6 Discussion and conclusions

The optical spectrum of the post-AGB star HD 101584 is rather complex. We find several emission lines and P-Cygni profiles indicating an ongoing mass-loss and the presence of a circumstellar gaseous envelope. From the analysis of the absorption lines we find the atmospheric parameters to be $T_{\rm eff}=8500$ K, log g=1.5, $V_{\rm t}=13$ km s^-1 and [Fe/H] = 0.0.

Carbon and Nitrogen are found to be overabundant indicating that material processed by triple alpha C-N and O-N cycles has reached the surface. Since our blue spectra are of relatively low resolution and because of the presence of emission and shell components it is difficult to estimate reliable abundances of s-process elements. The OI line at 6156 Å is blended with a weak FeI emission line. The OI triplet at 7777 Å is very strong and affected by NLTE. In any case it appears that the oxygen abundance is nearly solar. A NLTE analysis of the high resolution OI 7777 Å triplet may yield a more reliable oxygen abundance.

The nitrogen abundance is based on 6 lines in the 7440 Å and 8710 Å region. We have not used the strong nitrogen lines. Nitrogen seems to be clearly overabundant. The carbon abundance is based on two CII lines at 6578 Å and 6582 Å. There is a clear indication that carbon is overabundant. The abundance of Mg, Ti, and Fe are nearly solar. The Ti abundance is based on 15 lines and the Fe abundance is based on 6 lines. Many of the other atomic lines are affected by emission and shell components. In our opinion, the line at 5876 Å might be due to CI (Hibbert et al. 1991) and not to HeI, as previously suggested by Bakker et al. (1996a). We have not found any other HeI, NII or OII lines. Our analysis shows that the $T_{\rm eff}$ is $8500\pm 500$ K.

  

Table 2: Chemical composition of HD 101584

Bakker et al. (1996b) found small amplitude light and velocity variations and suggested that HD 101584 is a binary with an orbital period of 218 days. The radial velocity variations may be due to pulsation, macroturbulence motions or shock waves in the outer layers of the stellar atmosphere. Many post-AGB supergiants show small amplitude light and velocity variations (Hirvnak 1997). These variations may not be interpreted as due to the presence of a binary companion. Long term monitoring of the radial velocities is needed in order to understand the causes for these variations.

The spectrum and the brightness of HD 101584 appears to remain the same during last two or three decades. There is no evidence for significant variations in brightness similar to those observed in Luminous Blue Variables (LBVs). The chemical composition and all the available multiwavelength observational data collected during the last two decades by various observers indicates that HD 101584 is most likely a post-AGB star.

The presence of several P-Cygni lines with significant outflow velocities, the OH maser and CO emission profiles (Te Lintel Hekkert et al. 1992; Trams et al. 1990) and the IRAS infrared fluxes and colours (Parthasarathy & Pottasch 1986) indicates the possibility that HD 101584 is a post-AGB star with a bipolar outflow with a dusty disk. Since HD 101584 shows a strong H$\alpha$ emission line, high resolution imaging with the Hubble Space Telescope (HST) may reveal the bipolar nebula and the presence of a dusty disk similar to that observed in other post-AGB stars like IRAS 17150-3224 (Kwok et al. 1998) or IRAS 17441-2411 (Su et al. 1998).