Simultaneous iron and nickel isotopic analyses of presolar silicon … · 2017-12-18 · Simultaneous iron and nickel isotopic analyses of presolar silicon carbide grains ... Geochimica

Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 221 (2018) 87–108

Simultaneous iron and nickel isotopic analyses of presolarsilicon carbide grains

Reto Trappitsch a,b,c,⇑, Thomas Stephan a,b, Michael R. Savina d,b,c,Andrew M. Davis a,b,e, Michael J. Pellin d,a,b,e, Detlef Rost a,b,1, Frank Gyngard f,

Roberto Gallino g, Sara Bisterzo g,h, Sergio Cristallo i,j, Nicolas Dauphas a,b,e

aDepartment of the Geophysical Sciences, The University of Chicago, 5734 S Ellis Ave, Chicago, IL 60637, USAbChicago Center for Cosmochemistry, Chicago, IL, USA

cNuclear and Chemical Sciences Division, Lawrence Livermore National Laboratory, 7000 East Ave., Livermore, CA 94550, USAdMaterials Science Division, Argonne National Laboratory, 9700 S. Cass Avenue, Argonne, IL 60439, USA

eEnrico Fermi Institute, The University of Chicago, 5640 S Ellis Ave, Chicago, IL 60637, USAfLaboratory for Space Sciences and Department of Physics, Washington University, 1 Brookings Drive, St. Louis, MO 63130, USA

gDipartimento di Fisica, Universita di Torino, Torino 10125, Italyh INAF – Osservatorio Di Torino, Torino 10025, Italy

i INAF – Osservatorio Astronomico di Teramo, Teramo 64100, Italyj INFN – Istituto Nazionale Di Fisica Nucleare Sezione Di Perugia, Perugia 06123, Italy

Received 8 October 2016; accepted in revised form 20 May 2017; available online 29 May 2017

Abstract

Aside from recording stellar nucleosynthesis, a few elements in presolar grains can also provide insights into the galactic chem-ical evolution (GCE) of nuclides. We have studied the carbon, silicon, iron, and nickel isotopic compositions of presolar siliconcarbide (SiC) grains from asymptotic giant branch (AGB) stars to better understand GCE. Since only the neutron-rich nuclidesin these grains have been heavily influenced by the parent star, the neutron-poor nuclides serve as GCE proxies. Using CHILI,a new resonance ionization mass spectrometry (RIMS) instrument, we measured 74 presolar SiC grains for all iron and nickel iso-topes.With the CHARISMA instrument, 13 presolar SiC grains were analyzed for iron isotopes. All grains were alsomeasured byNanoSIMS for their carbon and silicon isotopic compositions. A comparison of the measured neutron-rich isotopes with modelsfor AGB star nucleosynthesis shows that our measurements are consistent with AGB star predictions for low-mass stars betweenhalf-solar and solarmetallicity. Furthermore, ourmeasurements give an indication on the 22Neða; nÞ25Mg reaction rate. In terms ofGCE, we find that the GCE-dominated iron and nickel isotope ratios, 54Fe=56Fe and 60Ni=58Ni, correlate with their GCE-dominated counterpart in silicon, 29Si=28Si. The measured GCE trends include the Solar System composition, showing that theSolar System is not a special case. However, as seen in silicon and titanium, many presolar SiC grains are more evolved for ironand nickel than the Solar System. This confirms prior findings and agrees with observations of large stellar samples that a simpleage-metallicity relationship for GCE cannot explain the composition of the solar neighborhood.� 2017 Elsevier Ltd. All rights reserved.

Keywords: Presolar grains; S-process; Stellar nucleosynthesis; Galactic chemical evolution; Isotope anomalies

http://dx.doi.org/10.1016/j.gca.2017.05.031

0016-7037/� 2017 Elsevier Ltd. All rights reserved.

⇑ Corresponding author at: Nuclear and Chemical Sciences Division, Lawrence Livermore National Laboratory, 7000 East Ave L-231,Livermore, CA 94550, USA.

E-mail address: [email protected] (R. Trappitsch).1 Present address: Department of Physics, University of Auckland, Auckland 1010, New Zealand.


mailto:[email protected]


http://crossmark.crossref.org/dialog/?doi=10.1016/j.gca.2017.05.031&domain=pdf

88 R. Trappitsch et al. /Geochimica et Cosmochimica Acta 221 (2018) 87–108

1. INTRODUCTION

After the Big Bang, the universe contained mainlyhydrogen, helium, and small amounts of lithium. Heavierelements were subsequently synthesized in stars and mixedinto the interstellar medium (ISM) when those starsexploded or lost their outer shells through stellar winds.Stars with initial masses larger than about 10 solar masses(M�) end their lives as core-collapse supernovae (CCSNe).These stars have shorter lifetimes than stars with initialmass � 10M�, and played a proportionally greater role inenriching the ISM with heavy elements early on. Type Iasupernovae form in binaries where both stars have initialmasses of � 10M�. The larger of the two stars lives its lifeand forms a white dwarf. When the companion star reachesthe red giant stage, it transfers mass to the white dwarf untilthe latter grows beyond the Chandrasekhar mass andexplodes. This requires that one star with a massK 10M� undergoes its full stellar evolution and become awhite dwarf. Heavier stars would not have resulted in awhite dwarf (see, e.g., Herwig, 2013). Type Ia SNe thereforecould not have contributed to the interstellar medium rightafter the Big Bang. The same is true for asymptotic giantbranch (AGB) stars – their low mass results in a longer life-time and hence in a later return of synthesized materialcompared to massive stars. A recent review on GCE wasgiven by Nomoto et al. (2013).

Presolar grains, recovered from meteorites, are mineralcondensates that formed in the outflows of dying stars, sur-vived their journey through the ISM, and were incorpo-rated into meteorite parent bodies at the time the SolarSystem formed. They represent a unique opportunity tostudy stellar nucleosynthesis as well as GCE in the labora-tory. For recent reviews, see Davis (2011) and Zinner(2014). Various types of presolar grains are known, amongwhich silicon carbide (SiC) grains, which can only condensefrom gases with a carbon-to-oxygen ratio >1, are the beststudied. Presolar SiC occurs in sizes from tens of nanome-ters to tens of micrometers and can contain relatively highlevels of trace elements (Amari et al., 1995). Silicon carbidegrains can be divided into several groups based on theirnitrogen, carbon, and silicon isotopic compositions(Zinner, 2014). The composition of each group along withother chemical, structural, and isotopic data, is indicativeof the type of parent star the grains came from. While main-stream, Y, and Z SiC grains came from AGB stars, X and Cgrains originated in CCSNe. The origin of AB grains isambiguous, and several possibilities including novae havebeen discussed in the literature (e.g., Amari et al., 2001a;Amari et al., 2001c).

Most presolar SiC grains carry the isotopic signature ofAGB stars. Most of the variation in isotopic compositionsfor elements heavier than iron can be attributed to a rangeof masses and/or metallicities resulting in a variety of neu-tron exposures and neutron-to-seed ratios in their parentAGB stars. However, certain elements in mainstreampresolar SiC grains, like silicon and titanium, vary in iso-topic composition in ways different from that expected fromAGB stars (e.g., Nittler, 2005; Huss and Smith, 2007).Isotopic anomalies in silicon and titanium are of larger

magnitude and correlate with one another in ways notexpected for AGB star nucleosynthesis, which can haveonly relatively small effects on silicon and titanium isotopes.These isotopic variations could be due to GCE. Claytonand Timmes (1997) showed that silicon isotopes in presolarSiC grains vary with metallicity of the parent star. In a stan-dard GCE picture where age and metallicity correlate,higher 29Si and 30Si relative to 28Si would indicate a laterformation. This, however, would mean that the Solar Sys-tem is a special case, since it looks isotopically less evolvedthan most presolar mainstream grains, i.e., it looks olderthan presolar SiC grains in the standard GCE model. A dif-ferent explanation was introduced by Lugaro et al. (1999).They argued that the silicon isotopic composition repre-sents heterogenous GCE. Nittler (2005) followed up on thisidea and showed that such a heterogenous evolution couldnot fully explain the silicon-to-titanium correlation. A thirdexplanation was introduced by Clayton (2003): a dwarfgalaxy merged with the Milky Way and subsequently, theSolar System and the parent stars of the presolar SiC main-stream grains formed from material that originated frommixing the materials from both galaxies. A galactic mergercould trigger star formation in this region and add an iso-topic composition different from the composition of theMilky Way prior to the merger. The isotopic compositionof presolar grains and the Solar System in this scenariowould be on a mixing line between the metal-poor andthe metal-rich component, i.e., between the merging galaxyand the Milky Way. The Solar System composition wouldnot be special and is simply one possible mixture. Focusingon analyses of Y and Z grains, Zinner et al. (2006) showedthat the silicon isotopic composition of presolar grains withrespect to the Solar System could be explained by GCE;however, a low-metallicity source, i.e., a CCSN, for 29Siand 30Si needs to be included into GCE models in orderto explain the observed correlation between d29Si andd30Si. Note that observations of stars in the solar neighbor-hood do not confirm the existence of an age-metallicityrelationship (e.g., Holmberg et al., 2007). Lewis et al.(2013) explored the ability of chemodynamical models(e.g., Kobayashi and Nakasato, 2011) to explain the siliconisotopic composition of presolar SiC grains. Lewis et al.(2013) proposed that the enrichment in 29Si and 30Si ofpresolar SiC grains is a bias effect, due to high-metallicitystars producing dust more efficiently than low-metallicitystars.

Models of s-process nucleosynthesis in AGB stars (e.g.,Gallino et al., 1998; Cristallo et al., 2011) predict that, sim-ilar to silicon isotopes, most of the iron and nickel isotopes(except for 58Fe and 64Ni) are little-affected by AGB nucle-osynthesis. Furthermore, iron and nickel are abundantenough in presolar SiC grains that these elements are suit-able to be measured and used as GCE proxies. Previousmeasurements of iron and nickel in presolar SiC grains wereperformed by Marhas et al. (2008) using a CAMECANanoSIMS 50. In addition, Ong and Floss (2015) measurediron isotopic ratios in presolar silicates using the sameinstrument. Iron-58 and 58Ni could not be resolved withthe NanoSIMS, and the peak at mass 58u was dominatedby 58Ni. Iron-58 was therefore not reported, and the

R. Trappitsch et al. /Geochimica et Cosmochimica Acta 221 (2018) 87–108 89

abundance of 58Ni was calculated by assuming a solar58Fe/56Fe ratio. Furthermore, NanoSIMS results for 54Fewere corrected for unresolved 54Cr by assuming a solar54Cr/52Cr ratio. Marhas et al. (2008) did not report datafor 64Ni due to interferences with 64Zn and the low naturalabundance of 64Ni. However, 58Fe and 64Ni are the isotopesthat are most strongly enhanced by s-process nucleosynthe-sis as predicted by AGB models. Understanding the AGBstellar contribution is essential in deciphering the GCErecord and therefore requires a technique that eliminatesisobaric interferences.

Here we report the results of simultaneous iron andnickel isotopic and elemental measurements of 74 presolarSiC grains. These measurements were performed with theChicago Instrument for Laser Ionization (CHILI) – a reso-nance ionization mass spectrometry (RIMS) instrument atthe University of Chicago (Stephan et al., 2016). We alsoinclude results of 13 presolar SiC grain measurements ofiron isotopes from Trappitsch et al. (2012), which were per-formed with an earlier generation RIMS instrument,CHARISMA, at Argonne National Laboratory (Savinaet al., 2003). All presolar grains were classified by measur-ing carbon and silicon isotopes with the CAMECA Nano-SIMS 50 at Washington University in St. Louis.

2. ANALYTICAL METHODS

2.1. Sample preparation and standards

Presolar grains analyzed in this study are from theRWB6 separation from the Murchison CM meteorite(Levine et al., 2009) and are mostly in the size range of2–4 lm. The grains were acid-cleaned prior to pressing intoa high purity gold foil, as described by Levine et al. (2009).For strontium and barium, acid-cleaning has been shown toremove surface contamination that may have occurred onthe meteorite parent body or during isolation of SiC fromthe host meteorite (Liu et al., 2014, 2015a). On the samemount, we deposited terrestrial Cr2C3 that fortuitously con-tained measurable amounts of iron and nickel. The Cr2C3

grains were used to normalize the iron isotopic composi-tions measured with CHARISMA. In preparation forNanoSIMS analysis, we deposited fine-grained syntheticSiC grains on the edges of the same mount. These grainsalso contained trace amounts of iron and nickel and couldtherefore be used to normalize all isotopic data measuredwith CHILI. Aside from the SiC standard measurements,we used CHILI to frequently measure the iron and nickelisotopic composition of SAE 304 stainless steel, of a sepa-rately mounted NIST SRM610 standard glass, as well asof the Cr2C3 grains. Cr2C3 was also measured withCHARISMA.

2.2. Resonance ionization mass spectrometry

Iron and nickel isotopic compositions of presolar SiCgrains were simultaneously measured with CHILI(Stephan et al., 2016). With CHARISMA, only iron iso-topes were measured, since not enough lasers were availableto measure iron and nickel simultaneously. In both cases,

atoms were desorbed from the sample surface using a351nm wavelength laser with a lateral resolution of�1 lm. For CHARISMA measurements, the desorptionlaser beam was rastered over the sample in order to hitthe entire grain. In CHILI, we performed spot analysissince the beam could be precisely located with the help ofthe built-in scanning electron microscope. A small fractionof the desorbed particles were already ionized and are apotential source of mass interferences. Therefore, they wereejected from the system by applying a strong early voltagepulse to the mass spectrometer’s extraction optics. Afterthis ejection pulse, the voltage was lowered to extractionlevels, i.e., to the level at which ions can pass through thetime-of-flight analyzer to the microchannel plate detector.In CHILI, the iron and nickel resonance ionization lasersdefined the start time of the respective ions. A time delayof 200 ns between these laser pulses allowed the separationof the isobaric interference at mass 58 u. The delay movesthe 58Ni peak by approximately 0.5 u with respect to the58Fe peak. This separation is enough to fully resolve 58Fefrom 58Ni and thus all iron and nickel isotopes were mea-sured simultaneously without interference (Stephan et al.,2016).

