Chemical Signatures of AGB Mass Transfer in Gaia White Dwarf Companions

TL;DR

This work investigates s-process enrichment in 160 Gaia WD+MS binaries with AU-scale separations to understand past MT from AGB donors. By combining high-resolution spectroscopy with Fe I/II excitation-ionization balance and synthetic spectral fitting, the authors identify 40 Ba dwarfs (36 new) and reveal correlations between Ba/Y abundances, donor/companion masses, and metallicity, with minimal dependence on current orbital separation. They demonstrate that variations in AGB donor mass, metallicity, and the MT epoch, along with dilution in the accretor’s convective envelope, can explain the observed s-process diversity; population modeling shows how MT history controls the Ba-dwarf fraction. The results establish Gaia WD+MS binaries as a powerful laboratory for constraining binary mass-transfer physics and the origins of chemically peculiar stars, including linking Ba dwarfs to CEMP-s stars at low metallicity. The study leverages FRUITY AGB yields and MESA models to connect observed abundances to the evolutionary state of donors and accretors, with implications for binary evolution and chemical tagging in the Galaxy.

Abstract

We present a homogeneous abundance analysis of 160 main-sequence stars in astrometric white-dwarf + main-sequence (WD+MS) binaries with orbits from Gaia DR3. These systems have AU-scale separations and are thought to have undergone mass transfer (MT) when the WD progenitor was an asymptotic giant branch (AGB) star. Using high-resolution spectroscopy, we measure chemical abundances of the MS stars, focusing on s-process elements. Since s-process nucleosynthesis occurs mainly in AGB stars, s-process enhancement in the MS star is a key signature of accretion from an AGB companion. We identify 40 barium dwarfs -- 36 of them newly discovered -- roughly doubling the known population in astrometric WD+MS binaries and extending it to lower metallicities than previously studied. The s-process abundances show large star-to-star variations that correlate with component masses and with metallicity but not with orbital separation. At the lowest metallicities, three barium dwarfs display strong CH and $\rm C_2$ absorption bands, linking them to CEMP-s stars and implying that AGB mass transfer usually leads to strong carbon enhancement at low metallicity. By comparing the observed abundance patterns to AGB nucleosynthesis models, we show that the diversity of s-process enhancements can be explained by variations in donor mass, metallicity, and most importantly, the number of thermal pulses the AGB star experienced before the onset of MT. Variation in the depth of the accretors' convective envelopes, with which accreted material is diluted, strengthens correlations with MS star mass and metallicity. Our results establish Gaia WD+MS binaries as a powerful laboratory for constraining binary mass-transfer physics and the origins of chemically peculiar stars.

Figures (17)

• Figure 1: Left: AMRF versus luminous star mass, $M_1$, for the full non-class I sample from Shahaf2024MNRAS (grey points). Our targets selected for follow-up are shown with colored points, separated by whether they were classified as NCE (pink diamonds) or RCE (blue circles) by Shahaf2024MNRAS. The dotted and dashed lines are boundaries separating systems in the three classes (Section \ref{['sec:selection']}). The shaded regions corresponds to the expected locations of systems hosting dark compact object secondaries with masses between $0.45-0.75\,M_{\odot}$ (yellow, lower) and $1.4-2.1\,M_{\odot}$ (orange, upper), corresponding to typical WD and NS masses. While sources classified as NCE are less likely to be triples, we believe that both the NCE and RCE sources in our follow-up sample are dominated by compact object companions (see text). Right: Color excess against [Fe/H] Zhang2023MNRAS for non-class I systems. The two quantities are anti-correlated, meaning that the NCE sample is biased against metal-poor primaries. We selected several RCE systems lying below the densely populated strip in this space, which are likely to host WDs.
• Figure 2: Comparison of best-fit stellar parameters obtained in this work (Section \ref{['ssec:sp_ew_balance']}) with those calculated using Gaia low-resolution spectra (Andrae2023ApJS; top row), as well as those from GALAH DR4 (Buder2025PASA; bottom row). Systems with secondary masses above and below the Chandrasekhar mass of $1.4\,M_{\odot}$ are plotted with red unfilled and blue filled markers respectively.
• Figure 3: Observed spectra of three targets (gray) compared with synthetic spectra generated using their best-fit stellar parameters (pink; Section \ref{['ssec:sp_ew_balance']}). We select targets covering a wide range of [Fe/H] with similar $T_{\rm eff}$ and log$(g)$, observed with different instruments. On the left, we plot the region around the magnesium triplet, and on the right, we move to redder wavelengths populated with iron lines. Visually, we see good agreement between the observations and the corresponding models, suggesting that our stellar parameters are reasonable. For the most metal-poor star (top row), the C$_2$ Swan band at $5165\,$Å is visible, which suggests carbon enhancement (Section \ref{['ssec:cemp_stars']}).
• Figure 4: Examples of strong lines of singly ionized barium and yttrium. The fitting region is shaded in yellow, and the best-fit model is plotted in red. The solid vertical line indicates the position of the line center provided in the linelist, while the dashed and dotted lines indicate the minima of the fitted Gaussians. The upper two panels are fits to HIRES spectra, while the lower left and right are fits to FEROS and MIKE spectra.
• Figure 5: Atmospheric parameters (A) and derived abundances of various elements (B-F). In the background of each panel, we plot the distribution of the corresponding parameters for sources in the GALAH DR4 catalog. Panel (A) shows that as expected, all targets are consistent with being on the MS, with more metal-poor stars being hotter. Panels (B) - (D) show that the expected trends in the $\alpha$-elements are seen in both GALAH and our sample. We target stars that likely accreted material from an AGB donor, explaining the greater fraction of stars that are rich in s-process elements, seen in panels (E) and (F).