HAL Id: insu-03197621
https://hal-insu.archives-ouvertes.fr/insu-03197621
Submitted on 14 Apr 2021
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
isotopic variations in Allende components: implications
for complementarity
Rayssa Martins, Marc Chaussidon, Zhengbin Deng, Francesco Pignatale,
Frédéric Moynier
To cite this version:
Rayssa Martins, Marc Chaussidon, Zhengbin Deng, Francesco Pignatale, Frédéric Moynier. A
con-densation origin for the mass-dependent silicon isotopic variations in Allende components:
impli-cations for complementarity. Earth and Planetary Science Letters, Elsevier, 2021, 554, pp.116678.
�10.1016/j.epsl.2020.116678�. �insu-03197621�
JID:EPSL AID:116678 /SCO [m5G; v1.297] P.1 (1-10)
Earth and Planetary Science Letters•••(••••)••••••
A
condensation
origin
for
the
mass-dependent
silicon
isotopic
variations
in
Allende
components:
implications
for
complementarity
Rayssa Martins
1,
Marc Chaussidon
∗
,
Zhengbin Deng
2,
Francesco Pignatale,
Frédéric Moynier
UniversitédeParis,InstitutdephysiqueduglobedeParis,CNRS,F-75005Paris,France
a
r
t
i
c
l
e
i
n
f
o
a
b
s
t
r
a
c
t
Articlehistory:
Received16April2020
Receivedinrevisedform28October2020 Accepted10November2020 Availableonlinexxxx Editor:W.B.McKinnon Keywords: siliconisotopes Allendechondrite condensation complementarity chondrules
Primitive chondrites have bulk compositions close to that of the solar photosphere, with however significantvariations ofelemental ratio relative tothe solar composition, dependingon thevolatility oftheelementsconsidered.Thisisclassicallyunderstoodasindicatingaprimarygeochemicalsignature duetotheformationofthecomponentsofchondrites(refractoryinclusions,chondrulesandmatrix),orof theirprecursors,throughcondensationofagasofnearsolarcomposition,plussecondaryvariationsdue toprocessessuchas(i)incompletevolatilizationofpresolarcomponents,(ii)complexhigh-temperature exchangesbetweencondensedphasesandthe nebulargas,and(iii)sortingandtransport ofgrains in theaccretiondiskbeforeaccretionofchondriteparentbodies.Becausemostofthemassofchondritesis madebyelementswhichcondenseathightemperatures,equilibriumcondensationproducesingeneral little isotopic fractionation for these elements. Silicon is however an exception with per mil level equilibrium isotopic fractionation at high temperature between the SiO gas and condensed silicates, allowingtouse siliconisotopes inchondritestoconstrain theorigin oftheircomponents and toput attestscenariosofcondensation.
Individualcomponents(chondrulefragments,isolatedolivinesinthematrix,andmatrixfragments)ofthe carbonaceouschondriteAllendewereseparatedandanalysedathigh-precisionfortheirsiliconisotopic composition.Largevariationshavebeenfoundamongchondrules (δ30Sifrom-0.86±0.162 s.e.to +0.04±0.03for11chondrules),isolatedolivines(δ30Sifrom-0.51±0.122s.e.to+0.20±0.10
for 12 olivines), and matrix (δ30Si from -0.95± 0.08 2 s.e.to -0.41± 0.01 for 17 matrix fragments).Thesevariationsdistributeonbothsidesofthebulk
δ
30SivalueofAllende(-0.43±0.032 s.e.,Armytageet al.,2011;Pringleet al.,2013,2014;SavageandMoynier,2013).Thereisaglobalpositive trendbetweenδ
30SivaluesandMg/Feratioforchondrulesandisolatedolivines.Thissystematicsappears inagreementwithwhatcanbemodeledforproducingAllendecomponents,ortheirprecursors,from fractionatedcondensationofasingle gaseousreservoirhavinginitiallythesiliconisotopiccomposition ofbulk Allende. Massbalance taking intoaccountthe meanabundances and δ30Sivalues of Allende componentsisconsistentwiththeiraccretionintheAllendeparentbodyintheproportionsproducedby thecondensationoftheparentparcelofnebulargas.Thissupportscomplementaritybetweenchondrules, olivinesandmatrixasbeingaprimaryfeature.However,thisconclusioncannotbedefinitivebecauseof theuncertaintiesindefiningmeanδ
30SivaluesforAllendecomponents.*
Correspondingauthor.E-mailaddress:chaussidon@ipgp.fr(M. Chaussidon).
1. Introduction
Duringthe formation ofthe solar System, mostchemical ele-ments were introduced in the accretion disk in the form of in-terstellardust grains, the gaseous phase being madedominantly ofaH2 +COgaswitha fractionofother volatilessuch as
nitro-genorraregases.Becausehightemperaturesdevelopintheinner accretiondiskdueto radiativeandviscous heating, the interstel-lardustcanbeeithervaporized,thermallyprocessedorpreserved, dependingwhetheritisinjectedinthediskcloseorfarfromthe
1 Presentaddress:DepartmentofEarthScienceandEngineering,ImperialCollege
London,LondonSW72AZ,UK.
2 Presentaddress:CentreforStarandPlanetFormation,GlobeInstitute,
Univer-sityofCopenhagen,Denmark.
formingSun(Pignataleet al.,2018).Thecomponentsmaking prim-itive chondrites (Ca-, Al-rich inclusions or CAIs, chondrules and matrix) formed at various temperatures and times in the accre-tiondisk(e.g.ScottandKrot, 2014). Theirprecursors canthusbe considered to be either (for the major part) the products of the condensationofagasmadebyvaporizationofinterstellardust,or interstellardustwhichescapedtotalevaporationandwasvariously processedorpreservedintheaccretiondisk.
Thefactsthat(i)primitivechondriteshavebulkchemical com-positions close to that of the solar photosphere with carbona-ceous chondrites and more specifically CI chondrites having the best match (Lodders, 2003), and that (ii) the differentchondrite groupsshow elementalfractionations relativeto CIdepending on temperatureofcondensationoftheelements (Palmeet al.,2014), are generally considered to indicate that the three major com-ponents of chondrites, ortheir precursors, formed fromthe con-densation of a gas close to solar in composition. Carbonaceous chondrites are enriched in refractory elements and depleted in volatile elements. Among carbonaceous chondrites, the CV chon-drites are the most fractionated with e.g. Al/Si ratio
≈
1.4 times higherthanCAI,Mg/SiratioslightlyhigherthanCI,andNa/Si ra-tio≈
0.5timesCI(Palmeet al.,2014).Thesimplestinterpretation ofsuch afractionationinchondrites wouldbe toconsiderthat it reflects various proportions ofmixing between high-temperature and low-temperature (i.e. early and late) products of a conden-sation sequence.However, the realprocesses at play behind this fractionationhavelongbeencontroversial(e.g.Anders,1977;Wai and Wasson, 1977; Wasson, 1977). Recently, it was shown that mixing of four components (a CAI-like component, a chondrule-likecomponent, ananhydrous matrix-likecomponent, andwater) could explain thediversity both in chemical fractionationand in non-mass-dependentisotopiccompositionsofcarbonaceous chon-drites(Alexander,2019).Mostofthe massofcondensableelements is madebyO, Mg, SiandFe:theMg/Siratiochangesfrom2to1betweenforsterite andenstatite,predictedtocondensesequentiallyfromasolargas (Grossman, 1972). In CV chondrites, the fact that the two ma-jor components hosting these elements (chondrules and matrix) have compositions (Mg/Si chondrules >Mg/Si CI >Mg/Si matrix) andproportions (45%vol forchondrules and40% volformatrix) resulting in a solar like bulk composition hasled to the idea of chemical complementaritybetweenchondrules andmatrix(Hezel andPalme,2010; Palmeet al.,2015).Thisseems verifiedevenat smallscales,i.e.formassesof600mgintheAllendeCV3chondrite forinstance(Palmeet al.,2015).Complementaritymeansthat ma-trixmineralsaremadefromthetotalcondensationoftheparcelof gasremainingafterthecondensationandextractionofthe precur-sorsofthechondrules,orthat matrixmineralsformedfromagas processedby theformationofchondrules(e.g.Bland et al.,2005; Friend et al., 2016; Hezel et al., 2018 and refs therein). Thus, a key implication ofcomplementarity is that matrixcannot have a CIcomposition, i.e.that oftheinitialgas.Thishasbeenobserved but its real meaning has been challengedarguing that the non-CIcompositionofthematrixisapost-accretionsecondaryfeature relatedtoparent-body fluid-assistedchemicalexchangesbetween chondrules andmatrix (Zandaet al.,2018). Inaddition,the rela-tive abundance of presolargrains andorganicmatter in the ma-trixof themostprimitivechondrites is CI-like(Alexander, 2005). However, complementarity seems present in different classes of CV chondrites despite large variations in abundance and chemi-cal compositions between matrix and chondrules between these classes(e.g.HezelandPalme,2010;Ebelet al.,2016).
