Based on observations with the NASA/ESA Hubble Space Telescope, obtained at the Space Telescope Science Institute, which is operated by AURA for NASA under contract NAS5-26555
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218 USA
Institut für Astronomie, ETH-Zentrum, CH-8092 Zürich, Switzerland
Claus Leitherer and Antonella Nota
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218 USA
NASA Goddard Space Flight Center, Code 681, Greenbelt, MD 20771 USA
Max-Planck-Institut für Astrophysik, Karl-Schwarzschild-Str. 1, D-85748 Garching b. München, Germany
Laurent Drissen and Carmelle Robert
Département de Physique, Université Laval, Quebec, PC G1K 7P4, Canada
Affiliated with the Astrophysics Division, Space Science Department of the European Space Agency
We derive wind terminal velocities between 400 and 500 from the P Cygni profiles of SiIV and CIV doublets.
We also derive effective temperatures between 30000 and 39000 K and mass loss values ranging from 2.4 x 10 to 6.6 x 10Myr from optical line strengths.
We discuss the fundamental properties of this sample of Ofpe/WN9 stars and we analyse their evolutionary status.
Keywords: stellar winds - mass-loss - Wolf-Rayet stars
Mass loss and strong stellar winds rule massive star evolution, since at high stellar masses the Core-H exhaustion and the loss of the H-rich envelope become comparable in time-scale (Maeder 1990). In this generally accepted scenario, spectroscopic observations have identified Ofpe/WN9 stars as a post-MS phase during which massive stars are turning into the WR configuration. In fact, Ofpe/WN9 stars display a hybrid spectral morphology characterized by the simultaneous presence of high (i.e., WR-like) and low (i.e., O-like) ionization emission lines at the same intensity (Bohannan & Walborn 1989). A detailed line profile and equivalent width analysis shows a smooth transition between Of and Ofpe/WN9 spectral types (Nota et al. 1995): spectral lines gradually change their absorption profile into a P-Cygni profile, increasing also their equivalent widths. The interpretation of this is an increasing mass loss rate.
These results suggest Ofpe/WN9 stars represent an evolutionary step beyond Of stars, when WR properties have already started to develop. Therefore, an evolutionary continuity seems to exist along the spectral types sequence O - Ofpe/WN9 - WR.
Ofpe/WN9 stars appear also to be related to the LBV phenomenon. The best known examples of this connection are R127 and AG Carinae. R127 had been originally classified as an Ofpe/WN9 star till the '80s when Stahl et al. (1983) observed it undergoing a typical LBV outburst. AG Carinae is a proven LBV star which assumes an Ofpe/WN9-like spectral morphology during its quiescent periods (Stahl 1986). As a consequence of these observations Ofpe/WN9 stars are now suspected to be dormant LBVs.
The above properties make Ofpe/WN9 stars ideal to verify stellar evolution models and to study the mass loss phenomenon. In particular, it is interesting to compare the mass loss parameters (such as mass loss rate, stellar radiation field and wind momentum) found for Ofpe/WN9 with the ones derived for O and WR stars. In this way, it becomes possible to compare the mass loss properties during the evolution from the main sequence to the early Wolf-Rayet stage.
We observed a sample of 7 LMC Ofpe/WN9 stars with the Faint Object Spectrograph (FOS) on board of the HST. The objects are listed in Table 1. All the stars except BE381 had been observed before COSTAR installment.
Table 1: The sample of Ofpe/WN9 stars
We used gratings H13 and H19 in order to cover two spectral ranges: 1150--1610Å at a dispersion of 1Å/diode and 1570--2330Å at 1.47Å/diode. We followed the usual STSDAS pipeline for FOS in order to match the flux and wavelength calibrations for each spectrum.
We normalized the observed spectra through the IRAF task CONTINUUM by fitting Chebyshev polynomials of variable order.
In Figure 1 we plot the spectral range between 1320 and 1690Å, which is characterized by a nicely developed P Cygni profile and a deep absorption profile in the SiIV 1393, 1403 and CIV 1548, 1551 resonance lines, respectively. In most spectra an FeIV absorption line region is also seen developing redward of = 1500Å.
Figure: Spectral range between 1320Å and 1690Å.
