Astron. Astrophys. 334, 618-632 (1998)

1. Introduction

The GW Vir instability strip is composed of hot, hydrogen deficient post-AGB stars whose prototype PG 1159-035 (=GW Vir) was detected in the Palomar Green Catalog (Green et al. 1986). The peculiar nature of this star became evident when McGraw et al. (1979) took first spectra revealing the high effective temperature from the detection of highly ionized carbon lines. Emission lines at He II 4686 Å and C IV 4660 Å led to speculations about helium mass transfer to a compact star in a new type of an interacting binary. A light curve was taken to verify this hypothesis but instead of the expected flickering a sinusoidal light curve with a period of about eight minutes and an amplitude of 0 02 was obtained suggesting that this star defines a new type of low amplitude, non-radial g-mode pulsator. The star PG 1159-035 became also the prototype of a new spectroscopic class, the PG 1159 stars, characterized by a broad absorption trough around 4670 Å composed of He II 4686 Å and several C IV lines. Today, 31 PG 1159 stars are known (see Dreizler et al. 1995 [DWH], Werner et al. 1996c). Nine of them were found to be variable, five belonging to the group of PNNVs (Planetary Nebulae Nuclei Variables) and four to the GW Vir stars ¹. Werner (1992) defined three spectral subtypes of PG 1159 stars (A, E, lgE) based on the presence or absence of photospheric emission lines and the width of the absorption wings. A fourth subtype (H) was introduced by Dreizler et al. (1996) to classify stars that show Balmer lines of hydrogen in addition to the characteristic He/C features. These definitions include several central stars of planetary nebula (CSPN) in the PG 1159 class. In fact, most members of the lgE (low gravity emission) subclass, such as NGC 246 and K 1-16, are central stars of PN. But CSPN are also found amongst the E (e.g. PG 1520 525), A (e.g. Abell 21) and H (e.g. Sh2-68) subclasses. About every other PG 1159 star resides in an old planetary nebula.

The quantitative spectral analysis had to await the development of new sophisticated non-LTE model atmosphere techniques (Werner 1986). These have been applied to the optical spectra of four PG 1159 stars and effective temperatures, gravities and abundances of C and O have been determined for the first time by Werner et al. (1991 [WHH]). Subsequently, several other PG 1159 stars have been analyzed in a similar way. The results are summarized in the reviews by DWH and Werner et al. (1997a).

The quantitative analyses confirmed that PG 1159 stars are amongst the hottest stars known and that their abundance patterns are indeed very peculiar. The PG 1159-phenomenon persists from effective temperatures of 75 000 K (HS 0704 6153, this work) up to =180 000 K (RX J 0122.9-7521, Werner et al. 1996c), a range considerably wider than thought previously and the region in the HRD overlaps with the DO white dwarfs (Dreizler & Werner 1996). Surface gravities also cover the wide range between log g=5.5-8.0. Typically, these stars have a surface composition of 33% He, 50% C, and 17% O by mass (PG 1159-035, WHH). In general, hydrogen is below the detection limit (Werner 1996a). The four PG 1159 stars of type H also display hydrogen lines (therefore also termed hybrid PG 1159 stars; Napiwotzki & Schönberner 1991, Dreizler et al. 1996). Nitrogen is also below the detection limit except for PG 1144 005 with N/He=0.01 (Werner & Heber 1991).

The analyses suggested that the PG 1159 stars provide an evolutionary link between the helium- and carbon-rich central stars of planetary nebulae (spectral type [WC]) and the sequence of helium-rich white dwarfs (DO and DB). The link to the [WC] central stars has been strikingly confirmed by observing the dramatic change of spectral type of the central star Lo 4 from a PG 1159 (lgE) spectrum to WC2 and back in less than half a year (Werner et al. 1992). The similarity of their surface compositions as determined from spectral analyses of [WC] stars (Koesterke & Hamann 1997, Leuenhagen et al. 1996) corroborated this link. The hot helium-rich white dwarfs (DO) are regarded as successors of the PG 1159 stars. Gravitational settling of the heavier elements, probably retarded by radiative acceleration, turns a PG 1159 star into a DO if no trace of hydrogen is left in the envelope. We emphasize that this trace amount is very difficult to determine since it requires high resolution spectroscopy. With existing spectra we can therefore not exclude that PG 1159 stars turn into hydrogen rich DA white dwarfs (see also Sect. 3.2). In the case of the type H PG 1159 stars, were hydrogen is clearly present, the transition into a DA is more likely.

The surface composition of PG 1159 stars cannot be reproduced by canonical evolution calculations, which predict a hydrogen-rich surface during the entire post-AGB phase. Therefore, strong mass loss, possibly caused by a late He shell flash (Iben 1984), must be invoked in order to explain the exotic surface abundance pattern which is typical for 3 processed matter of a former, double shell burning, thermally pulsing AGB star which has lost its entire hydrogen and (almost) its helium-rich envelopes. Evidence for such unexpected mass loss events is given by the above mentioned case of Lo 4.

