Origin of meteoritic stardust unveiled by a revised proton-capture rate of $^{17}$O

TL;DR

This study links the isotopic composition of meteoritic stardust grains to intermediate-mass stars by revising the proton-capture rate of $^{17}$O, providing evidence of their contribution to the Solar System's dust.

Contribution

It demonstrates that an updated $^{17}$O proton-capture rate explains the isotopic ratios in stardust, identifying intermediate-mass stars as their likely origin.

Findings

Revised $^{17}$O proton-capture rate matches observed isotopic ratios.

Intermediate-mass stars are confirmed as the source of certain meteoritic grains.

First direct evidence of these stars contributing to Solar System dust.

Abstract

Stardust grains recovered from meteorites provide high-precision snapshots of the isotopic composition of the stellar environment in which they formed. Attributing their origin to specific types of stars, however, often proves difficult. Intermediate-mass stars of 4-8 solar masses are expected to contribute a large fraction of meteoritic stardust. However, no grains have been found with characteristic isotopic compositions expected from such stars. This is a long-standing puzzle, which points to serious gaps in our understanding of the lifecycle of stars and dust in our Galaxy. Here we show that the increased proton-capture rate of $^{17}$O reported by a recent underground experiment leads to $^{17}$O/$^{16}$O isotopic ratios that match those observed in a population of stardust grains, for proton-burning temperatures of 60-80 million K. These temperatures are indeed achieved at the base of the convective envelope during the late evolution of intermediate-mass stars of 4-8 solar masses, which reveals them as the most likely site of origin of the grains. This result provides the first direct evidence that these stars contributed to the dust inventory from which the Solar System formed.

Tables

Table 1. Table 1: Summary of the physical properties of a selection of AGB models with initial masses (M, Column 1) from 4.5 to 8 M ⊙ subscript M direct-product {\rm M}_{\odot}  and metallicities (Z, Column 2) half-solar (0.007), solar (0.014), and twice-solar (0.03). Column 3: the total number of thermal pulses (TPs) of the He-burning shell; Column 4: the maximum temperature at the base of the convective envelope; Column 5: the maximum temperature in the inter-shell; Column 6: the mass lost during the whole evolution.

M ($M_{\odot }$)	Z	No. of TP	$T_{\mathrm{bce}}^{\max }$ (MK)	$T_{\mathrm{inter}-\mathrm{shell}}^{\max }$ (MK)	$M_{\mathrm{lost}}^{\mathrm{total}}$ ($M_{\odot }$)
5	0.007	59	82.9	357	4.11
4.5	0.014	31	63.5	356	3.64
5	0.014	41	75.4	354	4.12
6	0.014	53	85.5	365	5.08
8	0.014	67	100	376	6.94
5	0.03	26	54.2	345	4.13
5.5	0.03	31	64.7	348	4.62
6	0.03	33	71.2	352	5.09
8	0.03	63	94.0	373	6.95

Full text

∗Corresponding author {affiliations}

Konkoly Observatory, Research Centre for Astronomy and Earth Sciences, Hungarian Academy of Sciences, H-1121 Budapest, Hungary

Monash Centre for Astrophysics (MoCA), Monash University, Clayton VIC 3800, Australia

Research School of Astronomy and Astrophysics, Australian National University, Canberra, ACT 2611, Australia

Kavli Institute for the Physics and Mathematics of the Universe (WPI), University of Tokyo, Kashiwa, Chiba 277-8583, Japan

SUPA, School of Physics and Astronomy, University of Edinburgh, EH9 3FD Edinburgh, United Kingdom

Department of Terrestrial Magnetism, Carnegie Institution for Science, Washington, DC 20015, USA

Helmholtz-Zentrum Dresden-Rossendorf, Bautzner Landstr. 400, 01328 Dresden, Germany

Università di Napoli Federico II and INFN, Sezione di Napoli, 80126 Napoli, Italy

Gran Sasso Science Institute, INFN, Viale F. Crispi 7, 67100 L’Aquila, Italy

INFN of Padova, Via Marzolo 8, I-35131 Padova, Italy

Università degli Studi di Padova and INFN, Sezione di Padova, Via F. Marzolo 8, 35131 Padova, Italy

Università degli Studi di Genova and INFN, Sezione di Genova, Via Dodecaneso 33, 16146 Genova, Italy

Institute for Nuclear Research (MTA ATOMKI), PO Box 51, HU-4001 Debrecen, Hungary

