Antonella Nota and Mark Clampin
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218 USA
Affiliated with the Astrophysics Division, Space Science Department of the European Space Agency
UNAM, Instituto de Astronomia, Apdo Postal 70-264, 04510 Mexico
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218 USA
Max-Planck-Institut fur Astrophysik, Karl-Schwarzchild Str. 1, Postfach 1523, D-85740 Garching b. Munchen, Germany
Keywords: LBVs - nebulae
Very massive stars (M > 25 M) evolve at roughly constant bolometric luminosity throughout their various phases. Therefore, it is likely that O stars, WN stars, LBVs, and possibly also blue and red supergiants, represent different evolutionary phases of a star with a certain ZAMS mass. A primary goal in stellar physics is to investigate the correct order in time of these various phases. However, this goal has not yet been fully achieved. The sequence O star -- Of -- H rich WN star -- LBV -- H poor WN -- WC -- SN (Langer et al. 1994) seems to be at odds with current state-of-the-art massive star models, i.e., O star -- LBV -- WN star, and may require non-standard assumptions about the physics of massive star mass loss (Langer et al. 1994).
As an example, although Wolf-Rayet (W-R) stars lose mass at a very high rate in the form of strong stellar winds, their mass-loss rate integrated over their lifetime is insufficient to enable their direct evolution from O stars, except perhaps for stars more massive than 120 M. At some stage, O stars with masses between about 20 and 120 M have to undergo extremely high mass-loss rates, and lose several solar masses. This might happen in the very short-lived ( 10yr), evolutionary transition identified as the Luminous Blue Variable (LBV) phase. LBVs are indeed well known to be surrounded by ring nebulae composed of stellar ejecta, which contain several solar masses (Nota et al. 1995).
We have already carried out an extensive ground-based coronographic imaging program, to establish the basic parameters of those LBV nebulae that can be resolved from the ground. We recently compared (Nota et al. 1995) the morphologies and physical parameters for six LBV nebulae and found that, with the possible exception of P Cygni, they all show a degree of bipolarity, in addition to other similar properties.
We, therefore, proposed a model which consists of a stellar wind interacting with a pre-existing density contrast (may be the remnant of the pre-outburst phase) between the equatorial and polar directions, in similarity to the interacting wind model of planetary nebulae (Balick 1987).
A higher contrast would generate a pronounced bipolar structure, while a smaller density contrast would result in a more elliptical geometry. A simple, two-dimensional hydrodynamic simulation by Nota et al. (1995) has demonstrated that a morphology similar to that of R127 can be easily obtained from such a model. Similarly, a calculation with parameters appropriate for Eta Carinae reproduced very well the observed morphology (Frank, Balick, & Davidson 1995).
It has also been shown by García-Segura & MacLow (1995) using multi-dimensional hydrodynamic calculations for interacting steady winds, that the succession of massive star winds, which correspond to the different evolutionary stages of the central star, provides very different morphologies of the circumstellar nebulae for different evolutionary sequences.
These can be easily identified both in the large and in the small scale structure of the circumstellar nebulae. In the large scale structure, the circumstellar shell is being swept up by a slow wind being followed by a fast wind, e.g., during and after the LBV phase (García-Segura et al. 1995, Nota et al. 1995).
High resolution calculations show the occurrence of hydrodynamic instabilities, which break the wind shells into fragments, clumps or filaments, with a characteristic appearance depending on the nature of the instability (García-Segura & MacLow 1995, García-Segura et al. 1995). Since the occurrence of each type of instability requires specific conditions in the interacting gas components, the detection of small scale structures in circumstellar nebulae, together with the large scale dynamical behavior, can yield rigorous conclusions about the winds which have formed them.
AG Carinae is a well known galactic LBV, which has been extensively studied from the ground. AG Carinae is surrounded by an impressive shell nebula. From ground-based coronographic imaging and long slit spectroscopy, we have accumulated information on:
Figure: A composition of 2 300 s images of the AG Carinae nebula, taken with the HST/WFPC2 and the [NII] filter F658N.
Figure: A composition of 5 400 s images, of the AG Carinae nebula, taken with the HST/WFPC2 and the F547M filter. North is in the lower right corner. The field of view is equivalent to the entire PC chip, that is 36" 36".