Although the PG 1159 stars seem to be exotic, their number is sufficiently large to account for up to 50% of the transition objects from the hottest post-AGB phase to the white dwarf stage (Dreizler & Werner 1996). This makes these stars key objects for a complete understanding of post-AGB evolution. Even more interesting is the fact that the many observed pulsation modes in the GW Vir stars allow the powerful tools of asteroseismology to be applied as first demonstrated in the case of PG 1159-035 by Winget et al (1991). This provides an independent determination of stellar parameters, which are partly overlapping, partly complementary to the spectroscopical ones, as well as a direct insight into the structure and evolution of these stars. This was the motivation for extensive photometric monitoring of six PG 1159 stars with the Whole Earth Telescope (WET, Nather et al. 1990) as well as in several single site campaigns (PG 1159-035 Winget et al. 1991, PG 1707 427 Fontaine et al. 1991, RX J 2117 3412 Vauclair et al. 1993, PG 2131 066 Kawaler et al. 1995, PG 0122 200 Vauclair et al. 1995, O'Brien et al. 1996, NGC 246 Ciardullo & Bond 1996). Detailed asteroseismological analyses of WET data (PG 1159-035, PG 2131 066; Kawaler & Bradley 1994, Kawaler et al. 1995) provided many stellar parameters like mass, luminosity, effective temperature, surface composition, thickness of the surface layer, rotational velocities, and upper limits for magnetic field strength, e.g. the mass of 0.59 0.01 as derived from the g-mode period spacing of PG 1159-035 is in reasonably good agreement with the spectroscopic mass determination (WHH, see also Sect. 4.2).

While our existing spectroscopic analyses were sufficient to reduce the huge parameter space of the asteroseismological analysis by providing reliable starting values and to distinguish between multiple solutions (Kawaler et al. 1995), the question of the driving mechanism of the GW Vir pulsation requires more detailed spectroscopic investigations. Even though there is a general agreement among the pulsation theorists (see Gautschy 1997 and references therein) that cyclic ionization of carbon and oxygen in a layer slightly below the photosphere (10^-9 ) causes the instability, details are still unknown. The amount of helium (and hydrogen) in the driving region as well as the stellar radii are critical parameters to match the observed instability strip. Better observational constraints for the limits of the GW Vir instability strip are urgently needed.

Up to now the observational limits of the GW Vir instability strip in the HRD are essentially unknown. In fact a puzzle persists: Our spectral analysis of four PG 1159 stars showed them to be two spectroscopic twins (PG 1159-035 = GW Vir and PG 1520 525; PG 1707 427 = V817 Her and PG 1424 535) with identical atmospheric parameters (to observational limits). However, one star out of each pair pulsates whereas the other does not. Moreover, two of the lowest temperature stars (PG 0122 200 = BB Psc and PG 2131 066 = IR Peg) pulsate while two somewhat hotter objects do not. Hence the physical parameters which determine whether a star pulsates or not remain obscured.

Therefore high quality UV spectra of nine PG 1159 stars (all four known GW Vir pulsators and five non-variable stars ²) have been obtained with the Hubble Space Telescope in order to determine the photospheric parameters with much higher precision than from optical data alone, to determine the O abundance for the first time and to reveal the differences between the pulsating and the non-pulsating objects. Precise temperature and O abundance determinations are a prerequisite for any further discussion of the pulsation properties. Especially the oxygen abundance has to be known precisely because of its importance for the driving of the GW Vir pulsations. Unfortunately, only the hottest PG 1159 stars display O lines in their optical spectra from which O abundances have been derived.

In contrast to the optical the UV covers several ionization stages of carbon and oxygen so that sensitive temperature indicators become available for analyses. The UV also enables us to determine the oxygen abundance of the cooler ( 100 000 K, type A) PG 1159 stars by providing suitable spectral lines not available in the optical. All programme stars belong to the high gravity subtype (A or E), PG 2131 066 is a spectroscopic binary (Wesemael et al. 1985) and PG 1520 525 resides in a Planetary Nebula (Jacoby & van de Steene 1995). Seven of them are "cool" and therefore lack an O abundance determination. The two hot ones ( 140 000 K) form the spectroscopic twin pair PG 1159-035/PG 1520 525 discussed above. The former one was observed already in HST cycle 1 (Werner & Heber 1993) and its analysis demonstrated the achievable precision. It will therefore be the guide line for this analysis.

In the following sections we describe our observations (Sect. 2) as well as the employed model atmospheres (Sect. 3). Results are discussed in Sect. 4 and summarized in Sect. 5.

© European Southern Observatory (ESO) 1998

Online publication: May 15, 1998

helpdesk.link@springer.de