Observed isotopic variations, either mass-dependent or non mass-dependentofnucleosyntheticorigin,arenotdecisive regard-ingcomplementarity.Strontiumisotopicvariations(88Sr/86Srratio)
betweenchondrules(depletedin88Srbyupto
≈
-1.7)and
ma-trix(enrichedin88Srbyupto
≈
+0.7)observedinAllendehas
been interpreted as reflecting isotopic fractionation during fluid-assistedmetamorphismontheparentbody,butthelight88Sr/86Sr
ratio of the CAIs was considered as primary andreflecting frac-tionationduringeithercondensationorelectromagneticsortingof thepartiallyionizednebula gas(Moynier et al., 2010). RecentZn isotopic data (66Zn/64Zn ratio) on chondrules and matrix of the
CV3.1 Leoville (Van Kooten and Moynier, 2019) revealed signifi-cant variations between the olivine-rich core of chondrules and theirigneousrims(thecoresbeingdepletedin66Znby
≈
-0.4)
andbetweenthe matrixandtheigneousrimsofchondrules (the matrixbeing enriched in 66Znby
≈
0.2). These variations are
notinterpretedasreflectingcomplementaritybutasreflecting pri-marilyZnisotopic fractionations duetothe presenceof asulfide immisciblemelt during interactions between the nebula gas and thepartially meltedchondrules (VanKootenandMoynier, 2019). Atvariance,chondrulesandmatrixinAllendeappear complemen-taryfortheir183Wisotopeanomalies(183Wdepletionfrom-1.5to
-0.7ppminmatrix, 183W excessfrom+1.4 to+2.3 ppmin
chon-drules,andnoanomalyinbulk)inagreementwiththeirformation froma single reservoir ofdust in which the presolar carriers of the183W anomaliesweresortedbetweentheprecursors of
chon-drulesandofmatrix(Buddeet al.,2016).Alternatively,ithasbeen proposed that these variations in 183W betweenchondrules and
matrixcouldbeduetothepartialremobilizationofWfrom chon-druletomatrixduringhydrothermaloxidationofchondrule metal-blebs(Alexander,2019).Finally,therangeofFeisotopicvariations of chondrules from Allende andfrom the CM chondrite Murchi-son (variations of the 56Fe/54Fe ratio from
≈
-0.6 to≈
+0.4 aroundthebulkvaluesofAllendeandMurchison) isindicativeof formation of the chondrules from a single reservoir from which thematrix alsoderives, withmost oftheisotopic variabilitydue topre-accretion evaporation/recondensationprocesses andnot to post-accretionhydrothermalismandmetasomatism(Mullaneet al., 2005; Hezelet al.,2018).Noneoftheseisotopicstudies,withthe exception of Fe isotope studies, concern the major elements (O, Mg,Si)thatrepresentthemajormassofchondrulesandmatrix.Here we present a high-precision Si isotope studyof individ-ualcomponents inthe CV3chondrite Allende: single chondrules, Mg-rich olivines isolated in the matrix and bulk matrix sam-ples. Thisstudywas designedto search forsystematics in mass-dependent Si isotope variations among Allende components that could test and possibly further constrain (i) their originby con-densationprocesses fromasingle gaseous reservoirand(ii) their chemicalandisotopiccomplementarityacquiredbyfractional con-densationofthisinitialparentgaseousreservoir.Siisotopesshow largemass-dependent equilibriumisotopic fractionationsat high-temperaturebetweenSiOgasandsilicate(e.g.
30Sienstatite-gasand
30Siolivine-gas increasing from
≈
+1 to +2 with temperaturedecreasing from 2000 K to 1450 K, Clayton et al., 1978; Javoy et al.,2012;Méheutet al.,2009)sothatduringacondensation se-quencethecondensatesandremaininggaswilldevelopspecificSi isotopiccompositions.Allendewaschosen because(i) itisa case studyforcomplementarity (Hezel andPalme, 2010; Palmeet al., 2015), (ii) existing Si isotopic datafor theCa-, Al-rich inclusions (Clayton et al., 1988; Grossman et al., 2008) and for chondrules andmatrix(Armytage,2011;Kadlaget al.,2018;Villeneuveet al., 2020) show thepresenceoflarge isotopicvariations,(iii)the rel-ative abundances of the various components have been recently redetermined (Ebel et al., 2016), and (iv) it is, among CV chon-drites, the one richest in isolated Mg-rich olivinesin the matrix thatcan thusbe handpickedandanalysed individuallyforSi iso-topecomposition. Despite Allende being an oxidized CV of type >3.6(Bonalet al.,2006),detailedchemicalmappingshowsthatSi was not significantly redistributed by parent body processes be-tweenthedifferentcomponents(Ebelet al.,2016).
2. Methods
Chondrules,isolatedolivinesandmatrixfragmentswere hand-pickedfromaroughlycrushedfragmentoftheAllendeCV3 chon-drite. They were inspected microscopically and fragments were studied by scanning electron microscope (see supplementary for details).The isolated olivines, aswell asthe matrixsamples and aliquotsofthe brokenchondrules,were digested followingNaOH fusionprotocolsmodifiedafterGeorgetal.(2006).Theglasscakes produced were dissolved in Milli-Q H2O, and the solutions
pro-cessed through cation exchange resin to purify Si (Pringle et al., 2014).ThesesolutionsweremeasuredforSiisotopiccompositions on a Neptune plus (Thermo-Fisher) multi-collector inductively-coupled-plasmamassspectrometer(MC-ICP-MS)atIPGPwith pro-cedures dedicated to minimize blanks and analytical errors for small sample mass (see supplementary for all analytical details). 92runsoftheNBS-28standardsolutionat2ppmSiconcentration yielda2s.d.of
±
0.13anda2s.e.of±
0.01forδ
30Sivalues, whereδ
30Sirepresentsthepermildeviationofthe30Si/28SiratioofthesamplerelativetothatofNBS28standard.
3. Results
3.1. Range of Si isotopic variations
The11Allendechondrulesanalyzedshowalargerangeofbulk Siisotopicvariationswith
δ
30Si(whereδ
30Si=[(Rsample/RNBS28)-1]×
1000;R=30Si/28Si)from-0.86±
0.16(2s.e.)to+0.04±
0.03 (Table 1), inagreement withprevious data(range for10 Allende bulk chondrules from-0.71±
0.03 to-0.10±
0.03, Armytage, 2011).InsituanalysesinAllendechondrulesbyfemtosecondlaser ablationcoupledwithMC-ICP-MSalsoshowthepresenceoflarge Siisotopicvariationsfrom-1.28±
0.19to+0.32±
0.19 (Kad-lag et al., 2019).Evenlargerδ
30Si variations,from-3.41±
0.3 to +1.9±
0.3 werereported fromionmicroprobein-situ anal-yses (ata spatialscale of≈
10 μm)ofolivinesintype Iandtype II chondrules(311 spotsanalysedin17chondrules)fromAllende (Villeneuveet al.,2020;Fig.S4a).The12isolatedAllendeolivines studied show arangeof variations forδ
30Sivaluesfrom-0.51±
0.12to +0.15±
0.10(Table 1).Thisrange isalsolower than the rangefound fromionprobe analyses intwo Allendeisolated olivinesby Villeneuveet al. (2020) from-2.8±
0.3to0±
0.3 (17spotsin2isolatedolivines, withhoweveronespotat-4.5±
0.3).At variancewithbulk chondrulesandisolated olivines,the 17 samples of matrix show a significantly more negative range of
δ
30Si values from -0.95±
0.08 to -0.41±
0.01. The meanδ
30Sivalueforthematrixisof-0.65±
0.26(2s.d.).Thisis
in-distinguishablefromthevalueof-0.63
±
0.04(2s.e.)reported by Armytage (2011) for17analyses of ahomogenized sample of Allende matrix.Thepresentdatashow thepresence ofsmallbut significantSiisotopicvariationsinthematrix.Thesevariations, be-causeofthe matrixsamplingprocedure(see 2.Methods), cannot be attributed to thepresence of visiblefragments of chondrules, CAIsorisolatedolivines.Thesevariationsarehowevermuchmore restricted than the rangeofδ
30Sivalues from-1.33±
0.15 to +0.65±
0.21 found by in situ analyses in samples of matrix variouslyenrichedinrefractoryorvolatileelements (Kadlaget al., 2018).3.2. Mass-dependent isotopic variations
Notethatthepresentstudywas notdesignedtooptimize pre-cisionon
29Sibuttooptimizeprecisionon
δ
30Siforsmall sam-ples. Ina Sithree-isotopediagram (Fig. 1), all thepresentSi iso-topic variations appear mass-dependent within errors, exceptforFig. 1. Sithree-isotope diagramfor thepresent samplesofAllende components. Errorsshownare2s.e.Onlyonesample(ofisolatedolivine)showsamarginally significantnonmass-dependentisotopicvariation.Allotherssamplesdefineaδ29Si
versusδ30Silinewithaslopeof0.536(+0.017/-0.035)consistentwiththeslope ofthemassfractionationline(0.517).Notethatisolatedolivineshavehigherδ30Si valuesthanmatrix,whilechondrulesdistributeoverthewholerangeofvariations.
oneisolatedolivinewhichhasamarginallysignificant29Sideficit
with
29Si = -0.16
±
0.10 (29Si being defined as
δ
29Si – 0.517×
δ
28Si). Some Allende CAIs are known to contain large Si isotopeanomalieswith29Si from-0.12to+1.18(7among26 CAIsstudiedinClaytonet al.,1988).Thesenon-zero
29Sivalues can be considered to reflect the presence within CAI precursors of presolar components carrying Si isotopic anomalies of nucle-osyntheticorigin. Thus, except forone sample, all the presentSi isotopic variations reflect Si isotopic fractionations having taken placeduringtheformationofmatrixminerals,chondrulesand iso-latedolivines, oroftheir precursors.Ifpresolarcomponentswere presentintheirprecursors, theymusthavebeeneithervaporized orprocessed inthe nebular gas atsufficiently-high temperatures tolosetheirinitialSiisotopicsignature.Studiesofvariouskindof meteoritesshow thatthe presolarnebula inbulkdidnot contain anySiisotopicanomaly relativeto theEarthwithin 15ppm(e.g. Fitoussiet al.,2009;Zambardiet al.,2013;Pringleet al.,2013).
