Astron. Astrophys. 348, L25-L28 (1999)

4. Metal abundances

The metal lines are sufficiently isolated to derive abundances from their equivalent widths except for the crowded region from 4635 Å to 4660 Å which we analyse by detailed spectrum synthesis. Results are listed in Table 2 and plotted in Fig. 2. Although several O lines are available, it was impossible to determine the microturbulent velocity () in the usual way, i.e. by minimizing the slope in a plot of the O abundances versus equivalent widths. We adopted = 55 km/s which translates into small systematic abundance uncertainties of 0.05 dex for most ions. The analysis is done in LTE and we therefore used a model (= 33500 K, = 5.35) that is consistent with the He, N and Si ionization equilibria. A temperature uncertainty of =1000 K translates into abundance uncertainties of less than 0.1 dex. Hence systematic errors are smaller for most ions than the statistical errors given in Table 2.

Fig. 2. Abundances of PG 1605+072 relative to the sun. For Mg an uncertainty of 0.3 dex was assumed.

Table 2. Metal abundances for PG 1605+072 compared to solar composition. n is the number of spectral lines per ion.

Carbon and oxygen are depleted by 0.8-0.9 dex with respect to solar composition whereas nitrogen, magnesium and silicon are only slightly deficient (factor 2). Surprisingly neon and iron are solar to within error limits. This peculiar abundance pattern is probably due to diffusion, i.e. the interplay of gravitational settling and radiative levitation. The iron abundance is of special interest as an diffusive enrichment in subphotospheric layers is necessary to drive the pulsations. Its surface abundance is consistent with the diffusion calculations of Charpinet et al. (1997).

© European Southern Observatory (ESO) 1999

Online publication: July 26, 1999
helpdesk.link@springer.de