of the solar accretion disk inferred from Cr-Ti-O Earth and Planetary Science Letters

(1)

Contents lists available atScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

Early evolution of the solar accretion disk inferred from Cr-Ti-O isotopes in individual chondrules

Jonas M. Schneider

^a^,^∗

, Christoph Burkhardt

^a

, Yves Marrocchi

^b

, Gregory A. Brennecka

^c

, Thorsten Kleine

^a

aInstitutfürPlanetologie,UniversityofMünster,Wilhelm-Klemm-Straße10,48149,Germany bCRPG,CNRS,UniversitédeLorraine,UMR7358,Vandoeuvre-lès-Nancy,54501,France

cNuclearandChemicalSciencesDivision,LawrenceLivermoreNationalLaboratory,Livermore,CA94550,UnitedStatesofAmerica

a rt i c l e i n f o a b s t r a c t

Articlehistory:

Received30January2020

Receivedinrevisedform26August2020 Accepted12September2020

Availableonline23September2020 Editor:W.B.McKinnon

Keywords:

chondrules isotopicanomalies meteoritedichotomy nebularmixing

Isotopic anomalies in chondrules hold important clues about the dynamics of mixing and transport processesinthesolaraccretiondisk.The meaningoftheseanomaliesisdebatedand theyhavebeen interpretedto indicateeither disk-wide transportof chondrulesor localheterogeneities ofchondrule precursors. However, all previous studies relied on isotopic data for asingle element (either Cr, Ti, orO), whichdoes not allowdistinguishing betweensource and precursor signaturesas the causeof the chondrules’isotope anomalies. To overcome thisproblem, we obtained the ﬁrst combined O, Ti, and Cr isotope data for individual chondrules from enstatite, ordinary, and carbonaceouschondrites.

We ﬁndthat chondrules from non-carbonaceous (NC) chondriteshave relatively homogeneous¹⁷O,

ε⁵⁰Ti,andε⁵⁴Cr,whicharesimilartothecompositionsoftheirhostchondrites.Bycontrast,chondrules from carbonaceouschondrites (CC) have more variable compositions, someof which differfrom the host chondrite compositions.Although the compositions ofthe analyzed CC and NC chondrules may overlapforeitherε⁵⁰Ti,ε⁵⁴Cr,or¹⁷O,inmulti-isotope space,noneoftheCCchondrulesplotinthe compositional ﬁeld of NC chondrites, and no NC chondrule plots within the ﬁeld of CC chondrites.

Assuch, our data reveal afundamental isotopic differencebetween NC and CC chondrules,whichis inconsistent with adisk-wide transport ofchondrules acrossand between theNC and CC reservoirs.

Instead,theisotopicvariationsamongCCchondrulesreﬂectlocalprecursorheterogeneities,whichmost likely result from mixing between NC-like dust and a chemically diverse dust component that was isotopicallysimilartoCAIsandAOAs.Thesamemixingprocesses,butonalarger,disk-widescale,were likelyresponsibleforestablishingthedistinctisotopiccompositionsoftheNCandCCreservoirs,which representininnerandouterdisk,respectively.

©²⁰²⁰^TheÂuthor(s).^Published^byÊlsevier^B.V.^Thisîsânôpenâccessârticleûnder^the^CC^BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Chondritic meteorites provide fossilized remnants of the materials present in the early Solar System, and are key samples to unravel the early dynamical evolution of the solar accretion disk(Scott andKrot, 2013).Chondrites areunequilibratedassem- blagesofvarious low- andhigh-temperaturecomponents,includ- ingvolatile-rich,ﬁne-grainedmatrix,Fe-Nimetal,refractoryinclu- sions,andtheeponymouschondrules,whichconstituteup to90%

ofbulkchondrites(e.g.,Scott,2007).Althoughtheconditions,pro- cess(es), and chronology of chondrule formation remain debated (forareviewseeConnollyandJones,2016),theirubiquitouspres-

*

Correspondingauthor.

E-mailaddress:[email protected](J.M. Schneider).

enceinthe diskmakesthema primetracer oftransport,mixing, andaccretionprocessesintheearlySolarSystem.

Utilizingchondrulesinthismannerrequiresunderstandingthe genetic relationships among individual chondrules, and between chondrulesandtheir hostchondrite.Thisinformationmaybeob- tainedfromisotopeanomalies inindividual chondrules. Ofthese, O isotopes are the most extensively studied, but the O isotope variability among chondrules likely represents a combination of precursor heterogeneity, open-system exchange with the nebular gas,andﬂuid-assistedalterationontheparentbody(e.g.,Clayton, 2003; Kita et al., 2010; Pirallaetal., 2020). As such, the genetic heritageofchondrulescanbediﬃculttointerpretusingOisotopes alone.

Otherisotopetracersthathavebeenusedtoassessthegenetic heritageofchondrulesincludevariationsintherelativeabundance of⁵⁰Ti and⁵⁴Cr.Anomalies inthesetwo isotopes arenucleosyn-

https://doi.org/10.1016/j.epsl.2020.116585

0012-821X/©²⁰²⁰^TheÂuthor(s).^Published^byÊlsevier^B.V.^Thisîsânôpenâccessârticleûnder^the^CC^BY-NC-ND^license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

(2)

theticinoriginandare powerfultoolstoinvestigategeneticrela- tionships betweenindividual solids, bulk meteorites, and planets (e.g., Gerber et al., 2017; Trinquier et al., 2009; Warren, 2011).

Forinstance,isotopicanomaliesin ⁵⁰Tiand⁵⁴Cr,together withO isotopevariations,wereinstrumentalinidentifyingafundamental isotopic dichotomybetweennon-carbonaceous (NC)andcarbona- ceous(CC)meteorites(Warren,2011).Subsequentworkhasshown that this dichotomy extends to several other elements, most no- tably Mo(Buddeetal.,2016a;Pooleetal.,2017;Worshametal., 2017). The NC and CC meteorites represent two genetically dis- tinct disk reservoirs thatremained spatially separatedforseveral millionsofyears(Ma),mostlikelythrough theearlyformationof Jupiter(Kruijeretal.,2017).TheNCreservoiristhereforethought torepresenttheprimordialinner,andtheCCreservoirtheprimor- dialouterSolarSystem.