The repetition rate for measurements with both instru-ments was 1 kHz. With CHARISMA, 25,000 shots, i.e.,25 s of data, were recorded as a sum spectrum. If signalbursts (large signals that saturate the detector) occurredduring one of these intervals, the respective data wasrejected from further evaluation. Each sum spectrum wasthen dead-time corrected (Stephan et al., 1994), and indi-vidual spectra were added up before isotope ratios were cal-culated and normalized to those obtained for standards.With CHILI, the arrival time for each individual ion atthe detector was recorded. The data were automaticallyscreened for signal bursts after the measurement. If a burstoccurred, the corresponding interval of 1000 shots wasrejected. In addition, if an individual shot contains morethan ten ions, it is rejected as well. The screened data werefinally dead-time-corrected (Stephan et al., 1994). The peakintegrals were then evaluated for all added spectra, and theisotope ratios were normalized to the results of the standardmeasurements. Error bars for both data sets were based oncounting statistics of sample and standard measurements.

Throughout this work, we report isotopic ratios as d-values, which are permil deviations from the Solar Systemratios. Our d-values are, unless otherwise noted, normalizedto the major isotope (28Si, 56Fe, or 58Ni) of the given ele-ment. For example, d29Si is defined as followed:

d29Si ¼29Si28Si

� �measured

29Si28Si

� �standard

� 1

24

35� 1000 ð1Þ

Note that we further assume that the standard has SolarSystem isotopic composition, which is a reasonableassumption compared to the large anomalies we find inpresolar SiC grains.

Fig. 1 shows the resonance ionization schemes that wedeveloped for iron (left) and nickel (right) ionization. Bothionization schemes were developed using the wavelengthtables of Moore (1971). An efficient ionization scheme

Fig. 1. Resonance ionization schemes used for the ionization of iron (left) and nickel (right). After the desorption event, iron neutrals wereionized from the ground state, while nickel neutrals were only ionized, if they were in the low-lying excited state at 204.787 cm�1 (seeSection 2.2 for further discussion).


requires sufficiently high and stable laser power, stablewavelengths, as well as broad laser bandwidths. All theseparameters inevitably fluctuate from one laser pulse to thenext; the ideal ionization scheme yields isotope ratios thatare robust against these fluctuations. The iron and nickelionization schemes terminate on autoionizing states abovethe ionization potential utilizing spectroscopy data fromWorden et al. (1984) for iron and from Page andGudeman (1990) for nickel. An atom in such an autoioniz-ing state decays via electron emission and becomes ionized.To determine the optimal wavelength, we performed scansaround the literature values and settled on the states thatprovide high ion signal and high measurement precision.We made similar scans for the other transitions and mea-sured the isotopic precision and accuracy to ensure ourselected ionization scheme was sensitive, precise, androbust. Furthermore, saturation curves (ion signal versuslaser power) were obtained. While the first two transitionsin both ionization schemes were fully saturated, the ioniza-tion transitions were nearly, but not fully saturated. Iso-topic shifts of excitation levels are small in the case ofiron and nickel (generally less than 1 GHz2), and are signif-icantly smaller than the bandwidths of the Ti:sapphirelasers (�10 GHz2). Thus fractionation due to unequal ion-ization rates among the various isotopes was negligible.Nickel has a low-lying excited state at 204.787 cm�1, i.e.,just above the ground state. Using four lasers initially, wedetermined that slightly more laser-desorbed atoms are inthis low-lying excited state than are in the ground state.Since CHILI has six tunable Ti:sapphire lasers, and three

2 Note that the bandwidth in nm can be calculated as Dk ¼ Dm k2

c ,where Dm is the frequency bandwidth, k the center wavelength, andc the speed of light.

of them are required for the ionization of iron, we decidedto ionize the neutral nickel atoms from the low-lying state.This however means that our useful yield for nickel isreduced by roughly a factor of two compared to a four-laser scheme.

Since resonance ionization is orders of magnitude moreefficient than nonresonant multiphoton ionization, otherelements in the cloud of neutral atoms are mostly transpar-ent to the laser light and hence hardly ionized at all (Hurstet al., 1979). To show the elemental suppression factors, wemeasured iron and nickel isotopes in stainless steel with anatomic iron-to-nickel ratio of 9.0. The discrimination of oneelement versus the other is determined by acquiring massspectra using the ionization lasers of only one element,e.g., measuring only the nickel isotopes and looking for anonresonant iron signal. Using the iron scheme, weacquired a long-term mass spectrum, but did not detectany nickel isotopes. Thus the lower limit for the discrimina-tion against nickel isotopes in the iron scheme is

> 5:2� 104. Assuming an elemental ratio of one and terres-trial isotope ratios, the interference from nonresonant 58Nion 58Fe is <4.7‰. The elemental discrimination for iron in

the nickel ionization case is 3:6� 105. Assuming an elemen-tal ratio of one, the contribution of nonresonant 58Fe to the58Ni signal is 11.4 ppb. These contributions of 58Ni to 58Feand vice versa are negligible for our measurements and aretherefore not included in further considerations here.

Of all the presolar SiC grains analyzed in CHILI, twelveshowed molecular interference, which most likely origi-nated from TiO and/or other molecular species that werenonresonantly ionized. This interference was detected bythe occurrence of small but distinct peaks in between thenickel peaks that were caused by nonresonant ionizationby the iron resonance ionization lasers. Such nonresonant

3 http://fruity.oa-teramo.inaf.it.


ionization could have also been caused by the nickel reso-nance ionization lasers leading to interferences at 62Niand 64Ni. For all mass spectra in which interferences wereobserved, we reject and do not report 62Ni and 64Ni data.

2.3. Contamination

As described in Section 2.1, great care was taken toensure that the presolar grains were clean of residual mate-rial from the host meteorite and that they were not contam-inated with iron and nickel during the separation process.However, even after sputter-cleaning the surface of the goldfoil using a Ga+ ion beam, we found residues of iron andnickel in the gold. This contamination was not solely onthe surface but also in the bulk gold. We performed multi-ple blank measurements of the gold surface and found that(1) small surface contamination is present at the beginningof a measurement and (2) further contamination is foundwhen the desorption laser exceeds a certain threshold value.This threshold value is coincident with a clear craterappearing on the gold mount, i.e., material is removed fastfrom the surface. When measuring presolar grains, the sur-face contamination could simply be removed by excludingthe first several thousand shots from the data, i.e., by start-ing the data recording a few seconds after the desorptionlaser started desorbing material. To avoid further contam-ination, we ensured that measurements on the presolargrain were always done at desorption laser powers belowthe ‘‘cratering” threshold.

Unfortunately, these blank measurements cannot beused to determine a worst-case scenario for contamination.Presolar SiC grains as well as the gold foil were measuredby increasing the laser power until an ion signal wasdetected. However, the desorption laser couples very differ-ently to gold than to SiC. When measuring a grain, the des-orption beam was centered on the presolar grain, however,with increasing power, the wings of the laser beam couldalso move over the threshold power to release contamina-tion from the gold surface. Thus, quantifying the totalamount of contamination from the gold is not possible.

2.4. Scanning electron microscopy and classification of

presolar SiC grains

The SiC grain mount was imaged and mapped usingscanning electron microscopy. Prior to performing RIMSmeasurements, we also performed energy dispersive X-rayspectroscopy (EDS) to confirm that the grains on themount were SiC grains. After RIMS analysis, we imagedthe grains again to check for residual material. Three outof the 13 grains from the preliminary study with CHAR-ISMA did not have enough residual material left to be clas-sified by NanoSIMS. One of the grains analyzed withCHILI was missed during prior NanoSIMS classification.

Those grains that were not completely consumed duringanalysis with CHARISMA were measured for their carbonand silicon isotopic composition with the NanoSIMS atWashington University in St. Louis to assign them to thevarious presolar SiC types. In the same measurement ses-sion, we also classified additional grains in preparation

for subsequent analysis with CHILI. NanoSIMS analyticalprotocols were identical to those employed for the workdescribed in Barzyk et al. (2007) and Liu et al. (2015a). Alow current (<0.5 pA) Csþ beam was used for sputtering,and negative secondary ions of 12C, 13C, 28Si, 29Si, and30Si were detected simultaneously. A fine-grained syntheticSiC standard was used for normalization of the NanoSIMSisotopic results. The standard grains were placed outsidethe area containing the presolar grains and did not interferewith grain analyses. For grains measured with CHAR-ISMA, NanoSIMS measurement times were kept long (sev-eral minutes), in order to analyze as much of the leftovergrain as possible in order to achieve high enough precisionfor classification. For the grains to be measured subse-quently by CHILI, acquisition times were intentionallyshort (� 1 min), just long enough to determine the graintype and leave sufficient material for the subsequentanalyses.

3. MODELS

3.1. AGB star nucleosynthesis models

Fig. 2 shows an excerpt from the chart of the nuclides.We show the measured and calculated Maxwellian-average neutron cross-sections at 30 keV taken from theKADONIS v1.0 database (Dillmann et al., 2009) as wellas the half-lives of unstable isotopes at room temperature

and at a stellar temperature of 1� 108 K (if different fromroom temperature). The main path for the s-process is givenas thick arrows, minor reaction paths with thin arrows.Note that 54Fe and 58Ni do not lie on the main s-processpath. Iron-56 is by far the most abundant nuclide in thisgroup of elements and is considered the seed of the s-process. Aside from the p-only isotope 54Fe, the neutroncapture cross sections of neutron-even isotopes are typicallylower than the cross sections of neutron-odd isotopes, see,e.g., 57Fe in comparison to 58Fe in Fig. 2. The same is truefor nickel isotopes, with 61Ni having the highest neutroncapture cross section. While the main s-process path inthe iron isotope chain is composed of stable isotopes only,the main path in the nickel isotopic chain is interrupted bythe unstable 63Ni. The half-live is in a range where the s-process can pass through to 64Ni, however some 63Ni willdecay to 63Cu before capturing another neutron. Thisbranch-point is not of significant influence however, since64Cu has a short half-life and 61.5% of it decays to 64Ni.More specific details on 58Fe and 64Ni nucleosynthesis inAGB stars are given in Section 5.2.2.

We compare our results for iron and nickel isotopiccompositions to two different sets of nucleosynthesis mod-els: the Torino models used here are updated models asdescribed below and are based on the models by Gallinoet al. (1998) and Bisterzo et al. (2010, 2014). The FRUITYmodels3 are described by Cristallo et al. (2009, 2011, 2015a)and Piersanti et al. (2013). We also compare our data withupdated FRUITY models as described below. We compare

http://fruity.oa-teramo.inaf.it

Fig. 2. Section from the chart of the nuclides. The abundances of stable nuclides are given 4.56 Ga ago (Lodders et al., 2009). Half-lives forunstable nuclides at room temperature are given (in black). Half-lives at stellar temperatures, if different from the ones at room temperature,are given in red according to Takahashi and Yokoi (1987) at a temperature of 1� 108 K. The bottom-most number for each isotope (whereavailable) is the Maxwellian-averaged neutron capture cross section at 30 keV, taken from the KADONIS v1.0 database (Dillmann et al.,2009). Cross sections printed in italic show theoretically calculated values.


our data to models with stellar envelopes that becomecarbon-rich, i.e., reach a carbon-to-oxygen ratio greaterthan unity, since SiC grains can only form effectively insuch environments.

The Torino models are computed using a post-processing code, which is based on input data of full evolu-tionary FRANEC models (Chieffi et al., 1998; Stranieroet al., 2003). Because only a limited number of isotopesare considered in full evolutionary models, i.e., only theones that release a significant amount of energy in thenuclear reactions such that they influence the evolution ofthe star, post-processing calculations include an extendednetwork of nuclides. Since these calculations are fast com-pared to, e.g., the FRUITY calculations, the impact ofthe main model uncertainties on s-process nucleosynthesiscan be easily tested. The FRUITY models, on the otherhand, follow a more self-consistent approach, in which afull nuclear network is coupled to the stellar structure equa-tions. Although this method is more time-consuming thanthe post-processing method employed in the Torino mod-els, it allows detailed investigation of the possible physicalexplanations for the most promising post-processed cases.For full details about the input physics and the numericalmethod, we refer the reader to the above references.

Various initial metallicities were modeled. While theTorino models use an initial solar metallicity (Z�) of0.0154 by mass, which is based on Lodders et al. (2009),the FRUITY models use a value of 0.014 by mass for thesolar metallicity, which is based on Lodders (2003). Subse-quently, we report the metallicities of the models used as afraction of the solar metallicity that was used for the initialmodel. There is little difference in heavy element abun-dances between the two sets of initial solar abundances;the main difference is that Lodders et al. (2009) adoptedsomewhat higher abundances of carbon and oxygen com-pared to Lodders (2003).