3.3. Systematics among Allende components
AsshowninFig.2,allbuttwofragmentsofmatrixhave
δ
30Sivalueslower than bulk Allende (-0.43
±
0.102 s.d.,±
0.03 2 s.e., calculated for 10 different aliquots of Allende analysed in Armytage et al., 2011; Pringle et al., 2013, 2014; Savage and Moynier, 2013) while all isolated olivines (except 2) haveδ
30Si values higher than the bulk. Chondrules have bulkδ
30Si values distributingabove andbelowbulkAllendecomposition.Notethat CAIsfromCVchondrites(datafromClaytonet al.,1988and Gross-man et al., 2008) have very variableδ
30Si values ranging from -1.77to+5.96(averageof+1.42±
3.86),with12CAIs(out ofthe 28 CAIs with no Si isotopic anomaly) havingδ
30Si values within0±
1.Despite some outliers, there is a broad trend between bulk
δ
30Si values andMg# (Mg#= 100×
Mg/(Mg+Fe), with the Fe and Mgcontents in atomic%, see Table S1) for isolated olivines and chondruleswithδ
30SivaluesdecreasingwithdecreasingMg#(see Fig.5).Asimilartrendseemspresentinthein-situdatabyKadlag et al. (2019). The presentMg-rich isolated olivines have system-aticallyhigherδ
30Sivalues(averageof-0.03±
0.312s.d.for9 olivines)thanFe-richisolatedolivines(averageof-0.50±
0.052 s.d.for3olivines,Table1).Notethat,inallchondrulesbutone,the Mg#oftheolivinesarehigherthantheMg#ofthebulkchondrule fragment(TableS1),sothattheδ
30SivsMg#trendwouldbeeven better defined when considering isolated olivines and chondrule olivinestogether.IonprobedatabyVilleneuve et al.(2020) showTable 1
SiliconisotopedataforAllendecomponents.
Sample Type Mg#a δ29Si()b 2 s.d.c 2 s.e.d δ30Si()b 2 s.d. 2 s.e. ne
AL19A1-10 Mg-rich olivine 96.2 -0.01 0.16 0.08 -0.16 0.23 0.11 4 AL19F1-5 Mg-rich olivine 97.2 0.05 0.17 0.09 0.15 0.21 0.10 4 AL19F2-1 Mg-rich olivine 96.9 0.02 0.14 0.07 -0.06 0.38 0.19 4 AL19F2-2 Mg-rich olivine 95.2 0.05 0.06 0.03 0.08 0.13 0.06 4 AL19G2-4 Mg-rich olivine 96.6 -0.13 0.28 0.14 -0.29 0.47 0.24 4 AL19I2-1 Mg-rich olivine 96.1 -0.01 0.17 0.08 -0.05 0.20 0.10 4 AL19K2-5 Mg-rich olivine 96.5 -0.02 0.06 0.03 -0.05 0.16 0.08 4 AL19L1-1 Mg-rich olivine 97.0 -0.22 0.19 0.09 -0.11 0.31 0.15 4 AL19N1-3 Mg-rich olivine 98.7 0.14 0.08 0.04 0.20 0.22 0.11 4 AL19D1-3 Fe-rich olivine 91.4 -0.24 0.35 0.18 -0.51 0.24 0.12 4 AL19E2-2 Fe-rich olivine 84.8 -0.17 0.14 0.07 -0.47 0.18 0.09 4 AL19G1-4 Fe-rich olivine 90.0 -0.27 0.11 0.05 -0.51 0.02 0.01 4 Ch3 Chondrule 80.2 -0.31 0.05 0.02 -0.62 0.04 0.02 4 Ch6 Chondrule 92.0 0.05 0.03 0.02 0.04 0.05 0.03 4 AL19CH-1 Chondrule 88.0 -0.19 0.18 0.09 -0.39 0.10 0.05 4 AL19CH-2 Chondrule 82.3 -0.34 0.11 0.06 -0.77 0.22 0.11 4 AL19CH-3 Chondrule 90.8 0.21 0.12 0.06 0.33 0.05 0.02 4 AL19CH-4 Chondrule 89.6 -0.16 0.06 0.03 -0.39 0.04 0.02 4 AL19CH-6 Chondrule 85.6 -0.19 0.14 0.07 -0.49 0.07 0.04 4 AL19CH-7 Chondrule 83.5 -0.26 0.13 0.06 -0.71 0.05 0.03 4 AL19CH-9 Chondrule 83.7 -0.16 0.07 0.03 -0.40 0.16 0.08 4 AL19CH-11 Chondrule 85.0 -0.45 0.11 0.05 -0.86 0.32 0.16 4 AL19CH-12 Chondrule 86.6 0.18 0.04 0.02 0.30 0.06 0.04 3 Mtx1 Matrix 51.0 -0.21 0.05 0.02 -0.41 0.03 0.01 4 Mtx2 Matrix 53.8 -0.21 0.16 0.08 -0.46 0.10 0.05 4 AL19MX-2 Matrix 50.2 -0.31 0.07 0.04 -0.67 0.04 0.02 4 AL19MX-3 Matrix 50.7 -0.40 0.07 0.03 -0.74 0.12 0.04 8 AL19MX-5 Matrix 51.2 -0.33 0.06 0.03 -0.71 0.16 0.08 4 AL19MX-7 Matrix 55.5 -0.41 0.03 0.01 -0.81 0.07 0.04 3 AL19MX-9 Matrix 55.2 -0.30 0.05 0.03 -0.62 0.04 0.03 3 AL19MX-10 Matrix 53.7 -0.40 0.02 0.02 -0.68 0.07 0.05 2 AL19MX-11 Matrix 54.6 -0.31 0.08 0.03 -0.59 0.16 0.06 7 AL19MX-12 Matrix 53.4 -0.34 0.08 0.03 -0.66 0.14 0.06 6 Al-Mtx-1 Matrix n.d.f -0.33 0.05 0.02 -0.53 0.10 0.05 4 Al-Mtx-2 Matrix n.d. -0.47 0.10 0.05 -0.95 0.16 0.08 4 Al-Mtx-3 Matrix n.d. -0.37 0.04 0.02 -0.64 0.17 0.10 3 Al-Mtx-4 Matrix n.d. -0.42 0.09 0.05 -0.75 0.08 0.05 3 Al-Mtx-5 Matrix n.d. -0.30 0.12 0.06 -0.52 0.07 0.04 4 Al-Mtx-7 Matrix n.d. -0.36 0.08 0.04 -0.69 0.02 0.01 4 Al-Mtx-8 Matrix n.d. -0.36 0.14 0.08 -0.63 0.20 0.11 3
a Mg#=100×Mg/(Mg+Fe),MgandFecontentsinatomic%. b δXSi=[(R
sample/RNBS28)-1]×1000;R=XSi/28SiandNBS28theinternationalreferencestandard. c 2s.d.=2standarddeviation.
d 2s.e.=2standarderroronthemean. e n=numberofanalysesforthissample. f n.d.=notdetermined.
that
δ
30SivariationsaremorepronouncedinMg-richolivinesthan inMg-poor olivinesbutnotrendbetweenδ
30SiandMg#canbe identified(seeFig.S3forcomparisonbetweenthemeanδ
30Sival-uesforchondrulesfromionprobedataandthepresentbulk
δ
30Si values).4. Discussion
4.1. Silicon isotopic variations expected from kinetic isotopic fractionation
Previousstudies(analytical,experimentalandtheoretical)show that FUN-CAIs(fractionation andunidentifiednucleareffects) and “normal” CAIs exhibit large bulk Si isotopic variations with sev-eral
positiveandnegative variations ofδ
30Sivalueslikelydue to kinetic isotopic fractionations during Si evaporation and con-densation processes,respectively(e.g.Claytonet al., 1988; Gross-man et al.,2008;Mendybaevet al.,2013;Richter et al.,2002). In thesestudies,equilibriumcondensationisconsideredtohave pro-ducedonlysub-levelofSiisotopicvariations intheprecursors ofrefractoryinclusions.Presentdataandpreviousdata(Armytage, 2011;Kadlaget al.,2018)showthattherangeofbulkδ
30Sivalues in isolatedolivines, chondrules andmatrix ismuch morelimitedthan in refractory inclusions. This argues against kinetic isotopic fractionationprocessesplayingadominantroletocontrolthebulk Siisotopiccompositionofchondrules,isolatedolivinesandmatrix. However,thelarge
δ
30Siheterogeneities(from-7to +2.6) foundbyionmicroprobeatthescaleof≈
10 μmintype I chon-druleolivinesandlow-Capyroxenesandinafewisolatedolivines fromcarbonaceouschondriteshavebeeninterpretedtoshowthat, locallyinsome objects,theeffectsofkinetic Siisotopic fractiona-tionscanbeimportant(Villeneuveet al.,2020).Thiscanbeeither via the presenceof relict phases carryingkinetic light Siisotope enrichments such as olivinesderived from amoeboid olivine ag-gregaterefractoryinclusions(AOAshavebeenshowntohaveδ
30Sivalues from
≈
-9 to -1; Marrocchi et al., 2019b) or via Si kineticisotopicfractionationbetweenchondrulemeltsandthe am-bientgas(Villeneuveet al.,2020).Using17Ovalues(tracingthe
16O-richreservoirparentofAOAs;Krotetal., 2004) toidentifyin
chondrules relict AOAs olivineswithnegative
δ
30Sivalues shows thatthiswouldexplainaminorfractionofδ
30Sivariationin chon-drules(speciallyinAllende).Mostoftheintra-chondrulevariations isthusascribedtoSikinetic isotopicfractionationduring interac-tions between the ambient gas andchondrule melts (Villeneuve et al.,2020).JID:EPSL AID:116678 /SCO [m5G; v1.297] P.5 (1-10) R. Martins, M. Chaussidon, Z. Deng et al. Earth and Planetary Science Letters•••(••••)••••••
Fig. 2. Distributionofδ30SivaluesamongAllendecomponents:isolatedMg-rich olivinesfrom thisstudy, chondrulesfromthis studyand fromArmytage (2011), bulkAllendefromArmytageet al.(2011),CAIsfromClaytonet al.(1988) andfrom Grossmanet al.(2008).Toshowthedetailsoftheδ30Sivariationsinchondrules,
isolatedolivineandmatrix,thescalewaschosenfrom-1to+1. TwoCAIs withδ30Si<-1and15CAIswithδ30Si>+1,amongthe28CAIsinClaytonet al.
