5 Results

In what follows, we present derived elemental abundances for individual stars.

5.1 HD 725

This star is an IRAS source (00091+5659). It was classified as F5Ib-II by Griffin & Redman (1960) suggesting a temperature of $\sim$6900 K if we follow the calibration of Schmidt-Kaler (1982).

Using the uvby photometry of Perry (1969), Olsen (1983), Hauck & Mermilliod (1998), reddening free colours and our own unpublished calibrations for F-G supergiant stars, we estimated $T_{\rm eff} = 6900$ K. Balmer line fitting was not performed because the profiles display complex structure and underlying emission is suspected.

 

 

Table 3: Elemental abundances for HD 725

Species	$\log \epsilon_{\odot}$	[X/H]	s.d.	N	[X/Fe]
C I	8.55	-0.38	$\pm$0.03	3	-0.08
Na I	6.32	+0.21	$\pm$0.08	3	+0.51
Mg I	7.58	-0.18	$\pm$0.25	4	+0.12
Si I	7.55	+0.05	$\pm$0.03	2	+0.35
Si II	7.55	+0.12		1	+0.42
S I	7.21	-0.10	$\pm$0.17	3	+0.20
Ca I	6.35	-0.16	$\pm$0.10	12	+0.14
Sc II	3.13	-0.03	$\pm$ 0.23	7	+0.27
Ti II	4.98	-0.31	$\pm$0.08	6	-0.02
Cr I	5.67	-0.16	$\pm$0.17	7	+0.14
Cr II	5.67	-0.19	$\pm$0.16	11	+0.11
Mn I	5.39	-0.27	$\pm$0.12	6	+0.03
Fe I	7.51	-0.26	$\pm$0.14	51
Fe II	7.51	-0.33	$\pm$0.15	11
Ni I	6.25	+0.04	$\pm$0.14	3	+0.33
Y II	2.23	+0.12	$\pm$0.15	3	+0.42
Zr II	2.60	-0.07	$\pm$0.13	2	+0.22
Ba II	2.13	+0.15		1	+0.45
Ce II	1.58	-0.17	$\pm$0.13	2	+0.13

Notes - The solar abundances are taken from Grevesse et al. (1996).

- N is the number of lines included in the calculation.

The finally adopted parameters are $T_{\rm eff} = 7000$ K, $\log g = 1.0$ and $\xi_{\rm t} = 4.65$ km s^-1. We relied upon Fe II lines for calculating microturbulence velocity. A large number of Fe I lines were measured and used in estimating $T_{\rm eff}$. Also neutral and ionized lines of Ti and Cr were employed to derive a satisfactory pair of $T_{\rm eff}$ and $\log g$ values. A competing solution could have been $T_{\rm eff} = 7200$ K, $\log g = 1.5$ but the adopted parameters gave marginally better consistency in the abundances of neutral and ionized lines. HD 725 appears to be very marginally metal-poor (Table 3) and it has a moderately high radial velocity of -57 km s^-1. The elemental abundances relative to Fe, i.e. [X/Fe] = [X/H]-[Fe/H] , given in the last column of Table 3 and subsequent tables, were calculated adopting the average value of [Fe/H] from Fe I and Fe II lines. For HD 725 the [X/Fe] values are similar to those in normal unevolved stars for many elements. [Na/Fe] = +0.5 dex indicates relative enrichment of Na which is a well-known feature of A-F supergiants (Takeda & Takada-Hidai 1994; Venn 1995a,b). But Na I abundances are likely to be affected by non-LTE effect. Non-LTE analysis of Na I lines has been done by Gigas (1986), Takeda & Takada-Hidai (1994) and others. The errors introduced by the neglect of non-LTE becomes more severe for higher temperatures. According to Takeda & Takada-Hidai (1994), the $\triangle$log $\epsilon$ in Na I lines at 5682, 5688, 6154 and 6160 Å is -0.09, -0.10, -0.07 and -0.07 dex respectively at temperature 7500 K. At temperature 7000 K the correction is 0.01 dex smaller for all lines than the values mentioned above. According to Gigas (1986) the non-LTE correction could be +0.1 to +0.2 dex. The use of [Na/Ca] instead of [Na/Fe] is recommended by Lambert (1992) for LTE calculations. The [Na/Ca] of +0.4 dex found from our analysis shows that Na enrichment appears to be real. The Na might have been synthesized in the H-burning region where the NeNa cycle might operate together with the CNO cycle. The suggestion that the proton capture on ²²Ne could lead to enhancement of ²³Na is followed up by Langer et al. (1993). These authors used a nuclear reaction network to examine the changes in abundances caused by proton capture at T₉ =0.040. Mixing of this region just below the oxygen shell over a timescale of 30000 yr would cause enhancement of ¹⁴N, ²³Na and ²⁷Al at the expense of ¹⁶O, ²²Ne, ²⁵Mg and ²⁶Mg. These authors also suggested that the abundant ²⁰Ne could also be transformed into ²³Na on longer timescales, and pointed out that the depletion of ²⁵Mg and ²⁶Mg would not modify the Mg abundance. Globular cluster giants are known to display Na-O anticorrelation as reported by Sneden et al. (1991) and Kraft et al. (1992). Since our spectrum does not go to wavelengths longer than 6800 Å, we could not measure N I abundance nor could we measure the O abundance. In metal-poor stars, relative enrichment of $\alpha$ elements is to be expected but the effect becomes evident for [Fe/H] < -0.5 dex. At [Fe/H] -0.2 to -0.3 dex the spread in observed abundances is large and a large number of stars have [ $\alpha/{\rm Fe}]\sim0$. For HD 725, the $\alpha$ element Si shows small enhancement but the number of lines used are woefully small to make a definite claim.

