SF2A 2003
F. Combes, D. Barret and T. Contini (eds)
THE ATMOSPHERE OF RED SUPERGIANT STARS
Josselin, E. 1 and Plez, B.1
Abstract. Red supergiants (RSG) constitute a key-phase in the evolution
< initial mass < 40 M ), characterized by strong massof massive stars (10 ∼
∼
loss, and rich nucleosynthetic processes. They play a decisive role in the
chemical evolution of the Galaxy. Our knowledge of these objects is however
still limited, in particular because of the complexity of their atmospheres. We
present an analysis of high-resolution spectra of a sample of 20 RSG, probing
the velocity structure of their atmosphere. We interpret lineshifts as due to
motions of probable convective origin and we show that convection must
play an essential role in the mass-loss process. We also present preliminary
results from multi-epoch spectroscopy of some representative objects.
1
Introduction
Red supergiants (hereafter RSG) represent a key-phase in the evolution of massive
stars, with initial mass ranging from ∼10 to ∼40 M , preceding the explosion into
type II supernovae. The RSG stage is dramatically influenced by strong mass loss
(Ṁ ∼ 10−6 M yr−1 ; Castor 1993). Also, thanks to the various nucleosynthetic
processes taking place in RSG and (unknown) mixing processes (Rauscher et al.
2002 and references therein), RSG play a major role in galactic chemical evolution.
In this context, it is striking that the mass-loss process of RSG is still so
poorly known. Extrapolation of models based on pulsationally-driven winds coupled with radiation pressure on dust grains are irrelevant, as RSG are generally
low-amplitude, irregular variables. Alternative theories of mass loss, in particular those based on acoustic or Alfvén wave (Hartmann & Avrett 1984, Pijpers &
Hearn 1989, Cuntz 1997) driven winds, have to be considered again. Furthermore,
recent hydrodynamical simulations by Freytag et al. (2002) of the prototypical
< 10) evolving on a timescale of
RSG Betelgeuse show giant convective cells (∼
> 1 month with supersonic velocities. Acoustic energy generated by such peculiar
∼
convection should provide energy to heat stellar chromospheres and produce shock
waves capable to drive mass flows.
We present here an analysis of high-resolution visible spectra of a sample of
RSG which bring new constraints on the structure of their atmosphere and the
origin of mass loss.
1
GRAAL, cc072, Université Montpellier II, F-34095 Montpellier cedex 05, France
c EDP Sciences 2003
2
2
The atmosphere of red supergiant stars
Sample, observations and method of analysis
Our sample consists of 20 M supergiants selected in Humphreys (1978). Most of
them are members of galactic OB associations, so their distance and the interstellar
extinction affecting them are known. Fundamental parameters such as bolometric
luminosity and dust mass-loss rate were estimated by Josselin et al. (2000), based
on their infrared photometry and millimeter spectroscopy dataset.
A first set of high-resolution (λ/∆λ = 42000) spectra, covering the wavelength
range λλ 3900-6800 Å, were obtained in August 1999 with the echelle spectrograph
ELODIE, mounted on the T193cm at the Haute-Provence Observatory (Baranne
et al. 1996). We then started a time series follow-up with the same instrument of
those objects for which the fundamental parameters were best constrained.
In order to derive the characteristics of atmospheric motions, we applied to
these spectra the tomographic technique developed by Alvarez et al. (2001). The
spectra are correlated with numerical masks which probe layers at different atmospheric heights. The spectrum used to build the masks is a synthetic spectrum of
a giant with Tef f = 3500 K (type M5) computed with MARCS model atmosphere
(Plez 2003 and references therein). Eight masks are used, labelled C1 to C8, which
probe innermost to outermost layers, respectively. They are described in Alvarez
et al. (2001).
3
3.1
Atmospheric motions and mass loss
Results of the tomography
The cross-correlation functions (CCF) present a wide variety of shapes, for each
star from one mask to another, and between stars. A few RSG appear ”static”,
with symmetric profiles at the same velocity for all the masks. On the contrary,
many RSG have CCF with increasing asymmetry as one probes layers with decreasing depth. Simultaneously the peak of CCF shifts to the red, indicating downward
motions of the outer or cooler layers. Finally a second peak appears sometimes
in the outermost CCF. As the mask C8 contains essentially low-excitation and
resonance lines, this additional feature may correspond to a circumstellar feature
in these lines.