(1988) andGrossmanet al.(2008) withno29Sianomaly(seetext),arenotshown.
Similarly,AOAsolivineswhichhaveδ30Sivaluesfrom-9to-1(Marrocchiet al.,
2019b)arenotshown.
Wenotethat,whilekineticSiisotopefractionationduring con-densation of olivines seems effectively the only viable explana-tion for the very low
δ
30Sivalues of olivines fromAOAs, this is not the only possible explanation of theδ
30Si heterogeneity in olivines fromchondrules. This comes fromthe fact that Mg-rich olivines in AOAsare considered to haveformed, probably in the CAIs formingregion,by aggregationofhightemperature conden-satesofCAI,metalandMg-richolivine(Krotetal.,2004).Atthese high-temperatures, the major reservoir of Si is still the SiO gas (e.g.EbelandGrossman,2000;PetaevandWood,2005)sodespite significantequilibriumSiisotopicfractionationexistsathigh tem-perature (e.g.30Siolivine-SiO gas = +1.6 at1600K, Clayton et al.,
1978;Javoyet al.,2012)nostrongreservoireffectcanbeproduced bycondensationofolivines.Forinstancetheremovalof20%ofSi bythecondensationofCAIsandolivineswoulddecreasethe
δ
30Si of the remaining gas by only 0.4. This was pointed out early (Claytonet al.,1978).Thus,kineticSiisotopicfractionationis defi-nitelyrequiredtoexplainthelightSienrichmentsinAOAsolivines and this can constrain the duration of their condensation to be very short,daystoweeks(Marrocchi et al.,2019b).The situation isverydifferentforFe-Mgchondruleswhichcanbeconsideredto have among their precursors Mg-Fe olivines and pyroxenes con-densed attemperatureswhereaverysmallfractionofSiremains in the gas.At this stage, fractional equilibrium condensationcan leadto largelightSienrichments inthelastcondensates: infact, the distribution ofin-situδ
30Sivalues found by Villeneuve et al. (2020) is not better explained by kinetic isotopic fractionations during chondrule meltingthan by equilibrium isotopic fractiona-tions during condensationofchondruleprecursors (FigsS4a,b,c, d).Thus, while it is clear that
δ
30Si heterogeneity exists locally in chondrules and could result from non-equilibrium evapora-tion/condensation processesduring chondruleformation,itis un-likely that they could have resulted in the systematic variations (tenths of ofδ
30Sivalues,Fig.2) amongbulkisolatedolivines, chondrules and matrix. In the next sections, we investigate how these bulkδ
30Si variations could result from simple equilibriumcondensationofthesilicatedust formingtheprecursors of chon-drulesandofthematrix.
4.2. Limited silicon isotopic variations expected from full equilibrium condensation
In the following we consider the Si isotopic evolution of a parcel ofgas undergoing closed system full equilibrium conden-sation, to test if it can produce both the range of
δ
30Si values observed in Allende components,and the broad relationship be-tween Mg# andδ
30Si values of the condensates (Table 1). Theequilibriumcondensationsequence considersthat withinaparcel ofgas, atanytemperature during the condensation,the conden-sates areatchemical equilibriumwiththegas(Grossman, 1972). InacanonicalgaswithsolarcompositionandPtot=10−3–10−4atm,
thermodynamicsshowsthat siliconbegins tocondenseataround 1625Kinmelilite(Ca2Al2SiO7gehlenite- Ca2MgSi2O7 åkermanite
solidsolution)byreactionbetweentheSiOgasandtheCa-Al-rich phasescondensed athighertemperatures.Becausethesolar Al/Si andCa/Si ratios are low (both <0.1), there is not enough Al and CaavailabletocondenseallSiOfromthegas.SiOisthenremoved fromthegasatlowertemperatures(between
≈
1450Kand≈
1310 K) toformforsteriteandenstatite bycondensation andreactions withmorerefractoryphases.Thesetemperaturesofcondensationcanbesignificantlyhigher incaseofcondensationinnon-canonicalconditions.Dust-enriched systemshavebeenconsidered(PalmeandFegley,1990;Woodand Hashimoto,1993; EbelandGrossman,2000) becausethey would allowtosolvethediscrepancybetweentheobservationthat high-temperaturesilicates inprimitive chondrites have variableMg/Fe ratiosandthe predictionthat athightemperatureina canonical gas olivine andpyroxene are nearly pure forsterite andenstatite because Fe condenses as Fe-Ni metal (Grossman, 1972). Vapor-izing silicatedust in dust-enriched systems makes the gas more oxidizing,thusstabilizinguponcondensationsilicatesthatcan in-corporate some Fe and that have Mg/Fe ratios decreasing with temperature (Ebel and Grossman, 2000). This effect is enhanced ifwater-richCIdustisvaporized(FedkinandGrossman,2016).At a total pressure of10−3 atm, theMg# of olivine is predictedto
decrease from100 to 73% between 2000and 1650K for a dust enrichmentof
×
1000andfrom100to98%foradustenrichment of×
100(Ebel andGrossman,2000).Thus, condensation in dust-enrichedsystempredictMg#similartothoseobservedinAllende chondrulesandrefractory olivines(Table1), thefirstcondensates (madeatthehighesttemperature)beingthemostMg-rich.The condensates are expected to show significant Si isotopic variationsbecauseofthelargeequilibriumSiisotopicfractionation betweensilicate andSiO gas.The Si isotopic fractionation calcu-latedfromdensityfunctional theory (Javoy et al.,2012) between enstatiteandSiOgasisgivenby:
30Sienstatite-gas
=
4.
1877× (
106/
T2)
−
0.
056078× (
106/
T2)
2−
0.
0002044× (
106/
T2)
3 (1)implying that
30Sienstatite-gas increases from +1.58 to +2.42
for T decreasing from 1625 K to 1310 K. In this temperature range,no significantSiisotopic fractionationisexpectedbetween forsterite and enstatite with
30Siforsterite-enstatite ranging from
-0.03and-0.06
(Méheutet al., 2009). Calculationsdonotexist forrefractory silicatephasessuchasmelilite, butbecauseinthis temperature range30Si between silicates is expected to be 0
±
0.3 (Clayton et al., 1978; Méheut et al., 2009; Javoy et al., 2012), it will be assumed in the following that30Sisilicate-gas =
30Sienstatite-gasduringcondensation.
TheSiisotopiccomposition ofthebulksilicatecondensatecan becalculated fromsimplemassbalance ofSiisotopes. Incaseof
Fig. 3. Modelforacondensationsequenceatfullequilibriuminadust-enriched sys-tem(×100enrichmentinCIdust)foratotalpressureof10−3bar.Fig.3a:fractions
oftotalSiinCaO-MgO-Al2O3-SiO2(CMAS)liquid,MELTliquid,olivine,
orthopyrox-eneandtotalcondensateversusfractionofSiintheSiOgas(datafromEbeland Grossman,
2000
).Fig.3b:Siisotopiccompositionofthetotalcondensateandofthe SiOgasversusfractionofSiintheSiOgas,assumingfullisotopicequilibrium be-tweenthetwophases(seetext,SiisotopicfractionationfromJavoyet al.,2012
). DashedlinesshowtemperaturesinK(fromEbelandGrossman,2000
).Theyellow lineshowstheδ30SivalueofbulkAllende(-0.43)whichisassumedtobethatofthe initialgas.(Forinterpretationofthecoloursinthefigure(s),thereaderis referredtothewebversionofthisarticle.)
full isotopic equilibrium betweenthe condensed phases and the SiO gas, at any temperature during the condensation sequence, massbalanceisgivenby:
δ
30Si0=
fSi×δ
30Sigas+(
1−
fSi)
×(
30Sisilicate−gas+δ
30Sigas)
(2)
with fSi the fraction ofSi remaining inthe SiO gas atthis
tem-perature,
30Sisilicate-gas the equilibrium Si isotopic fractionation
betweensilicatecondensate andSiO gasatthis temperature (as-sumedtobe
30Sienstatite-gas),and
δ
30Si0 theSiisotopiccomposi-tionoftheSiO gasatthebeginningofcondensation(whenfSi=1).