Among Fe-peak elements, Ni, represented by 3 Ni I lines, appears to be enriched, with [Ni/Fe] of +0.3 dex. HD 218753 and HD 173638 of our sample show a positive [Ni/Fe] but the value is not above abundance errors. As such, for HR 725 the value is slightly above twice the standard error, nevertheless a more extensive analysis based on a larger number of lines is required to see if the enrichment is real. Luck & Bond (1983, 1985) have reported [Ni/Fe]$\sim 0$ for metal-poor star. Wheeler et al. (1989) on the other hand report very large scatter in [Fe/H] vs. [Ni/Fe] relation. These authors suggest that non-zero [Ni/Fe] can be observed at all metallicities.

Another interesting finding is [Y/Fe] of +0.4 dex. The three Y II lines used are quite well separated and have good estimates of oscillator strengths, but out of the three lines, one is a little strong. The same is true for the Ba II line used. Mild enrichment of s-process elements in the moderately metal-poor star HD 70379 has already been reported (Reddy 1996). Hence, our average [s/Fe] ratio +0.3 does not come as a surprise. In addition, its radial velocity of -57 km s^-1 lends support to our view that it is a low-mass evolved object. These results are highly suggestive though not conclusive indicators of evolution beyond the red giant branch. However, making use of its proper motions and parallax from the Hipparcos catalogue we have calculated the galactocentric velocities $\Pi = -32.6$ km s^-1, $\Theta = +190.3$ km s^-1 and Z = +5.4 km s^-1 that indicate a space velocity of 193.2 km s^-1with a small pitch angle of 9.7$^{\rm o}$ relative to the circular orbit and on the galactic plane. This indicates that the star is at a lower galactic latitude than the Sun and placed in a mildly eccentric orbit.

5.2 HD 9167

This star has infrared flux and is an IRAS source (01285+6115). It appears to be another moderately metal-poor star (Table 4). We did not find any remarkable abundance peculiarity for this object. With radial velocity of -45 km s^-1 and [Fe/H] of -0.3 dex, one could be optimistic of seeing positive [$\alpha$/Fe]. Unfortunately, we could not measure good Si I and Si II lines and Mg, Ca and Ti do not show any enrichment. But on the other hand, for stars with [Fe/H] in the range of 0.0 to -0.5 dex, the observed abundance ratios of $\alpha$-process elements have very large scatter, hence finding evolutionary changes is almost as probable as not finding them. Likewise HD 725, rather large radial velocity implies a space velocity of 208.2 km s^-1with a small pitch angle of 5.7$^{\rm o}$ relative to the circular orbit and on the galactic plane. The orbit is even less eccentric than that of HD 725 and it is at a lower galactic latitude than the Sun.

 

 

Table 4: Elemental abundances for HD 9167

Species	$\log \epsilon_{\odot}$	[X/H]	s.d.	N	[X/Fe]
Mg I	7.58	-0.37	$\pm$0.15	2	-0.04
Ca I	6.35	-0.22	$\pm$0.09	6	+0.12
Sc II	3.13	-0.28	$\pm$ 0.09	3	+0.05
Ti II	4.98	-0.48	$\pm$0.17	10	-0.15
Cr I	5.67	-0.19		1	+0.15
Cr II	5.67	-0.27	$\pm$0.21	9	+0.07
Mn I	5.39	-0.35		1	-0.02
Fe I	7.51	-0.32	$\pm$0.16	29
Fe II	7.51	-0.35	$\pm$0.21	10
Y II	2.23	-0.23	$\pm$0.27	3	+0.11
Ba II	2.13	-0.22	$\pm$0.09	2	+0.12

Notes - same as Table 3.

