The methods used to derive the stellar atmospheric parameters are similar to those discussed in previous papers concerning high-latitude B-type stars (see, for example, Paper I and references therein). All theoretical results are based on the grid of line blanketed model-atmospheres of Kurucz (1991), together with radiative transfer codes to derive the atmospheric parameters and chemical compositions. A normal Population I chemical composition was assumed in the atmospheric models and this was subsequently found to be appropriate for all of the stars (see Sect. 3.2).
3.1. Atmospheric parameters
Atmospheric parameters were derived for the bright, `normal' Population I comparison stars using the same procedures as for the programme objects. Stellar effective temperatures were primarily evaluated from the reddening free Strömgren indices using the calibration of Napiwotzki et al. (1993). In all cases, the photometric uncertainties of 0.01 magnitudes translated to an error of less than 1 000 K in effective temperature. Estimates of the surface gravities were then deduced by comparing computed profiles of the Balmer lines (H, H and H) with the extracted spectra. For three objects, viz. PG 0934+145, PG 2229+099 and PG 2345+241, it was possible to obtain a second effective temperature estimate from the Si II / Si III ionization equilibrium. Comparison of the two methods show that the latter temperatures (see Table 2) are consistently higher than the photometrically derived temperatures (presented in Table 3) by approximately 2 000 K. The calculation of the silicon line-strengths were undertaken in a non-LTE regime (see, for example, McErlean et al. 1998) and used unblanketed model atmospheres generated with the code TLUSTY (Hubeny et al. 1994). By contrast, the photometric temperature calibration was based on LTE line-blanketed models of Kurucz (1991). Hence, the systematic difference in the temperature estimates from the two methods may reflect the different treatment of line blanketing. Indeed, a comparison of the temperature-optical depth scales implied that for models with similiar structures, the effective temperature label of the unblanketed non-LTE model was typically 1500-2500 K higher than that for the Kurucz model. Here, we have adopted the photometric temperature estimates to ensure consistency with the analyses of the other programme stars. Microturbulent velocities (), used in the LTE analysis of the helium and metal-line spectra, were derived for two objects (PG 2229+099 and PG 2345+241) and for all comparison stars, by adjusting this parameter so as to remove the dependence of O II or S II abundance upon line-strength (see, for example, Gies & Lambert 1992). For the remaining targets, we assumed a value of the microturbulence to be similar to that of the corresponding comparison star. The final adopted atmospheric parameters for the programme stars are presented in Table 3, where they are grouped with the appropriate comparison object from the grid presented in Paper I.
Table 2. Silicon ionization equilibria based on non-LTE model atmosphere calculations and a microturbulence, km s-1.
Table 3. Adopted atmospheric parameters.
3.2. Photospheric chemical compositions
Equivalent widths for the non-diffuse helium and metal-lines were measured by normalizing regions of continuum around identified lines with low-order polynomials and then, employing the non-linear least-squares Gaussian fitting routines within the spectrum analysis package DIPSO (Howarth et al. 1994). The results are given in Tables 4 and 5 for all lines in the spectra of the target stars. These tables are not reproduced here, but can be obtained electronically by ftp from the Centre de Données Stellaire, Strasbourg (http://cdsarc.u-strasbg.fr/ ) or from the Astrophysics & Planetary Science Division, Queen's University of Belfast, World Wide Web Server (http://star.pst.qub.ac.uk/ ). Also listed are the LTE abundances derived assuming the atmospheric parameters given in Table 3. Atomic data were as discussed in Jeffery (1996); however, the choice of atomic data is not critical for the differential abundance analysis.
Mean absolute abundances (on the logarithmic scale with hydrogen 12.0) and normal Population I abundance values (taken from Gies & Lambert 1992, Kilian 1994) are collected in Table 6. The effect of changing the atmospheric parameters by their error estimates (see Table 3) were considered. These affected the absolute abundance estimates by typically less than 0.2 dex, and always by less than 0.3 dex. Mean differential abundances [X/H] (where the square brackets denote the abundance of the species with respect to that in the corresponding Galactic comparison) are presented in Table 7. It should be noted that in the differential analyses, the atmospheric parameters of the relevant comparison star were derived using the same technique as for the Palomar-Green targets. Such an approach should minimize the effects of systematic errors, while errors in the adopted atomic data will also be less important.
Table 6. Mean absolute LTE abundances.
Table 7. Mean differential LTE abundances.
3.3. Evolutionary parameters
Stellar masses, luminosities and evolutionary ages () have been deduced from the evolutionary tracks of Claret & Gimenez (1992) using the derived atmospheric parameters listed in Table 3. Values of the visual bolometric correction have been extracted from the grids of Kurucz (1979). Given the stellar mass and bolometric luminosity, we obtained a distance estimate for individual objects using the apparent visual magnitude, the visual bolometric correction and an assumed Galactic reddening law, where (Woltjer 1975) and extinction =. Given the relatively large Galactic latitudes of our targets, their reddening was small and hence not a significant source of error. Stellar projected rotational velocities () were also estimated by convolving synthesised spectra with rotational broadening functions until they matched the observations. Details of these parameters are given in Table 8.
Table 8. Kinematic and derived evolutionary parameters.
3.4. Kinematical analysis
We have undertaken a kinematical investigation in order to determine whether these high-latitude, normal Population I objects could be `runaway' stars - born and subsequently ejected from the known sites of star formation in the Galactic disk. Reliable proper motions are not readily available for our targets; therefore, in the first instance we have used the observed radial velocity of a star to constrain its evolutionary history. Our method of analysis is outlined briefly below; the full details can be found in Rolleston et al. (1997) and references therein.
The kinematical analysis is summarized in Table 8. Stellar radial velocities () were measured using the Doppler shifts of all lines identified in the high-resolution optical spectra, relative to the Local Standard of Rest in the solar neighbourhood. These have been transformed to a standard of rest defined by a star's local environment (), by correcting for the effects of differential Galactic rotation (Fich et al. 1989) assuming that the halo co-rotates with the disk. As a first approach, we have considered the scenario whereby a star has a zero velocity component parallel to the Galactic disk, ie. its space motion is perpendicular to the plane of the Galaxy. Flight-times () have therefore been determined using the estimated stellar space velocity () perpendicular to the plane of the Galactic disk, the inferred z-distance and the gravitational potential function of House & Kilkenny (1980). Any subsequent comparison with evolutionary timescales implicitly assume that the stars were ejected from the disk shortly after gravitational collapse, consistent with cluster ejection simulations (Leonard 1993).
© European Southern Observatory (ESO) 1999
Online publication: June 18, 1999