INFN, Laboratori Nazionali del Gran Sasso (LNGS), 67100 Assergi, Italy

Università degli Studi di Torino and INFN, Sezione di Torino, Via P. Giuria 1, 10125 Torino, Italy

Università degli Studi di Milano and INFN, Sezione di Milano, Via G. Celoria 16, 20133 Milano, Italy

INFN, Sezione di Roma La Sapienza, Piazzale A. Moro 2, 00185 Roma, Italy

Università degli Studi di Bari and INFN, Sezione di Bari, 70125 Bari, Italy

Osservatorio Astronomico di Collurania, Teramo, and INFN, Sezione di Napoli, 80126 Napoli, Italy

Institut für Experimentalphysik III, Ruhr-Universität Bochum, 44780 Bochum, Germany

Present address: South Dakota School of Mines, 501 E. Saint Joseph St., SD 57701 USA

Stardust grains found in meteorites (and also interplanetary dust particles and samples returned from comet Wild 2) represent the very small fraction of presolar dust that survived destruction in the protosolar nebula. They condensed in the atmospheres of evolved stars and in nova and supernova ejecta and were preserved inside meteorites[1]. Their isotopic compositions are measured with high precision (few percent uncertainties) via mass spectrometry and represent a direct record of their site of formation, providing us with deep insights into stellar physics and the origin of elements and of dust in the Galaxy. Identified stardust includes both carbon-rich (diamonds, graphite, silicon carbide) and oxygen-rich (e.g., Al-rich oxides, silicate) grains, with the former condensing from gas with C$>$O, and the latter from gas with C$<$O. Here we focus on oxide and silicate grains, which are classified into different groups mostly based on their oxygen isotopic compositions[10]. Group I grains make up the majority ($\sim$75%) of oxide and silicate grains and show excesses in 17O characteristic of the first dredge-up in red giant stars of initial mass roughly 1–3 ${\rm M}_{\odot}$, with a maximum 17O/16O$\sim$0.003. Their origin is generally well understood and attributed to the O-rich phases of the subsequent asymptotic giant branch (AGB), when large amounts of dust condense in the cool, expanding stellar envelopes[2]. Group II grains represent roughly 10% of all presolar oxide grains, although this is a lower limit since measured compositions may suffer from isotopic dilution during ion probe analysis. Like Group I grains they display excesses in 17O (with 17O/16O up to 0.0015), but are also highly depleted in 18O, with 18O/16O ratios down by two orders of magnitude relative to the solar value. The initial ratio of the radioactive 26Al (half life, T${}_{1/2}=0.7$ Myr) to 27Al is inferred from 26Mg excesses and in Group II grains reaches 0.1, almost an order of magnitude higher than in Group I grains, on average. While this composition is the indisputable signature of H burning activating proton captures on the oxygen isotopes and on 25Mg (the 25Mg(p,$\gamma$)26Al reaction), hypotheses on the site of formation of Group II grains are still tentative.

Hydrogen burning affects the surface composition of massive ($>$ 4 ${\rm M}_{\odot}$) AGB stars when the base of the convective envelope becomes hot enough for proton-capture nucleosynthesis to occur[7] (“hot bottom burning”, HBB, Figure 1). These are the brightest AGB stars, and the fact that they mostly show C/O$<$1 is attributed to the operation of the CN cycle, which depletes carbon[11], in contrast to their less bright counterparts, which mostly show C/O$>$1 as a result of the dredge-up of He-burning material rich in carbon. Characteristic temperatures of HBB exceed $\sim$60 MK and, thanks to the fast convective turn-over time ($\approx 1$ yr), the composition of the whole envelope is quickly transmuted into the H-burning equilibrium abundances produced at the base of the envelope. Massive AGB stars are observed to generate significant amounts of dust and based on current models of Galactic dust evolution are expected to have contributed almost half of the O-rich dust of AGB origin in the Solar System[2, 3]. However, no stardust grains have been found to show the signature of HBB because, although Group II grains show the highly depleted 18O/16O ratios qualitatively expected from HBB, their 17O/16O ratios are roughly two-times lower than predicted[4, 5] using the available reaction rates[12].