5.3 HD 158616

This star is an IRAS source (17279-1119) with significant infrared fluxes at shorter wavelengths.

The spectral type as listed in the SAO catalogue is F8, which suggests a temperature between 6100-6200 K (Schmidt-Kaler 1982), depending upon the luminosity class. No photometric data are available and hence no other temperature estimate was made. Two spectra were obtained for this star at the OHP. The S/N for these two spectra are 34 and 53. Since the spectrum from July 7 is better, it was decided to use it to determine $T_{\rm eff}$, $\log g$ and $\xi$$_{\rm t}$, complete the abundance analysis and then simply use the same parameters on the lower S/N spectrum from July 6 to verify the abundance pattern. The results are given in Table 5. The second entries for each species are for the lower S/N spectrum. One can see that the results from both spectra are in a good agreement.

 

 

Table 5: Elemental abundances for HD 158616

Species	$\log \epsilon_{\odot}$	[X/H]	s.d.	N	[X/Fe]
C I	8.55	-0.25	$\pm$0.20	4	+0.33
		-0.23	$\pm$0.14	4	+0.35
O I	8.87	-0.54	$\pm$0.02	2	+0.04
		-0.55		1	+0.03
Na I	6.32	+0.05	$\pm$0.01	3	+0.63
		-0.05		1	+0.49
Mg I	7.58	-0.40	$\pm$0.11	3	+0.18
		-0.56		1	-0.02
Si I	7.55	+0.03	$\pm$0.17	3	+0.61
Si II	7.55	-0.06	$\pm$0.17	2	+0.52
		+0.04	$\pm$0.47	2	+0.58
S I	7.21	+0.08	$\pm$0.12	6	+0.66
		+0.10	$\pm$0.01	2	+0.64
Ca I	6.35	-0.36	$\pm$0.16	12	+0.22
		-0.34	$\pm$0.31	9	+0.20
Sc II	3.13	+0.00	$\pm$ 0.27	8	+0.58
		-0.09	$\pm$0.24	4	+0.45
Ti I	4.98	+0.08	$\pm$ 0.20	2	+0.66
		+0.18		1	+0.72
Ti II	4.98	-0.01	$\pm$0.19	13	+0.57
		-0.13	$\pm$0.11	8	+0.41
Cr I	5.67	-0.55		1	+0.03
Cr II	5.67	-0.39	$\pm$0.14	14	+0.19
		-0.47	$\pm$0.22	7	+0.07
Mn I	5.39	-0.17		1	+0.41
Fe I	7.51	-0.58	$\pm$0.15	39
		-0.61	$\pm$0.18	20
Fe II	7.51	-0.57	$\pm$0.19	19
		-0.51	$\pm$0.19	7
Ni I	6.25	-0.41	$\pm$0.08	4	+0.17
		-0.39		1	+0.19
Zn I	4.60	-0.48	$\pm$0.02	2	+0.10
		-0.40		1	+0.18
Y II	2.23	+0.37	$\pm$0.25	4	+0.95
		+0.48	$\pm$0.03	2	+1.02
Ba II	2.13	+0.04	$\pm$0.22	3	+0.62
		+0.10	$\pm$0.20	2	+0.64
La II	1.21	-0.11		1	+0.47
Ce II	1.58	+0.13	$\pm$0.22	5	+0.71
		+0.21		1	+0.75

Notes - same as Table 3.
- Second entries are results from a lower S/N spectrum.

