X-Shooting ULLYSES: Massive stars at low metallicity XIV. Properties of SMC late-O and B supergiants reveal the metallicity dependence of winds in the Magellanic Clouds

Abstract

Considering the physics of radiation-driven winds of massive stars, the wind properties should depend on the metal content of the stellar atmosphere. Therefore, studying the winds of massive stars in different metallicities provides a sanity check on prescriptions that are widely used in evolutionary calculations. We obtained the stellar and wind properties of a sample of 20 late-O and B supergiants in the Small Magellanic Cloud (SMC) from a quantitative combined UV and optical spectroscopic analysis using CMFGEN. By comparing these properties with those of a Large Magellanic Cloud counterpart study, which has a similar sample and data, and employed the same modelling techniques used in this study, We derived a metallicity-dependent recipe for wind momentum, which is applicable for $5.4 \leq \log{L_{\rm bol}/L_{\odot}} \leq6.1$ and $14 \leq T_{\rm eff}/{\rm kK} \leq 32$. We find a significant dependence of the wind momentum on the metallicity, which is largely due to the mass-loss rates. We do not find any evidence of a discontinuity in either the mass-loss rate or the ratio of the terminal wind velocity to the escape velocity between $25$ and $21$~kK, which could be attributed to the bi-stability jump. Stellar parameters are consistent across different methods and radiative transfer codes, whereas mass-loss rates differ significantly, with our values being generally lower. We find a discrepancy between the evolutionary and spectroscopic masses in $40\%$ of our sample, with the evolutionary mass usually being systematically higher. The mass-loss rates of blue supergiants are far too low to strip the stellar envelope and the subsequent formation of classical Wolf-Rayet (WR) stars, leading to the conclusion that luminous blue variable eruptions or binary interactions are necessary to explain the characteristics of the WR population in the SMC.

Figures (16)

• Figure 1: Hertzsprung-Russell diagram for our sample. Overlaid in solid black lines are non-rotating SMC isochrones for different ages ($\approx0{\rm -}10$ Myr). Dashed black lines are the SMC rotating evolutionary tracks for stellar masses in the range of $\approx10{\rm -}60~M_{\odot}$ with a rotational velocity of $110~{\rm km\,s^{-1}}$. Both the isochrones and evolutionary tracks are adopted from brott2011. The shaded area is defined by the HD limit humphreysdavidson1979smith2004davies2018.
• Figure 2: Relative (percentage) difference of the evolutionary and spectroscopic masses $(M_{\rm evo}-M_{\rm spec})/M_{\rm spec}$. $M_{\rm evo}$ was obtained using a Bayesian inference method applied to SMC evolutionary models of brott2011. The diagonal black crosses indicate the presence of a mass discrepancy.
• Figure 3: Best fits (solid red lines) to XShootU H$\alpha$ (+ $\textup{C\,ii}~\lambda6578$) profiles (solid black lines). Magellan/MIKE spectra are also shown (dashed blue lines), where available, to illustrate line variability.
• Figure 4: $\Delta\log{\dot{M}}$ vs $T_{\rm eff}$. $\Delta\log{\dot{M}}$ is the difference between our derived $\Delta\log{\dot{M}}$ and those obtained from numerical recipes from vinksander2021 (yellow dots), bjorklund2023 (blue dots), and krticka2024 (red dots).
• Figure 5: $\varv_{\infty}$ vs $T_{\rm eff}$. The green dots represent our results. The green line is the linear fit. The dashed magenta line is the Z-dependent $\varv_{\infty}$-$T_{\rm eff}$ relation from hawcroft2024. The solid magenta line was obtained from fitting hawcroft2024 results for O and B supergiants only. The solid blue line was obtained from fitting the results of bernini2024. The grey crosses are $\varv_{\infty}$ calculated from vinksander2021 recipe. The black triangles represent $\varv_{\infty}$, calculated from the recipe in krticka2021.