Untangling the Sources of Abundance Dispersion in Low-metallicity Stars II: Neutron Capture Elements

TL;DR

This work quantifies intrinsic abundance scatter in 23 elements, including nine neutron-capture species, for 84–86 low-metallicity subgiants using high-resolution, high-S/N spectra and 1D-LTE abundances from KORG. By modeling [X/Fe] as metallicity-dependent trends and decomposing scatter into photon-noise and intrinsic components, the authors show that heavy elements exhibit substantially larger intrinsic scatter than light elements, with Eu displaying the largest dispersion. They explore whether stochastic sampling of supernova yields (rotating CCSN, MRD SNe, and AGB contributions) can reproduce the observed abundances and scatter, finding that rotating CCSN help for some first-peak s-process elements but cannot alone explain the heaviest species, while MRD SNe can raise heavy-element production but tend to overproduce certain nuclei and fail to match all constraints. The work highlights the need for densely sampled, low-metallicity SN yield grids to robustly interpret early Galactic chemical evolution and the origin of neutron-capture elements.

Abstract

We present the abundances of 23 elements, including 11 heavy elements (Cu, Zn, Sr, Y, Zr, Ba, La, Ce, Nd, Sm, Eu) for up to 86 metal-poor (-2 < [Fe/H] < -1) subgiants. We use KORG, a state of the art spectral synthesis package, to derive 1D-LTE abundances from high-SNR and high-resolution spectra taken by the Large Binocular Telescope with the Potsdam Echelle Polarimetric and Spectroscopic Instrument. These precise spectra and abundance measurements minimize the impact of photon-noise (<0.06 dex), allowing us to robustly measure the intrinsic abundance scatter in [X/Fe]. After removing two stars with exceptional s-process enhancement, we find that the intrinsic scatter among the s- and r-process elements tends to be larger than for the lighter elements, with heavy element scatter ranging from 0.11 (Zn) to 0.27 (Eu) dex. Intrinsic abundance scatter could have multiple origins, including star-to-star variations in the ratios of nucleosynthetic sources as well as stochastic sampling of the progenitor supernovae properties, such as mass, rotation, and magnetic field strength. We explore the expected abundance scatter signature caused by stochastic sampling, finding that a fraction of both rapidly rotating CCSN and magnetorotationally driven SN are needed to reach the observed abundances and intrinsic scatter. This analysis is limited by the restrictive parameter spaces spanned by existing yield sets. A diverse, finely sampled grid of supernovae yields is needed to robustly model stochastic abundance scatter.

Figures (11)

• Figure 1: Heavy element line windows for three stars with [Fe/H] of $-2.12$ (green, 2MASS J04315411-0632100), $-1.58$ (blue, 2MASS J15581861+0203059), and $-1.04$ (purple, 2MASS J17140534+1407170). Spectra have been offset by 0.1 for clarity. We show one window for each heavy element, with line centers labeled, including blended lines for Zr and Eu.
• Figure 2: Comparison of [X/H] abundances derived in this work using Korg synthesis and [X/H] abundances derived in G23 using MOOG synthesis. Error bars represent photon-noise estimates on individual abundances, as discussed in Section \ref{['subsec:photon-noise']}. In the bottom right corner of each panel we include the average difference in abundances, where $\Delta [\rm X/H] = [X/H]_{\rm Korg} - [X/H]_{\rm MOOG}$. Note that the axes scales vary between subplots.
• Figure 3: [X/Fe] vs. [Fe/H] abundances for the heavy element abundances derived in this work (dark purple points). Error bars represent the photometric-noise (Section \ref{['subsec:photon-noise']}) Stars for which only upper limits can be measured are shown as light purple downward triangles. In each panel, we note the number of stars with a detected abundances of a given element out of the 86 total low-metallicity stars. For comparison, we also include a sample of high S/N stars from GALAH DR4 (grey 2D histogram; buder2025) and low-metallicity stars from the MINCE survey (teal xs; cescutti2022francois2024).
• Figure 4: Magnitude of the photon-noise scatter about the two-parameter model (dark purple, $\sigma_{\rm phot, \,2-param}$) and intrinsic scatter about the two-parameter model (blue, $\sigma_{\rm intrin,\,2-param}$) for our sample of [X/Fe] abundances. We show the intrinsic scatter including the two outlying Ba stars as the light blue bars, and the intrinsic scatter with their exclusion as the foreground dark blue bars.
• Figure 5: Pearson correlation coefficient ($r$) between the [X/Fe] deviations from the two-parameter model (Equation \ref{['eq:two-param']}) for pairs of elements. The size and shade of the circle represents the magnitude of the correlation, with positive values in green and negative values in purple. Correlations between pairs of the same elements are one by definition. Pairs where the correlation coefficient is strong ($|r|>0.4$) are highlighted with a white background, while pairs where the correlation is weaker ($|r|<0.4$) have a gray background.