HD 158616 was also studied by Van Winckel (1995, 1997). While he had better data for C, N, O due to extended coverage in the long wavelength region, for other elements, our spectra contained more lines per element, and more elements are included. This star is Fe-poor by a factor of 4 or so and shows very clear indications of CNO processing and enrichment of s-process elements. We get ${\rm C/O}\sim 1$ whereas Van Winckel (1995) got this ratio significantly larger than one (see Table 14). His carbon abundance is based on more lines in the 7100 Å region. His choice of $T_{\rm eff}$ is also hotter than our adopted value that can also account for the higher carbon abundance derived. Van Winckel (1995) found [N/Fe] of +0.2 dex. This clearly shows that the star has gone through CNO cycle and its products have been brought to the surface. The carbon enrichment could be caused by the helium-shell burning when oxygen is also manufactured. The enhancement of Si and S could be present in ISM from which the star is formed. With the [Fe/H] of -0.6 dex, the star is moderately metal-poor and therefore it is possible that the parent ISM might have received ejecta from type II SNe. Although the effect of departure from LTE has been discussed for the light elements C, N, O and also for Mg by different investigators, we did not come across similar discussion on Si and S. Hence, at this stage, we chose not to offer any detailed explanation. The most interesting feature of the derived abundances is definitely the enrichment of s-process elements Y, Ba, La and Ce. This star is undoubtedly a post-AGB star that has brought the products of helium burning as well as elements formed by s-processing to the surface. Our analysis covers the light s-process element Y (one of the three elements referred as ls) and the heavy s-process elements Ba, La and Ce, representing heavy (hs) s-process elements. We have made an estimate of [hs/ls]. We have not corrected the ls-index for the lack of light elements Sr and Zr since for light elements the odd-even effect is not strong as pointed out by Van Winckel & Reyniers 2000). We estimated [hs/ls] = -0.3 and [ls/Fe] of +1.0. When plotted on the [hs/ls] vs. [ls/Fe] plot of Busso et al. (1999), (their Fig. 7 giving theoretical predictions), we found $\tau_{0} \sim 0.22$ mbarn^-1. Incidentally, the data point falls near the thick line which for solar metallicity would indicate C/O = 1 but for [Fe/H] of -0.6, it indicates a C/O $\sim$ 3, whereas we get C/O $\sim$ 1. This reduction of carbon abundance, while hs and ls indices point to larger C/O added to small but significant enhancement of nitrogen, as reported by Van Winckel (1995), strongly favour the hot bottom burning scenario described and discussed in Sect. 5.5. More extensive coverage of s-process elements will enable a meaningful comparison with third dredge-up models developed by Straniero et al. (1995) and Busso et al. (1995).

Another star of similar temperature and showing a similar trend in abundances is HR 6144 (Luck et al. 1990), however, in this star, the s-process enhancement is not so significant and C/O is less than 1.

Figure 1: H$_\alpha$ and H$_\beta$ variations exhibited by HD 172324 in two spectra taken 4 years apart: June 1995 at McDonald and July 1999 at OHP

5.4 HD 172324

This star has been classified as B9Ib by Morgan & Roman (1950) which suggests a temperature of 10280 K (Schmidt-Kaler 1982). The Strömgren colours (Hauck & Mermilliod 1998) however point towards higher temperature: 13100 K using the calibration of Napiwotzki et al. (1993). An independent temperature calibration of the Geneva photometric system (Cramer & Maeder 1979) gives a temperature of 13215 K.

The He I lines have been used to estimate $T_{\rm eff}$ and luminosity class of the star. Didelon (1982) gives very useful plots of the dependence of many He I, Si II, Mg II, C II, O II and N II line strengths on the spectral type and luminosity class. The equivalent widths in our spectrum for the He I lines at 4120, 4143, 4387 and 4471 Å very clearly suggest a spectral type of B9. A luminosity class Ib was suggested by the strengths of Si II line at 4128 Å, Mg II line at 4481 Å and C II feature at 4267 Å, this is in good agreement with the spectral type above.

An estimate of $T_{\rm eff}$ has been made by requiring the He I lines to give solar abundance. The best estimate is 11500 K. Given the class Ib, $\log g$ must not be very different from 2.0. In any case, at the high temperature end the Kurucz (1993) models do not reach very low gravities. Fortunately, Mg I and Mg II lines are present in the OHP spectrum and Si II and Si III lines can be measured on the McDonald spectrum. Our derived $T_{\rm eff}$ and $\log g$ appear to be good estimates for the epoch of OHP spectrum. However there is indication that at the epoch at which the McDonald spectrum was taken, the gravity was somewhat lower.