The existence oftheNC-CC dichotomy amongbulk meteorites raisesthequestionofwhetherthisdichotomyalsoexistsatthein- dividualchondrulescale.Thismightbeexpected,giventhatthedi- chotomywaslikelycausedbydistinctdustcompositionsbetween the tworeservoirs. As chondrules providea snapshot ofthe dust composition ata particular time and location in the disk, chondrules fromthe NC and CC reservoir should have fundamentally different isotopic compositions, at least for elements that deﬁne theNC-CCdichotomy.However,theisotopiccompositionofchon- drules may also vary because the chondrules themselves derive fromdifferentareasofthediskandweresubsequentlytransported to the accretion region of their host chondrite. Currently available CrandTi isotopic datado not allowdistinguishing between thesetwodisparateinterpretationsandhavebeeninterpretedboth assourcesignatures(Olsen etal., 2016) andlocalheterogeneities amongchondruleprecursors(Gerberetal.,2017).

A major problemcommon to previous isotopic studies on individual chondrules is that the isotopic datawere only obtained for a single element. Although such studies reveal isotopic het- erogeneities amongchondrulesforeachinvestigatedelement, the interpretation of a single isotopic signature in terms of large- scale disk processes remains inherently uncertain. To overcome thisshortcoming,weobtainedcombinedCr,Ti,andOisotopicdata foracomprehensivesetofindividualNCandCCchondrules.These data provide new and ﬁrm constraints on the origin of isotope anomaliesinchondrulesandtheirrelationtotheNC-CCdichotomy observedamongbulkmeteorites.Combined,thesenewconstrains haveimportantimplicationsforunderstandingtheearlyevolution ofthesolaraccretiondisk.

2. Materialandmethods 2.1. Samples

Allsamples ofthisstudywere partofthe Tiisotopestudy by Gerberetal.(2017) andincludesinglechondrulesaswellasbulk rocksfromthecarbonaceouschondrites Allende(CV3),GRA06100 (CR2), and NWA 801 (CR2), the enstatite chondrite MAC02837 (EL3) andthe ordinary chondrite Ragland (LL3.4). Forthis study, sevenchondrulesfromAllende,sixfromGRA06100,onefromNWA 801,ninefromRagland,andfourfromMAC02837wereselected.In addition,matrixandpooledAllendechondruleseparates,consist- ingofhundredsofchondrulesandchondrulefragments(Buddeet al.,2016a),werealsoanalyzed.

Aftergentlycrushingpiecesofbulkchondriteinanagatemor- tar,individual chondrules were hand-pickedwithtweezersunder abinocularmicroscope.Wheneverpossible,whole,sphericalchon- drules wereselected.Adheringdustwas removedfromchondrule rimsbysonicationinacetonefor15–30min.Theindividualchon- drules were photographed witha Keyence VHX-500F digital microscope(SMFig.S1),andthenbrokeninanagatemortar.Afrag-

ment comprising between 5-20 wt.% of the bulk chondrule was embedded into epoxy for petrographic characterization by SEM (Fig. S1) andin-situ Oisotopic analyses by SIMS. The remaining chondrule material, comprising 80-95 wt.% of the original bulk chondrule, was dissolved in puriﬁed mineral acids, and Ti was separatedby ion exchange chromatography(Gerber etal., 2017).

TheCrisotopemeasurementsofthisstudyweremadeonpuriﬁed Cr fromthe column washes of the Ti chemistry of Gerber et al.

(2017).Chromiumconcentrationsweredeterminedonthecolumn washespriortoCrseparationbyionexchangechromatography,using aThermoScientiﬁc X-SeriesII quadrupole inductivelycoupled plasma-massspectrometer(ICP-MS).

2.2.OxygenisotopemeasurementsbySIMS

Thinsections of the chondrulefragments were examined mi- croscopicallyintransmitted andreﬂectedlight. Scanningelectron microscope observations were performed at CRPG-CNRS (Nancy, France)usingaJEOLJSM-6510with3nAprimary beamat15kV.

Cathodoluminescence (CL) imaging of chondrules was performed usingan RGBCL-detection unit attached to a ﬁeld emission gun secondary electron microscope JEOL J7600F at the Service Com- mun de Microscopie Électronique (SCMEM, Nancy, France). We measured the O isotopic compositions of chemically character- izedolivine crystals withaCAMECA ims1270 E7atCRPG-CNRS.

16O⁻,¹⁷O⁻,and¹⁸O⁻ ionsproducedbyaCs⁺ primaryionbeam (∼^{15 μm,} ∼⁴^nA) ^were ^measured ⁱⁿmulti-collection mode with two off-axis Faraday cups (FCs) for ^16,18O⁻ and the axial FC for 17O⁻(see Marrocchietal.,2018a).Toremove¹⁶OH⁻interference on the ¹⁷O⁻ peak and to maximize the ﬂatness atop the ¹⁶O⁻ and¹⁸O⁻peaks,theentranceandexitslitsofthecentralFCwere adjusted to obtain a mass resolution power (MRP) of ∼⁷⁰⁰⁰ ^for

17O⁻.The multi-collection FCswere seton slit1 (MRP= ^2500).