In low mass AGB stars (K 3M�), the13C-pocket, a thin

region enriched in 13C, where most of neutrons available forthe s-process nucleosynthesis are produced (Straniero et al.,1995), represents one of the major issues of stellar models.Indeed, the possible physical mechanisms leading to the13C-pocket formation (e.g., rotation, magnetic fields, grav-ity waves, overshooting) are poorly understood, making the13C and 14N profiles inside the pocket and the involvedmass quite uncertain (see discussion by Liu et al., 2015a).The Torino models artificially introduce the 13C pocket intheir computations, and different assumptions are adoptedto investigate the characteristics of the internal structureof the 13C-pocket. The 13C and 14N abundances are treatedas free parameters, starting from a specific shape and size ofthe 13C pocket called the st case, which represents theamount of 13C in the 13C-pockets 1:5M� to 3M� AGBmodels with half-solar metallicity that, when mixedtogether, best reproduce the Solar System s-process iso-topes (Arlandini et al., 1999; Bisterzo et al., 2014). A setof Torino models is usually obtained by dividing the 13C(and 14N) abundances of the st case by different factorsand leaving the mass of the 13C-pocket constant. Thesecases, denoted by D, represent tests where the mass fractionof 13C is lower compared to the st case (we use the abbrevi-ations D1.5, D2, D3, D6 and D12, with D12 having the low-est mass fraction of these models). The number is theindicator of how much lower the mass fraction of 13C iswith respect to the st case. For example, the D3 case hasa 13C concentration that is three times lower than the st

case. On the other hand, FRUITY models obtain the 13Cpocket self-consistently by introducing an exponentiallydecaying profile of convective velocities at the base of thestellar envelope.

Low-mass AGB stars (K 3M�) experience a marginal

activation of the 22Neða; nÞ25Mg reaction during thermalpulses, which produces a short neutron exposure with high

Fig. 3. Expected GCE spread in the relevant silicon, iron, andnickel isotopes according to the model by Kobayashi et al. (2011).Metallicity [Fe/H] is as defined in Eq. (2). The GCE model predictsan almost identical evolution for the ratios 29Si/28Si and 30Si/28Si.


neutron density. This second neutron exposure influences

the iron isotopic predictions. The 22Neða; nÞ25Mg rate firstadopted in the Torino models was from Kappeler et al.(1994), after excluding the contribution by the elusive reso-nance at 635keV and taking for the resonance at 828keVthe strength factor xc ¼ 200� 64 leV at the 1r lower limit.Also, the exact determination of the resonance energydepends on the accuracy of the experimental measurements.A few years later Arlandini et al. (1999) made a full analysisof the s-process branchings using the st case and making the

choice of dividing the 22Neða; nÞ25Mg rate by a factor oftwo (case ne22d2). Serendipitously, this new rate wasalmost identical to a new experimental evaluation byJaeger et al. (2001), the one adopted in the FRUITY data-

base. There is still uncertainty in the 22Neða; nÞ25Mg ratebecause of possible subthreshold resonances (Bisterzoet al., 2015) and the rate is estimated to be accurate within‘‘a factor 2 to maybe 4”.

Liu et al. (2015a) showed that the strontium and bariumisotopic composition of mainstream SiC grains can beexplained well when doubling the total mass of the13C-pocket while keeping the concentration of 13C in the13C-pocket the same. We compare our measurements withthese models as well and denote them with the labelpocketx2, since the total mass of the 13C-pocket wasmultiplied by a factor of two.

Previous generations of Torino AGB models, includingthe models discussed in Liu et al. (2015a) treated the massloss of the envelope based on the physical quantitiesextracted from the FRANEC code (Chieffi et al., 1998) inthe form of an interpolation (Straniero et al., 2003). Sincethen, several upgrades have been introduced in the mostrecent FUNS code (Full Network Stellar evolutionarycode: the code used to calculate the FRUITY models,Straniero et al., 2006). As shown by Straniero et al.(2006) and Cristallo et al. (2009) the introduction of anexponentially decaying profile at the inner base of the con-vective envelope of convective velocities results in anincreased third dredge-up efficiency (Straniero et al., 2006;Cristallo et al., 2009). On the other hand, the introductionof carbon-enhanced molecular opacities leads to lower sur-face temperatures inducing enhanced mass-loss rates(Cristallo et al., 2007). It is worth mentioning that the the-oretical luminosity function of carbon stars constructedwith FUNS models agrees nicely with a reanalysis ofnear- and mid-infrared photometric observations ofcarbon-rich stars (Guandalini and Cristallo, 2013). TheTorino models presented in this work introduce the TDUFmodification, in which the mass loss rate and the amount ofmaterial dredged up in each third dredge-up episode aresimilar to those in the FRUITY calculations.

3.2. Galactic chemical evolution models

Isotopes that are only slightly influenced by the AGBstar can be used as proxies for GCE. The best isotope ratiosof the analyzed elements to use are 29Si/28Si, 54Fe/56Fe, and60Ni/58Ni. For example, in the Torino model (D6 case) for astar with 2M� and 0:44Z� with the mass of the 13C-pocket

multiplied by two (pocketx2), the deviations for the AGBstar in the last thermal pulse compared to the solar abun-dances are expected to be +2‰, �5‰, and +39‰ ford29Si, d54Fe, and d60Ni, respectively. This shows that theseisotopes are minimally influenced by the AGB star, hencetheir composition should represent the initial compositionof the star. A large spread in the silicon isotopic composi-tion has been observed in presolar SiC grains from AGBstars, spanning a total range from approximately �100‰to +200‰ in d29Si and from approximately �100‰ to+160‰ in d30Si (Hynes and Gyngard, 2009). It has beenargued (e.g., Alexander and Nittler, 1999; Lugaro et al.,1999; Zinner et al., 2006) that the variability in silicon iso-topic composition is due to the GCE in the region where theparent stars of the measured presolar grains formed – otherscenarios were elaborated in the introduction. If this argu-ment is correct, a similar isotopic variation is expected ford54Fe, and a slightly larger spread is expected for d60Ni.This is shown in Fig. 3, which plots the expected isotopiccompositions for metallicities between ½Fe=H� ¼ �2:5 andsolar ([Fe/H] = 0) for the GCE predictions of Kobayashiet al. (2011), which are based on their previous models(Kobayashi et al., 2006). The metallicity notation [Fe/H]is defined as followed:

½Fe=H� ¼ log10NFe

NH

� �star

� log10NFe

NH

� ��

ð2Þ

Here, NX is the atomic abundance of the given element,either in a given star/environment, or in the Sun. Note thatwe have normalized the isotope ratios of the GCE model atsolar metallicity to the Solar System composition, as pro-posed by Clayton and Timmes (1997). Such a normaliza-tion is necessary due to the uncertainties of absolute


abundance predictions in the GCE models. For presolargrains, metallicities of interest are expected to be above½Fe=H� ¼ �0:5, which is equivalent to � 2 Ga prior toSolar System formation (Kobayashi et al., 2011) assumingan age-metallicity relationship predicted by standard GCEmodels. Fig. 3 shows that the spread that is seen in d29Si– if due to GCE – should be positively correlated withd54Fe and of similar magnitude, as well as negatively corre-lated with d60Ni and of lower magnitude for the metallicityrange of interest.

4. RESULTS

4.1. Standards

Fig. 4 shows the results for the measurements of thestandards with CHILI: clearly visible in all isotope mea-surements is the odd–even effect. The isotopes with oddmass numbers, 57Fe and 61Ni, have a higher sensitivity inthe resonance ionization process and therefore have posi-tive d-values (Fairbank et al., 1989; Wunderlich et al.,1992). Within uncertainty, all standard measurementsexcept for the stainless steel measurement agree with eachother. Aside from the odd–even effect, which probablyresults from ionization transitions that are not fully satu-rated, the measurements also seem to be mass-dependently fractionated compared to the measurementsof the other standards. We address the mass-dependentfractionation issue by using matrix-matched standards tonormalize our measurements. The odd–even effect can alsobe corrected in the same way, since the results for the stan-dard measurements were constant throughout the wholemeasurement period. All CHILI measurements were nor-malized to the SiC standard material; all CHARISMAmeasurements were normalized using the Cr2C3 standard,which we have established as having similar mass-dependent fractionation behavior to SiC.

Fig. 4. Standard measurements, normalized to terrestrial values. Error57Fe=56Fe ¼ 0:02310, 58Fe=56Fe ¼ 0:00307 (Taylor et al., 1992), 60Ni64Ni=58Ni ¼ 0:01360 (Gramlich et al., 1989).

4.2. Carbon, silicon, iron, and nickel isotopic measurements

Table 1 shows all results of our isotopic measurements.Uncertainties are 2r, based on counting statistics. In Fig. 5,we compare our results for d57Fe versus d54Fe (a) and d62Niversus d61Ni (b) with literature values by NanoSIMS byMarhas et al. (2008) for SiC grains and by Ong and Floss(2015). The uncertainties of the iron measurements withCHILI are about a factor of two smaller than in previousmeasurements with CHARISMA and, for iron and nickel,about a factor of four smaller than in the measurements byMarhas et al. (2008) and Ong and Floss (2015) for presolarsilicates (iron isotopes only). In general, the data agree wellwith published measurements, however, our study alsodetermined 58Fe and 64Ni, which could not be measuredby NanoSIMS due to isobaric interferences or low concen-trations. Aside from the molecular interferences mentionedin Section 2.2, no other interferences were detected in theRIMS measurements that could potentially affect the ironand nickel measurements.

Fig. 6 shows the silicon isotopic measurements of allpresolar SiC grains. Out of the 13 grains that were analyzedwith CHARISMA, we found nine mainstream grains, oneAB grain, and three grains that did not have enough mate-rial left for NanoSIMS analyses. Out of the 74 grains mea-sured with CHILI, we found 67 mainstream grains, one Xgrain, two Y grains, one Z grain, two AB grains, and thereis one grain for which we do not have NanoSIMS data.About 93% of all presolar SiC grains are mainstreamgrains, about 1% are Y and about 1% are Z grains(Zinner, 2014). These three types are in agreement withan origin in AGB stars. The chance that all four unclassifiedgrains come from AGB stars is therefore 81%, in addition,their iron and nickel isotopic compositions agree with theother mainstream grains that we measured. We thereforeinclude data from these unclassified grains in our discussionof AGB stars and GCE.

bars are 2r. The terrestrial values used are 54Fe=56Fe ¼ 0:06370,=58Ni ¼ 0:38520, 61Ni=58Ni ¼ 0:01674, 62Ni=58Ni ¼ 0:05339, and

Table 1Carbon, silicon, iron, and nickel isotopic measurements for all presolar SiC grains. Uncertainties are 2r. Grain numbers starting with the letter A were measured with CHARISMA at ArgonneNational Laboratory, grain labels starting with N with CHILI. SiC70 is the X grain. Grains were classified using the usual definitions, MS for mainstream, X, Y, Z, and AB. Grains that had notenough material left for classification are treated as unclassified (U). Iron and nickel concentrations are given in parts per million by weight, whenever presolar grain sizes were measured.