Bulk carbonaceous chondrites show little variations in
δ
30Si val-ues with an average of -0.48±
0.10 (2 s.d.; Armytage et al., 2011),thebulkδ
30Sivalue ofAllendebeingof-0.43±
0.10(2 s.d.; Armytageet al., 2011;Pringleet al., 2013, 2014; Savageand Moynier,2013).Taking this value of-0.43
for the initial Si isotope compo-sition of the SiO gas, theδ
30Si values of the equilibriumsilicate condensatescanbe calculatedinacanonicalnebula orina dust-enrichedsystem.Inacanonicalcondensationsequence(Grossman, 1972),condensateswillshowδ
30Sivaluesdecreasingfrom+1.13 to-0.43fortemperaturesfrom1625K(takingfSi=0.99)to1310K (taking fSi=0), respectively. At variance, for condensation in a
dust-enriched system (Ebel and Grossman, 2000), fSi decreases
from1 at T >
≈
2200K to≈
0.38at T≈
1700 K (for dust en-richment×
100) or to 0.45at T≈
1950K (for dust enrichment×
1000) (Fig.3a andS5a).Thus, thehigh-temperature condensate willhavepositiveδ
30Sivaluesfrom≈
+0.42forthefirst conden-satesto+0.12at1700K(dustenrichment×
100),(or-0.06 at1950Kfordustenrichment×
1000,seeFigs. 3bandS5b).The differenceinδ
30SiofthesilicatesforasimilarfSifordifferentdustenrichmentsisduetothechangeof
30Sisilicate-gaswith
tempera-ture.
By definition of mass balance, when fSi=0, the
δ
30Si value ofthetotalcondensate is
δ
30Si0.Thus,iffull isotopicequilibriumismaintained betweenthe condensates and the gas, the
δ
30Si val-ueslower than -0.43 observed inchondrules, isolated olivines and the matrix cannot result from isotopic fractionation during condensation.Note also, that the presence of CAIs ora rangeinδ
30Sivaluesamongchondrules andisolatedolivinesare inconsis-tentwithfullequilibriumcondensation.4.3. Large silicon isotopic variations expected from fractional condensation
Ifequilibriumisnotmaintainedbetweenthecondensedphases andthe gas, the condensation is fractional. This can arise either fromcondensationwithtimescalesshorterthanthoserequiredfor re-equilibrationof solidswith thegas by diffusion,or from sim-pleremoval ofsolidsfromthe gasby e.g. vertical settling ofthe dustinthe disk.In termsofchemical andisotopicmassbalance, thesetwo processes are equivalent. Pignataleet al. (2016) calcu-lateda simple caseoffractionation in threesteps of a canonical solargas(withPtot=10−3atmandsolarcomposition)duringa
clas-sical condensationsequence. Inthismodel,refractories withiron andminorsilicatescondensedbetween1850K(fSi=1)and1443K
(fSi=0.88) arefirst fractionatedat1443K,ironwithforsteriteand
minorsilicatescondensed between1443Kand 1380K(fSi=0.55)
are fractionated at1380 K, andthe silicates condensed at lower temperatureareenstatitewithaminoramountofforsterite.From massbalanceand(1),the
δ
30Sivaluesofthephasessequentially fractionatedcanbecalculatedtodecreasefrom+1.4forthe mi-nor silicates fractionated at 1443 K, to +0.5 for the forsterite fractionatedat1380K,andfinallytoarangefrom+0.5to-1.7 fortheenstatitecondensatesfractionatedbelow1380K.Obviously, suchafractionalcondensationproducesaδ
30Sirange correspond-ingtotherangeobservedinAllendecomponents(Fig.2).However,asexplainedintheprevious section,condensationin dust-enrichedsystemswouldberequiredtoexplaintherange ob-served for both
δ
30Si values and Mg#. Calculations by Ebel and Grossman(2000) showthat,upon condensation,theMg#of con-densed silicates and fSi will decrease together with temperature,thus predicting a positive correlation between
δ
30Si values and Mg#.FirstorderpredictionscanbemadeforSiisotopicvariations during fractional condensation using the approach of condensa-tionwithisolation(PetaevandWood,1998,2005).InPetaev and Wood(1998) asingleparameterisusedtoquantifythefractionof solidsthat are fractionated(or isolated): the isolation degree (ξ
) simply definedas thepercentage per Kelvin ofcondensed solids withdrawn from further reaction with the residual gas at lower temperature.Fractional condensationcanbemodeled forthetwo cases studied above (Ptot = 10−3 atm and dust enrichments of×
100,×
1000; Ebel and Grossman, 2000) by considering that a fractionofCAIsisisolatedfromtheCMASmelt(fCAIs=ξ
×
fCMAS)andthatafractionofsilicateisisolatedfromthesilicatesat equi-librium with the silicate melt (MELT) and the gas (fSil-isol=
ξ
×
fSil-eq, subscripts Sil-isol and Sil-eq referring to isolated silicates
JID:EPSL AID:116678 /SCO [m5G; v1.297] P.7 (1-10) R. Martins, M. Chaussidon, Z. Deng et al. Earth and Planetary Science Letters•••(••••)••••••
Fig. 4. Modelforacondensationsequencewithisolation(ξ=0.25%/K,seetext)inadust-enrichedsystem(×100enrichmentinCIdust;EbelandGrossman,
2000
)foratotal pressureof10−3bar.Thecurvesshowhowtheδ30Siofthephasesatequilibrium(inredthemeltofCMAScomposition,inbluealltheequilibriumsilicates,i.e.meltandcrystals)withtheSiOgas(greycurve)changewithdecreasingtemperatureduringcondensation.TheinsertisahistogramshowingthefractionoftotalSihostedbythe isolated(orfractionated)solids(CAIsandolivines)versustheirδ30Sivalues.DashedlinesshowtemperaturesinK(fromEbelandGrossman,
2000
).δ
30SivaluesoftheSiOgasandofthephasescondensed(at equi-libriumorisolated)canbecalculatedatanystep(i)oftemperature dropfromtheδ
30Sivaluesatstep(i-1)andthefollowingmass bal-anceequation:δ
30Si0=
fSii× δ
30Siigas+ (
fC M A Si+
fM E LT Si+
fSili −eq)
× (
30Siisilicate−gas
+ δ
30Siigas)
+
i−1 x=0 fC A Ix× δ
30SixC A I+
i−1 x=0fSilx−isol
× δ
30SixSil−isol,
(3)with fAi and
δ
30SiiA the fractions of Si in phase A and itsδ
30Si value at step (i),30Siisilicate−gasthe equilibrium Si isotopic frac-tionationbetweensilicateandgasatthetemperature correspond-ing tostep (i).Fig. 4showstheresultsfor
ξ
=0.25%/K and dust enrichmentof×
100(see Fig.S6forξ
=0.25%/K anddust enrich-mentof×
1000).Thisvalueof0.25%/K forξ
isrequiredtoreachat theendofthecondensationalowδ
30Sivaluefortheequilibrium condensatesof-0.64thatissimilartothebulkδ
30Sivalueof Al-lendematrix(thelowestbulkδ
30SivalueofAllendecomponents). When fSi=0, the equilibriumcondensates host47.2% of all the Sicondensed,while3.4%isintheisolatedCAIsand49.4%inthe iso-lated silicates (Fig. 4). A value of 0.25%/K for
ξ
is also required fordustenrichmentsof×
1000togetaδ
30Sivalue-0.65atthe endofcondensation,butthefractionsofSihostedinthedifferent phases are slightly different (7.6% in isolated CAIs, 32.6% in iso-latedsilicates,59.8%inequilibriumsilicates;FigS6).Inbothcases, the isolated CAIs show the highestδ
30Si values (from +0.65 to +0.70 fordust enrichment×
100, Fig. 4) while the isolated silicates show a large range ofδ
30Si values (from +0.41 to -0.43fordustenrichment×
100, Fig.4). ThecovariationofMg# andδ
30Sivaluespredictedby themodelforcondensed phasesisshowninFig.5fordifferentvaluesof
ξ