Totalmeasurementtimeswere240s (180smeasurement+⁶⁰^s pre-sputtering).We usedthree terrestrialstandardmaterials (San Carlosolivine,magnetite,anddiopside)todeﬁnetheinstrumental massfractionation lineforthe threeoxygen isotopes andcorrect for instrumental mass fractionation due to the matrix effect of olivine. Toincrease the precision onthe measurements, we con- secutivelyanalyzedfourstandards,eightchondruleolivinecrystals, andfourstandards.TypicalcountratesobtainedontheSanCarlos olivine standards were 2.5×¹⁰⁹ ^cps ^for ¹⁶^O, ¹.0×¹⁰⁶ ^cps ^for

17O, and5.4×¹⁰⁶ ^cps ^for¹⁸Ô. ^Data âre^givenⁱⁿ ^the^delta^notation δ^17,18O (=^[(^17,18Ô/¹⁶Ôsample/^17,18O/¹⁶OSMOW)−^1] ×^1000), ând capital delta notation ¹⁷O (=δ¹⁷O−^0.52×δ¹⁸Ô). ^Depending ôn samplehomogeneityandsize,between2and25individual measurements were obtained on each chondrule fragment. Only one measurementperolivinegrainwascarriedoutandeachSIMSspot was then checkedby SEM.Analyses fromcracks orgrain bound- aries were not included. The 2

σ

errors were ∼^0.2 for δ¹⁸O,

∼^0.4^forδ¹⁷O,and0.9^for¹⁷O.Theerroron¹⁷Owascal- culatedbyquadraticallysummingtheerrorsonδ¹⁷Oandδ¹⁸Oand thestandarddeviationof¹⁷Ovaluesofthefourterrestrialstan- dards.

2.3.ChromiumseparationandisotopicmeasurementbyTIMS

Chromium was separated from the sample matrix using a three-stageionexchangechromatography.Theﬁrsttwostepswere similarto the procedures fromYamakawaetal.(2009) andTrin- quier et al. (2008), employing both anion and cation exchange resins.Theelutionprocedureonthethirdcolumn(cationexchange resinAG50W-X8,200-400 mesh)was slightlymodiﬁed compared to the two aforementioned studies. The Cr cut from the second columnwasloaded in0.5MHNO₃ andresidual matrixelements wererinsedoffthecolumnusing0.5MHNO3,0.5MHF,and6ml

(3)

1MHCl,beforeCrwas elutedwith10mlof2MHCl.Toremove residualorganics, theCrcut was thentreatedwith2×⁴ ^drops^of aquaregia at120^◦C, 2×2 dropsof concentrated HNO3 at120^◦C, and2×¹^drop ^of^H2O2 at80^◦C.The sample was then converted withconcentratedHCltwice beforeit wastakenup inan appro- priate volumeof6MHCltoobtaina∼^{500 μg/g}^loading^solution.

The Cr yield of the full chemical procedure was determined on aliquotsbyICP-MS,andwasgenerally>70%.Aliquots oftheNIST 3112astandardprocessedthroughthesamechemistryyieldedin- distinguishableisotopicresultsasunprocessedstandards.

TheCrisotopemeasurementswereperformedonaThermoSci- entiﬁcTritonPlusthermalionizationmassspectrometer (TIMS)at theInstitutfürPlanetologieinMünster.Approximately1 μgCrper sample was loaded onto single outgassed Re ﬁlaments and cov- ered with an Al-Si-gel emitter. The emitter was made from 3 g of Aerosil 300 (Evonik) Si-Gel in 60 ml MQ water, a portion of which was then mixed 1:1 with a 5 g/L H3BO3 solution and a 1000 μg/g Al solution. Ion beams were simultaneously collected instaticmodeforallfourchromiumisotopes(⁵⁰Cr,⁵²Cr,⁵³Cr,and 54Cr)usingFaraday cups connectedto10¹¹ feedbackresistors.

The signalintensitywas usually ∼¹⁰ ^V^on⁵²^Cr^and^200-250 ^mV on⁵⁴Cr.Possibleinterferencesfrom⁵⁰Ti,⁵⁰V,and⁵⁴Feweremon- itored using ⁴⁹Ti, ⁵¹V, and ⁵⁶Fe beams in cups L4, L1, and H4, respectively. However, Ti and V signals were not detected above baselineduring themeasurementsofthisstudy,andinterferences fromFewereminimal.Measured⁵⁶Fe/⁵²Crratiosweretypically<

10⁻⁵,wellbelowthe10⁻⁴thresholdforsuccessfulFeinterference correction(Qinetal.,2010).WetestedtheaccuracyoftheFeinter- ference correction bymultipleanalyses ofFe-dopedCrstandards.

Repeatedmeasurementofindividualﬁlaments showedadecrease intheFesignalovertime,butnochangeintheﬁnalinterference- corrected⁵⁴Cr/⁵²Crratio.

Dataacquisitionconsistedof50blocksof30cycleseach(1500 totalcycles),4s integrationtimepercycle, anda baselineandre- focuseverythreeblocks.Instrumentalmassfractionationwascor- rectedbyinternalnormalizationto⁵⁰Cr/⁵²Cr= ^0.051859^(Shields et al., 1966) usingthe exponential law(Russell et al., 1978). All data are reported in

ε

-units as the parts-per-ten-thousand deviation from the ⁵³Cr/⁵²Cr and ⁵⁴Cr/⁵²Cr measured for the NIST SRM3112aCrstandardineachsession.Repeatedmeasurementsof the NISTSRM3112a Crstandard ineach session(n = ^7–14)^provided an external reproducibility (2s.d.) of ±^0.2

ε

⁵³Cr and ±^0.3

ε

⁵⁴Cr. Depending on the total amount of Cr available, samples were loaded on 1-3 ﬁlaments and each ﬁlament was measured multipletimesifpossible,resultinginuptosevenindividualmea- surements for individual chondrules. For the pooled chondrule separates,matrixfractions,andbulk Allende,thenumberofmea- surementswaslarger asmoreCrwasavailable forthesesamples.

Reported uncertainties for sample measurements are 95% conﬁ- dencelimitsofthemean(n≥^4),^or^twice^the^standard^deviation obtained for the NIST SRM3112a Cr standard that was analyzed duringthesamesession(n<4).