Grain Type 12C/13C d29 Si d30Si d54Fe d57Fe d58Fe d60Ni d61Ni d62Ni d64Ni

A8 U �28 ± 14 +40 ± 24 �38 ± 63A55 U �26 ± 22 �60 ± 43 +66 ± 93A56 MS 17.6 ± 0.3 +41 ± 14 +40 ± 11 �25 ± 18 �11 ± 27 +136 ± 88A60 U �43 ± 24 �40 ± 36 +3 ± 170A63 MS 55.7 ± 0.8 +8 ± 14 +13 ± 12 �11 ± 18 �61 ± 30 +67 ± 84A64 MS 68.7 ± 1.1 �34 ± 13 �17 ± 12 �44 ± 19 +13 ± 44 �7 ± 87A65 MS 49.0 ± 0.7 +16 ± 13 +15 ± 11 �5 ± 16 +32 ± 26 �51 ± 70A66 MS 64.8 ± 1.0 +28 ± 14 +35 ± 12 +47 ± 17 �14 ± 32 �30 ± 73A68 MS 59.0 ± 0.9 +71 ± 14 +54 ± 11 +46 ± 18 �17 ± 31 �111 ± 74A69 MS 41.6 ± 0.6 +158 ± 15 +120 ± 12 +36 ± 24 �7 ± 43 +51 ± 144A70 MS 42.6 ± 1.6 +159 ± 16 +121 ± 13 �39 ± 17 �39 ± 29 +37 ± 73A73 MS 42.9 ± 1.2 +101 ± 28 +71 ± 23 0 ± 16 �9 ± 22 +29 ± 85A75 AB 2.9 ± 0.1 �57 ± 29 �66 ± 28 �77 ± 25 �28 ± 37 �105 ± 116N14 AB 3.6 ± 0.1 �36 ± 30 �22 ± 29 �14 ± 18 �1 ± 27 +76 ± 87 +90 ± 12 +309 ± 56N21 MS 56.3 ± 1.8 +58 ± 29 +29 ± 26 +66 ± 8 +43 ± 16 +421 ± 47 �25 ± 7 +260 ± 38 +126 ± 19 +817 ± 49N22 MS 48.1 ± 1.4 +30 ± 27 +30 ± 22 �87 ± 9 �10 ± 19 +280 ± 61 +134 ± 8 +382 ± 39 +237 ± 20 +1185 ± 54N23 MS 47.0 ± 1.7 +61 ± 31 +61 ± 29 0 ± 10 �24 ± 15 �25 ± 48 �34 ± 26 +4 ± 117 +48 ± 67 +109 ± 140N24 MS 101.4 ± 3.4 +14 ± 27 +34 ± 24 �8 ± 16 +5 ± 26 +102 ± 81 +66 ± 14 +270 ± 65 +166 ± 35 +959 ± 86N25 MS 49.9 ± 1.6 +76 ± 29 +66 ± 26 +26 ± 15 �7 ± 25 �26 ± 79 �14 ± 10 +14 ± 42 +13 ± 24 +255 ± 50N26 MS 78.2 ± 2.4 +18 ± 27 +16 ± 22 +5 ± 25 �2 ± 40 +69 ± 124 +51 ± 14 +296 ± 67 +170 ± 36 +1208 ± 94N27 MS 81.4 ± 2.6 +18 ± 27 +37 ± 24 +5 ± 10 �9 ± 18 +14 ± 53 +59 ± 8 +277 ± 36 +150 ± 19 +983 ± 50N28 MS 81.6 ± 2.8 +21 ± 28 +18 ± 25 +7 ± 10 �20 ± 15 �12 ± 52 +69 ± 10 +302 ± 48N29 MS 64.1 ± 2.3 +45 ± 29 +40 ± 26 +13 ± 28 �5 ± 42 +97 ± 137 +16 ± 14 +219 ± 66N30 MS 11.0 ± 0.3 +5 ± 27 +3 ± 24 �7 ± 23 +10 ± 35 +93 ± 110 +71 ± 16 +248 ± 73 +130 ± 39 +564 ± 86N31 MS 67.8 ± 2.1 �9 ± 26 �8 ± 23 �19 ± 14 �18 ± 22 +33 ± 67 +40 ± 18 +32 ± 81 +93 ± 44 +383 ± 92N32 MS 55.2 ± 1.9 +68 ± 31 +50 ± 28 +13 ± 30 +10 ± 47 +54 ± 159 �17 ± 16 +202 ± 78 +103 ± 41 +640 ± 96N33 MS 51.6 ± 1.6 +87 ± 29 +74 ± 24 +35 ± 23 +16 ± 35 +339 ± 173 �8 ± 8 +366 ± 40 +144 ± 20 +1067 ± 55N34 MS 41.3 ± 1.3 +39 ± 28 +19 ± 24 �1 ± 9 �10 ± 14 +41 ± 49 �4 ± 13 +40 ± 58 +38 ± 31 +266 ± 65N37 Z 40.4 ± 1.2 �24 ± 26 +25 ± 23 �8 ± 12 �14 ± 20 �44 ± 64 +51 ± 15 +67 ± 65N38 MS 93.7 ± 3.1 �37 ± 26 �9 ± 24 +10 ± 10 �5 ± 16 �5 ± 63 �4 ± 9 +17 ± 45 +17 ± 20 +197 ± 41N39 MS 93.7 ± 3.1 �37 ± 26 �9 ± 24 +8 ± 11 �11 ± 20 �5 ± 55 �11 ± 9 +75 ± 41 �1 ± 22 +202 ± 45N40 MS 88.6 ± 2.6 +6 ± 26 0 ± 21 �6 ± 23 +8 ± 34 +12 ± 103 �7 ± 17 +58 ± 76 �29 ± 40 +105 ± 78N41 MS 52.9 ± 1.6 +113 ± 29 +69 ± 24 �2 ± 12 �5 ± 19 +3 ± 63 �26 ± 10 +229 ± 48 +98 ± 26 +599 ± 59N42 MS 42.8 ± 1.4 +123 ± 33 +89 ± 30 +14 ± 11 +5 ± 19 �50 ± 58 �10 ± 16 +55 ± 74 +35 ± 39 +272 ± 81N43 MS 73.2 ± 2.3 +37 ± 28 +42 ± 24 �1 ± 17 �6 ± 27N44 MS 68.2 ± 2.1 +36 ± 27 +29 ± 23 +11 ± 15 �18 ± 23 �54 ± 74 �3 ± 19 �34 ± 78N45 MS 102.3 ± 3.3 �14 ± 26 +16 ± 23 �2 ± 8 �17 ± 14 �21 ± 39 +23 ± 10 +31 ± 41 +18 ± 23 +291 ± 48N46 MS 56.8 ± 1.8 +45 ± 29 +46 ± 25 �9 ± 10 �2 ± 17 +17 ± 57 +3 ± 8 +44 ± 35N47 MS 45.3 ± 1.3 +85 ± 28 +67 ± 23 +10 ± 13 �31 ± 20 �47 ± 62 �5 ± 12 �18 ± 51 �9 ± 28 +93 ± 55N48 MS 65.0 ± 1.9 +12 ± 29 +17 ± 26 +4 ± 10 �12 ± 15 +17 ± 48 +4 ± 30 +180 ± 143N50 AB 8.0 ± 0.2 +171 ± 14 +113 ± 16 +41 ± 39 �40 ± 55 �142 ± 172 �96 ± 18 +79 ± 83 +42 ± 47 +259 ± 96

(continued on next page)

R.Trap

pitsch

etal./

Geochim

icaet

Cosm

ochim

icaActa

221(2018)

87–10895

Table 1 (continued)

Grain Type 12C/13C d29 Si d30Si d54Fe d57Fe d58Fe d60Ni d61Ni d62Ni d64Ni

N51 MS 52.4 ± 1.9 +61 ± 31 +56 ± 29 +17 ± 24 +18 ± 34 +253 ± 120 �10 ± 12 +287 ± 59 +149 ± 32 +798 ± 76N52 MS 52.8 ± 1.7 +71 ± 29 +56 ± 25 +22 ± 20 �13 ± 30 +124 ± 121 +16 ± 10 +309 ± 50 +132 ± 26 +903 ± 66N53 MS 55.3 ± 1.8 +71 ± 29 +46 ± 25 �15 ± 42 �19 ± 62 +161 ± 216 �26 ± 18 +260 ± 85 +104 ± 45 +786 ± 108N54 MS 61.5 ± 2.1 +19 ± 29 +17 ± 26 +1 ± 34 �8 ± 50 +253 ± 172 +100 ± 24 +264 ± 104 +124 ± 56 +997 ± 141N55 MS 46.6 ± 1.4 +56 ± 29 +27 ± 25 �1 ± 16 �9 ± 24 +96 ± 89 +13 ± 12 +217 ± 54 +77 ± 29 +649 ± 67N56 MS 56.0 ± 1.7 +33 ± 27 +33 ± 23 +13 ± 16 +2 ± 25 +157 ± 91 +19 ± 8 +380 ± 40 +181 ± 21 +946 ± 53N59 MS 40.7 ± 1.2 +131 ± 29 +99 ± 24 +24 ± 28 �8 ± 42 �92 ± 139 �24 ± 15 +273 ± 72 +145 ± 38 +800 ± 91N60 MS 52.6 ± 1.2 +6 ± 14 +2 ± 16 �15 ± 34 �16 ± 52 +38 ± 172 �30 ± 24 +110 ± 108 +23 ± 60 +185 ± 120N61 MS 17.5 ± 0.3 +52 ± 11 +42 ± 12 +4 ± 33 �55 ± 47 +65 ± 152 +15 ± 29 +24 ± 121 +38 ± 69 +336 ± 148N62 MS 49.6 ± 1.1 +96 ± 13 +60 ± 15 �2 ± 58 �70 ± 84 �18 ± 268 �111 ± 90 �239 ± 336N63 MS 91.9 ± 2.5 �19 ± 13 �34 ± 15 �45 ± 62 �126 ± 88 �149 ± 273 +10 ± 91 �120 ± 571 +9 ± 213 +310 ± 505N64 MS 64.1 ± 1.4 +3 ± 11 +4 ± 12 +14 ± 30 �39 ± 45 �73 ± 140 +2 ± 33 �8 ± 144N65 MS 37.8 ± 0.8 +156 ± 13 +103 ± 14 +22 ± 20 +17 ± 30 +131 ± 101 �55 ± 14 +149 ± 69 +93 ± 35 +457 ± 77N66 MS 46.4 ± 1.1 �20 ± 13 +5 ± 15 +10 ± 24 �30 ± 35 +73 ± 115 +81 ± 31 +72 ± 130N67 MS 63.6 ± 1.6 +65 ± 15 +38 ± 17 �2 ± 44 �57 ± 63 +21 ± 204 �41 ± 35 +168 ± 160 +35 ± 86 +350 ± 188N68 MS 71.5 ± 1.9 +50 ± 16 +19 ± 18 +16 ± 16 +20 ± 25 +467 ± 96 �27 ± 10 +346 ± 55 +109 ± 26 +934 ± 67N69 MS 59.1 ± 1.5 +65 ± 15 +42 ± 17 +1 ± 6 �9 ± 11 �16 ± 30 �2 ± 10 +40 ± 48 +1 ± 21 +95 ± 42N70 MS 48.6 ± 1.1 +125 ± 13 +91 ± 14 +26 ± 24 +30 ± 37 +347 ± 136 �4 ± 16 +290 ± 77 +106 ± 41 +611 ± 92N71 MS 59.9 ± 1.3 +50 ± 12 +40 ± 14 �2 ± 13 �1 ± 21 +51 ± 69 +26 ± 11 +248 ± 52 +109 ± 28 +709 ± 66N72 MS 90.7 ± 1.9 +6 ± 11 +24 ± 12 �20 ± 18 +8 ± 27 +117 ± 97 +74 ± 12 +384 ± 58 +219 ± 30 +1318 ± 82N73 MS 55.4 ± 1.3 �48 ± 12 �14 ± 14 �43 ± 22 �9 ± 34 +27 ± 169 +186 ± 17 +460 ± 75 +246 ± 39 +1208 ± 100N74 MS 69.5 ± 1.9 +4 ± 16 �19 ± 19 �7 ± 11 +38 ± 17 +290 ± 62 +30 ± 6 +319 ± 33 +169 ± 17 +985 ± 45N75 MS 42.4 ± 0.9 +119 ± 12 +85 ± 13 �16 ± 23 +6 ± 36 +34 ± 114 +9 ± 25 �14 ± 103 +115 ± 64 +233 ± 123N76 MS 42.3 ± 0.9 +8 ± 11 +16 ± 12 �49 ± 40 +57 ± 64 +123 ± 239 +76 ± 15 +399 ± 72 +259 ± 39 +866 ± 92N78 MS 65.3 ± 1.4 +35 ± 12 +37 ± 13 �13 ± 21 �17 ± 32 +64 ± 115 +14 ± 12 +283 ± 57 +140 ± 30 +858 ± 75N79 MS 42.1 ± 0.9 +147 ± 14 +115 ± 15 +31 ± 31 �12 ± 46 +137 ± 168 �50 ± 17 +155 ± 80 +105 ± 45 +550 ± 100N80 MS 49.4 ± 1.0 +62 ± 11 +48 ± 12 �7 ± 17 �18 ± 26 +46 ± 107 +3 ± 17 +138 ± 74 +31 ± 40 +348 ± 86N81 MS 56.8 ± 1.2 +87 ± 11 +82 ± 12 +3 ± 13 �26 ± 19 +32 ± 64 +37 ± 11 +276 ± 52 +116 ± 27 +892 ± 68N82 Y 108.5 ± 2.8 �7 ± 12 +12 ± 14 �20 ± 12 �6 ± 18 +98 ± 60 +92 ± 8 +324 ± 39 +211 ± 21 +1244 ± 59N83 MS 68.6 ± 1.7 +11 ± 13 +33 ± 15 �10 ± 15 +2 ± 23 +2 ± 73 +33 ± 12 +149 ± 51 +83 ± 28 +428 ± 60N84 MS 70.4 ± 1.6 +24 ± 12 +34 ± 14 +1 ± 17 �18 ± 26 �42 ± 81 +48 ± 15 +205 ± 66 +103 ± 36 +802 ± 88N85 MS 43.8 ± 0.9 +111 ± 11 +90 ± 12 +22 ± 31 �4 ± 47 +123 ± 155 +6 ± 24 +153 ± 109 +69 ± 58 +365 ± 124N86 MS 88.8 ± 2.0 �13 ± 11 �23 ± 12 +8 ± 16 �29 ± 24 +41 ± 81 +2 ± 18 �25 ± 74 +39 ± 43 +107 ± 83N87 MS 59.5 ± 1.3 +50 ± 13 +33 ± 14 +4 ± 19 �13 ± 31 �7 ± 96 �5 ± 22 +196 ± 101 +19 ± 52 +533 ± 121N89 MS 67.1 ± 1.7 +33 ± 14 +33 ± 16 +4 ± 27 +18 ± 41 +67 ± 141 +46 ± 10 +337 ± 49 +155 ± 26 +1206 ± 69N90 MS 52.1 ± 1.2 +85 ± 14 +51 ± 15 �6 ± 15 �18 ± 24 +12 ± 73 +5 ± 10 +245 ± 48 +126 ± 26 +713 ± 61N91 MS 46.0 ± 1.1 �41 ± 13 �7 ± 16 +8 ± 12 �16 ± 21 �36 ± 69 +8 ± 12 +10 ± 50N92 MS 53.2 ± 1.6 +85 ± 29 +62 ± 24 �15 ± 27 �13 ± 41 +140 ± 144 +19 ± 21 +304 ± 100 +118 ± 53 +749 ± 124N93 AB 8.3 ± 0.2 +181 ± 37 +101 ± 34 +22 ± 17 +6 ± 27 +42 ± 94 �108 ± 9 +131 ± 43N94 Z 46.6 ± 1.5 �71 ± 25 +46 ± 25 �91 ± 29 +2 ± 45 �59 ± 141 +1145 ± 28 +571 ± 86 +595 ± 50 +1536 ± 121N95 Y 117.2 ± 4.0 �27 ± 26 �5 ± 24 �11 ± 18 +2 ± 28 +60 ± 107 +141 ± 11 +468 ± 54 +248 ± 29 +1447 ± 78N106 U �43 ± 24 �5 ± 38 +223 ± 125 +61 ± 9 +349 ± 48 +187 ± 24 +1245 ± 65N107 MS 68.1 ± 1.6 �24 ± 12 +1 ± 14 �18 ± 10 �2 ± 18 +245 ± 57 +72 ± 8 +276 ± 38 +144 ± 19 +696 ± 46N108 MS 96.6 ± 2.5 �15 ± 13 +10 ± 15 �10 ± 8 �2 ± 13 0 ± 58 +70 ± 10 +161 ± 42 +131 ± 23 +848 ± 57N109 MS 51.8 ± 1.1 +65 ± 12 +47 ± 14 +20 ± 20 �1 ± 30 +59 ± 103 +4 ± 13 +255 ± 61 +112 ± 32 +772 ± 76SiC70 X 923 ± 34 �359 ± 17 �513 ± 11 �5 ± 12 +5 ± 19 0 ± 70 +60 ± 18 +1411 ± 120 +362 ± 50 +1217 ± 120

96R.Trap

pitsch

etal./

Geochim

icaet

Cosm

ochim

icaActa

221(2018)

87–108

Fig. 5. Comparison of our results for iron (a) and nickel (b) isotopic measurements with literature values by Marhas et al. (2008) and Ong andFloss (2015). Error bars are 2r.

