Extensive Observational Evidence for Massive Star Stellar Wind Variability at Low Metallicities: implications for mass-loss rate determination

TL;DR

This study extends the understanding of massive-star wind variability to low-metallicity environments by analyzing UV spectra from 20 LMC/SMC OB stars using SEI-derived radial optical depths $ au_{ m{rad}}(w)$. It confirms the ubiquity of large-scale wind structure and narrow absorption components (NACs) at subsolar metallicity and quantifies the inherent uncertainty in single-epoch mass-loss estimates via $rac{\sigma( au)}{ au}$, finding results comparable to Galactic stars (e.g., $1\sigma$ lower bounds ranging roughly 13–40% and up to tens of percent for cooler stars). The presence and evolution of NACs/DACs significantly affect wind-profile interpretation and thus mass-loss inferences, underscoring the need for high-cadence multi-epoch UV spectroscopy at low metallicities. The paper also provides new insights into binarity in AzV 75, illustrating how companion stars can alter UV wind diagnostics and mass-loss determinations. Overall, the work highlights substantial wind-structure-driven uncertainties in mass-loss rate derivations at low metallicity and calls for time-series observations to calibrate wind-feedback prescriptions in galaxy evolution models.

Abstract

Mass-loss from massive stars is fundamental to stellar and galactic evolution and enrichment of the interstellar medium. Reliable determination of mass-loss rate is dependent upon unravelling details of massive star outflows, including optical depth structure of the stellar wind. That parameter introduces significant uncertainty due to the nearly ubiquitous presence of large-scale optically thick wind structure. We utilize suitable available ultraviolet spectra of 20 Large and Small Magellanic Cloud (LMC, SMC) OB stars to extend existing Galactic results quantifying uncertainty inherent in individual observations to lower metallicity environments. This is achieved by measuring standard deviations of mean optical depths of multiple observations of suitable wind-formed absorption profiles as a proportion of their mean optical depths. We confirm earlier findings that wind structure is prevalent at low metallicities and demonstrate that quantifying the consequent uncertainty is to some extent possible, despite the near-complete absence of time series UV spectroscopic observations in those environments. We find that the uncertainty inherent in any single observation of stellar wind optical depth at low metallicity is of similar magnitude to that already identified at Galactic metallicity (up to 45% for cooler OB stars). We further demonstrate how the effect of varying narrow absorption components in wind-formed UV spectral profiles is unlikely to be properly accounted for in existing mass-loss models. We present further evidence of a binary companion to the SMC O-type giant star AzV 75. The importance of obtaining high cadence multi-epoch, or genuine time series, UV spectroscopic observations at low metallicities is highlighted.

Figures (67)

• Figure 1: Comparison of two observations of the SMC star AzV 70, made at an interval of 47.954 days, the first (dark violet) spectrum obtained on 2020 June 28 and the second (light green) spectrum obtained on 2020 August 15. The significant changes in the absorption profiles of each of the NV, SiIV and CIV resonance line profiles can be seen, in particular, the enhanced absorption toward higher stellar wind velocities in the second observation may be noted. Prominent narrow absorption component (NAC) features are also seen for each species and in each observation and these are highlighted, although the NAC for the red element of the SiIV profile is obscured by the emission profile of the blue element of that doublet. Note that the separations of the NACs in each case equate to the separations of the doublet elements, confirming their nature.
• Figure 2: Comparison of the optical depth inherent ncertainty results from this work against those for Galactic stars presented by Massa2024. The plot shows derived lower bounds on $\sigma(\tau)/\tau$ against stellar effective temperature for the LMC and SMC targets considered in this work. Uncertainties on the results for Galactic stars are also from Massa2024. Targets with observations at a cadence longer than the stellar rotation period are distinguished from those where the cadence is shorter than the stellar rotation period. Targets with more than two observations used are highlighted. The result for AzV 75, although included in Table \ref{['tab:Alldata']}, is not plotted here due to the significant effect that the binary nature of AzV 75, discussed below, has on the inherent uncertainty derived in this work.
• Figure 3: Comparison of 10 HASP COS UV spectra for the LMC star Sk -67 166, spectral type O4 If, concentrating on the SiIV resonance line doublet. Plotted in the rest frame of the star (rest wavelength of the blue element of the doublet is at zero velocity). Considerable variation in the overall absorption profile shapes and the simultaneous increases and decreases in optical depth in different regions of the stellar wind can be observed. The average time between each observations is approximately 21 hours.
• Figure 4: Comparison of one STIS and 24 COS UV spectra for the SMC star AzV 75, spectral type O3.5 III(f) highlighting the CIV resonance line doublet. Plotted in the rest frame of the star (rest wavelength of the blue element of the doublet is at zero velocity). The considerable variation in the blue edge of the absorption feature, in the shape of the absorption and, most unusually, in the emission features can be clearly observed. Note that not all observations extend to this portion of the spectrum, so the applicable colours are skipped in the sequence to maintain consistency and comparability with other plots.
• Figure 5: Time-ordered comparisons of multiple UV spectra of (a) the excited OIV 1338.61 Å and (b) the excited NIII 1183.03, 1184.55 Å photospheric lines observed in AzV 75. The colour sequence of the observations is the same in each plot and matches that in Figure \ref{['fig:75_compare_select']}, however some observations do not extend to the wavelength of the NIII feature and/or the OIV feature, so the applicable colours in the sequence are skipped to maintain consistency and comparability.