3. Results

3.1. PetrographyandOisotopes

Petrographic characteristics of the individual chondrules are provided in Table 1and inFig. S1. Exceptforone barred olivine (BO)chondrule,and oneAl-rich chondrule,all investigatedchon- drules fromtheCVchondrite Allendeareporphyriticolivine (PO) andolivine-pyroxene(POP)type I(lowFeO)chondrules.Allinves- tigatedchondrulesfromtheCRchondritesGRA06100andNWA081 are porphyritic olivine-pyroxene-rich (POP)type Ichondrules. All butonechondrulefromtheordinarychondritesRaglandaretypeII chondrules(75<Mg#<90)andincludeporphyriticchondrules(PO,

Fig. 1.Oxygenthree-isotopeplot.CCchondrulesplotalongtheCCAMlinewitha slopeofapproximately1.Thecomparativelylargeerrorsareduetointra-chondrule heterogeneitywithsingleolivinegrainsbeingisotopicallydistinctfromoneanother.

OCandECchondrulesdonotdisplaysigniﬁcantintra-chondrulevariationandplot onorabovetheterrestrialfractionationline(TFL)withaslopeof0.52.

POP),radialpyroxenechondrules(RP,withsystematicpresence of smallolivine grains)andone crypto-crystalline (C) chondrule.Fi- nally,thefourinvestigatedchondrulesfromtheenstatitechondrite MAC02837 aretype Iporphyriticpyroxene(PP) andradialpyrox- ene (RP) chondrules. The diameters of the analyzed chondrules rangefrom0.7to 1.9mm forCC chondrules,from1.0to2.7mm forOC,andfrom0.9to1.7mmforECchondrules.

TheOisotopiccompositionsofthechondrules areprovidedin Table 1, and plotted in Fig. 1. The data of the individual analy- sesfor each chondrule are givenin the supplementary materials (SM, Table S1). For most chondrules, measured olivine grainsof agivenchondruleexhibithomogeneous ¹⁷Owithinuncertainty.

Forthesesamples averagingthe individual olivine measurements providesagood estimateofthebulk chondrule’sOisotopiccom- position(TableS1).However,threeCVandtwoCRchondrulescon- tain relict grains,which are deﬁnedashavinga ¹⁷Ovalue that differsfromthatoftheirhostchondruleoutsideof3

σ

(Ushikubo etal., 2012). An SEM image of the CV chondrule fragment with themost¹⁶O-richrelictgrain(¹⁷O= −¹⁴)isshowninFig.2, alongwithanCLimageofthesamesample.Forthesampleswith relictgrains,averagingoftheolivine¹⁷Ovaluesprovidesabiased estimateofthebulkchondrulecompositiontowardsthecomposi- tion of the relict grains. This is because we have not randomly sampledthe chondrules, andchondrules do not consist solelyof olivine. Prior studies haveshown that low-Capyroxenes are iso- topicallyhomogeneousforO, andthatthey havethesameOiso- topiccompositionsasthehostolivines(Tenneretal.,2018,2015).

Wethereforecalculatedthe“bulk”O-isotopiccompositionofchon- drules having relict grainsby considering the host olivines only.

Importantly,theOisotopic compositionsofhostmineralsreﬂects contributionsfromOinheritedfromchondruleprecursors(through relict mineraldissolution) andfromO comingfrom the gas(i.e., SiO,H₂O;LibourelandPortail,2018;Marrocchi etal.,2019). Our bulk estimate, therefore, accountsfor the effectof relict mineral dissolutionandshouldbeclosetotheactualbulkOisotopiccom- position(TableS1).

In an O three-isotope diagram the bulk CC chondrules plot alongaslope∼¹^line^below^theterrestrialfractionationline(TFL), whereas the NC chondrules are clustered on or above the TFL (Fig. 1). The ordinary chondrite chondrules display indistinguishable¹⁷Ovaluesaveragingat1.4±^0.7^.^Similarly,^the^enstatite chondritechondrulesexhibithomogeneous ¹⁷Owithanaverage

(4)

Table 1

Chromium,titaniumandoxygenisotopicandconcentrationdataforterrestrialsamples,bulkchondritesandchondrules.

Sample Type^a ∅

[mm]^b

Weight [mg]^c

Cr [ppm]^d

n^e ε⁵³Cr±2σ^f ε⁵⁴Cr±2σ^f Ti [ppm]^b

ε⁵⁰Ti^b±2σ ¹⁷O±2σ

Terrestrial

DTS2b Silicates^g 27 1551 8 0.12±^0.07 ^0.16±^0.13

DTS2b Chromites^g 7 −0.03±0.20 −0.02±0.34

JA-2 161 423 5 0.03±^0.16 ^0.13±^0.28

BIR1a 240 392 3 0.08±0.17 0.07±0.30

Average 0.05±^0.13 ^0.08±^0.16

OC (Ragland - LL3.4)

MSc42 Type II, POP 2.7 17 4108 1 0.40±^0.10 −^0.24±^0.24 ⁶⁰¹ −^0.43±^0.14 ^1.28±^1.24

MSc43 Type II, RP 2.1 14 4613 5 0.45±^0.17 −^0.11±^0.18 ³²⁷ −^0.41±^0.13 ^0.82±^0.63

MSc44 Type II, RP 2.4 13 4693 3 0.25±0.15 −0.34±0.31 550 −0.48±0.13 1.58±2.25

MSc45 Type II, RP(O) 2.6 18 3959 4 0.37±^0.16 −^0.14±^0.34 ⁸⁰³ −^1.00±^0.12 ^1.83±^0.97

MSc49 Type II, RP 1.4 1.8 4316 3 0.74±0.13 −0.39±0.32 439 −0.67±0.56 1.43±0.60

MSc50 Type II, PO 1.4 2.4 3870 741 −^0.53±^0.08 ^1.15±^0.39

MSc51 Type II, C 1.0 1.2 4593 2 0.47±^0.13 −^0.51±^0.32 ⁵⁵³ −^0.58±^0.08 ^1.79±^0.83