Fig. 6. Silicon isotopic measurements. The inset figure shows anoverview, including the X grain. The isotopic ratios of themainstream grain are consistent with literature values (see, e.g.,Hynes and Gyngard, 2009).

Fig. 7. Mixing lines between the observed extreme of the SiC dataand CI for a variety of iron-to-nickel elemental ratios.


5. DISCUSSION

5.1. Considerations on presolar SiC grain contamination

The highest isotope anomalies measured in presolar SiC

mainstream grains were d58Fe ¼ 467� 96‰ in grain N68

and d64Ni ¼ 1318� 82‰ in grain N72. These anomaliesare in good agreement with model predictions for AGBstars (see below). However, the isotopic compositions ofmany grains are closer to the Solar System value, whichposes the question of whether some or all of the analyzedgrains were significantly contaminated with solar material.Using the CI iron-to-nickel ratio and CI isotopic composi-tion, we calculated mixing curves between a solar compo-nent and an assumed presolar grain component. For thepresolar component, we used +600‰ for d58Fe and+1300‰ for d64Ni, both round values that are in good

agreement with the highest anomaly measured as well aswith the AGB star predictions. Fig. 7a shows a series ofmixing calculations between the Solar System and ourassumed presolar grain component. Stellar envelopes ofAGB stars are predicted to have iron-to-nickel ratios thatare in the same range as the CI ratio. However, Marhaset al. (2008) reported that presolar grains are highlyenriched in nickel compared to iron. It is thus unclearhow iron and nickel condense into presolar SiC grains.Therefore, the elemental ratio of an uncontaminated SiCgrain is not known. We modeled a range of iron and nickelelemental ratios, from the lowest measured ratios byMarhas et al. (2008) to about the ratio expected in AGBstar envelopes (which is very close to the CI chondriteratio). As Fig. 7 shows, contamination with Solar System


material could explain some of the measured isotopic data,but the full range of iron-to-nickel ratios in uncontami-nated grains is required. Since the CI iron-to-nickel ratiois usually higher than the ratio in the measured presolargrains, it is easier to contaminate a presolar grain with ironthan with nickel.

In general, we cannot exclude contamination with SolarSystem material. Acid cleaning of presolar grains has beenshown to mostly remove contamination for barium andstrontium (Liu et al., 2014, 2015a). For iron and nickelhowever, we cannot say whether acid cleaning completelyeliminates contamination; on the contrary, Fig. 7 showsthat some contamination, especially with iron, is likely tohave remained even after cleaning.

5.2. Comparison with AGB star models

5.2.1. Comparison of measurements with updated Torino and

FRUITY models

Figs. 8 and 9 compare our data with the FRUITY andTorino models. We chose stellar models of 1.5 and 2M�with various metallicities as a representative sample oflow-mass AGB stars to explain the presolar SiC grain mea-surements. The Torino models shown here were computedwith the st case for the concentration of 13C in the 13C-pockets (see Section 3.1). Note that the st case is not thepreferred case; however, it can be used to compare theeffects of varying the stellar mass and metallicity. In Sec-tion 5.2.2 we will discuss the preferred model as well asthe influence of the 13C concentration in the 13C-pocket.While the 13C mass fraction in the 13C-pocket has a largeinfluence on 58Fe and 64Ni, the case used here primarilyserves to compare models with different masses and metal-licities. We plot envelope compositions for every thermal

Fig. 8. Comparison of all iron isotopic measurements with the updated Tthe envelopes of AGB stars are plotted for all thermal pulses (connectedunity are marked by additional symbols.

pulse of the AGB star and connect them with lines. Sym-bols on these lines indicate when the carbon-to-oxygenratio in a given thermal pulse is larger than unity(carbon-rich). Note that some models predict a carbon-rich envelope after only a few thermal pulses, and have onlyminor anomalies in the oxygen-rich envelope phase. If thecarbon-to-oxygen ratio in the envelope after a given ther-mal pulse is smaller than one, all of the carbon is boundin stable gas-phase CO, and no carbon is left to formSiC. Presolar SiC grains can therefore only form whenthe envelope has a carbon-to-oxygen larger than one (see,e.g., Lodders and Fegley, 1999). In addition, about halfof the total envelope mass loss occurs after the last thermalpulse with a third dredge-up. The composition of presolarSiC grains from a given AGB star is most likely to matchwith this last pulse, assuming that the pure AGB star com-ponent is measured. The 1:5M�; Z� Torino and FRUITYmodels do not predict carbon-rich thermal pulses and aretherefore only plotted as lines.

The iron isotopic composition measured in some preso-lar SiC grains can be explained by 2M� Torino models with0:44Z� and Z� metallicity. The st case here however under-predicts the excess in 58Fe for many presolar SiC by several100‰. The plotted FRUITY models with 2M� explain thehigher excesses in 58Fe.

Fig. 9 compares our nickel measurements with the samemodels. Some of the measured presolar SiC grains fromAGB stars show slightly larger anomalies in 64Ni and in58Fe than what is predicted in the Torino models (st case).The FRUITY models severely underpredict the excesses in64Ni and cannot explain the measurements.

While grains with lower than predicted excesses in 58Feand 64Ni can be explained by contamination with Solar Sys-tem material, grains with larger anomalies than predicted

orino and the FRUITY AGB star models. Isotopic compositions inline). Thermal pulses for which the carbon-to-oxygen ratio exceeds

Fig. 9. Comparison of all nickel isotopic measurements with the FRUITY and the updated Torino AGB models. Isotopic compositions in theenvelopes of AGB stars are plotted for all thermal pulses (connected line). Thermal pulses for which the carbon-to-oxygen ratio exceeds unityare marked by additional symbols.


require another explanation. In the next two sections wewill first explore the Torino models and discuss various13C sizes, as well as new and updated FRUITY models.

This comparison mostly shows that lower metallicitystars predict larger anomalies in 58Fe and 64Ni than modelswith higher metallicities. The same is the case for heaviermass stars over lower mass stars. Next we will discuss theinfluence of the 13C-pocket on the nucleosynthesis of theseisotopes.

Fig. 10. Comparison of our measurements with the Torino models for thin the 13C-pocket for the pocket with the standard mass previously used (

5.2.2. d58Fe and d64Ni as indicators for the 13C-pocket

Fig. 10 compares our measurements with the2M�; 0:44Z� model. We use this model as a representativecase for a star that forms presolar SiC grains and look atthe effect of different 13C-pockets on the production of58Fe and 64Ni. Fig. 10a shows various cases of 13C concen-tration in the 13C-pocket for the standard size 13C-pocket,i.e., the size that has previously been used in the Torinomodels. The right panel, on the other hand, shows the same

e 2M�; 0:44Z�. We show the various cases of concentrations of 13Ca) and for a pocket that has twice the standard mass (b; pocketx2).


cases of 13C concentrations, however, for a 13C-pocket oftwice the mass. We focus on 64Ni and 58Fe since these iso-topes show the largest anomalies in the measurements aswell as in the model predictions.

The standard size 13C-pocket model (Fig. 10a) underpre-dicts the presolar SiC grains with the largest anomalies ind58Fe and d64Ni. Models with a slightly lower 13C concen-tration than the standard case (D1.5 to D3) seem to onlyslightly underpredict the measured values of d58Fe by� 100‰ and d64Ni by � 200‰. The model with a twiceas massive 13C-pocket (panel b) requires a lower 13C con-centration in the 13C-pocket to explain the measured iso-tope ratios. The D6 case however closely predicts themeasured d58Fe and d64Ni values. Note that the rest ofthe modeled concentrations of 13C in the more massive13C-pocket do not predict the measured anomalies. Liuet al. (2015a) did not use the TDUF modification of theTorino models, but did find success in matching data withTorino models with more massive 13C-pockets with lower13C amounts. We found that the TDUF modification doesnot significantly change the predictions for more massive13C-pockets, but this will be explored further in a futurepublication. It is unclear if grains that show lower anoma-lies require a different stellar source or if these measure-ments are simply the result of contamination. Fig. 7ashows that contamination with solar material could explainthe measured isotopic compositions.

Both models in Fig. 10 show a curious behavior in d58Feand d64Ni. In terms of 13C concentration in the 13C-pocket,the models predict a steady rise in d58Fe with falling 13Cconcentrations. Note that the st case has the highest andthe D12 case the lowest concentration of 13C in the 13C-pocket. For d64Ni however, the behavior is quite different.With falling 13C concentration, the value of d64Ni first

Fig. 11. Comparison of the nickel isotopic measurements wi

grows up to case D2, and then falls again. Both modeledmasses of the 13C-pocket show the same behavior. Themodels show the same behavior for d60Ni, d61Ni, andd62Ni, however at a much lower amplitude. With rising13C concentration in the 13C-pocket, more neutrons areavailable to be captured on the iron and nickel seed nucleito form 58Fe and 64Ni. More available neutrons howeveralso lead to the formation of more heavy elements. At13C concentrations higher than case D2 in the 13C-pocket,the higher amount of heavy elements capture proportion-ally more neutrons than the iron and nickel seed nuclei.Thus, fewer neutrons are available to form 58Fe and 64Ni,which leads to a lower anomaly in these isotopes for thecases with the highest 13C concentration in the 13C-pocket. In addition, 64Ni has a lower neutron capture crosssection than 58Fe and 58Ni has a higher neutron capturecross section than 56Fe, which yields to a larger predictedanomaly in d64Ni than in d58Fe.

Fig. 11 compares the best fitting models from Fig. 10

with the regular 13C-pocket mass and for the pocketx2 case.As discussed before, the pocketx2 case predicts better themeasured anomalies in 58Fe and 64Ni. In addition, this casealso yields a better agreement with the measurements ofd60Ni, d61Ni, and d62Ni.

5.2.3. In-depth comparison with FRUITY models

The comparison between grain measurements andFRUITY models done in the previous Section shows thata rather poor agreement can be attained. The choice of adifferent initial stellar mass (1.5 or 3M�) would not signifi-cantly improve the situation. Therefore, different solutionswere explored.

The difficulty in computing reliable and robust theoret-ical stellar models mostly derives from the lack of knowl-

th the two preferred Torino model cases from Fig. 10.


edge of many physical phenomena that are at work duringthe AGB phase, the most important being convection androtation. Both processes may affect the formation and evo-lution of the 13C pocket in AGB stars. Cristallo et al.(2015b) discussed the effects that a different handling ofthe radiative/convective interface during a third dredge-upepisode has on s-process nucleosynthesis (see alsoCristallo et al., 2016). In particular, they substituted theirprevious criterion to fix the maximum penetration of pro-tons (based on the variation of the local pressure scaleheight) with a new criterion based on convective velocities.Thus, they determined the velocity of the most internal con-vective mesh point (or position) by checking that a furtherpenetration would produce a negligible variation of the sur-face s-process nucleosynthesis (such a velocity is equal to

10�11 � vSC , where vSC is the velocity of the convective meshpoint defined by the Schwarzschild criterion). This new pre-scription stretches the 13C pocket, leading to the formationof an extended 13C tail. In Fig. 12a, we report for a starwith initial mass of 2M� different models computed withthis new treatment of the inner border of the convectiveenvelope (labeled as TAIL). It can be seen that, with respectto the reference FRUITY model, there is a slight increase ofthe d58Fe, while there is a consistent increase of the d64Ni byalmost a factor of three. This can be easily understood bylooking at Fig. 12b, where we report the iron and nickel iso-topic abundances in the stellar layers where the first 13Cformed. The abundances are taken shortly before the onsetof the next thermal pulse. While 56Fe and 58Ni aredestroyed, 58Fe and 64Ni are produced. This figure showsthat 56Fe and 58Ni are more greatly depleted and 58Feand 64Ni are more greatly enhanced in the TAIL case thanin the standard FRUITY case. The larger d64Ni withrespect to d58Fe arises from the fact that the radiative 13Cburning increases the local 64Ni by about a factor 100, whilethe growth of 58Fe is limited to a factor ten. Another pro-cess which may significantly affect isotopic abundances isrotation. Piersanti et al. (2013) showed how the inclusionof rotation in an AGB model modifies its s-processnucleosynthesis. The major effect induced by rotation is

Fig. 12. (a) Comparison of all measured presolar SiC with the FRUITY58Fe, 58Ni, and 64Ni in the stellar layers where the first 13C-pocket form

the partial mix between the 13C and the 14N pockets, thelatter being the major neutron poison of the s-process viathe resonant 14N(n,p)14C reaction. Later, Liu et al.(2015a) demonstrated that rotating AGB models withextended 13C tails provide the best fit to strontium and bar-ium isotopic measured in presolar grains, which can befound in the framework of the FUNS code. We thus calcu-lated a model with extended 13C tails and with an initialrotation velocity v = 60 km/s (see Fig. 12a). In this model,the d64Ni further increases, reaching a value around900‰. Thus, we confirm the previous result of Liu et al.(2015a). The last parameter to be explored is the initialmetallicity; therefore, we ran two additional models withextended 13C-pockets: the first with solar metallicity(Z ¼ 0:0138) and the second with a lower metallicity(Z ¼ 0:006, corresponding to 0:4Z�). While the first casedoes not improve the situation, the second model shows lar-ger d64Ni values (up to 1100‰), possibly suggesting that themost enriched grains formed in AGB stars with about halfof the solar metallicity. It has to mentioned that the grainsmost enriched in d64Ni cannot be matched by FUNS mod-els. Moreover, it has to be noted that the agreementbetween laboratory measurements and theory is still ratherpoor for d58Fe. However, in this case a possible contamina-tion with solar material cannot be excluded, considering thedetection of grains extremely rich in d64Ni but showingalmost solar d58Fe.

