Mass loss through a stellar wind is an important physical process that steers the evolution of massive stars and controls the properties of their end-of-life products, such as the supernova type and the mass of compact remnants. To probe its role in stellar evolution over cosmic time, mass loss needs to be studied as function of metallicity. For mass loss to be accurately quantified, the wind structure needs to be established jointly with the characteristics of small-scale inhomogeneities in the outflow, which are known as wind clumping. We aim to improve empirical estimates of mass loss and wind clumping for hot main-sequence massive stars, study the dependence of both properties on the metal content, and compare the theoretical predictions of mass loss as a function of metallicity to our findings. Using the model atmosphere code Fastwind and the genetic algorithm fitting method Kiwi-GA we analyzed the optical and ultraviolet spectra of 13 O-type giant to supergiant stars in the Small Magellanic Cloud galaxy, which has a metallicity of approximately one-fifth of that of the Sun. We quantified the stellar global outflow properties, such as the mass-loss rate and terminal wind velocity, and the wind clumping properties. To probe the role of metallicity, our findings were compared to studies of Galactic and Large Magellanic Cloud samples that were analyzed with similar methods, including the description of clumping. We find significant variations in the wind clumping properties of the target stars, with clumping starting at flow velocities $0.01 - 0.23$ of the terminal wind velocity and reaching clumping factors $f_ cl = 2 - 30$. In the luminosity ($ L / odot = 5.0 - 6.0$) and metallicity ($Z/Z_ odot = 0.2 - 1$) range we considered, we find that the scaling of the mass loss $ M $ with metallicity $Z$ varies with luminosity. At $ odot = 5.75$, we find $ M Z^m$ with $m = 1.02 0.30$, in agreement with pioneering work in the field within the uncertainties. For lower luminosities, however, we obtain a significantly steeper scaling of $m > 2$. The monotonically decreasing $m(L)$ behavior adds a complexity to the functional description of the mass-loss rate of hot massive stars. Although the trend is present in the predictions, it is much weaker than we found here. However, the luminosity range for which $m$ is significantly larger than previously assumed (at $ odot 5.4$) is still poorly explored, and more studies are needed to thoroughly map the empirical behavior, in particular, at Galactic metallicity.
Read full abstract