MSc52 Type II. POP 1.4 2.8 800 −0.60±0.09 1.09±1.29

MSc53 Type II, RP 1.5 1.2 472 −^0.24±^0.15 ^1.07±^1.09

Average 1.8 4307 0.45±0.33 −0.29±0.31 587 −0.55±0.42 1.34±0.69

Std8 Whole rock 51 3034 4 0.21±^0.06 −^0.31±^0.19 ⁶⁰⁶ −^0.70±^0.13

EC (MAC02837 - EL3)

MSc55 Type I, PP 1.3 1.3 2200 443 −0.13±0.11 −0.20±0.54

MSc57 Type I, PP 1.2 0.9 2217 546 −^0.21±^0.13 −^0.02±^0.44

MSc58 Type I, RP 0.9 0.7 1865 408 −1.47±0.19 −0.08±0.81

MSc61 Type I, PP 1.7 4.4 796 1 0.18±^0.06 ^0.01±^0.19 ¹³⁷ −^0.28±^0.08 ^0.38±^0.94

Average 1.3 1769 384 −^0.52±^1.27 ^0.02±^0.51

Std9 Whole rock 54 2843 1 0.16±^0.06 ^0.02±^0.19 ⁴⁸⁶ −^0.40±^0.12

CV (Allende - CV3)

MSc2 Type I, PO 1.5 1.5 2493 5 0.12±0.25 1.24±0.14 1190 4.85±0.16 −5.21±1.29

MSc3 Type I, PO(P) 1.2 1.7 2670 5 0.07±^0.16 ^0.71±^0.19 ⁷⁹³ −^0.15±^0.14 −^5.95±^0.75

MSc5 Type I, BO 1.2 1.3 3295 4 0.05±0.17 1.16±0.21 1401 1.73±0.18 −4.82±1.22

MSc8 Type I, POP 1.9 5.7 3121 7 0.13±^0.10 ^1.43±^0.16 ⁷⁷⁵ −^0.05±^0.08 −^2.98±^1.21

MSc10 Type I, PO 1.4 2.0 2844 6 0.07±^0.12 ^0.82±^0.15 ¹⁰¹⁵ ^1.30±^0.11 −^4.99±^1.90

MSc12 Type I, PO 1.8 2.9 3307 6 0.07±0.12 0.09±0.11 953 1.56±0.20 −5.43±1.45

MSc14 Al-rich 1.8 4.2 2987 3 −^0.10±^0.13 ^0.44±^0.32 ¹⁷⁶⁰ ^7.32±^0.09 −^5.74±^1.13

Average 1.5 2960 0.06±0.15 0.84±0.95 1127 2.36±5.48 −5.02±1.97

AL 25 Chondr. fraction 0.25−^1.6 ³⁰² ¹⁷ ^0.07±^0.09 ^0.55±^0.23 ¹²⁰⁴ ^2.50±^0.12

AL 32 Chondr. fraction 0.5−^1.0 ³⁸⁴ ¹⁰ ^0.03±^0.08 ^0.65±^0.19 ¹¹⁶¹ ^2.17±^0.15

AL 36 Chondr. fraction 0.5−1.0 518 15 0.05±0.08 0.55±0.18 1029 2.24±0.14

Average 0.05±^0.04 ^0.58±^0.11 ¹¹³¹ ^2.30±^0.35

AL 26 Matrix fraction 506 14 0.30±0.07 1.18±0.14 656 2.68±0.07

AL 27 Matrix fraction 394 17 0.15±^0.04 ^0.98±^0.13 ⁶⁶³ ^1.91±^0.19

AL 42 Matrix fraction 469 8 0.17±^0.09 ^1.01±^0.19 ⁶⁹⁶ ^2.25±^0.15

Average 0.21±0.16 1.06±0.22 672 2.28±0.78

Allende 2 Whole rock 36 6 0.12±^0.12 ^0.96±^0.22

Allende 2 Whole rock 41 4 0.11±0.10 1.06±0.30

Ti1 Whole rock 37 3970 4 0.13±^0.20 ^1.06±^0.34 ⁷³¹ ^3.31±^0.14

Ti2 Whole rock 32 4013 12 0.20±0.06 0.99±0.11 826 3.22±0.34

Std1 Whole rock 10 3198 6 0.19±^0.07 ^1.18±^0.18 ⁷²⁹ ^3.20±^0.16

Average 3727 0.15±^0.08 ^1.05±^0.17 ⁷⁶² ^3.24±^0.12

CR (GRA06100 - CR2)

MSc72 Type I, POP 0.9 0.9 3984 2 0.06±^0.13 ^1.25±^0.30 ⁸⁷³ ^2.21±^0.16 −^2.28±^0.60

MSc73 Type I, POP 0.7 0.4 4952 842 2.13±0.26 −2.58±1.69

MSc74 Type I, POP 1.1 0.9 3880 2 0.27±^0.13 ^1.39±^0.30 ⁶²⁴ ^1.98±^0.23 −^1.72±^0.47

MSc75 Type I, POP 1.1 1.0 5617 977 1.79±^0.16 −^3.65±^0.72

MSc76 Type I, POP 1.2 1.9 3778 2 0.21±0.13 1.20±0.30 782 1.41±0.15 −3.88±0.86

MSc77 Type I, POP 1.8 4.6 3706 2 0.24±^0.13 ^1.37±^0.30 ⁷⁸⁹ ^1.21±^0.08 −^0.89±^1.00

Average 1.1 4320 0.20±0.19 1.30±0.18 814 1.79±0.80 −2.50±1.14

Std10 Whole rock 51 3096 3 0.26±^0.13 ^1.23±^0.30 ⁷¹⁵ ^3.26±^0.09

CR (NWA 801 - CR2)

MSc78 Type I, POP 1.6 1 3393 642 3.39±^0.17 −^1.75±^1.10

a TypeIchondruleshaveFa-content<10,typeIIchondrules>10;PP—porphyriticpyroxene,PO—porphyriticolivine,POP—porphyriticolivine-pyroxene,BO—barredolivine, RP—radialpyroxene,C—cryptocristalline.