5.3. Implication for GCE of iron and nickel isotopes

Fig. 13 shows relationships of the GCE-dominated iso-tope ratios d29Si, d54Fe, and d60Ni to each other as wellas in comparison with normalized GCE predictions byKobayashi et al. (2011). These GCE models range fromsolar metallicity, which plots at the origin, towards½Fe=H� ¼ �0:5 and apply to the solar neighborhood.Unweighted linear regressions were calculated for all main-stream grains. The measurements show a large scatter,likely due to inhomogeneities in the interstellar medium;therefore the fits represent GCE trends rather than strict

and TAIL models for various metallicities. (b) Abundance of 56Fe,ed in the FRUITY and TAIL models for the 2M� star.

Fig. 13. Comparison of the GCE-dominated isotope ratios d29Si with d54Fe (a), d29Si with d60Ni (b), and d54Fe with d60Ni (c) with GCEmodels by Kobayashi et al. (2011) (dashed line). The GCE models are normalized to the Solar System as described in the text. The linear fitsare unweighted linear regressions of the mainstream grain measurements and the calculated slope is given as m, the y-intercept as b. The graybands are 95% confidence intervals.


linear relationships. GCE trends in stellar elemental abun-dances also show considerable scatter, likely for the samereason.

The trends found in the measurements are consistentwith previous findings on GCE for silicon and titanium iso-topes (Alexander and Nittler, 1999; Huss and Smith, 2007).While the d54Fe versus d29Si trend has a positive slope, thetrends d60Ni versus d29Si and d54Fe versus d60Ni have neg-ative slopes. This is expected from the GCE models(Kobayashi et al., 2011) as shown in Fig. 3. The iron andnickel nuclides are mostly produced in CCSNe and typeIa SNe, with the latter type having a delayed start beforecontributing to the GCE. The low metallicity supernovamodel of Kobayashi et al. (2011) (18M�; ½Fe=H� ¼ 0:04),i.e., a star that contributes early on to GCE, predicts

d54Fe ¼ þ121‰ and d60Ni ¼ �530‰. Supernovae type Iamodels by Nomoto et al. (1997), which are the models usedfor the GCE model by Kobayashi et al. (2011), predictd54Fe between +350‰ and +1600‰ and d60Niaround �700‰. Type Ia supernovae however only start

contributing to the GCE later on in galactic history. The s-process on the other hand is expected to drive the ratios inthe opposite direction, however, only minimally. The Torinomodel 2M�; 0:44Z�, case D6, pocketx2 star is expected toinfluence the ratios by �5‰ and +39‰ in d54Fe andd60Ni, respectively. These shifts are small compared to thepredictions by SNe. Thus we expect the d54Fe in the galaxyto increase with time while the d60Ni is expected to decrease.Our measurements clearly show this trend, thus agreeingqualitatively with the GCE model by Kobayashi et al.(2011). Note that our measurements only agree with themodel because we renormalized the GCE model such thatit ends up at the solar composition for solar metallicity.The actual model however predicts a d54Fe of +724‰ forsolar metallicity, showing that the model predictionsrequire significant improvements in order to agree with theconstraints of presolar grain measurements.

The modeled GCE trends as shown in Fig. 13c agreerather well with the measured correlation between d54Feand d60Ni. A slight preferential contamination of the preso-


lar SiC grains with iron, as noted above in the comparisonswith the AGB models, could even explain why the mea-sured trend has a shallower slope than the models byKobayashi et al. (2011). However, the model does not agreequantitatively with the measured correlations betweend54Fe and d60Ni with d29Si. Fig. 13a shows that the GCEmodel overproduces 29Si with respect to 54Fe and 60Ni.Note that, since we renormalized the GCE model to gothrough the solar composition at a metallicity of [Fe/H]= 0, the ‘‘overproduction” in 29Si can only be discussed rel-ative to 54Fe and 60Ni. Contamination in iron and nickel,which are both trace elements, without contamination insilicon, one of the major elements in SiC, could explainthe difference in the measured versus the modeled slopes.Why the GCE model otherwise overpredicts the d29Si isunclear, however, for unnormalized values, there has beena long-standing problem with underproducing 29Si in super-novae (Timmes and Clayton, 1996). The models byKobayashi et al. (2011) seem to suffer from the same issuesas previous models. Normalizing the GCE model to solarcomposition requires a significant correction; the GCEmodel predicts for solar metallicity a d29Si value of�534‰. It is therefore not surprising that the trendsbetween d54Fe and d60Ni versus d29Si do not perfectly align,however, it is interesting that the models predict about themeasured trend between d54Fe and d60Ni.

The linear regressions through the presolar SiC main-stream grain data in Fig. 13 do not include the Solar Systemaverage, however, the plotted data still include the AGBstar contribution to the isotope abundance. The Torino2M�; 0:44Z�, case D6, pocketx2 model predicts isotope

anomalies of d54Fe ¼ �5‰ and d60Ni ¼ þ39‰. In the sili-con isotopes the total enhancement of 29Si/28Si from theAGB star is approximately �2‰. These enhancementsare model-specific, however, on average Fig. 13 shows thatour measured GCE trend line includes the Solar Systemaverage composition within uncertainty. Thus, the SolarSystem is not a special case with regard to its iron andnickel isotopic composition.

Testing the validity of heterogeneous GCE, a galacticmerger model, or other scenarios for the measured correla-tions is outside of the scope of this work and will beaddressed in future studies. However, we would like topoint out the agreement of the measured d54Fe and d60Nitrend with the GCE models by Kobayashi et al. (2011).Such relationships between multielement isotopic systemsmeasured in individual presolar SiC promise a great testingground for GCE models.

5.4. AGB star origin of the presolar SiC Y, Z, and AB grains

It has been argued that presolar SiC Y grains come fromlow-metallicity stars (Amari et al., 2001b). The isotopicanomaly in 64Ni in the two Y-grains measured is slightlyhigher than in the mainstream grain with the highest anom-aly. However, the Y grains show no significant anomaly in58Fe, which implies that they are significantly contaminatedwith iron. While the 2M�; 0:44Z�, case st, pocketx2 Torinomodel predicts a slightly lower anomaly in 64Ni and sincecontamination seems to be a significant problem with the

Y grains, it is possible that these grains originated in a starwith lower metallicity. As shown previously, lowering themetallicity increases the anomalies in the neutron-rich ironand nickel isotopes.

Presolar AB grains have a more ambiguous origin(Amari et al., 2001b,c), however, their iron and nickel iso-topic compositions agree with the AGB star models dis-cussed here. These grains show small anomalies in theirnickel and none in their iron isotopic composition. Thiscould be caused by severe contamination with solar ironand nickel, but a higher mass star with close to solar metal-licity would also have small iron and nickel isotopeanomalies.

At first glance, the presolar SiC Z grain N94 shows ironand nickel isotopic compositions that are in agreement withan origin in an AGB star, however, its d62Ni ofþ595� 50‰ is much higher than expected from an AGBstar. While a low-metallicity AGB star could account forlarger anomalies in d61Ni, d62Ni, and d64Ni, the large anom-aly in d60Ni of þ1145� 28‰ is two orders of magnitudehigher than the expected anomaly in a low-mass AGB star.While the d30Si of this grain is positive, which allowed us toclassify it as a Z grain, we have to consider that this grainwas misclassified and might actually be from a core collapsesupernova; we will refer to it as the ‘‘odd” Z-grain. The sec-ond measured Z grain N37 on the other hand agrees wellwith an origin in AGB stars. It does not seem to requirea particularly low-metallicity star, however, the mostly nor-mal isotopic ratios measured in this grain could be due tocontamination.

5.5. Potential CCSN origin and mixing calculations for two

SiC grains

Contrary to observations for X grains by Marhas et al.(2008), the SiC X grain in our study has normal iron iso-topic composition, which makes it difficult to explain allmeasured isotope ratios using SNe models. The nickel iso-topic composition, on the other hand, shows

d60Ni ¼ þ60� 18‰, d61Ni ¼ þ1411� 120‰, d62Ni ¼þ362� 50‰, and d64Ni ¼ þ1217� 120‰. The excesses in61Ni and 64Ni have also been observed in supernova grainsby Kodolanyi et al. (2018), but the excesses in the single Xgrain we analyzed exceed those found by them in 25 Xgrains. In our grain, we measured low iron and nickel con-centrations of 6 ppm and 1 ppm, respectively. A small con-tamination with meteoritic or terrestrial metal couldtherefore explain the measurement of a normal iron iso-topic composition while we measure anomalies in nickelisotopes. We will thus not try to fit the iron isotopic compo-sition to SNe model calculations.

Fig. 14 shows the iron and nickel isotope abundancesinside a 15M� SN model (Rauscher et al., 2002), assumingthat all the radioactive species, except 60Fe, have decayedfor 4.567 Ga. The green curves show 60Fe and 60Ni 1.4aafter the model, i.e., around the time when dust condensesaccording to observations by Wooden (1997). Certain zoneswithin the SN have more 60Fe than 60Ni at this point andare shaded in green. Also shown are the names of the differ-ent zones we used for the following mixing calculation.

Fig. 14. Iron and nickel isotope abundances inside the 15M� SN model (Rauscher et al., 2002) assuming that all the radioactive species ofinterest, except 60Fe, have decayed for 4.567Ga. The green curves show the 60Fe and 60Ni 1.4years after the model ends, i.e., 60Fe has not yetfully decayed and is in certain zones more abundant than 60Ni (green shaded areas).


Using this model, we found a mixture that can describethe carbon, silicon, and nickel isotopic composition well,however it does not explain the iron isotopic measurements.Since the iron isotopic composition has no anomalies, con-tamination of the presolar SiC grain with iron (but notnickel) is a likely explanation for the observation. In orderto explain the silicon isotopic composition, parts of the Si/Szone must be mixed into the grain. This zone however con-tains large amounts of 54Fe resulting in a positive d54Feanomaly. The measured values along with the predictedvalues from the SN model are shown in Table 2. Alsoshown are the proportions of the respective zone admix-tures for our optimized fits, as well as the carbon-to-oxygen ratio at the time of condensation. The iron isotopicmeasurements were not used to fit the model to the data.The model predicts well the carbon isotopic ratios and also

Table 2Presolar SiC X grain (SiC70) in comparison with the optimum zonemixture from the 15M� model by Rauscher et al. (2002). Ironisotopic measurements, given in italic, were not considered for thisfit.

Measured SN prediction

Si/S zone (%) 0.22O/Ne zone (%) 0.57O/C zone (%) 0.09He/C zone (%) 61.97He/N zone (%) 37.15C/O (atoms/atoms) 3.6512C/13C (atoms/atoms) 923 ± 34 899d29Si (‰) �359 ± 17 �426d30Si (‰) �513 ± 11 �467d54Fe (‰) �5 ± 12 1075

d57Fe (‰) 5 ± 19 953

d58Fe (‰) 0 ± 70 2954

d60Ni (‰) 60 ± 18 0d61Ni (‰) 1411 ± 120 1711d62Ni (‰) 362 ± 50 577d64Ni (‰) 1217 ± 120 976

shows the correct trends in the nickel isotopic composi-tions. The silicon isotopic ratios however are not ade-quately reproduced. The result of the SN mixingcalculation for d60Ni is zero, however, a positive d60Nianomaly was measured. This could be explained if moreiron condenses into the presolar SiC grain than nickel, pro-ducing after condensation a positive d60Ni anomaly due to60Fe decay. The rest of nickel isotopic measurements fit wellwith the mixing calculation. The carbon-to-oxygen atomicratio when the grain condenses is 3.65, i.e., a SiC graincould condense from such a mixture.

The supernova grain discussed above has large anoma-lies in d62Ni and d64Ni of 1411‰ and 1217‰, respectively.The odd Z grain on the other hand has similarly highanomalies, but in d60Ni and d64Ni. In addition, it hasanomalies of around 600‰ in d61Ni and d62Ni. While wecannot explain the nickel isotopic composition of the oddZ grain using a low-metallicity AGB star, we will explorethe possibility that this grain requires reclassification asan X grain and thus should be explainable by supernovamodels. Explaining this grain using the same SN model asused for the X grain is not possible. The silicon isotopiccomposition with a negative d29Si and a positive d30Si aredifficult to explain. Explaining the iron and nickel isotopesproves difficult as well. Especially hard to explain is thed54Fe value of �91� 29‰. Adding material from the Si/Szone, which could explain the silicon isotopic ratios, wouldresult in positive d54Fe values. Also the nickel isotopic com-position is difficult to explain. We measured d60Ni andd64Ni anomalies >1000‰ and d61Ni and d62Ni anomalies� þ600‰. Table 3 shows our best fitting mixing calculationexcluding (Mix 1) and including (Mix 2) the iron isotopicconstraints from the fit.