 

 

Table 6: Elemental abundances for HD 172324

Species	$\log \epsilon_{\odot}$	[X/H]	s.d.	N	[X/Fe]
He I	10.99	-0.08	$\pm$0.35	4	+0.54
He I*	10.99	-0.19	$\pm$0.35	12	+0.44
C II*	8.55	-1.30		syn	-0.68
O I	8.87	+0.41	$\pm$0.20	2	+1.03
O I*	8.87	+0.26	$\pm$0.19	4	+0.89
Ne I*	8.09	+0.05		1	+0.68
Mg I*	7.58	-0.63		1	0.00
Mg II	7.58	-0.66	$\pm$0.17	2	-0.04
Mg II*	7.58	-0.65	$\pm$0.12	3	-0.02
Al II	6.47	-0.59		1	+0.03
Al II*	6.47	-0.65		1	-0.02
Si II	7.55	-0.16	$\pm$0.03	2	+0.47
Si II*	7.55	-0.29	$\pm$0.14	6	+0.39
Si III	7.55	+0.25		1	+0.88
S II	7.21	-0.19		1	+0.43
S II*	7.21	-0.47		1	+0.16
Ti II	4.98	-0.03	$\pm$0.24	8	+0.59
Ti II*	4.98	-0.28	$\pm$0.33	3	+0.35
Cr II	5.67	-0.43	$\pm$0.19	11	+0.19
Cr II*	5.67	-0.48	$\pm$0.28	3	+0.15
Fe II	7.51	-0.62	$\pm$0.17	17
Fe II*	7.51	-0.63	$\pm$0.11	11

Notes - same as Table 3.

- $T_{\rm eff}$ = 11000 K, $\log g$ = 2.5 and $\xi_{\rm t}$ = 7.00 km s^-1 for McD spectrum.

- $T_{\rm eff}$ = 11500 K, $\log g$ = 2.5 and $\xi_{\rm t}$ = 5.40 km s^-1 for OHP spectrum.

* Result from OHP spectrum.

This star appears to be deficient in Fe by a factor of 3-4 and has a large radial velocity of -110 km s^-1 making it a likely halo or old disk object. Carbon abundance is very important in ascertaining the evolutionary status. We could observe only the blend at 4267 Å while supposedly strong lines at 6578 and 6582 Å were too weak to be measured. By computing the blend of two C II lines at 4267 Å we find [C/H] of -1.3 dex, for which Takeda et al. (1996) found zero non-LTE correction. In the neighbourhood of $T_{\rm eff}$ 10000 K the non-LTE abundance correction for carbon using C I lines is $\sim$-0.4 dex (Venn 1995b). Our iron abundance is based on Fe II lines that are not strongly affected by non-LTE effects. For O I line at 6156-6158 Å the non-LTE correction is $\sim$-0.3 dex (Takeda & Takada-Hidai 1998).

For A-type supergiants of solar metallicity, Venn (1995a) found a mean enrichment of $\sim$0.6 dex for Na and $\sim$0.3 dex for S. Venn (1995b) reported CNO abundances for the same sample after applying non-LTE corrections to the derived abundances. The mean values found were: [C/H] = -0.4 dex, [N/H] = +0.08 dex and [O/H] = -0.2 dex. Nevertheless, the hottest star of Venn's sample, HD 161695, is about 1500 K cooler than HD 172324.

McErlean et al. (1999) have plotted LTE and non-LTE line profiles for a range of temperatures 10000 K to 35000 K. At 11000 K the non-LTE correction for C II lines at 4267 Å is very small, in agreement with the prediction of Takeda et al. (1996). Hence our values [C/H] = -1.3 dex and [C/Fe] = -0.68 dex are realistic estimates.

For an A-type star of solar metallicity, a mean value [C/H] = -0.4 was reported by Venn (1995b), whereas for B supergiants McErlean et al. (1999) report mean values of [C/H] = -0.35 dex and [O/H] = -0.32 dex. After applying non-LTE correction for oxygen we get [O/H] of +0.1 dex and [O/Fe] of +0.7 dex. Our derived carbon and oxygen abundances in Table 6 are much different from what is expected for a young B9 supergiant. It has been suggested by Boothroyd & Sackmann (1999) that the carbon deficiency can also be caused by deep circulating mixing below the base of the convective envelope followed by cool bottom processing (CBP) of the CNO isotopes. They also showed that the CBP became more extensive at reduced metallicities or at low masses. The high radial velocity for HD 172324 (more than 100 kms^-1) and significantly low [Fe/H] makes it very likely that it has experienced CBP.

The only trace of an odd Z element we could observe was a single line of Al I, providing [Al/Fe] $\sim$ 0. One does not see [Al/Fe] $\sim$ +0.30 dex reported by Edvardsson et al. (1993), but with single line being used in our study not much significance could be attached to our estimated [Al/Fe].

Hot stars at high galactic latitude were studied by Conlon et al. (1993a) who found very large Fe and C deficiencies in them. But the stars studied by these authors are much hotter than HD 172324. Neverthless, the abundance pattern of HD 172324 is similar to those studied by Conlon et al. (1993a,b) and McCausland et al. (1992). This star being most likely an oxygen-rich post-AGB star deserves an extended analysis covering CNO and as significant spectral variations can be seen in four years (see Fig. 1), a monitoring of hydrogen line profiles over long time scales will be of interest.