b DatafromGerberetal.(2017).

c Massdenotesdigestedsamplemass,forchondrulesthisrepresentsbetween80and95%ofchondrulemass.

d ErroronCrconcentrationmeasurementsisestimatedtobe10%.

e NumberofindividualCrisotopemeasurements.

f Chondruledata:forn≥⁴^the^95%^confidenceîntervalîsûsed,^forⁿ≤⁴^the^2s.d.ôf^the^NIST3112a^standardôf^the^respective^sessionîsûsed,^forâverages^the^2s.d.îs^given.

g BulkpowdersofDTS2bweredigestedinaquaregiaandsubsequentHF-HNO3.TheresidualchromitesweredigestedinParrbombswith2mlconc.HNO3at190^◦Cfor4 days.

(5)

Fig. 2.(a)Back-scatteredelectronimageofPOchondruleMSc10fromAllende(CV3) showingthelocationsofionprobemeasurements(coloredellipses).Thecolorsin- dicatetherangeof¹⁷Ovalues.(b)CLimagerevealinginternalstructuresofolivine grains.¹⁶O-richrelictolivinegrainsshownoCLemissionduetolowabundancesof CLactivators(e.g.,Ca,Al,Ti)whilehostolivinegrainsshowCLemissioninducedby higherconcentrationofCa,Al,andTi.

value of 0.02±^0.51^. ^The ^CR ^chondrules ^display ^more ^variable ¹⁷Owithvaluesfrom−^3.9^to−^1.5^,ând^for^chondrules^from Allende¹⁷Ovariesfrom−³^to−⁶^.Îtîsîmportant^to^recognize that theuncertaintyonthebulk¹⁷Oforsomeoftheindividual CCchondrules islargerbecausethechondrulesareinternallyhet- erogeneous.Thus, withinandamongindividual CCchondrules(in particularCVchondrules)thereisOisotopevariability.Bycontrast, different grains within individual NC chondrules are isotopically homogeneous,andvariabilityamongNCchondrulesislimited.

3.2. Chromiumisotopes

The Cr isotope data are reported in Table 1. In addition to the chondrules andbulk chondrites, we also report data for the terrestrialrockstandardsBIR1a,DTS2b,andJA-2.Thesewerepro- cessed through the full analytical procedures alongside the me- teorite samples to assess the accuracy and precision of the Cr isotopemeasurements.The

ε

⁵³Crvaluesoftheterrestrialsamples are indistinguishablefromNIST SRM3112a,buttheir

ε

⁵⁴Cr seems tobeoffsettoslightlyhighervalues.Thisisconsistent withsimi- larsmalloffsetsreportedformeasured

ε

⁵⁴Crofterrestrialbasalts inpriorstudies(Mougeletal.,2018;Qinetal.,2010).Similaroff- setsofnaturalrocksfromstandardsolutionsareobservedforother

Fig. 3.ε⁵⁴Crvaluesofindividualchondrules.Compositionsofrespectivebulkchon- dritesareshown assolid lines;shaded areasaround theselinesindicate2s.d.

uncertaintiesofthebulkvalues.Boldsymbolsrepresentdatafromthisstudy,open symbolsareliteraturedata(Trinquieretal.,2008;Yamashitaetal.,2010;Connelly etal.,2012;Olsenetal.,2016;Zhuetal.,2019).Bulkchondritevaluesarefromthis study(EC,OC,CR,andCV),andQinetal.(2010) (CB,CO).

elements(e.g.,Ti,Mo) andarelikelyduetonon-exponential fractionation processes, possibly induced during standard production (Zhangetal.,2011;Buddeetal.,2019).

TheCrisotopicdataforthebulkchondritesamplesarealsoin good agreement with previously reported data for samples from thesamechondritegroups(Trinquieretal.,2007;Qinetal.,2010;

Göpeletal.,2015).Combined,thedataforterrestrialsamplesand bulkchondritesthereforeshowgoodinter-laboratoryagreement.

Individualordinarychondritechondrulesdisplayvariable

ε

⁵³Cr rangingfrom∼^0.25^to∼^0.74, ^which^for^most^chondrules^are^also higher than the bulk ordinary chondrite value. Chondrules from the other chondrite groups investigated in this study show less variabilityand are more similar to the valuesof their respective hostchondrites. Thevariable

ε

⁵³Cr amongthe OCchondrules al- mostcertainlyreﬂectsradiogenic variationsresultingfromthede- cay of short-lived ⁵³Mn (t1/2∼³.7 Ma). However, asthe Mn/Cr ratios of the chondrules were not determined in this study, the magnitudeofpotentialradiogenic

ε

⁵³Crvariationscannotbequan- tiﬁed. Regardless, the main focus of this study is on nucleosyn- thetic

ε

⁵⁴Cr variations, and the

ε

⁵³Cr will therefore not be dis- cussedfurther.

(6)

The

ε

⁵⁴Cr values ofthe EC, OC, andCC chondrules andtheir hostbulk chondrites areshowninFig. 3. TheOCchondrules dis- playnoresolvable

ε

⁵⁴Crvariationsandtheirmean

ε

⁵⁴Crareindis- tinguishable from thehost chondrite’s

ε

⁵⁴Cr. Likewise, the

ε

⁵⁴Cr ofthe singleEC chondruleofthisstudy isindistinguishablefrom the

ε

⁵⁴Cr of bulk enstatite chondrites. Compared to the EC and OCchondrules,allCRchondruleshaveelevated

ε

⁵⁴Cr,andmostof them areindistinguishable fromthe

ε

⁵⁴Cr ofbulk CRchondrites.