In order to derive the wind terminal velocities of our sample we have used the SEI code developed by Lamers et al. (1987) with the SiIV, CIV and NV 1239, 1243 resonance lines, and the CIII 1176 and NIV 1718 excited lines. The SEI (Sobolev with Exact Integration) code calculates the source function in the Sobolev's approximation and exactly solves the equation of radiative transfer. Once the wind velocity and opacity laws are specified, analytical lines profiles are calculated for comparison with the observations.
In the case of the Ofpe/WN9 stars we have adopted the velocity law:
where = 1 in agreement with observations of O stars (Groenewegen & Lamers 1989) and v/v = 0.03 in order to reach the sound speed at the base of the wind.
The results of the fit are summarized in Table 2.
All the stars (except R99) exhibit a wind terminal velocity of 400--500 50 , a factor 4--5 less than the wind velocities measured for O and WR stars of comparable temperature. Uncertainties on v are mainly due to the low dispersion of our spectra and to the SEI computational method which does not give a unique fit for each line. The wind of R99 is instead characterized by v = 900 which is certainly unusual for this spectral type.
Table 2: Wind terminal velocities
R99 shows several anomalies, such as a very strong infrared emission excess (see Stahl et al. 1984) and the lack of broad FeIV absorptions regions at 1500Å (see Figure 1), possibly due to a lower metallicity or a higher ionization degree of its wind. The presence of the HeII 1640 line (with a strong P Cygni profile) supports this suggestion.
We have also re-scaled Schmutz et al. (1989) stellar atmosphere models (originally calculated for v = 2500 ) to the reference v value of 400 in order to derive the stellar parameters and mass loss rates for our sample.
These models consider a non-LTE spectrum formation in a spherically expanding atmosphere where the velocity field of the wind is represented by the analytical law defined above. The temperature structure is derived from the assumption of radiative equilibrium for the grey LTE case. The theoretical results are in the form of contour plots which give for a fixed line equivalent width the effective temperature, the mass loss rate and the absolute B and V fluxes (from which stellar radii are derived).
Table 3: Temperatures, radii, luminosities and mass loss rates for Ofpe/WN9 stars
As input for the new stellar grids we have used the optical H and He lines equivalent widths obtained by Nota et al. (1995) during a comprehensive, ground-based spectroscopic survey of the same sample. Table 3 summarizes the stellar properties.
Effective temperatures and stellar luminosities place Ofpe/WN9 stars in the upper HR diagram between the areas occupied by WR and LBV stars in agreement with early observational findings (see Sect. 1). The link between Ofpe/WN9 and WR stars is also supported by the mass loss rates which are of the order of 10Myr for both spectral types.
In the radiation pressure driven wind scenario the stellar radiation field supports the mass loss. Stellar photons interact with matter through one scattering event which transfers momentum from the radiation to the wind. Nevertheless, a discrepancy exists between the observed and predicted wind momentum for extreme O stars. The predicted momentum becomes smaller than the observed value for stars with increasing wind density. The observed momentum transfer is then more efficient.
This could be due to the ionization stratification of the wind which depends on its density and allows more than one scattering event between each photon and the particles (Lamers & Leitherer 1993). In this case (not yet implemented in the calculations) every photon transfers to the wind an amount of momentum larger than h/c up to 4h/c (Abbott & Lucy 1985, Schulte-Ladbeck et al. 1995). The same v discrepancy is found enhanced in the case of WR stars.
How do Ofpe/WN9 stars fit into this picture?
We have calculated for the stars in our sample the observed wind momentum v), the predicted one (derived from Eqn. 22 by Lamers & Leitherer 1993) and the ratio = v)c/L. This parameter, known as efficiency, measures the amount of radiation momentum transformed into kinetic momentum of the wind. Thus, if the stellar wind is entirely radiation-supported, is expected to be 1.
These quantities are plotted in Figure 2 where open dots and triangles represent O and WR stars of Lamers & Leitherer's (1993) database. Black dots are the Ofpe/WN9 stars of our sample.
In the upper panel of Figure 2, the observed wind momentum is plotted against the radiative momentum. O stars form the linear distribution which is expected when the wind is driven by the radiation pressure (Lamers & Leitherer 1993), while in the case of WR stars the wind momentum does not correlate with the stellar radiation field. Ofpe/WN9 stars mostly follow the edge which separates O and WR stars distributions.