5.5 HD 172481

This star shows light variations of very small amplitude, $\pm$0.15 mag (Van Winckel 1995). The spectral energy distribution of this star has an unusual multi-peaked shape (Bogart 1994). It was included in the study of Van Winckel (1995) who, by fitting the observed flux distribution to that of Kurucz (1993) model atmosphere, derived $T_{\rm eff}$ = 7000 K. Van Winckel also found that the strength of emission components in hydrogen lines varied strongly. From the study of radial velocity variation spread over a decade, he reported that there is not enough evidence for the binarity of the star. However, his spectra show considerable line splitting attributed to the passage of a shock. Fortunately, we obtained the spectrum of HD 172481 when the atmosphere of the star was stable and therefore no line splitting was observed, as can be seen in the Fig. 2.

 

 

Table 7: Elemental abundances for HD 172481

Species	$\log \epsilon_{\odot}$	[X/H]	s.d.	N	[X/Fe]
Li I	1.16	+2.54:		1	+3.16:
C I	8.55	-0.62	$\pm$0.21	12	-0.01
N I	7.97	-0.63	$\pm$0.10	3	-0.02
O I	8.87	-0.58	$\pm$0.05	2	+0.04
Mg I	7.58	-0.13	$\pm$0.21	2	+0.48
Mg II	7.58	-0.06	$\pm$0.54	3	+0.55
Si I	7.55	-0.05	$\pm$0.17	6	+0.57
Si II	7.55	-0.10	$\pm$0.06	3	+0.52
S I	7.21	-0.04	$\pm$0.16	7	+0.58
K I	5.12	-0.28		1	+0.34
Ca I	6.35	-0.28	$\pm$0.19	12	+0.34
Sc II	3.13	-0.20	$\pm$ 0.23	6	+0.41
Ti I	4.98	-0.31		1	+0.31
Ti II	4.98	-0.28	$\pm$0.30	6	+0.34
V II	4.00	-0.00	$\pm$0.14	2	+0.62
Cr I	5.67	-0.34	$\pm$0.05	3	+0.28
Cr II	5.67	-0.41	$\pm$0.16	10	+0.21
Mn I	5.39	-0.51	$\pm$0.11	3	+0.11
Fe I	7.51	-0.62	$\pm$0.15	43
Fe II	7.51	-0.61	$\pm$0.14	13
Ni I	6.25	-0.31	$\pm$0.22	6	+0.30
Zn I	4.60	-0.23	$\pm$0.08	2	+0.39
Y II	2.23	+0.08	$\pm$0.26	4	+0.69
Zr II	2.60	-0.14		1	+0.46
Ba II	2.13	+0.03	$\pm$0.20	2	+0.65
La II	1.21	-0.55	$\pm$0.14	2	+0.07
Ce II	1.55	-0.24	$\pm$0.19	7	+0.38
Nd II	1.50	+0.03	$\pm$0.09	3	+0.64
Eu II	0.51	+0.24		1	+0.86

Note - same as Table 3.

Sadly, the S/N ratio of our spectrum is not very high. We could measure very few lines of important elements like C and O, hence we were not satisfied with the limited data to carry out abundance analysis. But it was known beyond doubt that the star is highly evolved since we could identify lines of C I and also saw Li I line at 6708 Å. However, the Li I feature was falling at the end of our CCD frame and was showing distinct doubling. One component coming nearest to Li I wavelength gave log $\epsilon$(Li) of 1.8. But S/N being very poor at the end of the CCD frame, we decided to follow this object with more spectra. At our request, David Yong of McDonald Observatory got a spectrum using the 2d coudé echelle spectrograph at the 2.7 m telescope of McDonald Observatory on May 13, 2000. The spectrum has resolution of 30000 and wide spectral coverage from 3900 Å to 10200 Å. In this spectrum, we found Li I feature to be single with some asymmetry in the blue wing but much deeper than in July 1999. In all HD 172481 spectra, the H$_\gamma$ has a broad absorption, a narrow absorption and a possible red shifted emission component. The H$_\beta$ also has a broad absorption and a narrow absorption at the centre of broad absorption. There is no indication of emission component. The H$_\alpha$ has a complex profile with one shallow absorption, one deep absorption that has emission components in both the wings. The blue emission component is stronger than the red one.