By contrast,individual Allende chondrulesexhibit highly variable

ε

⁵⁴Cr, with a continuum from ∼^0.1 ^to ∼^1.4

ε

⁵⁴Cr (Fig. 3). The

ε

⁵⁴CrresultsforCRandCVchondrules areconsistent withthose of Olsen et al.(2016), although these authors reported a some- what larger range of

ε

⁵⁴Cr valuesamong CR andCV chondrules (Fig.3).Finally,thethreepooledchondruleseparatesfromAllende, each consisting of hundreds of chondrules (Budde etal., 2016b), areindistinguishablefromoneanotherandhaveanaverage

ε

⁵⁴Cr of0.58±^0.12.^This^value^is^resolved^from^the^bulk^CVcomposition (

ε

⁵⁴Cr ∼¹⁾ ând^fromÂllende^matrix,^which^hasân âverage

ε

⁵⁴Cr of∼^1.

3.3. Titaniumisotopes

To facilitate direct comparison of the O and Cr isotopic data withpreviouslyobtainedTiisotope dataon thesamechondrules, theTi concentrationand

ε

⁵⁰TidatafromGerberetal.(2017) are also provided in Table 1,and the resultsare brieﬂy summarized here.WhereasECandOCchondrulesexhibitonlysmall

ε

⁵⁰Tivari- ations relative to the composition of their host chondrites, CC chondrules are characterized by a broad range of

ε

⁵⁰Ti values, rangingfromtheslightlynegative

ε

⁵⁰TitypicallyobservedforOC chondrules tothelarge

ε

⁵⁰Ti excessesknownfromCa–Al-richin- clusions (CAIs) (Fig. 4). The CC chondrules also display a broad correlation of

ε

⁵⁰Ti andTi concentration, indicating the addition ofisotopicallyheterogeneousandTi-rich—i.e.,CAI-like—materialto enstatiteandordinarychondrite-likechondruleprecursors(Gerber etal.,2017).

4. Discussion

WhilethenewOandCrisotopedataforsingle chondrulesare consistent with results of previous studies, the major advantage of thisstudy is that,for theﬁrst time, the Cr, O, andTi isotopic data havebeen obtainedforthe same individual chondrules.We willshow belowthat althoughtheisotopicdataforeachelement alone reveal isotopic heterogeneity among chondrules, only the combineddataofallthreeelementsmakeitpossibletodisentan- gletheprocess(es)responsiblefortheseheterogeneities.

ThekeyobservationsofthecombinedCr,Ti,andOisotopicdata maybesummarizedasfollows(Fig.4).First,OCandECchondrules displayrelativelyhomogeneous ¹⁷O,

ε

⁵⁰Ti,and

ε

⁵⁴Cr,whichare indistinguishable from the compositions of their respective host chondrites.Second,CCchondruleshave,onaverage,moreelevated

ε

⁵⁰Tiand

ε

⁵⁴Cr,andlower¹⁷OcomparedtoNC(i.e.,OCandEC) chondrules. Although the compositionsof CC and NCchondrules mayoverlapforeither

ε

⁵⁰Ti,

ε

⁵⁴Cr,or¹⁷O,inmulti-isotopespace noneoftheinvestigatedCCchondrulesplotsinthecompositional ﬁeldofNCchondrites,andnoneoftheinvestigatedNCchondrules plotswithintheCCﬁeld(Fig.4a-c).Finally,incontrasttoNCchon- drules, CC chondrules—and in particular CV chondrules—exhibit variableisotopeanomaliesforallthreeelements.Thedistinctiso- topic compositionsof NC andCC chondrules for a range ofgeo- andcosmochemiallydifferentelements,combinedwiththeobser- vation of isotopic homogeneity in NC chondrules versus isotopic heterogeneity in CC chondrules represents a fundamental differ- encebetweenNCandCCchondrites.

Below, we ﬁrst attempt to identify the carrierof the isotopic anomaliesamongthechondrules,andthenusethecombinediso- topicdatatoevaluate mixingandtransport processesbefore and duringchondrite parentbody accretion.Finally,we willplacethe chondrulecompositionsintothebroadercontext oftheevolution ofdistinctdiskreservoirs,namelytheNCandCCreservoirs.

4.1. Originofisotopicheterogeneitiesamongchondrules

When the

ε

⁵⁴Cr data for chondrules from NC and CC chon- dritesare consideredtogether,chondrulesfromchondriteswitha higherabundanceofCAIs(i.e.,CV,CO)exhibitmore

ε

⁵⁴Crvariabil- ity compared to chondrites with lower CAI abundances (i.e., EC, OC,CR). The same observation holds for

ε

⁵⁰Ti variations among chondrules,whichhavebeenattributedtotheheterogeneousdis- tribution of CAIs among the chondrule precursors (Gerber et al., 2017). Speciﬁcally, the

ε

⁵⁰Ti variations amongtheCV chondrules are correlated withtheir Ti content, consistent withthe variable additionof⁵⁰Ti- andTi-rich CAIstochondruleprecursors(Gerber etal.,2017).Bycontrast,thereisnoobviouscorrelation between

ε

⁵⁴Cr andCr concentration amongthe CV chondrules, neitherin thedata ofthis study,nor when datafor CVand COchondrules frompreviousstudiesarealsoincluded(Olsenetal.,2016;Zhuet al., 2019). The

ε

⁵⁴Cr anomalies inthe chondrulesmay, therefore, reﬂectadmixtureof anisotopicallyhighly anomalous component, in which case small amounts of admixed material are suﬃcient togeneratetheobserved

ε

⁵⁴Crvariations,whiletheCrconcentra- tionswouldremainunaffected.Alternatively,theadmixedmaterial isisotopicallylessanomalous andhada Cr concentration similar tothechondrulesthemselves.