Both mixtures only poorly fit the odd Z grain, especiallythe measured d60Ni. Fig. 14 shows that the upper part of

the O/C zone has a higher abundance of 60Fe than 60Ni.We split this zone further up into two different zones, tryingto achieve a better fit for d60Ni. The only way of producingthe observed d60Ni abundance is by enhancing the iron

Table 3Presolar SiC Z grain (N94) in comparison with several zone mixtures from the 15M� model by Rauscher et al. (2002). For mix 3 and 4, theO/C2 zone is used instead of the whole O/C zone (see text). In addition, we assume that the condensation ratio of nickel to iron is 15 for mix 3and 18 for mix 4. For all mixtures, the carbon-to-oxygen ratio prior to condensation is 1.0. Values in italic were not considered in fitting themeasurements.

Measured Mix 1 Mix 2 Mix 3 Mix 4

Si/S zone (%) 0.003 0 0.01 0.002O/Si zone (%) 0.001 0.001 0.01 0.01O/C zone (%) 0.72 0.59 0.88 0.72He/C zone (%) 4.64 3.77 2.31 1.91He/N zone (%) 94.64 95.64 96.8 97.3612C/13C (atoms/atoms) 46 ± 1.5 57 47 53 44d29Si (‰) �71 ± 25 15 31 �33 16d30Si (‰) 46 ± 25 51 63 34 82d54Fe (‰) �91 ± 29 5 �11 47 2d57Fe (‰) 2 ± 45 140 123 117 105d58Fe (‰) �59 ± 141 851 694 852 699d60Ni (‰) 1145 ± 28 59 48 1159 1143d61Ni (‰) 571 ± 86 583 475 627 519d62Ni (‰) 595 ± 50 421 341 297 246d64Ni (‰) 1536 ± 121 951 771 1078 885

Fig. 15. Comparison of nickel isotope anomalies in bulk meteorites(Steele et al., 2011, 2012) with the measured presolar grain values.


condensation rate into the grain with respect to nickel. Mix3 and 4 in Table 3 show special mixtures, for which we alsosubdivided the O/C zone into two parts: O/C1 in which theabundance of 60Ni is larger than 60Fe and O/C2 in whichthe abundance of 60Ni is smaller than 60Fe at condensation.In addition, we had to enhance the iron condensation by afactor of 15 (Mix 3) and 18 (Mix 4) compared to all otherelements in order to optimize the results. This significantlyinfluences the final 60Ni abundance while not influencingany other isotope ratios. These two mixtures show potentialin explaining the composition of the measured Z grainusing SNe models. However, the negative d54Fe anomalyremains very difficult to explain.

While the supernova mixing calculation cannot explainthis grain, it has to be noted that these types of calculationshave much larger uncertainties than the AGB models. Thestandard picture is that Z grains come from AGB stars withlower metallicity than Y grains, however, the extremeanomalies in d60Ni cannot be explained in this context.

As shown before, 60Ni/58Ni ratios are only changed by afew permil in AGB stars, thus we cannot explain this grainin the context of a low-mass star. We therefore propose toreclassify this grain as an X grain. Newer supernova models(Pignatari et al., 2013a; Pignatari et al., 2013b; Pignatariet al., 2015) should be used in the future to better explainpresolar SiC X grains. A more detailed model discussionhowever needs a larger set of presolar SiC X grains mea-sured for their iron and nickel isotopic composition. Sucha set with an evaluation was recently acquired and is pre-sented in Kodolanyi et al. (2018).

5.6. Origin of nickel isotopic anomalies in the Solar System

Steele et al. (2011, 2012) measured nickel isotopicanomalies in bulk iron meteorites and bulk chondrites,respectively. To remove mass dependent fraction from thedataset, these authors normalized the reported isotopicratios internally to 58Ni and 61Ni to the terrestrial 61Ni/58Niratio, using an exponential mass fractionation law. Steele

et al. (2011, 2012) found that the dð64Ni=58NiÞ58Ni=61Ni and

dð62Ni=58NiÞ58Ni=61Ni isotopic values correlate very well with

each other across all meteoritic samples. Fig. 15 shows thesebulk measurements in comparison with the measuredpresolar grains. In order to directly compare our measure-ments, we also internally normalize our data using an expo-nential fractionation law. It should be noted that correctionof the X grain to the terrestrial 61Ni/58Ni ratio involved


very large corrections to d60Ni, d62Ni, and d64Ni, so muchso that the large positive d62Ni, and d64Ni values in theuncorrected data turned into large negative ones. WhileSteele et al. (2011) concluded that the isotopic anomaliesin iron meteorites are consistent with an origin in type IaSNe, Steele et al. (2012) concluded that type Ia SNe arenot a likely source of the measured nickel anomalies. Theyexplained the isotopic anomalies by CCSNe with contribu-tions from the Si/S zone. Indeed, the observed trend goesfrom the Solar System composition towards the measuredpresolar X grain. Our work therefore agrees with the argu-ment, that the nickel isotopic anomalies of

dð64Ni=58NiÞ58Ni=61Ni and dð62Ni=58NiÞ58Ni=61Ni observed in

bulk meteorites (Steele et al., 2011, 2012; Tang andDauphas, 2014) are in agreement with a CCSN origin.

6. CONCLUSIONS

We measured 74 presolar SiC grains for all of their ironand nickel isotopes simultaneously using CHILI at theUniversity of Chicago. In addition, we reported the resultsof a previous study (Trappitsch et al., 2012) that measured13 presolar SiC grains for their iron isotopic compositionwith CHARISMA at Argonne National Laboratory. Theiron and nickel isotope measurements performed withCHILI have uncertainties that are about a factor of twosmaller than the ones measured with CHARISMA andabout a factor of four smaller than previous analysis donewith the CAMECA NanoSIMS 50 (Marhas et al., 2008;Ong and Floss, 2015). This shows the high sensitivity of thisnew instrument, since uncertainties are mostly limited bycounting statistics.

Our isotopic studies, especially for presolar grains inwhich iron and nickel were measured, can be used to con-strain the composition and size of the parent star by lookingat the neutron-rich isotopes. The isotopic composition of themeasured presolar SiC mainstream grains agrees best withan origin in low-mass AGB stars. We showed that our mea-surements for d58Fe and d64Ni agree with the models pro-posed by Liu et al. (2014, 2015a), i.e., the Torino models

when adopting half the standard 22Neða; nÞ25Mg reactionrate by Kappeler et al. (1994), twice the 13C-pocket masscompared to older models, as well as a low 13C concentrationin the pocket (caseD6). In addition, we updated these Torinomodels to include a similar mass loss prescription and a sim-ilar amount in material dredged up in each third dredge-upepisode as the FRUITY models. While our measurementsfor d58Fe and d64Ni do not agree with the standard FRUITYmodels, FRUITY models with an extended 13C-pocket,which are more similar to those characterizing the Torinomodels, better reproduce the measured values.

The Y grains agree with an origin in a low-mass, low-metallicity AGB star assuming their record was contami-nated with solar material. The measured AB grains andone of the measured Z grains agree well with low-massAGB stars that have close to solar metallicity. It cannotbe excluded here that the measured compositions are dueto contamination with isotopically normal material or ifthese compositions actually represent the nucleosynthetic

fingerprint of the environment in which the grains formed.The second measured ‘‘odd” Z grain cannot be explainedby any AGB star model due to its extremely high anomalyin d60Ni and d64Ni. We thus propose to classify this grain asan X grain rather than as a Z grain, as it seems to haveCCSNe affinities.

The neutron-poor isotopes are dominated by GCE andtherefore represent valuable proxies to study the trends ofGCE in the solar neighborhood. We showed that GCEmodels by Kobayashi et al. (2011) agree well with the mea-sured trends for d54Fe and d60Ni, however, the measuredd29Si trend does not agree with the model. The observedGCE trends agree well with the average Solar System com-position showing that the Solar System is not special in ironand nickel. This means that GCE does not evolve linearlywith time, i.e., a linear age-metallicity relation cannot beinvoked to explain the isotopic composition. This findingagrees with astronomical observations (i.e., Casagrandeet al., 2011) as well as with modeling attempts (Lewiset al., 2013). The trends found between the GCE-dominated isotopes however offer an exciting insight intoGCE and will be used in future work for in-depth testingof heterogeneous GCE scenarios (Lugaro et al., 1999;Nittler, 2005), the galactic merger hypothesis (Clayton,2003), as well as other models that include chemodynamicalGCE (e.g., Lewis et al., 2013).

ACKNOWLEDGEMENTS

We dedicate this paper to the memory of Ernst Zinner, grandmaster of presolar grains and their implications for astrophysics,and a great friend to all of the authors. We would like to thankKevin McKeegan (editorial handling), Gary Huss, and two anony-mous reviewers for carefully reading our manuscript and providingus with constructive reviews that significantly improved the manu-script. This work was supported by the National Aeronautics andSpace Administration through a NASA Earth and Space SciencesFellowship (to RT, NNX12AL85H), and the NASA Cosmochem-istry program through grant NNX09AG39G. Construction anddevelopment of CHILI was supported by the NASA SampleReturn Laboratory Instruments and Data Analysis Program andthe Laboratory Analysis of Returned Samples Program throughgrants NNX07AL94G, NNX11AC21G, and NNX15AF78G, bythe University of Chicago, and by Argonne National Laboratory.Part of this work was performed under the auspices of the U.S.Department of Energy by Lawrence Livermore National Labora-tory under Contract DE-AC52-07NA27344. LLNL-JRNL-723517.

REFERENCES

Alexander C. M. O’D. and Nittler L. R. (1999) The galacticevolution of Si, Ti, and O isotopic ratios. Astrophys. J. 519,222–235.

Amari S., Hoppe P., Zinner E. and Lewis R. S. (1995) Trace-element concentrations in single circumstellar silicon carbidegrains from the Murchison meteorite. Meteoritics 30, 679–693.

Amari S., Gao X., Nittler L. R., Zinner E., Jose J., Hernanz M.and Lewis R. S. (2001a) Presolar grains from novae. Astrophys.J. 551, 1065–1072.

Amari S., Nittler L. R., Zinner E., Gallino R., Lugaro M. andLewis R. S. (2001b) Presolar SiC grains of type Y: origin from

http://refhub.elsevier.com/S0016-7037(17)30316-2/h0005












low-metallicity asymptotic giant branch stars. Astrophys. J. 546,248–266.

Amari S., Nittler L. R., Zinner E., Lodders K. and Lewis R. S.(2001c) Presolar SiC grains of type A and B: their isotopiccompositions and stellar origins. Astrophys. J. 559, 463–483.

Arlandini C., Kappeler F., Wisshak K., Gallino R., Lugaro M.,Busso M. and Straniero O. (1999) Neutron capture in low-massasymptotic giant branch stars: cross sections and abundancesignatures. Astrophys. J. 525, 886–900.

Barzyk J. G., Savina M. R., Davis A. M., Gallino R., Gyngard F.,Amari S., Zinner E., Pellin M. J., Lewis R. S. and Clayton R.N. (2007) Constraining the 13C neutron source in AGB starsthrough isotopic analysis of trace elements in presolar SiC.Meteorit. Planet. Sci. 42, 1103–1119.

Bisterzo S., Gallino R., Straniero O., Cristallo S. and Kappeler F.(2010) s-Process in low-metallicity stars – I. Theoreticalpredictions. Mon. Not. R. Astron. Soc. 404, 1529–1544.

Bisterzo S., Travaglio C., Gallino R., Wiescher M. and Kappeler F.(2014) Galactic chemical evolution and solar s-process abun-dances: dependence on the 13C-pocket structure. Astrophys. J.10 (14pp.).

Bisterzo S., Gallino R., Kappeler F., Wiescher M., Imbriani G.,Straniero O., Cristallo S., Gorres J. and deBoer R. J. (2015) Thebranchings of the main s-process: their sensitivity to a-inducedreactions on 13C and 22Ne and to the uncertainties of thenuclear network. Mon. Not. R. Astron. Soc. 449, 506–527.

Casagrande L., Schonrich R., Asplund M., Cassisi S., Ramırez I.,Melendez J., Bensby T. and Feltzing S. (2011) New constraintson the chemical evolution of the solar neighbourhood andgalactic disc(s). improved astrophysical parameters for theGeneva-Copenhagen survey. Astron. Astrophys. A138 (21pp).

Chieffi A., Limongi M. and Straniero O. (1998) The evolution of a25M� star from the main sequence up to the onset of the ironcore collapse. Astrophys. J. 502, 737–762.

Clayton D. D. (2003) A presolar galactic merger spawned the SiC-grain mainstream. Astrophys. J. 598, 313–324.

Clayton D. D. and Timmes F. X. (1997) Implications of presolargrains for galactic chemical evolution. AIP Conf. Proc. 402,237–264.

Cristallo S., Straniero O., Lederer M. T. and Aringer B. (2007)Molecular opacities for low-mass metal-poor AGB starsundergoing the third dredge-up. Astrophys. J. 667, 489–496.

Cristallo S., Straniero O., Gallino R., Piersanti L., Domınguez I.and Lederer M. T. (2009) Evolution, nucleosynthesis, and yieldsof low-mass asymptotic giant branch stars at different metal-licities. Astrophys. J. 696, 797–820.

Cristallo S., Piersanti L., Straniero O., Gallino R., Domınguez I.,Abia C., Di Rico G., Quintini M. and Bisterzo S. (2011)Evolution, nucleosynthesis, and yields of low-mass asymptoticgiant branch stars at different metallicities. II. The FRUITYdatabase. Astrophys. J. Suppl. 17, 21pp.

Cristallo S., Straniero O., Piersanti L. and Gobrecht D. (2015a)Evolution, nucleosynthesis, and yields of AGB stars at differentmetallicities. III. intermediate-m a ss models, revised low-m a ssmodels, and the ph-FRUITY interface. Astrophys. J. Suppl. S.40, 14pp.