Currently, the preferred suggestion for the origin of Group II grains is that they formed in AGB stars of low mass ($<$1.5 ${\rm M}_{\odot}$) that did not dredge-up enough carbon to become C-rich but experienced extra mixing below the bottom of the convective envelope (“cool bottom processing”, CBP[13, 14], Figure 1). In this scenario, material from the bottom of the convective envelope penetrates the thin radiative region located between the base of the convective envelope and the top of the H-burning shell where the temperature and density increase steeply with mass depth and proton captures can occur (Figure 1). While mechanisms have been proposed to explain the physical process driving this extra mixing[15] the current modelling of the CBP is parametric: both the rate of the extra mixing and the depth reached are treated as a free parameters, with the depth adjusted in order to reach temperatures in the range 40–55 MK.

Whichever scenario we consider, the equilibrium 17O/16O ratio produced by H burning is determined by the competition between the processes that produce and destroy 17O. Specifically, it depends on the ratio between the rate of the 16O(p,$\gamma$)17F reaction, which produces 17O following the beta decay of 17F (T${}_{1/2}=64$ s), and the rate of the 17O(p,$\alpha$)14N reaction, which destroys 17O. (Note that the 17O(p,$\gamma$)18F is comparatively negligible at all temperatures considered here). The former is known to within 7%[5, 12]; the latter has recently been determined[6] from a direct measurement of the strength of the 64.5 keV resonance that dominates the 17O(p,$\alpha$)14N reaction rate at temperatures between 10 and 100 MK[6], i.e., over the entire range of interest here. The experiment took place at the Laboratory for Underground Nuclear Astrophysics (LUNA) at Gran Sasso, Italy, where improved experimental procedures and a 15-times lower background for $\alpha$ particles detection than in surface laboratories allowed for the most sensitive measurement to date[6]. The new rate is 2–2.5 times higher than previous evaluations[12, 16]. At the temperatures of 40–55 MK typical of CBP the new rate reproduces only the lowest 17O/16O values observed in Group II grains (Figure 2). At the temperatures of 60–80 MK typical of HBB, instead, the new rate reproduces most of the observed 17O/16O range, revealing for the first time the signature expected from HBB in stardust grains. HBB temperatures higher than $\sim$ 80 MK are excluded for the parent stars of the grains.

Although the initial stellar mass and metallicity ranges at which HBB occurs as well as the AGB lifetime are model dependent[7, 8, 9], our result is robust because any massive AGB model experiencing HBB with temperatures between 60 and 80 MK will necessarily produce 17O/16O ratios in agreement with those observed in most Group II grains. Figure 3 shows the surface evolution of the oxygen isotopic ratios for three AGB models (of initial mass 4.5, 5, and 6 ${\rm M}_{\odot}$) of solar metallicity that experience HBB (see Methods section), compared to observed isotopic ratios in Group II stardust grains. The models evolve through the first and second dredge-ups at the end of core H- and He-burning, respectively, which increase the 17O/16O ratio by roughly a factor of five. During the subsequent AGB phase, HBB quickly (e.g., after about 1/5 of its total TP-AGB lifetime for a 6${\rm M}_{\odot}$ star) shifts the oxygen isotopic composition to the equilibrium values corresponding to the burning temperature. Using the LUNA rate, the 17O/16O ratio produced by HBB is roughly a factor of 2 lower than that obtained with the previous rate by Iliadis et al.[12] and nicely reproduces those observed in Group II grains, when AGB material is diluted with material of solar composition. The dilution is required because HBB strongly depletes 18O. This is in accordance with the non-detection of 18O in bright O-rich AGB stars[17], but results in 18O/16O ratios more than two orders of magnitude lower than observed in Group II grains. Dilution with solar material is particularly effective in increasing the 18O/16O ratio: for example, 99% of HBB material mixed with only 1% of solar material increases the 18O/16O by two orders of magnitudes. On the other hand, dilution has a comparatively minor effect on the other isotopes measured in the grains because 17O, 25Mg, and 26Al are produced rather than destroyed in massive AGB stars. For example, it takes dilution with 50% of solar system material to decrease the 17O/16O and 25Mg/24Mg ratios by a factor of 2.

Dilution can be wrought by percent-level traces of contaminant oxygen (e.g., from terrestrial or non-presolar material) during isotopic measurements, which can result in 18O/16O up to $\sim$10*-4*; however, laboratory contamination cannot easily explain grains with higher 18O/16O values. For these, a dilution of the HBB signature composition with solar system material at the level of a few tens of percent is required. Even higher dilution would result in a fraction of Group I grains also originating from massive AGB stars. Possible processes may involve dilution with previously ejected gas within the dust formation region; dilution with material in the interstellar medium; and/or a significantly lower value of the 18O(p,$\alpha$)15N reaction rate. A study of this reaction has recently been completed at LUNA and data analysis is in progress.