(0,0.1,0.25%/K).Note that in this model of fractionated condensation for Si isotopes it is assumed that the phase diagram and phase com-positionscalculatedforfullequilibriumcondensationbyEbeland Grossman(2000) arestillvalid.Thisisofcourseincorrectbecause, ifaphase isfractionatedfromthecondensationsequenceto pre-vent re-equilibration of its Si isotopic composition with the SiO
gas,thechemicalcompositionofthebulkremainingsystemwould be modified. Thiswouldpossibly change phaserelationships and compositionsforthefollowingpartofthecondensationsequence whenisolationdegree
ξ
>0.20%/K (PetaevandWood,1998).Itis, however,beyondthe scopeofthepresentstudyto modelthatin detail.Anyway, due to thelarge Si isotopic variability ofAllende components,thetrendbetweenδ
30SiandMg#(Fig.5)is not de-fined precisely enough to be matched to a unique condensation scenario.Noteinadditionthattheoxidationstateofthegas (con-trolledbytheenrichmentfactorinCIdustandpossiblywaterice) isthemajorcontrolofMg#valueofthetotalcondensateatagiven temperature(Fig.5).Thus,therangeofbulkSiisotopiccompositionsobservedin Al-lendeisolatedolivines,chondrulesandmatrixcanresultfromthe fractionatedcondensationofagashaving initiallythesilicon iso-topiccompositionofbulkAllende.Thisrangeiscausedby(i) reser-voireffectsforSiandby(ii)changeswithtemperatureofthe equi-libriumSiisotopicfractionationbetweensilicatesandSiOgas.The modelpredicts,thatthemostrefractorysilicateswillhavepositive
δ
30Sivaluesandthatlessrefractoryoneswillhavelowerδ
30Si val-ues.Anisolationdegreeξ
of0.25%/Kisrequiredtoexplainthelowδ
30Sivaluesofthelessrefractorysilicates.Thisyieldsafractionof Si condensed in CAIs (from 3.4% to 7.6%) consistent with obser-vation in Allende (3.8%in CAIs and2.1% in AOAs; Table 2; Ebel et al.,2016).ThelargefractionofSicondensedinisolatedsilicates (from32.6%to49.4%)isalsoconsistentwiththeprevalencein Al-lendeandCVchondritesoftypeI(Mg-rich)chondrules(Scottand Krot,2014)containinginmanycasesrelictMg-richolivineswhich escapedre-equilibrationwiththesurroundinggas duringthe last chondrulemeltingevent(e.g.Chaussidonet al.,2008;Rudraswami et al.,2011; Marrocchiet al.,2019a,andrefstherein).Finally,the largefractionofSicondensedinequilibriumsilicates(from47.2% to59.8%)isinagreement withtheobservationsthat alarge frac-tionofolivinesintypeIchondrulesareformedatnear-equilibrium withthesurroundinggas,asimpliedbytheircomplexzoningand texture(LibourelandPortail,2018)andbydemonstratedgas-melt exchanges(Tissandieret al.,2002;Libourelet al.,2006;Marrocchi andChaussidon, 2015; Friend et al., 2016; Barosch et al., 2019). Notethat, at the larger scale ofthe disk, fractionationof olivine from the gas was proposed to be responsible of silicon isotopicFig. 5. VariationsoftheSiisotopiccompositionofisolatedolivines,chondrules,andmatrixversustheirMg#(datafromthisstudyandfromArmytage,
2011
).Thebulkδ30SivalueofAllendeiscalculatedfromdatainArmytageet al.,
2011
;Pringleet al.,2013
,2014
;SavageandMoynier,2013
.Notethatthegeneraltrendofdecreaseofδ30Si valueswithdecreasingMg#(between100and≈75)isbroadlyconsistentwithpredictionsmadeforthemodelsofacondensationsequencewithisolationofphasesin dust-enrichedsystemsasdescribedinFig.4for×100enrichmentandinFig.S6for×1000enrichment.Eachdotonthetwocondensationtrends(calculatedforisolationdegreeξ=0.25%/K)correspondstoincrementsoftemperatureof50Kfromminimumtemperatureof1650K(for×100enrichment)andof1450K(for×1000enrichment).For dustenrichment×1000,thetwoothercurvescorrespondtocondensationtrendscalculatedforξ=0.10%/Kandξ=0%/K(uppercurve,correspondingtothecaseofequilibrium condensation).Fordustenrichment×100,curvesforξ=0.25,0.1and0%/Karemostlysuperimposed.
Table 2
SidistributionandisotopiccompositionsamongAllendecomponents. Component fraction total Sia δ30Si 2 s.d. 2 s.e.
CAIs 0.038 1.42b 3.86 0.73 AOAs 0.021 n.d. n.d. n.d. Isolated olivines 0.017 -0.15c 0.52 0.16 Al-rich chondrules 0.025 n.d. n.d. n.d. Chondrules 0.351 -0.41d 0.64 0.14 Matrix 0.548 -0.65e 0.25 0.06 Bulk Allende -0.43f 0.10 0.03 a FractionoftotalSifromEbelet al.(2016).ThefractionofSihosted
byisolatedolivinesiscalculatedfromtheirabundanceandSiO2
con-tentsassuminganaverageMg#of0.95.
b Average of the 28 out of 36 CAIs from Clayton et al. (1988)
and Grossman et al. (2008) that haveno largeSi isotopeanomaly (29Si<0.1).
c Averageofthe isolatedolivinesfrom thisstudy thathavenoSi
isotopeanomaly(11outof12analysed).
d Averageof21chondrules(11fromthisstudyandof10from
Army-tage(2011)).
e Averageof17samplesofmatrix fromthisstudy andof1from
Armytage(2011).
f Calculatedfor10differentaliquotsofAllendeanalysedinArmytage
et al.,
2011
;Pringleet al.,2013
,2014
;SavageandMoynier,2013
.variations amongplanetaryobjects(Dauphas et al.,2015).Finally, the low Mg# and
δ
30Si valuesof the matrixsilicates can be ex-plainedby thepredominanceinthematrixofsilicatescondensed at lower temperatures than the precursors of chondrules andby thepresence inthematrixofmetal condensedtogether withthe silicates(EbelandGrossman,2000).Mg-Feexchangebetween ma-trixsilicatesandmetalgrainsatlowertemperaturewillnotchange theδ
30Sisignatureacquiredbythesilicatesduringcondensation.4.4. Silicon isotopic complementarity between Allende components
Allendecomponentswouldbetruly complementaryforSi iso-topes if they were presentin Allende with the
δ
30Sivalues and the proportions modeled for a fractional condensation sequence (see 4.3.). This would imply that they were accreted in relative abundancesequaltothatproducedbythecondensationsequence.Notethat,evenifCAIsprecursorsandchondruleprecursorsderive from different parent reservoirs (as suggested for instance from their different 16O enrichments), they are predictedto show the samesystematicsfor
δ
30Sivaluesaspredictedintheprevious sec-tion because (i) no significantδ
30Si variations are anticipated inthe accretiondisk(Armytage et al., 2011) and (ii) the condensa-tionsequencewouldbesimilarindifferentreservoirs.
Themean
δ
30SivaluesoftheAllendecomponentsarepossibly consistentwithpredictionsmadeforfractionalcondensationfrom agaseous reservoir havinginitiallythe bulkδ
30Si ofAllende(see Figs. 4,5 and previous section). Thus, one way to test the com-plementarityforSiisotopes is toverifythat the bulkδ
30Si value measured for Allende is the same as calculated from the meanδ
30Sivaluespredictedfromfractional condensationforthe differ-entcomponents.Ifnot,thiscouldindicateeither(i)thatthemeanδ
30Si valueused foroneorseveralcomponents isnot correct, or (ii) that the components are not derived froma single reservoir havingtheδ
30SivalueofbulkAllende.Inbothcases,Allendecom-ponentswouldnot betruly complementary forSi isotopes.There appearstobeseverallimitationsinmakingthisexercise.