Cristallo S., Abia C., Straniero O. and Piersanti L. (2015b) On theneed for the light elements primary process (LEPP). Astrophys.J. 801, 53 (14pp.).

Cristallo S., Karinkuzhi D., Goswami A., Piersanti L. andGobrecht D. (2016) Constraints of the physics of low-massAGB stars from CH and CEMP stars. Astrophys. J. 181

(12pp.).Davis A. M. (2011) Stardust in meteorites. Proc. Natl. Acad. Sci.

108, 19142–19146.

Dillmann, I., Plag, R., Kappeler, F., Rauscher, T., 2009. In:Proceeding of the Workshop ‘‘EFNUDAT Fast Neutrons -Scientific Workshop on Neutron Measurements, Theory &Applications, held on April 28–30 2009 at Geel, Belgium.

Fairbank, Jr., W. M., Spaar M. T., Parks J. E. and Hutchinson J.M. R. (1989) Anomalous odd- to even-mass isotope ratios inresonance ionization with broad-band lasers. Phys. Rev. A 40,2195–2198.

Gallino R., Arlandini C., Busso M., Lugaro M., Travaglio C.,Straniero O., Chieffi A. and Limongi M. (1998) Evolution andnucleosynthesis in low-mass asymptotic giant branch stars. II.Neutron capture and the s-process. Astrophys. J. 497, 388–403.

Gramlich J. W., Machlan L. A., Barnes I. L. and Paulsen P. J.(1989) Absolute isotopic abundance ratios and atomic weight ofa reference sample of nickel. J. Res. Natl. Inst. Stan. 94, 347–356.

Guandalini R. and Cristallo S. (2013) Luminosities of carbon-richasymptotic giant branch stars in the Milky Way. Astron.

Astrophys. A120, 7pp.Herwig F. (2013) Evolution of solar and intermediate-mass stars.

In Planets, Stars Stellar Syst. 4 Stellar Struct. Evol. (eds. T. D.Oswalt and M. A. Barstow). Springer Reference, Dordrecht,pp. 397–445.

Holmberg J., Nordstrom B. and Andersen J. (2007) The Geneva-Copenhagen survey of the Solar neighbourhood II - new uvby

calibrations and rediscussion of stellar ages, the G dwarfproblem, age-metallicity diagram, and heating mechanisms ofthe disk. Astron. Astrophys. 475, 519–537.

Hurst G. S., Payne M. G., Kramer S. D. and Young J. P. (1979)Resonance ionization spectroscopy and one-atom detection.Rev. Mod. Phys. 51, 767–819.

Huss G. R. and Smith J. B. (2007) Titanium isotopic compositionsof well-characterized silicon carbide grains from Orgueil (CI):implications for s-process nucleosynthesis. Meteorit. Planet.

Sci. 42, 1055–1075.Hynes, K.M., Gyngard, F., 2009. The presolar grain database:

http://presolar.wustl.edu/�pgd. Lunar Planet. Sci. Conf. 40,#1198.

Jaeger M., Kunz R., Mayer A., Hammer J. W., Staudt G., KratzK. L. and Pfeiffer B. (2001) 22Ne(a,n)25Mg: The key neutronsource in massive stars. Phys. Rev. Lett. 87, 202501 (4pp).

Kappeler F., Wiescher M., Giesen U., Gr¨res J., Baraffe I., El EidM., Raiteri C. M., Busso M., Gallino R., Limongi M. andChieffi A. (1994) Reaction rates for 18O(a; c)22Ne, 22Ne(a; c)26

Mg, and 22Ne(a, n)25Mg in stellar helium burning and s-processnucleosynthesis in massive stars. Astrophys. J. 437, 396–409.

Kobayashi C., Umeda H., Nomoto K., Tominaga N. and OhkuboT. (2006) Galactic chemical evolution: carbon through zinc.Astrophys. J. 653, 1145–1171.

Kobayashi C., Karakas A. I. and Umeda H. (2011) The evolutionof isotope ratios in the Milky Way Galaxy. Mon. Not. R.

Astron. Soc. 414, 3231–3250.Kobayashi C. and Nakasato N. (2011) Chemodynamical simula-

tions of the Milky Way galaxy. Astrophys. J. 16, 16pp.Kodolanyi, J., Stephan, T., Trappitsch, R., Hoppe, P., Pignatari,

M., Davis, A.M., Pellin, M.J., (2018). Iron and nickel isotopecompositions of presolar silicon carbide grains from super-novae. Geochim. Cosmochim. Acta 221, 127–144.

Levine J., Savina M. R., Stephan T., Dauphas N., Davis A. M.,Knight K. B. and Pellin M. J. (2009) Resonance ionization massspectrometry for precise measurements of isotope ratios. Int. J.Mass Spectrom. 288, 36–43.

Lewis K. M., Lugaro M., Gibson B. K. and Pilkington K. (2013)Decoding the message from meteoritic stardust silicon carbidegrains. Astrophys. J. Lett. L19(5pp).







































































































http://presolar.wustl.edu/<ucode type=

http://presolar.wustl.edu/<ucode type=




































Liu N., Savina M. R., Davis A. M., Gallino R., Straniero O.,Gyngard F., Pellin M. J., Willingham D. G., Dauphas N.,Pignatari M., Bisterzo S., Cristallo S. and Herwig F. (2014)Barium isotopic composition of mainstream silicon carbidesfrom murchison: constraints for s-process nucleosynthesis inasymptotic giant branch stars. Astrophys. J. 66(20pp).

Liu N., Savina M. R., Gallino R., Davis A. M., Bisterzo S.,Gyngard F., Kappeler F., Cristallo S., Dauphas N., Pellin M. J.and Dillmann I. (2015a) Correlated strontium and bariumisotopic compositions of acid-cleaned single mainstream siliconcarbides from murchison. Astrophys. J. 12(23pp).

Lodders K. (2003) Solar System abundances and condensationtemperatures of the elements. Astrophys. J. 591, 1220–1247.

Lodders, K., Fegley, B., Jr., 1999. Condensation chemistry ofcircumstellar grains. In: Le Bertre, T., Lebre, A., Waelkens, C.(Eds.), Asymptotic Giant Branch Stars, Volume 191 of IAU

Symposium, p. 279.Lodders K., Palme H. and Gail H.-P. (2009) Abundances of the

elements in the solar system. In Landolt Bornstein, vol. VI/4B(ed. J. E. Trumper). Springer-Verlag Berlin Heidelberg, pp.560–630 (Landolt-Bornstein, New Series, chapter 4.4).

Lugaro M., Zinner E., Gallino R. and Amari S. (1999) Si isotopicratios in mainstream presolar SiC grains revisited. Astrophys. J.527, 369–394.

Marhas K. K., Amari S., Gyngard F., Zinner E. and Gallino R.(2008) Iron and nickel isotopic ratios in presolar SiC grains.Astrophys. J. 689, 622–645.

Moore C. E. (1971). .Nittler L. R. (2005) Constraints on heterogeneous galactic chemical

evolution from meteoritic stardust. Astrophys. J. 618, 281–296.Nomoto K., Iwamoto K., Nakasato N., Thielemann F.-K.,

Brachwitz F., Tsujimoto T., Kubo Y. and Kishimoto N.(1997) Nucleosynthesis in type Ia supernovae. Nuc. Phys. A

621, 467c–476c.Nomoto K., Kobayashi C. and Tominaga N. (2013) Nucleosyn-

thesis in stars and the chemical enrichment of galaxies. Ann.Rev. Astron. Astroph. 51, 457–509.

Ong W. J. and Floss C. (2015) Iron isotopic measurements inpresolar silicate and oxide grains from the Acfer 094 ungroupedcarbonaceous chondrite. Meteorit. Planet. Sci. 50, 1392–1407.

Page R. H. and Gudeman C. S. (1990) Completing the iron period:double-resonance, fluorescence-dip Rydberg spectroscopy andionization potentials of titanium, vanadium, iron, cobalt, andnickel. J. Opt. Soc. Am. B 7, 1761–1771.

Piersanti L., Cristallo S. and Straniero O. (2013) The effects ofrotation on s-process nucleosynthesis in asymptotic giantbranch stars. Astrophys. J. 774, 98, 11pp.

Pignatari M., Hirschi R., Wiescher M., Gallino R., Bennett M.,Beard M., Fryer C., Herwig F., Rockefeller G. and Timmes F.X. (2013a) The 12C + 12C reaction and the impact on nucle-osynthesis in massive stars. Astrophys. J. 31, 23pp.

Pignatari M., Zinner E., Bertolli M. G., Trappitsch R., Hoppe P.,Rauscher T., Fryer C., Herwig F., Hirschi R., Timmes F. X.and Thielemann F.-K. (2013b) Silicon carbide grains of type Cprovide evidence for the production of the unstable isotope 32Siin supernovae. Astrophys. J. Lett. 771, L7, 5pp.

Pignatari M., Zinner E., Hoppe P., Jordan C. J., Gibson B. K.,Trappitsch R., Herwig F., Fryer C., Hirschi R. and Timmes F.X. (2015) Carbon-rich presolar grains from massive stars:Subsolar 12C/13C and 14N/15N ratios and the mystery of 15N.Astrophys. J. Lett. L43, 6pp.

Rauscher T., Heger A., Hoffman R. D. and Woosley S. E. (2002)Nucleosynthesis in massive stars with improved nuclear physics.Astrophys. J. 576, 323–348.

Savina M. R., Pellin M. J., Tripa E. C., Veryovkin I. V., CalawayW. F. and Davis A. M. (2003) Analyzing individual presolar

grains with CHARISMA. Geochim. Cosmochim. Acta 67, 3215–3225.

Steele R. J., Elliott T., Coath C. D. and Regelous M. (2011)Confirmation of mass-independent Ni isotopic variability iniron meteorites. Geochem. Cosmochim. Acta 75, 7906–7925.

Steele R. J., Coath C. D., Regelous M., Russell S. and Elliott T.(2012) Neutron-poor nickel isotope anomalies in meteorites.Astrophys. J. 59, 21pp.

Stephan T., Zehnpfenning J. and Benninghoven A. (1994) Correc-tion of dead time effects in time-of-flight mass spectrometry. J.Vac. Sci. Technol. A 12, 405–410.

Stephan T., Trappitsch R., Davis A. M., Pellin M. J., Rost D.,Savina M. R., Yokochi R. and Liu N. (2016) CHILI – theChicago Instrument for Laser Ionization – a new tool forisotope measurements in cosmochemistry. Int. J. Mass Spec-

trom. 407, 1–15.Straniero O., Gallino R., Busso M., Chieffi A., Raiteri C. M.,

Limongi M. and Salaris M. (1995) Radiative 13C burning inasymptotic giant branch stars and s-processing. Astrophys. J.Lett. 440, L85–L87.

Straniero O., Domınguez I., Cristallo S. and Gallino R. (2003)Low-mass AGB stellar models for 0:003 ( Z ( 0:02: basicformulae for nucleosynthesis calculations. Publ. Astron. Soc.

Aust. 20, 389–392.Straniero O., Gallino R. and Cristallo S. (2006) s process in low-

mass asymptotic giant branch stars. Nucl. Phys. A 777, 311–339.

Takahashi K. and Yokoi K. (1987) Beta-decay rates of highlyionized heavy atoms in stellar interiors. Atom Data Nucl. Data

36, 375–409.Tang H. and Dauphas N. (2014) 60Fe� 60Ni chronology of core

formation in Mars. Earth Planet. Sci. Lett. 390, 264–274.Taylor P., Maeck R. and De Bievre P. (1992) Determination of the

absolute isotopic composition and atomic weight of a referencesample of natural iron. Int. J. Mass Spectrom. 121, 111–125.

Timmes F. X. and Clayton D. D. (1996) Galactic evolution ofsilicon isotopes: application to presolar SiC grains frommeteorites. Astrophys. J. 472, 723–741.

Trappitsch R., Savina M. R., Willingham D. G., Liu N., Pellin M.J., Dauphas N. and Davis A. M. (2012) Iron isotopicabundances in presolar grains. Lunar Planet. Sci. Conf. 43,#2497.

Wooden, D.H., 1997. Observational evidence for mixing and dustcondensation in core-collapse supernovae. In: Bernatowicz, T.J., Zinner, E.K. (Eds.), Astrophys. Implic. Lab. Study Presolar

Mater. AIP, Woodbury, New York AIP Conference Proceedings,vol. 402, pp. 317–376.

Worden E. F., Comaskey B., Densberger J., Christensen J.,McAfee T. M., Paisner T. A. and Conway J. G. (1984) Theionization potential of neutral iron, Fe I, by multistep laserspectroscopy. J. Opt. Soc. Am. B 1, 314–316.

Wunderlich R. K., Wasserburg G. J., Hutcheon I. D. and Blake G.A. (1992) Systematics of the odd-even effect in the resonanceionization of Os and Ti. Inst. Phys. Conf. Ser. 128, 127–130.

Zinner E., Nittler L. R., Gallino R., Karakas A. I., Lugaro M.,Straniero O. and Lattanzio J. C. (2006) Silicon and carbonisotopic ratios in AGB stars: SiC grain data, models, and thegalactic evolution of the Si isotopes. Astrophys. J. 650, 350–373.

Zinner E. (2014) Presolar grains. In Meteorites and Cosmochemical

Processes, vol. 1, second ed (eds. A. M. Davis, H. D. Hollandand K. K. Turekian). Elsevier, Oxford, pp. 181–213 (Treatiseon Geochemistry).

Associate Editor: Anders Meibom





































































































































Documents

Simultaneous iron and nickel isotopic analyses of presolar silicon … · 2017-12-18 · Simultaneous iron and nickel isotopic analyses of presolar silicon carbide grains ... Geochimica