The other isotopic pairs measured in Group II grains are also consistent with an origin in massive AGB stars. The 25Mg/24Mg ratios are enhanced in massive AGB stars by the third dredge-up of material from the He inter-shell, where the 22Ne($\alpha$,n)25Mg reaction is activated, and such a signature is seen in some presolar spinel (MgAl2O4) grains (Figure 4a). Specifically, the two times solar 25Mg/24Mg value observed in a spinel grain named 14-12-7[18] is close to that obtained in the final composition of the 5 ${\rm M}_{\odot}$, model. However, grain OC2[4] and the majority of the other grains show a spread in the 25Mg/24Mg ratio from 1 to 1.5 times solar, i.e., lower than predicted by the dilution computed using the final AGB composition. This may reflect partial equilibration of Mg isotopes in the grains themselves[19]. Alternatively, the lower 25Mg/24Mg ratios may be explained by truncating the AGB evolution to one-half or one-third of the total computed evolution (as illustrated in Figures 3 and 4). This could result from a higher mass-loss rate and/or the effect of binary interactions. Another solution allowed within current model uncertainties is a less efficient third dredge-up than calculated in our models. Finally, the high 26Al/27Al ratios up to $\sim$0.1 typical of Group II grains are also consistent with HBB (Figure 4b), although an accurate analysis is currently hampered by the uncertainties in the 25Mg and 26Al proton-capture rates[12, 20].

Our evidence that some meteoritic stardust grains exist whose O, Mg, and Al isotopic composition is best accounted for by H-burning conditions characteristic of massive AGB stars proves that these stars were dust contributors to the early Solar System. It further provides us with a new tool to deepen our understanding of uncertain physical processes in massive AGB stars, for which observational constraints are still scarce.

Stellar models

Stellar structure models with metallicities (Z) from half to double solar (where solar is 0.014${}^{\mbox{\scriptsize 22}}$ were selected from the large set presented by Karakas${}^{\mbox{\scriptsize 23}}$ computed with the Monash-Stromlo code${}^{\mbox{\scriptsize 24}}$. No mass-loss was assumed on the red giant branch and the Vassiliadis & Wood${}^{\mbox{\scriptsize 25}}$ mass-loss formulation was used on the AGB. The C-rich and N-rich low-temperature opacity tables were taken from Marigo & Aringer${}^{\mbox{\scriptsize 26}}$. Convection was approximated using the mixing length theory with a mixing-length parameter of 1.86 in all calculations. No convective overshoot was applied, although the algorithm described by Lattanzio${}^{\mbox{\scriptsize 27}}$ was used to search for a neutrally stable point for the border between convective and radiative zones.

From the models presented by Karakas${}^{\mbox{\scriptsize 23}}$, in the Supplementary Table 1 we present a selection with initial masses from 4.5 to 8 ${\rm M}_{\odot}$ with canonical values of the He content. More details on the physical quantities of the models can be found in Table 1 of Karakas${}^{\mbox{\scriptsize 23}}$. In Supplementary Table 1 we only report a summary of those that are most relevant here: the total number of thermal instabilities of the He-burning shell (thermal pulses, TPs); the maximum temperature at the base of the convective envelope (${\rm T}^{\rm max}_{\rm bce}$); the maximum temperature achieved in the inter-shell (${\rm T}^{\rm max}_{\rm inter-shell}$); and the mass lost during the whole evolution (${\rm M}_{\rm lost}^{\rm total}$). All the models experienced ${\rm T}^{\rm max}_{\rm bce}$ high enough to activate HBB, except for the 5 ${\rm M}_{\odot}$  model of Z$=0.03$. It should be noted that the mass and metallicity ranges at which HBB occurs are model dependent: for the same mass and metallicity, models using more or less efficient convection, for example, via a different mixing length parameter or different mixing schemes, result in different temperatures[7, 8]. All our stellar models also experienced efficient third dredge-up, i.e., C-rich material being carried from the He-rich inter-shell to the convective envelope. This is also model dependent. Overshooting at the base of the convective region associated with the TPs is not included in our models. In combination with the third dredge-up it would enrich the envelope in 16O. However, this would not change the oxygen isotopic ratios at the stellar surface since HBB efficiently brings them to their equilibrium values, similarly to the case of the carbon isotopic ratios. In the lower-mass stars where HBB is not activated, efficient third dredge-up of 16O would be accompanied by efficient third dredge-up of 12C, producing a C-rich envelope where the oxide and silicate grains considered here do not form.