Materialsknown to have signiﬁcant ⁵⁴Cr excesses andwhich may be capable of shifting bulk isotopic compositions include presolargrainsandCAIs.Forinstance,Cr-bearingnanospinelsiden- tiﬁedinthematrixofprimitivecarbonaceouschondriteshaveel- evated ⁵⁴Cr/⁵²Cr ratios of up to 50x the solar composition (Qin etal., 2010; Dauphaset al., 2010; Nittler etal., 2018). Toassess whether these grains can account for the observed

ε

⁵⁴Cr varia- tionsamongchondrules, wecalculatedthe numberofnanospinel grainsthat wouldneed to be added toan averageCV chondrule togeneratetheobserved1-2

ε

⁵⁴Crvariations.Usingtypicalvalues forCV chondrules(1 mg; 3000 μg/gCr)andpresolarnanospinels (80-200 nm; 2.5x solar ⁵⁴Cr/⁵²Cr; Dauphas et al., 2010; Nittler etal., 2018), we ﬁnd that tens of thousands ofgrains would be neededtogeneratethe observed 1-2

ε

⁵⁴Cr variations amongthe chondrules. The basicpicture doesnot change even when grains withthe highest reported anomalies (56x solar) are used in the calculation,sincethe ⁵⁴Cr anomaliesofthenanospinelgrainsare inverselycorrelatedwiththeirsize(Nittleretal.,2018).Therefore, if presolar nanospinels were the only source of

ε

⁵⁴Cr variations amongchondrules,thentheamountofpresolargrainsmixedinto differentchondrules wouldhave to varyby tens of thousandsof grainsfromone chondrule tothe next. Such a scenariois highly unrealisticintermsofmixingdynamicsofdustin thesolarneb- ula,andcontrasts withtheoverall rarityofpresolargrainsinthe meteoriterecord.Assuch,presolarnanospinelsdonotseemtobe responsibleforthe

ε

⁵⁴Crvariabilityamongindividualchondrules.

Aswas shownbyGerber etal.(2017), Tiisotope variationsin CCchondrulesweremostlikelycausedbyincorporationofvariable amountsofrefractory inclusionssuch asCAIs.However, although CAIshavewellresolved

ε

⁵⁴Crexcesses,theycontaintoolittleCrto signiﬁcantlyaffecttheCrisotopic compositionofchondrules, and soadmixtureofCAIstochondruleprecursorsresultsinastrongly curvedmixinghyperbolain

ε

⁵⁰Tivs.

ε

⁵⁴Crspacewithlittleover- allvariationsin

ε

⁵⁴Cr(Fig.4d).Accordingly,thechondrulesofthis study do not plot on a mixing line betweenNC chondrules and CAIs (Fig. 4d-f). Moreover, there are chondrules with

ε

⁵⁴Cr ex-

(7)

Fig. 4.¹⁷O,ε⁵⁰Ti,andε⁵⁴CrofindividualchondrulesincomparisontobulkNCandCCmeteorites(a-c)andCAIs/AOAs(d-f).TheNC(red)andCC(blue)fieldsrepresent therangeofanomaliesmeasuredforbulkmeteoritesfromtheNCandCCreservoir,respectively.TheNCandCCfieldsaswellasrepresentativeCAIisotopiccompositions arebasedondatacompiledinBurkhardtetal.(2019).Bulkchondritedatapoints(opensymbols)areonlyshownforsamplesfromwhichindividualchondrules(filled symbols)wereanalyzedinthisstudy.(a)-(c)NCchondrulesdisplayrelativelyhomogeneous¹⁷O,ε⁵⁰Ti,andε⁵⁴Cr,whichareindistinguishablefromthecompositionsof theirrespectivehostchondrites,whileCCchondruleshavemorevariableε⁵⁰Ti,ε⁵⁴Cr,and¹⁷O.NotethatalthoughthecompositionsofCCandNCchondrulesmayoverlap foreitherε⁵⁰Ti,ε⁵⁴Cr,or¹⁷O,inmulti-isotopespacenoCCchondruleplotswithintheNCfield,andnoNCchondruleplotswithintheCCfield.Alsonotethatmost CCmeteoritesareaffectedbyaqueousalterationwhichmaymodifytheirpristine¹⁷Othroughtheinteractionwith¹⁶Opoorwater(-ice)(Pirallaetal.,2020).Alteration corrected¹⁷Ovaluesforbulkcarbonaceouschondriteswouldthusplotatslightlylower¹⁷O.(d)-(f)Thecompositionofallchondrulescanbereproducedbymixing betweenanNCcomponentwithchondriticCr/Ti/OratiosandanisotopicallyCAI/AOA-likecomponent(IC)having morevariableCr/Ti/Oratios.Crossesonmixinglines indicate20%stepsofICcomponentinNC-ICmixture.(Forinterpretationofthecolorsinthefigure(s),thereaderisreferredtothewebversionofthisarticle.)

cesses butno

ε

⁵⁰Ti anomalies,indicating that the

ε

⁵⁴Crexcesses are unrelatedtotheadditionofCAIs.Theseobservationsruleout CAI admixture as a signiﬁcant driver for

ε

⁵⁴Cr variations among chondrules,andshowthat the⁵⁰Tiand⁵⁴Cranomaliesarecarried bydistinctmaterials.

Amoeboid olivine aggregates (AOAs) are another component that isabundant incarbonaceous chondrites andmayberespon- sible for the

ε

⁵⁴Cr variations among chondrules. Of note, AOAs are characterized by similar

ε

⁵⁴Cr and

ε

⁵⁰Ti excesses than CAIs

(Trinquieretal.,2009),butcanhavemuchhigherCr/Ti(Komatsu et al., 2001; Sugiura et al., 2009), reﬂecting their less refractory character comparedto CAIs.Moreover, there isgrowing evidence thatCCchondrulesinheritedprimitive¹⁶O-richprecursorsinclud- ingnotonlyCAIsbutalsoAOA-likerelictsilicategrains,suggesting admixtureofnon-CAIbutisotopicallyanomalousmaterialtochon- druleprecursors(Kita etal.,2010;Marrocchietal., 2019,2018a).

This isalso consistent with the observation that samples ofthis studyincludechondrulescontainingrelictolivinegrainswithneg-