We have observed the AG Carinae nebula with HST/WFPC2 in June 1994, in the light of the V continuum (Figure 1) and of [NII] (Figure 2). The amount of information available, and the new high resolution HST images prompted us to perform hydrodynamical modelling of the nebula, with the objective to constrain, if possible, the pre-outburst phase of AG Carinae and its past mass loss history.
The simulations were performed with the hydrodynamic code ZEUS (developed by M. L. Norman and D. A. Clarke), which is a finite-difference, fully explicit Eulerian code. It integrates the hydrodynamic equations for an ideal gas, ignoring viscosity and resistivity.
We used spherical coordinates for our simulations, with a symmetry axis at the pole, and reflecting boundary conditions at the equator and the polar axis. Our models have grids of zones, with a radial extent of 0.5 pc, and an angular extent of 90. The innermost radial zone lies at 0.0125 pc. The set up of the winds is based on the studies of Bjorkman & Cassinelli (1993), which have the advantage of not assuming the structure of the wind.
We started from the observed parameters which describe the present phase of the system, namely the terminal velocity of the present stellar wind ( 250 ), the number of ionizing quanta (10), the ionized gas mass ( 4 M), the expansion velocity ( 70 ), the nebular radius (0.3--0.4 pc), and the dynamical timescale.
The observed parameters can be translated into wind properties, and into a mass-loss time history which develops through three phases: (i) a cool phase, characterized by a slow stellar pre-outburst wind ( 50 ), high mass loss (1.1 10 Myr), (ii) an outburst cooler phase, with even higher mass-loss (0.13 Myr, but slower wind velocity (15 ), and (iii) the present phase, with a faster stellar wind ( 300 ), lower mass loss, (4 10 Myr), with a duration which is given by the dynamical timescale.
The simulation reproduces very well the overall morphology of AG Car. In Figure 4, the model of the upper right quadrant is to be compared to Figure 3, where a ratio of the [NII]/V image illustrates the relative positions of the thick ionized shell (Figure 2) and of the more dense and neutral sub-structures embedded in the shell (Figure 1). The simulation shows the behavior of the ionization and how the radiation generates self-shielding regions. The ionized gas pushes the neutral gas in the self-shielding regions and shapes the observed radial filaments, producing the comet-like features observed by the HST (Figure 1, right upper quadrant).
Figure: Multi-dimensional hydrodynamical simulation of the same region as in Figure 3. Notice the elongated structure of the gas shell and the neutral filaments as in Figure 3.
Figure: A blow up of the upper right region (Figures 1 and 2 ) of the AG Carinae nebula, where a ratio [NII/V] illustrates the relative position of the gas and dust distribution
Extensive ground-based coronographic observations of nebulae around LBVs suggest that an interacting wind scenario can explain most of the morphologies observed, both on large and small spatial scales.
A comparison of high resolution HST images with the predictions of multi-dimensional, hydrodynamical models can quantitatively establish 1) the wind velocity and mass loss properties of the pre-outburst phase, and 2) the duration of the pre-outburst phase.
The large scale nebular structure provides information on the LBV outburst mechanism and related intervening factors (rotation, binarity), while the small scale features provide information on the local instabilities, such as anisotropies of the wind.
A recent experiment conducted on recent HST images of the galactic LBV AG Carinae has proven the method is effective and has demonstrated its potential as a diagnostic tool, to empirically investigate the mass loss history and, thereby, the evolutionary history of the central star.
Balick, B. 1987, AJ, 94, 671
Bjorkman, J. E. & Cassinelli, J. P. 1993, ApJ, 409, 429
Frank, A., Balick, B., & Davidson, K. 1995, ApJ, 441, L77
García-Segura, G. & MacLow, M. 1995, ApJ 455, 145
García-Segura, G., Langer, N. & MacLow, M. 1996, A&A 305, 229
Humphreys, R. M., Lamers, H., Hoekzema, N., & Cassatella, A. 1989, A&A 218, L17
Langer, N., Hamann, W.-R., Lennon, M., Najarro, F., Pauldrach, A. W. A., & Puls,J. 1994, A&A, 290, 819
Nota, A., Livio, M., Clampin, M., & Schulte-Ladbeck, R. 1995, ApJ 448, 788
Smith, L. 1991, in IAU Symp. 143, Wolf-Rayet Stars and Interrelations with other Massive Stars in Galaxies, ed. K.A. van der Hucht & B. Hidayat (Dordrecht: Kluwer), 385