Because of both the large dilution and the effect of HBB, most of the models lead to O-rich surfaces – the condition for the formation of oxide and silicate grains of interest here – during their whole evolution, except for the 4.5 ${\rm M}_{\odot}$  model of Z=0.014 and the 5 ${\rm M}_{\odot}$  model of Z=0.007. These latter become C-rich after the second last TP and the last TP, respectively, which results in 40–50% of the material ejected to be C-rich. For all the models, a relatively large fraction of the envelope material (20–30%) is still present when our calculations stopped converging. The abundances we calculated for the last model are either lower limits or a good approximation to the final enrichment, depending on possible further occurrence of third dredge-ups episodes beyond the point where our models stop converging.

We fed the computed stellar structure into the Monash post-processing code to calculate the detailed nucleosynthesis by solving simultaneously the abundance changes wrought by nuclear reactions and by convection using a “donor cell” advective scheme with two-streams (up and down) mixing. The simultaneous treatment of mixing and burning is required to model HBB in detail because nuclear reactions occur that may have timescales similar or shorter than the mixing timescales, also as function of the location in the envelope. In these cases it is not possible to make the assumption of istantaneous mixing at an average burning rate. Essentially our method couples mixing and burning together in the post processing to obtain the nucleosynthesis, while the energetic feedback of HBB is taken from the structure calculations performed using instantaneous mixing. The nucleosynthesis of elements up to Pb and Bi from the complete set of the models of Karakas${}^{\mbox{\scriptsize 23}}$ with He canonical abundance can be found in Karakas & Lugaro[9], together with a full discussion of the results. Briefly, in models that experience HBB, ${\rm T}^{\rm max}_{\rm bce}$ is the main feature controlling the composition of the stellar surface, and specifically the oxygen and aluminium ratios that are measured in oxide and silicate stardust grains. In massive AGB stars of metallicity around solar the Mg composition is affected mainly by the activation of the 22Ne($\alpha$,n)25Mg and 22Ne($\alpha$,$\gamma$)26Mg reactions in the He-rich inter-shell – where ${\rm T}^{\rm max}_{\rm inter-shell}$ is well above the activation temperature (T $\approx 300$ MK) of these reactions in all the models – and the subsequent third dredge-up of this material to the stellar surface. With respect to Karakas & Lugaro[9], we updated the 22Ne+$\alpha$ reaction rates from Iliadis et al.[12] to Longland et al.${}^{\mbox{\scriptsize 28}}$ and the 25Mg+$\gamma$ reaction rates from Iliadis et al.[12] to Straniero et al.[20]. Also, in the present study, we limited our calculations to a small network of 77 nuclear species, from neutrons to sulfur, plus the elements around the Fe peak, as described, e.g., in Karakas${}^{\mbox{\scriptsize 24}}$. This choice allowed us to run each model in a few hours and test different values of the 17O(p,$\alpha$)14N reaction rate: the recommended, the upper limit, and the lower limit from both Iliadis et al.[12] and LUNA[6]. For the 16O(p,$\gamma$)17F rate we used the value recommended by Iliadis et al.[12], which has an uncertainty of 7%[5]. For the initial composition, we used Asplund et al.${}^{\mbox{\scriptsize 22}}$ for the solar metallicity models, scaled down or up by factor of two for the Z=0.007 and Z=0.03 models, respectively. While we calculated detailed model predictions for all the models listed in the Supplementary Table 1, in the figures and discussion we focus on the 4.5, 5, and 6 ${\rm M}_{\odot}$  models with Z=0.014 only, for sake of clarity. Models with different metallicities in the same mass range have similar ${\rm T}^{\rm max}_{\rm bce}$ and provide similar results, except for the 5 ${\rm M}_{\odot}$model of Z=0.03, which does not experience HBB but remains O-rich due to the low efficiency of the third dredge-up combined with the high initial O abundance. On the other hand, the 8 ${\rm M}_{\odot}$  models have too high ${\rm T}^{\rm max}_{\rm bce}$ to provide a match with the grain data (see Figure 2). Stellar population synthesis models are needed to assess whether a number of Group I grains may also have originated from super-solar metallicity O-rich massive AGB stars that do not experience HBB.