ThefirstlimitationisthatAllendecomponentsshowarangeof
δ
30Sivaluesmaking thedefinitionofa meanδ
30Siforeachcom-ponentdifficult.ThisisexemplifiedbyCAIswhichshowthelargest rangeof
δ
30Sivariations,from-1.77to+14.26forthe36CAIs fromClaytonet al. (1988) andGrossmanet al. (2008).Thisrange isreduced butstill large, from-1.77 to +5.96 (mean of+1.4±
3.92 s.d.,Table 1),ifconsidering onlythe CAIswithsmall29Sianomalies(<0.1).TheCAIshavinglargeSiisotope anoma-lies(withamean
δ
30Sivalueof+4.7;Claytonet al.,1988)most likelyformed fromamix ofprecursors containing manypresolar grainswhichwerenottotallyevaporated.Thus,theyarenotreally relevant fortesting condensation.Because oftheir smallfraction (27% of all CAIs) and of the small fraction of total Si hosted in all CAIs (3.8%, Table 2), the contribution of CAIs to bulk Allende isnotdominant.Forinstance,thebulkδ
30SicalculatedforAllende wouldincreaseby0.14onlyiftakingameanδ
30SivalueforCAIs of+5insteadofthevalueof+1.4corresponding tothemean of“normal”CAIs.Onewaytounderstandthismeanδ
30Sivalue of +1.4 (Table 2) isto considerthat it represents themean valueJID:EPSL AID:116678 /SCO [m5G; v1.297] P.9 (1-10) R. Martins, M. Chaussidon, Z. Deng et al. Earth and Planetary Science Letters•••(••••)••••••
ofCAIsprecursorsproducedatthebeginningofthecondensation sequence(predictedtorangefrom+0.2
to+0.7ifthestarting gashasaδ
30Siof-0.43,seeFig.4andFig.S6)modified bythe kineticisotopicfractionationsduetoevaporationandcondensation duringthecomplexre-heatinghistoryofCAIsandtheirprecursors (e.g. Mendybaev et al., 2013; Richter et al., 2002; Richter, 2004; Grossmanet al.,2008).Thesecondlimitationcomesfromthefactthatthe
δ
30Sivalues ofAOAshavenot beenstudiedpreciselyinAllende.ThoughAOAs host a small fraction of total Si (2.1%), this represents however around twice more Si than “anomalous” CAIs. In addition, theirδ
30Si values could be very negative as shownby recention mi-croprobe data from three CV, CO andCM chondrites (Kaba, MIL 07342andNWA5958, respectively)showingaδ
30Sirangefor69individual olivine grainsin 7AOAs from -9.7
±
0.5(2 s.d.) to -1.0±
0.2andanaverageδ
30Sivalueforthe7AOAsof-4.3±
1.7 (Marrocchiet al.,2019b).Using-4.3insteadof+1.4(thevalue of“normal”CAIs)decreasesthebulkδ
30Sivaluecalculatedfor Al-lendeby0.12.Bearing inmind these limitations,a bulk
δ
30Si value of-0.55±
0.13 can be calculated for the sum of all the components listedinTable2usingavalueof-4.3forAOAs(Marrocchiet al., 2019b) andassuming that normal chondrules and Al-rich chon-drules sharethesameaverageδ
30Sivalue. Thisvalue seemscon-sistent within errorswiththebulk
δ
30Sivalue of-0.43±
0.03 forAllende.Notethattheδ
30SivalueforthesumofAllende com-ponents wouldbe even closerto thebulk ofAllende ifusingfor AOAsthemeanδ
30Sivalue ofCAIs:thiswouldgiveaδ
30Sivalue of -0.43±
0.13. Tomake this test asstringentas possiblewe used the smallest possible error on theδ
30Si of the sumof Al-lende components,i.e.the2 s.e.errors onthe meanδ
30Sivalues ofeachcomponent(Table2).Ifconsideringinadditionanerroron thefractionofSihostedbyeachcomponent,oriftaking2s.d. er-rorsonδ
30Sivaluesof thecomponentsinstead of2s.e., thetest wouldofcoursebeverifiedbutnotreallysignificantanymore be-causeoftheverylargeerrors.5. Conclusions
The silicon isotopic variations present among Allende compo-nents are consistent with resulting from a sequence of conden-sation. In such a scenario, early (CAIs and chondrule precursors, isolatedolivines)andlate(matrixminerals)condensateshave
δ
30Si valueshigherandlowerthan theinitialgas,respectively, andare accreted togetherinmoreorlesstheproportionsissuedfromthe condensation sequence. This results in the bulkδ
30Si of Allende beingthatoftheinitialgas.ThefactthatthedifferentAllendecomponentsappear comple-mentaryrelativetothecondensationprocessforamajorelement such as Si, which has a 50% condensation temperature of 1310 K (Lodders, 2003), does not mean that complementarity is obli-gatorily verified for highly refractory orhighly volatileelements. However, because it implies that all the major Allende compo-nentswereaccretedmoreorlessintheproportionsinwhichthey (or their precursors) were produced by condensation, it is very likely that complementarity mayalso be observed for other ele-ments that were fractionatedbetween early-condensed and late-condensedphases.
CRediTauthorshipcontributionstatement
MC,FMandZDdesignedtheproject.FMprovidedsamplesfor study. RMP preparedthe samples andperformedthe SEM study. ZD andRMP developed the technique, did the isotopic analyses, reduced the data and produced sample description and data ta-bles. FP advised on condensation calculations. MC, FM, ZD, RMP
andFPparticipatedtotheinterpretationofthedata.MCconceived andrealizedthemodel.ZD andRMPwrote thedescriptionofthe analyticaltechniques.MCwrotetherestofthepaper.
Declarationofcompetinginterest
Theauthorsdeclarethattheyhavenoknowncompeting finan-cialinterestsorpersonalrelationshipsthatcouldhaveappearedto influencetheworkreportedinthispaper.
Acknowledgements
Thiswork was supported by ANR-15-CE31-0004-1(ANR CRA-DLE),theUnivEarthSLabexprogramatSorbonne ParisCitè (ANR-10-LABX-0023andANR-11-IDEX-0005-02) andtheRégion Île-de-France through the DIM-ACAV+ project “HOC - Origine de l’eau etdu carbone dansle Système Solaire”. Parts of this work were supportedbyIPGPmultidisciplinary programPARI,andby Region île-de-FranceSESAMEGrant(no.12015908).WethankAndrewM. DavisandDominikHezelfortheirreviewsandhelpfulcomments.
Appendix A. Supplementarymaterial
Supplementarymaterialrelatedtothisarticlecanbefound on-lineathttps://doi.org/10.1016/j.epsl.2020.116678.
References
Alexander,C.M.O’D.,2005. Re-examiningthe roleofchondrulesinproducingthe elementalfractionationsinchondrites.Meteorit.Planet.Sci. 40,943–965. Alexander,C.M.O’D.,2019.Quantitativemodelsfortheelementalandisotopic
frac-tionationsinchondrites: thecarbonaceous chondrites.Geochim. Cosmochim. Acta 254,277–309.
Anders,E.,1977.Critiqueof“Nebularcondensationofmoderatelyvolatileelements andtheirabundancesinordinarychondrites”byChienM.WaiandJohnT. Was-son.EarthPlanet.Sci.Lett. 36,14–20.
Armytage,R.M.G.,2011.Thesiliconisotopiccompositionofinnersolarsystem ma-terials.UnpublishedPhDthesis.UniversityofOxford.212p.
Armytage,R.M.G.,Georg,R.B.,Savage,P.S.,Williams,H.M.,Halliday,A.N.,2011. Sili-conisotopesinmeteoritesandplanetarycoreformation.Geochim.Cosmochim. Acta 75,3662–3676.
Barosch,J., Hezel, D.C., Ebel, D.S.,Friend, P., 2019. Mineralogically zoned chon-drulesinordinarychondritesasevidenceforopensystemchondrulebehaviour. Geochim.Cosmochim.Acta 249,1–16.
Bland,P.A.,Alard,O.,Benedix,G.K.,Kearsley,A.T.,Menzies,O.N.,Watt,L.E.,Rogers, N.W.,2005.Volatilefractionationintheearlysolarsystemandchondrule/matrix complementarity.Proc.Natl.Acad.Sci. 103,13755–13760.
Bonal,L.,Ouirico,E.,Bourot-Denise,M.,Montagnac,G.,2006.Determinationofthe petrologictypeofCV3chondritesbyRamanspectroscopyofincludedorganic matter.Geochim.Cosmochim.Acta 70,1849–1863.
Budde,G., Kleine,T.,Kruijer, T.S.,Burkhardt,C., Metzler,K.,2016.Tungsten iso-topicconstraints ontheageandoriginofchondrules.Proc.Natl.Acad.Sci. 113, 2886–2891.
Chaussidon,M., Libourel,G.,Krot,A.N.,2008.Oxygenisotopicconstraintsonthe originofmagnesianchondrulesandonthegaseousreservoirsintheearlySolar System.Geochim.Cosmochim.Acta 72,1924–1938.
Clayton,R.N.,Mayeda,T.,Epstein,S.,1978.IsotopicfractionationofsiliconinAllende inclusions.In:Proc.Lunar.Planet.Sci.Conf.9th,pp. 1267–1278.
Clayton,R.N.,Hinton,R.W.,Davis,A.M.,1988.Isotopicvariationsintherock-forming elementsinmeteorites.Philos.Trans.R.Soc.Lond.A 325,483–501.
Dauphas,N.,Poitrasson,F.,Burkhardt,C.,Kobayashi,H.,Kurosawa,K.,2015. Plan-etaryandmeteoritic Mg/Si and δ30Sivariations inherited from solar nebula
chemistry.EarthPlanet.Sci.Lett. 427,236–248.
Ebel,D.S.,Grossman, L.,2000. Condensationindust-enrichedsystems. Geochim. Cosmochim.Acta 64,339–366.
Ebel,D.S.,Brunner,C.,Konrad,K.,Leftwich,K.,Erb,I.,Lu,M.,Rodriguez,H., Crapster-Pregont,E.J.,Friedrich,J.M.,Weisberg,M.K.,2016.Abundance, majorelement compositionandsizeofcomponentsandmatrixinCV,COandAcfer094 chon-drites.Geochim.Cosmochim.Acta 172,322–356.
Fedkin,A.V.,Grossman,L.,2016.Effectsofdustenrichmentsonoxygenfugacityof cosmicgases.Meteorit.Planet.Sci. 51,843–850.
Fitoussi,C.,Bourdon,B.,Kleine,T.,Oberli,F.,Reynolds,B.C.,2009.Siisotope system-aticsofmeteoritesandterrestrialperidotites:implicationsforMg/Si fractiona-tioninthesolarnebulaandforSiintheEarth’score.EarthPlanet.Sci.Lett. 287, 77–85.
Friend,P.,Hezel,D.C.,Mucerschi,D.,2016.Theconditionsofchondruleformation, partII:opensystem.Geochim.Cosmochim.Acta 173,198–209.
Georg,R.B.,Reynolds,B.C.,Frank,M.,Halliday,A.N.,2006.Newsamplepreparation techniquesforthedeterminationofSiisotopiccompositionsusingMC-ICPMS. Chem.Geol. 235,95–104.
Grossman, L., 1972. Condensation inthe primitive solar nebula. Geochim. Cos-mochim.Acta 36,597–619.
Grossman,L.,Simon,S.B.,Rai,V.K.,Thiemens,M.H.,Hutcheon,I.D.,Williams,R.W., Galy,A.,Ding,T.,Fedkin,A.V.,Clayton,R.N.,Mayeda,T.,2008.Primordial com-positionsofrefractoryinclusions.Geochim.Cosmochim.Acta 72,3001–3021. Hezel, D.C.,Palme, H.,2010. Thechemical relationshipbetweenchondrules and
matrixandthechondrulematrixcomplementarity.EarthPlanet.Sci.Lett. 294, 85–93.
