N 9 3 - t 6 7 2 0* THE ULTRAVIOLET SIGNATURE OF MASSIVE STARS IN STARBURST GALAXIES CARMELLE ROBERT, GLAUS LEITHERER1, AND TIMOTHY M. HECKMAN- Space Telescope Science Institute, 3700 Sail Martin Dr.. Baltimore, MD 21218 ^ Affiliated with the. Astrophysics Division of ESA 2 Johns Hopkins University, Dept. of Physics and Astronomy ABSTRACT We present a progress report of a study of the massive star population in starburst galaxies using the UV spectral region. INTRODUCTION Studies in different spectral regimes of starburst galaxies clearly indicate the presence of hot massive stars (see STScI Sym. 5: Massive Stars in Starbur.sts, 1991). However, only the UV spectral region can be used to directly identify the spectroscopic signature of these stars. The typical contributor to the integrated continuum at ~ HOO A of a starburst is from the hot B stars More massive stars (0 stars and their descendants, the Wolf-R.ayet [VV-R] stars) arc responsible for the majority to the UV lines which are broad (compared to the interstellar [IS] lines) photospheric absorption lines and wind emission or P Cygni profiles. POPULATION SYNTHESIS MODELS The line profiles of SilV A1397 and CIV A1549 are synthesized with the latest generation of stellar evolution models (Maeder 1990. A&AS 84. 139), stellar atmospheres (Kurucz 1992 IAU L(* p 225; Schmutz ft «/. 1992, PAST in press), and a line profile library based on IUE high dispersion spectra, of massive stars (except for the B stars for which we use low dispersion spectra at the moment). Stellar populations for continuous and instantaneous bursts are calculated given a certain slope (a) and cut-off masses (A/, and M ) for the initial mass function (IMF). An interpolation is u done for a mass interval of 1 M* and a time resolution of 104 yr. RESULTS Figure 1 shows the relative number of massive stars in an instantaneous burst. As one can see, the most massive stars, 0 and BIV-V VV-R stars, disappear after ~ 107 yr. The number of BV-IV stars slightly decreases OIV-V with time while the number of B I-III stars is increasing. oi-m In figure 2, we present the SilV and CIV S profiles synthesized for the stellar population described in figure 1. At t ~ 10'35 yr. the SilV doublet has a P Cygni type profile (over which the IS component is superposed). This 6.5 wind profile shape is due to an increased log(t [yr]) number of 0 supergiants and W-R stars. At t > 10'5S yr. the SilV doublet is broad but FIGURE 1. Number of stars as a function not blue shifted anymore. B stars, which of time for an instantaneous burst of total have become the dominant contributors, are mass io6 A/,:, (Z = Zr... a = 2.35. ,\j, [ = responsible for this profile shape (along with Mr,, and M = 120 M,-.,). u 34 the IS medium). Massive main sequence stars have a CIV doublet which already shows a P Cygni type profile. This profile increases in strength and become narrower as the winds from 0 slipergiants becomes more important (t ~ 1065 yr). Later on. B stars and the IS mecUum are responsible for a weak CIV feature. SilV A1397 CIV A1549 V (km s-1 ) 1.6 - 1.6r^/\ r^~^^ £ 100° 1 0.5 0 0 1.6 1 0.5 0.6 0 0 0 -1000 1.5 - M.= 120 I 1 -y -2000 M.=60 U. ' M.=30 U. - 0.5 - 6.5 7 7.5 0 i_ 1360 1380 1400 1420 1440 1500 1520 1540 1560 1560 1600 log(t [yr]) WAVELENGTH (A) WAVELENGTH (A) FIGURE 3. Synthetic SilV and CIV FIGURE 2. Synthetic SilV and CIV line profiles velocity shift as a function of time of a burst at three evolutionary phases. (Model for three upper cut-off mass limits. (Model parameters are as in fig. 1.) parameters are as in fig. 1.) The velocity shifts described in the synthetic profile of figure 2 are quantitatively presented in figure 3 (solid line for M = 120 M.-,). One can clearly see the blue shift of SilV after ~ 1U'3^ yr u due to stellar winds from massive stars. CIV is shifted back to the laboratory wavelength as time passes, i.e. as massive stars are dying. hi figure 3, the velocity shifts are presented for 3 models using different values of M,,. For M = 60 M,-.>, the main sequence star contribution to CIV is reduced at early phases. For u jl/,i = 30 M-,. most massive stars are absent and the profiles are virtually mishitted. No important dependences of the SilV and CIV velocity shifts with a (i.e. from 1.5 to 3) are observed. OBSERVATIONS OF STARBURST GALAXIES We have selected a sample of 49 starburst galaxies from IUE archives based on their high sigiial- to-uoise ratio. For this sample, we calculate an average velocity shift of-490 and -<45 km s"1 for SilV and CIV. respectively. No significant shifts (larger than ±200 km s"1 ) are seen for simple lilies formed in the IS medium. Based on our models, the large blue shifts observed for SilV and CIV clearly indicate the presence of massive stars. > 30 M:,, in most of our galaxies. In the case of NGC3256. for which Joseph (1991, Massive Stars in Starbursts. p. 259) estimated AI» = 25-30 M,-.< based on IR data, we obtain large blue velocity shifts for SilV and CIV (-611 and -112!) km s-i ). This result argues in favor of a large value for M . u The full potential of our method will be exploited on the basis of Hubble Space Telescope data of several galaxies available in a near future. We gratefully thank Kimiey, Bohlin. Prinja, Howarth. St.-Louis, and Faiielli for the use of their reduced IUE spectra, T. H. and C. L. acknowledge NASA for support of this research in the form of ADP and LTSA grants. C. R. thank CRSNG (Canada) for a post-doctoral fellowship and STSci's Director's Research Fund for travel support. 35