Early-B hypergiants belong to the most luminous stars in the Universe. Their optical spectra are distinguished by strong P Cygni-type profiles of HI and HeI lines indicating strong mass loss.
The B hypergiants represent a short-lived evolutionary phase and are of particular interest in connection with current evolutionary theories of massive stars with mass loss (Meynet et al. 1994; Schaerer et al. 1996a, b).
Quite recently, Kudritzki (1995) found an empirical relationship between the expression (the product of mass-loss rate, terminal wind velocity, and photospheric radius) and the luminosity of O, B, and A supergiants. This relation makes luminous hot stars potentially useful for measuring the cosmological distance scale. Since the B hypergiants represent extreme cases of supergiants, they are especially important to test the validity of this relation over a wide luminosity range. Since these stars are also visually bright enough to be easily observed in extragalactic systems, they can be used to test the metallicity dependence of this relation.
The best-investigated early-B hypergiant is the prototype Sco (HD 152236, B1.5 ), which is a member of the Sco OB1 association. Radio (Bieging et al. 1989) and infrared observations (Leitherer & Wolf 1984; Bertout et al. 1985) have shown that Sco loses mass at a very high rate of the order of . A similar mass-loss rate was derived by Leitherer (1988) from the P Cygni-type profile. The satellite-UV spectrum of Sco was first investigated in detail by Wolf & Appenzeller (1979) and Appenzeller & Wolf (1979) using high-resolution IUE observations. Sco has a high mass-loss rate and a low terminal velocity of only 400 50 (cf. Prinja et al. 1990; Lamers et al. 1995). Therefore the wind density is high and the UV spectrum shows in addition to the typical wind indicators (strong UV resonance lines) also P Cygni profiles in various lines from transitions of metastable lower states (mainly of FeIII).
Lamers et al. (1995) found that the terminal velocities of stellar winds show a transition from fast to slow winds at the temperatures of early-B stars. The ratio drops from a mean value of about 2.6 at to 1.3 at within one spectral type. The stars Sco and HD 190603 are, with =1.1 and 1.2, respectively, at the low side of this "bi-stability jump". Lamers et al. did not study the third star in our sample, HD 169454, but the similarity of the three stars suggests that this star is on the low side of the jump, too. (A selection of the UV wind profiles of Sco and HD 190603 is published in "The Atlas of Ultraviolet P Cygni profiles" by Snow et al. (1994)).
Among the most interesting characteristics of early-B hypergiants are the various kinds of photometric and spectroscopic variations. Sterken (1977) and Burki et al. (1982) discovered photometric variations of Sco of about 0.1 mag on typical time scales of 11 to 16 days. Similar results were obtained for HD 169454.
Extensive high-dispersion spectroscopic observations of Sco in the optical range were performed during several campaigns between 1972 and 1975 with the Coudé spectrograph at ESO (Sterken & Wolf 1978, 1979). These authors found pronounced line-profile variations of the P Cygni-type lines and radial-velocity variations with peak-to-peak values of about 20 . These variations of photospheric lines indicate considerable mass-loss variations and the presence of time- and depth-dependent velocity fields in the atmosphere. Again, similar variations were found for HD 169454. Moreover it should be noted that all lines of Sco are asymmetric with blue wings extending to to . This indicates that practically no line is formed in a pure hydrostatic photosphere (cf. Wolf & Appenzeller 1979).
Burki et al. (1982) investigated the variability of the UV resonance lines of Sco from several high-resolution IUE spectra. They found profile-variable discrete absorption components, which they ascribed to a discontinuous mass loss in addition to the continuous global flow.
These snap-shot observations clearly showed that the mass loss of early-B hypergiants is variable. On the other hand it became evident that a better understanding of the physical processes in the atmosphere and wind of these extreme stars requires long-term spectroscopic monitoring campaigns, with good resolution both in wavelength and time, and covering a wide wavelength range. Using our fiber-linked echelle spectrograph (Mandel 1988) such observing campaigns were carried out very successfully by our group in connection with LBVs (Wolf et al. 1993), P Cygni (Stahl et al. 1993, 1994), and with B and A supergiants (Kaufer et al. 1996a, b).
We therefore decided to monitor the early-B hypergiants Sco (B1.5 ), HD 169454 (B1 ), and HD 190603 (B1.5 ) during several years. Since the typical time scales determined from photometric variability are of the order of weeks we observed the objects during several months to cover typically more than three cycles. Our experience with BA supergiants showed that the repetition of such campaigns over several years is important to provide information on the stability of such cycles and to investigate the variability on still longer time scales.
© European Southern Observatory (ESO) 1997
Online publication: July 3, 1998