The extensive spectral coverage of McDonald spectrum enabled us to measure a large number of unblended lines for several light and heavy elements. As one can see from the Table 7, we could carry out a very comprehensive analysis of this object.

Figure 2: Three spectral regions in the spectrum of HD 172481 clearly showing no line splitting. Hence we believe the star was in a stable phase

From Fe I and Fe II lines we derived $T_{\rm eff}$ = 7250 K and $\log g$ = 1.5 and was supported by Si I, Si II, Mg I and Mg II lines. We derived the Li abundance by spectrum synthesis. All the four components of Li I feature at the 6708 Å region were included. We find log $\epsilon$(Li) of +3.7, which is surprisingly high. We were puzzled by the change in appearance of Li I feature at 6707, so on August 11 one more spectrum was obtained with the same set-up and by the same observer. The spectrum again showed a suggestion of doubling at the line core though the components were not well separated. It is not clear if the line is splitting periodically or the emission at the core is causing it to appear double. The overall strength of Li I feature in the August spectrum has reduced. For May 2000 spectrum, our estimated Li abundance (derived by synthesis) is shown in Table 7 with a colon. In the light of the large variation exhibited by Li I feature in strength as well as in profile shape, this value should be regarded with caution. However, the presence of Li I feature in the spectrum cannot be refuted. Another Li I feature at 6103 Å fell in between the echelle orders and therefore could not be used.

We could measure a large number of C I lines to estimate the carbon abundance. The three N I lines in the near infrared also enabled us to derive the nitrogen abundance. We get [N/Fe]$\sim 0$ for HD 172481 whereas for HD 158616 [N/Fe]= +0.3 is reported by Van Winckel (1995). The estimated C/O is +0.43. The element Li is considered a fragile element that gets destroyed in the course of evolution. The primordial abundance of Li (based on population II objects) is considered near log $\epsilon$(Li) = 2.2. The excess abundance must therefore be caused by AGB evolution of HD 172481 or might owe its origin to binarity.

Our analysis covers light s-process elements (ls) Y and Zr and heavy s-process elements (hs) Ba, La, Ce and Nd. We derive a mean [ls/Fe] of 0.6 and mean [hs/Fe] of 0.4. That leads to [hs/ls] of -0.2. With this value of [hs/ls] and [Fe/H] of -0.6 this star is remarkably similar to IRAS 04296+3429 and IRAS 19500-1709 studied by Van Winckel & Reyniers (2000), although [ls/Fe] and [hs/Fe] are much larger for these two stars.

The stars mentioned above and HD 158616 are very interesting post-AGB objects as they are, most likely, evolved from intermediate-mass stars (IMS). According to Travaglio et al. (1999), stars in the mass range 4-8 $M_{\odot }$, activate ²²Ne($\alpha$, n)²⁵Mg reaction during their TP-AGB phase more effectively than the low-mass stars due to the high temperatures reached at the bottom of the convective pulse ($T_{\max}$ $\leq$ 3.5 10⁸ K). IMS contribute more to the first s-peak represented by the elements Sr, Y and Zr. Actually, the number of known post-AGB IMS is very small. For IMS, the formation of a ¹³C pocket is less certain because of the reduced mass of the H-He intershell by about one order of magnitude. It is shown by Straniero et al. (1997) and Gallino et al. (1998) that the ¹³C neutron source active during the interpulse phase of low-mass TP-AGB accounts for most of the production of the second s-peak elements Ba, La, Ce, Sm and Eu. It is obvious from the relative enhancement of Y and Sr presented in Tables 5 and 7 that HD 158616 and HD 172481 belong to the relatively rare post-AGB IMS. We would therefore try to understand the observed abundances of HD 172481 in the framework of AGB models developed by Lattanzio and others for IMS.

Lattanzio (1997) in his AGB calculation has predicted that for stars more massive than 4 $M_{\odot }$ the bottom of the convective envelope penetrates into the hotter regions of the envelope where proton captures already modified the original CNO composition. This is called "hot bottom burning'' (HBB) and results in many important changes in chemical composition of the envelope. Lattanzio (1997) predicted the production of Li by the Cameron-Fowler mechanism operating at bottom of the envelope. Sackmann & Boothroyd (1992) showed that log $\epsilon$(Li) $\sim$ 4.5 could be produced in stars with bolometric magnitude between -6 and -7 when the temperature at the base of the convective envelope exceeds 50 10⁶ K. From the study of AGB stars in SMC and LMC, Smith et al. (1995) found them to have bolometric magnitudes in the range -6 to -7.2 and to show lithium values of log $\epsilon$(Li) $\sim$ 1.0 - 4.0. However, these are cool luminous S stars. Lattanzio (1997) also predicted the destruction of carbon via CN cycle that could prevent C/O ratio from exceeding one. Theoretical models for AGB stars of mass 4, 5 and 6 $M_{\odot }$ computed by Boothroyd et al. (1993) encountered HBB with a temperature at the base of the convective envelope reaching 80 10⁶ K. These models predict C/O of 0.4 to 0.5 for $\sim$10³ yr on the AGB.