We thus derive two parameters from these CCF. First, the atmospheric velocity
amplitude ∆vatm corresponds to the velocity difference between the peak of the
CCF obtained with mask C8 and that with mask C1. Second, the circumstellar
expansion velocity vexp is estimated as the difference between the secondary peak in
mask C8 and the peak in mask C1 (vexp may be smaller than the ”true” expansion
velocity, as the circumstellar absorption probably occurs in a region where the
terminal velocity is not reached). We get ∆vatm in the range [0,-20] km s−1 and
vexp in the range [0,14] km s−1 .
The atmosphere of red supergiant stars
3.2
3
Diagnostics of convection and relation with mass loss
We interpret the observed atmospheric motions as the signature of convective motions. Indeed, as suggested by Schwarzschild (1975) and shown by 3D hydrodynamical simulations by Freytag et al. (2002), the convective pattern of supergiants
may consist of very few cells (typically less than 10), so that each given line would
probe very few velocity components and not result into a ”C”-shaped bisector as
for the Sun (Asplund et al. 2000). Another clue to this peculiar convection comes
from the line shapes compared with synthetic spectra computed with MARCS
models. A remarkable result is that the ”macroturbulent broadening” is exponential rather than gaussian, suggesting again purely horizontal motions, as produced
if we would see a unique convective cell (Huang & Struve 1953).
How can convective motions account for the observed mass loss ? As shown
in Fig. 1, the dust mass-loss rate increases with ∆vatm . We also observe an
increase of ∆vatm with the stellar radius (Josselin et al. 2003). An increase
of stellar radius then results in a strong increase of the ratio ∆vatm /vescape and
consequently, should favor mass ejection. We thus propose that convection plays
a key-role in the mass-loss process in RSG.
Fig. 1. Relation between the atmospheric velocity amplitude and the dust mass-loss rate.
3.3
Variability of line profiles
Finally, we emphasize that the heterogeneity of the sample suggested by the variety
of CCF shapes may be only apparent. Indeed, our spectroscopic follow-up suggests
on the contrary that all the range of shapes are met at different epochs for each
stars, but with different velocity amplitudes. Figure 2 illustrates this potential
strong changes in the case of SW Cep (M3.5 I). The variations seem to be very
irregular, with time scales of a few months, compatible with the results of Freytag
4
The atmosphere of red supergiant stars
et al. simulations.
Fig. 2. Comparison of average line profiles (in fact CCF profiles) of SW Cep at two
epochs (full line: August 1999; dashed line: May 2003). Each panel is labelled with the
mask used, the masks C1 to C8 probe deep to high layers (Alvarez et al. 2001).
References
Alvarez R., Jorissen A., Plez B., et al., 2001, A&A 379, 288
Asplund M., Nordlund Å, Trampedach R., et al., 2000, A&A 359, 729
Baranne A., Queloz D., Mayor M., et al.,1996, A&AS 119, 373
Castor J.I., 1993, in: “Massive Stars: Their Lives in the Interstellar Medium”, ASP Conf.
Ser. Vol. 35, p. 297
Cuntz M., 1997, A&A 325 709
Freytag B., Steffen M., & Dorch B., 2002, Astron. Nachr. 323, 213
Hartmann L., & Avrett E.H., 1984, ApJ 284, 238
Huang S., and Struve O., 1953, ApJ 118, 463
Humphreys R.M., 1978, ApJS 38, 309
Josselin E., Blommaert J.A.D.L., Groenewegen M.A.T., et al., 2000, A&A 357, 225
Josselin, E. Plez, B., Mauron, N., 2003, in: ”Modelling of stellar atmospheres”, eds. N.E.
Piskunov, W.W. Weiss, D.F. Gray, ASP, in press
Pijpers F.P., & Hearn A.G., 1989, A&A 209, 198
Plez B., 2003, in ”GAIA Spectroscopy, Science and Technology”, ASP Conf. Ser. vol.
298, 189
Rauscher T., Heger A., Hoffman R.D., & Woosley S.E., 2002, ApJ 576, 323
Schwarzschild M., 1975, ApJ 195, 137