Clumping in hot-star winds as function of metallicity: first theoretical predictions and implications for mass-loss diagnostics


We explore the impact of metal content on clumping in the winds of massive stars. Using numerical hydrodynamics with radiative cooling, limb darkening, and base perturbations, we simulate a wind with a fixed stellar mass and luminosity and vary the wind strength \(\bar{Q}\), which varies directly with metal content. Analysis of the radial distribution of clumping in these simulations shows that the clumping factor varies with metallicity as \(f_{cl} \simeq Z^{0.3}\), implying that observationally determined H alpha mass loss diagnostics that assume metallicity-independent clumping have overestimated by 20-30% the exponent of the mass loss rate to metallicity power law relationship.


  • O stars importance, mass loss rate diagnostics

  • how the X-rays work

  • Clumping

Massive stars are thought to be important in reionization of the early universe, and also represent an important force in galaxy evolution through both their supernovae and strong, radiation-driven winds. These dense winds, driven by line scattering, can impact massive star evolution, removing in some cases more than 10% of an O star’s mass over its lifetime. The impact of metallicity on mass loss is especially interesting, since reionization and galaxy evolution involve massive stars in widely varying – and often very low metal – environments. Castor, Abbot, & Klein (1975) gave a first quantitative description of this line driving by assuming a smooth, steady-state wind, and extensions of this theory have had success explaining a variety of wind properties, including the dependence of mass loss rate on metal content (cite).

There is now substantial empirical evidence (cite Evansberg, Puls, Cohen, Njarro, Sundqvist, Bouret; see Sundqvist 2013) suggesting that these winds are not smooth, but instead substantially “clumped”. Theory also predicts that the line driving of these winds has an intrinsic instability, the line-deshadowing instability (LDI), which would generate clumped structure. This has important implications on commonly-used mass loss rate diagnostics like H-\(\alpha\), which typically assume a smooth wind, and work on clumping has provided evidence that mass-loss rates of massive stars have in general been overestimated by as much as a magnitude.

A recent multiwavelength study assuming a constant multiplicative factor for clumping correction (Mokiem 2007) produced an empirical power law relation between mass-loss rate \(\dot{M}\) and metal content \(Z\), with \(\dot{M} \propto Z^m\) and \(m = 0.83 \pm 0.16\), consistent with theoretical predictions by Vink et al. 2001. This model applies a uniform clumping correction to stars where the assumption of wind clumping improves line fitting. However, evidence (CITE THIS) suggests all massive star winds have some degree of clumping, and metallicity should have an effect on clumping since the LDI and the winds themselves are driven by bound electron transitions in metal ions. O stars in the LMC and SMC thus might have substantially different clumping properties, which impacts any study using mass loss rate diagnostics which depend on clumping.

This paper follows up on Mokiem 2007’s empirical relation by providing first theoretical predictions on metal content’s effects on wind clumping, using state-of-the-art hydrodynamic simulations to measure metal content on clumping. Focusing on two radial regions which produce the majority of \(H_{\alpha}\), we follow Mokiem et al. 2007 and investigate an array of metal contents which span the estimated metal contents of the galaxy at \(Z=1.0\), the Large Magellanic Cloud (LMC) with \(Z = 0.5\), and the Small Magellanic Cloud (SMC) with \(Z = 0.2\).

Numerical Simulations