In the lower panel of Figure 2 the ratio between the predicted and observed wind momentum is plotted as a function of the efficiency . O and WR stars are distributed diagonally indicating that the v discrepancy increases with from the O to the WR spectral type. As the wind density increases, the radiation momentum is converted into wind kinetic momentum more and more efficiently (for some WR stars 1) and the theory fails to reproduce the observed and v values (see Lamers & Leitherer 1993). Ofpe/WN9 stars confirm this trend; they fill the gap between O and WR stars previously discovered by Lamers & Leitherer (1993).
We can draw two main conclusions from Figure 2. First, Ofpe/WN9 stars exhibit mass loss properties which are intermediate between O and WR stars. In agreement with the suggestions by Nota et al. (1995), we find that also the wind physics indicates that a smooth transition from O to Ofpe/WN9 exists which may correspond to an evolutionary sequence: massive stars, evolving off the main sequence, would enter the Ofpe/WN9 phase during which they would begin to develop WR characteristics. Accurate determinations of the chemical abundances are needed to confirm this evolutionary link. Second, we find the theory of radiatively driven winds in its present formulation underestimates all along this sequence so that = log[)/ )] increases with .
Figure: Upper panel: the observed wind momentum as a function of the radiative momentum. Lower panel: the ratio between the predicted and observed wind momentum plotted against the efficiency .
Several solutions to this discrepancy are possible. First, the Fe atomic physics (concerning in particular FeV and FeVI) is poorly known so that the iron opacity, which drives the wind, is not properly represented in the calculations (Pauldrach et al. 1994). Alternatively, if the ionization stratification of the wind is responsible of amplifying the photon-ion interaction, it has to grow stronger during the evolution in order to support the WR final stage winds. Finally, the stellar radiation field might not be the unique engine driving the stellar winds, but other mechanisms could intervene, such as stellar radial pulsations, towards the final evolutionary phases to drive and enhance the stellar winds. The identification of these mechanisms or the correct formulation of the wind ionization law (which expresses the ionization degree as a function of the distance in the wind) is then crucial to make effectively work the radiative pressure driven wind theory.
In a forthcoming paper we will discuss in detail the ionization status of Ofpe/WN9 stars wind and their evolutionary status also including the analysis of their surface chemical composition.
Support for this work was provided by NASA through grant number G005.76100 from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555. AP acknowledges partial support from the STScI DDRF for the duration of the work. WS acknowledges the financial support and hospitality of STScI.
Abbott, D. C. & Lucy, L. B. 1985, ApJ, 288, 679
Bohannan, B., & Walborn, N. R. 1989, PASP, 101, 520
Castor, J. I., Abbott, D. C. & Klein, R. I. 1975, ApJ, 195, 157
Lamers, H. J. G. L. M., Cerruti-Sola, M. & Perinotto, M. 1987, ApJ, 314, 726
Lamers, H. J. G. L. M. & Leitherer, C. 1993, ApJ, 412, 771
Maeder, A. 1990, A&AS, 84, 139
Nota, A., Pasquali., A., Drissen. L., Leitherer, C., Robert, C., Moffat, A. F. J. & Schmutz, W. 1995, , in press
Pauldrach, A. W. A., Feldmeier, A., Puls, J. & Kudritzki, R. P. 1994, in Evolution of massive stars. A confrontation between theory and observation, D. Vanbeveren, W. Van Rensbergen & C. de Loore, Dordrecht: Kluwer, 105
Schmutz, W., Hamann, W.-R. & Wessolowski, U. 1989, A&A, 210, 236
Schmutz, W., Leitherer, C., Hubeny, I., Vogel, M., Hamann, W.-R. & Wessolowski, U. 1991, ApJ, 372, 664
Schulte-Ladbeck, R. E., Eenens, P. R. J. & Davis, K. 1995, ApJ in press
Stahl, O. 1986, A&A, 164, 321
Stahl, O., Wolf, B., Klare, G., Cassatella, A., Krautter, J., Persi, P. & Ferrari-Toniolo, M. 1983, A&A, 127, 49
Stahl, O., Wolf, B., Leitherer, C., Zickgraf, F.-J., Krautter, J. & de Groot, M. 1984, A&A, 140, 459
Wolf, B., Stahl, O. & Seifert, W. 1987, A&A, 186, 182