Astron. Astrophys. 326, 1069-1075 (1997)

4. Results

The results are detailed in Tables 5, 6 and 7 (available in electronic format from CDS), where the solar abundances, except for Fe, come from Anders and Grevesse (1989). The Fe abundance is derived from Johansson et al. (1994). These are shown graphically in Figs. 3, 4 and 5, which show the relative abundances of the RV Tau variables and their red-giant companions, where:

The filled circles denote elements for which there are at least two spectral lines measured from each star. The open circles, without error bars, are for those elements with only one spectral line in one or both of the stars. For comparison purposes, differential abundances are also displayed with the metal poor star BD+11 2998.

Fig. 3. Relative metal to iron abundances between the RV Tau star M2_V11 and a the red giant M2_HII_89, b the metal poor star BD+11 2998

Fig. 4. Relative metal to iron abundances between the RV Tau star M13_SH_11 and a the red giant M13_I_13, b the metal poor star BD+11 2998

Fig. 5. Relative metal to iron abundances between the RV Tau star M56_V6 and a the red giant M56_I_60, b the red giant M56_I_96, c the metal poor star BD+11 2998

The errors in the figures illustrate only the line to line variations. Errors due to uncertainties in the physical parameters are of the order of for most elements, although higher error of up to , are observed in some of the lighter elements. An illustration of possible modelling errors is presented in Table 4, for the particular case of M13_SH11. Notice how much more robust the metal-to-iron ratios are, compared with the metal-to-hydrogen ratios.

Table 4. Modelling errors from uncertainties in physical paremters

4.1. M2

The RV Tau variable M2_V11 is particularly enhanced in abundances of the odd-elements Na and Al, relative to the red giant M2_HII_89, as well as in the the -element Si. The elements Na and Al are diagnostics of H-burning at the high temperatures of the ON-cycle. These, together with the slight over abundances of some of the s-process elements - Y and Ba, may be an indication that normal third dredge-up has occurred in the RV Tau variable.

Since the Fe abundance is within 0.2dex of the red giant's, and Sc is of perfectly normal abundance, it seems that neither contention of Luck and Bond (1989) are supported. Comparison with the control star BD+11 2998 confirms all the results derived from the red giant differential analysis.

One possible alternative view is that RV Tau variables are post-AGB stars showing evidence for deposition on to grains, as was found for IW Car by Giridhar et al. (1994). Key abundance changes expected in this case would be a reduction in [Fe/H] by 1.0dex, a reduction in [Sc/H] and [Ca/H] by 2.0dex, and a slight enhancement in [Zn/H] by 0.2dex. None of these were found.

4.2. M13

The RV Tau variable M13_SH_11 shows much less scatter about the normal Fe abundance, relative to the red giant M13_I_13, than for M2_V11. There are four elements with obvious overabundances in M13_SH_11, the odd elements Na and Al (as for M2_V11), and Ba. The fact that Ba is again significantly high is interesting. Unfortunately, the abundances of Ba are derived from strong to very strong lines, and as such, are not dependable. At the very least, there is no evidence for any s-process element underabundance. Nor is there any underabundance observed for Sc. Since the Fe abundance is within 0.05dex of the M13_I_13 Fe abundance, again there is no support for either contention of Luck and Bond (1989), nor is there any s-process underabundance. In addition, there is no sign of deposition onto dust grains.

When comparisons are made with the control star BD+11 2998, we see that, although ZrI becomes severely underabundant, the neutron capture elements beyond Ba, are systematically enhanced. The enhancement in the heaviest elemental abundances was also observed in M2.

4.3. M56

The most puzzling star of the sample studied, was the RV Tau variable star M56_V6. Most lines of Fe were in agreement with a massive 1.7dex underabundance relative to the two red giants studied from M56. Excellent agreement was observed with this value and the Cr, Mn, and Ni abundances. However, four lines of Fe, and several other element abundances, including Sc, Ti and all the s-process elements, are vastly enhanced in abundance relative to the derived Fe abundance, and close to the abundances in the red giant companions (within 0.3dex).

The five Fe lines in question are those at 4930A; 5060A; 5247A; 5250A; and 5373A. The line at 4930A is particularly strong, indicating an abundance of 6.12, but the other four lines are closely grouped, with an average abundance of 5.41 0.05dex. The five Fe lines show no systematic property that may distinguish them from the majority of the lines. Two lines, including the one at 4930A, have a high excitation potential, while the other three have very low excitation potentials. The equivalent widths span the full range from 8mA to 96mA.

Is it possible that we are witnessing here the effects of grain deposition? The elements Fe, Cr and Ni are depleted equally, and are consistent with grain deposition. The critical element in this case, however, is Ti. According to Federman et al. (1993), Ti should be as depleted as Fe, or more so. The evidence remains, that Ti is much less depleted than Fe in M56_V6.

© European Southern Observatory (ESO) 1997

Online publication: April 8, 1998
helpdesk.link@springer.de