An Emerging Class of Dust Obscured Stars

All the newly identified stars have luminosities within a narrow range of $\log L/L_\odot \simeq 5.5$-$6.0$ (see Figure10), which roughly corresponds to initial stellar masses of $M_{ZAMS}\simeq 25$-$60 M_\odot$ (see Section 4 of de Jager1998, and references therein). Local examples of evolved stars in this luminosity range are the Yellow Hypergiants (YHGs) such as IRC$+$10420, $\rho$Cas and HR8752 (de Jager & Nieuwenhuijzen1997; Smith & Owocki2006), many of which are also partially obscured by dust ejecta. There is no means of cleanly surveying the Galaxy for these objects and they are so rare that samples in the Galaxy and the Magellanic Clouds do not provide good statistics for their abundances, life times or total mass loss. Our well-defined sample of likely extragalactic analogs provides a means of addressing some of these questions.

If we assume these objects are similar to stars like IRC$+$10420, their expansion velocities will be more like $50$ km/s than the $100$ km/s of the typical LBV shell. Hence, it seems more appropriate to scale the results to $t_d = 500 t_{d500}$ years. This also matches the estimated age of the phase of dusty mass loss by IRC$+$10420 (Tiffany et al.2010). With 18 candidates, this detection period then leads to a median estimate that $F_e = 0.20 t_{d500}^{-1}$ with $0.086 < F_e t_{d500} < 0.55$. If we associate these with the mass range from $25$ to $60 M_\odot$, they represent a fraction of $f_e \simeq 0.15$ of massive stars, so the average number of episodes per star, $N_e=F_e/f_e \simeq 1.3 t_{d500}^{-1}$ with a possible range of $0.58 < N_e t_{d500} < 3.7$, although this does not include the uncertainties in $f_e$

Figure9 shows that the median mass causing the obscuration is $M_e \sim 0.5 M_\odot$. The total mass lost in all the eruptions is then of order $N_e M_e$, which would be of order $0.3$- $1.9 t_{d500}^{-1} M_\odot$. This implies that the periods of optically thick (dusty) mass loss cannot dominate the overall mass loss of the star. To make the mass lost in these phases dominate either requires that we have grossly overestimated $t_d$, or that the mass range of the stars is much narrower. A related point is that these phases represent a negligible fraction of the post-main-sequence life times of the stars, at most lasting a few thousand years.