Hezel,D.C.,Wilden,J.S.,Becker,D.,Steinbach,S.,Wombacher,F.,Harak,M.,2018.Fe isotopecompositionofbulkchondrulesfromMurchison(CM2):constraintsfor parentbodyalteration,nebulaprocessesandchondrule-matrix complementar-ity.EarthPlanet.Sci.Lett. 490,31–39.
Javoy,M., Balan,E.,Méheut,M., Blanchard,M.,Lazzeri, M.,2012.First-principles investigationofequilibriumisotopicfractionationofO- andSi-isotopesbetween refractorysolidsandgasesinthesolarnebula.EarthPlanet.Sci.Lett. 319–320, 118–127.
Kadlag,Y.,Tatzel,M.,Frick,D.A.,Becker,H.,Kühne,P.,2018.Chondrule-matrix com-plementarityintheAllendeCV3chondrite–aSiperspective.In:Lunar Plane-taryScienceConference.#1092(abstract).
Krot,A.N.,Petaev,M.I.,Russel,S.S.,Itoh,S.,Fagan,T.J.,Yurimoto,H.,Chizmadia,L., Weisberg,M.K.,Komatsu,M.,Ulyanov,A.A.,Keil,K.,2004.Amoeboidolivine ag-gregatesandrelatedobjectsincarbonaceouschondrites:recordsofnebularand asteroidprocesses.Chem.Erde 64,185–239.
Libourel,G.,Krot,A.,Tissandier,L.,2006.Roleofgas–meltinteractionduring chon-druleformation.EarthPlanet.Sci.Lett. 251,232–240.
Libourel,G.,Portail,M.,2018.Chondrulesasdirectthermochemicalsensorsofsolar protoplanetarydiskgas.Sci.Adv. 4(7),eaar3321.
Lodders,K.,2003.Solarsystemabundancesandcondensationtemperaturesofthe elements.Astrophys.J. 591,1220–1247.
Marrocchi, Y., Euverte, R., Villeneuve, J., Batanova, V., Welsch, B., Ferrière, L., Jacquet,E.,2019a.FormationofCVchondrulesbyrecyclingofamoeboidolivine aggregate-likeprecursors.Geochim.Cosmochim.Acta 247,121–141.
Marrocchi,Y.,Chaussidon,M.,2015.Asystematicfor oxygenisotopicvariationin meteoriticchondrules.EarthPlanet.Sci.Lett. 430,308–315.
Marrocchi,Y.,Villeneuve,J.,Jacquet, E.,Piralla,M.,Chaussidon,M.,2019b.Rapid condensation of the first solar system solids. Proc. Natl. Acad. Sci. 116, 23461–23466.
Méheut,M.,Lazzeri,M.,Balan,E.,Mauri,F.,2009.Structuralcontroloverequilibrium siliconand oxygenisotopic fractionation:a first-principlesdensity-functional theorystudy.Chem.Geol. 258,28–37.
Mendybaev,R.A.,Richter,F.M.,Georg,R.B.,Janney,P.E.,Spicuzza,M.J.,Davis,A.M., Valley,J.W.,2013.ExperimentalevaporationofMg- andSi-richmelts: implica-tionsfortheoriginandevolutionofFUNCAIs.Geochim.Cosmochim.Acta 123, 368–384.
Moynier,F.,Agranier,A.,Hezel,D.C.,Bouvier,A.,2010.Srstableisotope composi-tionofEarth,theMoon,Mars,Vestaandmeteorites.EarthPlanet.Sci.Lett. 300, 359–366.
Mullane,E.,Russell,S.S.,Gounelle,M.,2005.Nebularandasteroidalmodificationof theironisotopecomposition.EarthPlanet.Sci.Lett. 239,203–218.
Palme,H.,FegleyJr.,B.J.,1990.High-temperaturecondensationofiron-richolivine inthesolarnebula.EarthPlanet.Sci.Lett. 101,180–195.
Palme,H.,Hezel,D.C.,Ebel,D.S.,2015.Theoriginofchondrules:constraintsfrom matrix compositionandmatrix-chondrulecomplementarity.EarthPlanet.Sci. Lett. 411,11–19.
Palme,H.,Lodders,K.,Jones,A.,2014.Solarsystemabundancesoftheelements.In: Holland,H.D.,Turekian,K.K.(Eds.),secondedition.TreatiseonGeochemistry, Vol.2.Elsevier,Oxford,pp. 15–36.
Petaev,M.I.,Wood,J.A.,1998.Thecondensationwithpartialisolation(CWPI)model ofcondensationinthesolarnebula.Meteorit.Planet.Sci. 33,1123–1137. Petaev, M.I.,Wood,J.A., 2005.Meteoritic constraintson temperatures, pressures,
coolingrates,chemicalcompositions,andmodesofcondensationinthesolar nebula.In:Krot,A.N.,Scott,E.R.D.,Reipurth,B.(Eds.),Chondritesandthe Proto-planetaryDisk.In:ASPConferencesSeries,vol. 341,pp. 373–406.
Pignatale,F.C.,Maddison,S.T.,Liffman,K.,Brooks,G.,2016.2Dcondensationmodel fortheinnersolarnebula:anenstatite-richenvironment.Mon.Not.R.Astron. Soc. 457(2),1359–1370.
Pignatale,F.C.,Charnoz,S.,Chaussidon,M.,Jacquet,E.,2018.Makingtheplanetary materialdiversityduringtheearlyassemblingofthesolarsystem.Astrophys. J. 867,L23–L27.
Pringle,E.A.,Savage,P.S.,Badro,J.,Barrat,J.-A.,Moynier,F.,2013.Redoxstateduring coreformationonasteroid4-Vesta.EarthPlanet.Sci.Lett. 373,75–82. Pringle, E.A., Moynier, F., Savage, P.S., Badro, J., Barrat, J.-A., 2014. Silicon
iso-topesinangritesandvolatilelossinplanetesimals.Proc.Natl.Acad.Sci. 111, 17029–17032.
Richter, F.M., 2004. Timescales determining the degree of kinetic isotope frac-tionation by evaporation and condensation. Geochim. Cosmochim. Acta 68, 4971–4992.
Richter,F.M.,Davis,A.M.,Ebel,D.S.,Hashimoto,A.,2002. Elementalandisotopic fractionationoftype Bcalcium-,aluminum-rich inclusions:experiments, the-oreticalconsiderations,and constraintsontheir thermalevolution.Geochim. Cosmochim.Acta 66,521–540.
Rudraswami,N.G.,Ushikubo,T.,Nakashima,D.,Kita,N.T.,2011.Oxygenisotope sys-tematicsintheAllendeCV3chondrite:highprecisionionmicroprobestudies. Geochim.Cosmochim.Acta 75,7596–7611.
Savage,P.S.,Moynier,F.,2013.Siliconisotopicvariationinenstatitemeteorites:clues totheiroriginandEarth-formingmaterial.EarthPlanet.Sci.Lett. 361,487–496. Scott,E.R.D.,Krot,A.N.,2014.Chondritesandtheircomponents,inmeteoritesand cosmochemicalprocesses.In:Holland,H.D.,Turekian,K.K.(Eds.),Treatiseon GeochemistrySecondEdition,Vol.1.Elsevier,pp. 65–137.
Tissandier,L.,Libourel,G.,Robert,F.,2002.Gas–meltinteractionsandtheirbearing onchondruleformation.Meteorit.Planet.Sci. 37,1377–1389.
VanKooten,E.,Moynier,F.,2019.Zincisotopeanalysesofsingularly small sam-ples(<5ng Zn):investigating chondrule-matrixcomplementarityinLeoville. Geochim.Cosmochim.Acta 261,248–268.
Villeneuve,J.,Marrocchi,Y.,Jacquet,E.,2020.Siliconisotopiccompositionof chon-drule silicates in carbonaceous chondrites and the formation of primordial solidsintheaccretiondisk.EarthPlanet.Sci.Lett. 542,116318.
Wai,C.M.,Wasson,J.T.,1977.Nebularcondensationofmoderatelyvolatileelements andtheirabundancesinordinarychondrites.EarthPlanet.Sci.Lett. 36,1–13. Wasson,J.T.,1977.ReplytoEdwardAnders:Adiscussionofalternativemodelsfor
explainingthedistributionofmoderatelyvolatileelementsinordinary chon-drites.EarthPlanet.Sci.Lett. 36,21–28.
Wood,J.A.,Hashimoto,A.,1993.Mineralequilibriuminfractionatednebular sys-tems.Geochim.Cosmochim.Acta 57,2377–2388.
Zambardi,T.,Poitrasson,F.,Corgne,A.,Méheut,M.,Quitté,G.,Anand,M.,2013. Sili-conisotopevariationsintheinnersolarsystem:implicationsforplanetary for-mation,differentiationandcomposition.Geochim.Cosmochim.Acta 121,67–83. Zanda,B.,Lewin,E.,Humayun,M.,2018.Thechondriticassemblage.In:Russell,S., ConnollyJr.,H.,Krot,A.(Eds.),Chondrules:RecordsofProtoplanetaryDisk Pro-cesses.CambridgeUniversityPress,Cambridge,pp. 122–150.