Analysis of mass-transferring binary candidates in the Milky Way

TL;DR

This work targets Hertzsprung-gap donor systems in the Milky Way to identify active mass-transfer binaries potentially leading to luminous red novae. It combines Gaia XP-based Balmer-emission filtering, mid-IR excess, and variability with multi-epoch photometry and spectroscopy to assemble and characterize a 67-source candidate sample, uncovering several Be/Oe stars and at least a handful of mass-transfer candidates, including systems with compact companions. A key finding is that extinction handling drives the sample purity: applying SHBoost extinction corrections reduces the HG-like pool from 67 to 25, refining the search toward genuinely yellow HG objects. The study also provides a refined HG catalog (308 objects) using improved extinction, enabling more robust future identifications of unstable mass-transfer binaries and guiding follow-up observations for merger-progenitor studies. Overall, the approach advances identifying potential LRN precursors and constraining mass-transfer physics in evolved binaries, with implications for compact-binary formation and gravitational-wave progenitors.

Abstract

Mass transfer between stars in binary systems profoundly impacts their evolution, yet many aspects of this process (especially the stability, mass loss, and eventual fate of such systems) remain poorly understood. One promising avenue to constrain these processes is through the identification and characterisation of systems undergoing active mass transfer. Inspired by the slow brightening preceding stellar merger transients, we worked on a method to identify Galactic mass-transferring binaries in which the donor is a Hertzsprung gap (HG) star. We constructed an initial sample of HG stars using the Gaia EDR3 contribution Starhorse catalogue, and we identified candidate mass-transferring systems by selecting sources that exhibit Balmer emission features (using the low-resolution Gaia XP spectra), mid-infrared excess (from WISE photometry), and photometric variability (inferred from the error in the Gaia G-band magnitude). This multi-criteria selection yielded a sample of 67 candidates, which we further analysed using complementary photometric and spectroscopic data. Among our candidates, we identified at least nine eclipsing binaries and some sources that are potential binaries as well. Three sources in our sample are strong candidates for mass-transferring binaries with a yellow component, and three more are binaries with a Be star. Notably, four sources in our sample are strong candidates for hosting a compact companion, based on their ultraviolet or X-ray signatures. The main sources of contamination in our search are hot but highly reddened stars (primarily Oe and Be stars). As an additional outcome of this work, we present a refined catalogue of 308 bona fide HG stars, selected using improved extinction corrections and stricter emission-line criteria. This enhanced sample is expected to contain a significantly higher fraction of scientifically valuable mass-transferring binaries.

Figures (17)

• Figure 1: Initial selection of our sample of mass-transferring candidates, based on a) the Gaia EDR3 StarHorse colour-magnitude diagram, b) the Gaia XP spectra, c) the ALLWISE colour-colour diagram, and d) the Gaia$G$-band variability. In a, c, and d, we show the final sample of 67 candidates as orange stars. The red lines in a show the cuts used to select stars between the MS and the RGB based on MESA Isochrones & Stellar Tracks. In b, we show the Gaia XP spectrum of one of the sources in our sample as an example, and we compare it with our follow-up spectrum taken with CAFOS (see Section \ref{['od:spec']}), and the $\oplus$ symbols represent telluric absorption (absorption by the Earth’s atmosphere). The red dashed lines in panel c indicate the zero levels and divide the plot into four regions: warm dust, hot dust, both hot and warm dust, and no hot or warm dust. The shaded green region in d corresponds to the 0.9 percentile of the initial HG sample.
• Figure 2: Overview of the selection process with the number of sources at each stage of the selection.
• Figure 3: Sky map showing the position of the 67 sources in our initial sample with reference to the Galactic density of stars, constructed with a Gaia DR3 random data set containing 100 000 sources. Different ranges of absolute magnitude are shown with different markers.
• Figure 4: Optical spectra for the subsample of spectroscopically characterised sources. Each spectrum, except for the Mookodi spectrum, is de-reddened for visualisation purposes using the best extinction values (see Sect. \ref{['disc:extinc']}, Tab. \ref{['longtable1:data']}), with Fitzpatrick99 dust extinction function and $R_V = 3.1$. The main emission and absorption lines are indicated. The areas with strong telluric absorption are shown by the shaded rectangles. On the right, the number in parenthesis indicates the index associated with each source (see Table \ref{['table:spec']}), and this index is followed by the instrument used to take each spectrum: CAFOS (CAF) from the Calar Alto Observatory 2.2 m telescope, HERMES from the Mercator telescope, FIES and ALFOSC (ALF) from the Nordic Optical Telescope, and Mookodi from the South African Astronomical Observatory 1 m Lesedi telescope.
• Figure 5: Flux normalised H$\alpha$ and H$\beta$ velocity profiles for the subsample of observed sources. All sources are shown on the same scale. The index corresponding to each source (see Table \ref{['table:spec']}) is shown in the bottom left corner of each spectrum. Medium- and high-resolution spectra have been binned using 10 $km/s$ bins for improved visual representation.