 

 

Table 8: Elemental abundances for HD 173638

Species	$\log \epsilon_{\odot}$	[X/H]	s.d.	N	[X/Fe]
C I	8.55	-0.16	$\pm$0.13	5	-0.09
Mg I	7.58	-0.05	$\pm$0.12	5	+0.03
Si I	7.55	+0.36	$\pm$0.12	2	+0.44
Si II	7.55	+0.41		1	+0.49
Ca I	6.35	+0.04	$\pm$0.25	12	+0.12
Ca II	6.35	-0.03		1	+0.05
Sc II	3.13	+0.16	$\pm$ 0.18	6	+0.24
Ti I	4.98	+0.01		1	+0.09
Ti II	4.98	-0.1	$\pm$0.18	17	-0.04
Cr I	5.67	+0.08	$\pm$0.07	3	+0.16
Cr II	5.67	-0.02	$\pm$0.18	15	+0.06
Mn I	5.39	+0.00		1	-0.08
Fe I	7.51	-0.07	$\pm$0.13	62
Fe II	7.51	-0.08	$\pm$0.14	19
Ni I	6.25	+0.06	$\pm$0.17	7	+0.14
Zn I	4.60	-0.06	$\pm$0.05	2	+0.02
Y II	2.23	+0.02	$\pm$0.22	5	+0.10
Zr II	2.60	-0.02		1	+0.06
Ba II	2.13	+0.15	$\pm$0.04	2	+0.23
Ce II	1.58	+0.02		1	+0.10
Nd II	1.48	+0.11		1	+0.19

Note - same as Table 3.

The abundance pattern of HD 172481 bears some resemblance to those found for HR 7671 though the latter is more metal-poor. Interestingly, HR 7671 also shows the Li I feature though Li is not as overabundant as in HD 172481. The estimated C/O for HR 7671 is 0.4. The mean [$\alpha$/Fe] is lesser than what we find for HD 172481 but [s/Fe] is comparable.

The existence of objects like HD 172481 and HR 7671 lend further support to the HBB scenario, put forward to explain the paucity of carbon-rich stars among AGB stars. The observed variation of Li I feature in HD 172481 makes it a very promising candidate for binarity search.

While the present paper was under the refereeing process, the referee called our attention to the then unpublished work on HD 172481 by Reyniers & Van Winckel (2001). Thus a comparison with their results is most appropriate. These authors have derived atmospheric parameters ( $T_{\rm eff}= 7250$ K and $\log g$ = 1.5) that are in excellent agreement with those derived by us. A small difference in $\xi_{\rm t}$ of 0.6 km s^-1 is seen. These authors also find a large lithium abundance similar to our finding. The [Fe/H] and other Fe-peak elements have very good agreement. The s-process elements La, Nd and Eu however, show some disagreement. We have measured 3 lines of Nd II so we are surprised by the lower limits placed by these authors. For Eu, we measured a fairly clean line at 6645 Å whereas Reyniers & Van Winckel used spectrum synthesis. It should be noted that the oscillator strengths used by these authors are quite different from those employed by us, that can possibly explain the abundance differences for the s-process elements. The gf values for the lines of these elements are known to have larger uncertainties compared to those of the Fe-peak elements. Another important finding by Reyniers & Van Winckel is the detection of a red luminuous companion, found from the presence of TiO bands in the red and also from the observed spectral energy distribution. As mentioned above, the presence of the companion might explain the Li I feature variations observed by us.

5.6 HD 173638

This star (HR 7055) was observed in the 13-Colour photometric system and its temperature was calculated by various approaches by Bravo Alfaro et al. (1997). The photometric temperature estimates range between 7486 K and 8100 K. The spectrum being of S/N = 95, we could measure a large number of lines covering many important elements like C, $\alpha$-process elements and Fe-peak elements. From the ionisation equilibrium of Fe I/Fe II, Cr I/Cr II, Ti I/Ti II, Si I/Si II and even Ca I/Ca II the atmospheric parameters are well-determined and listed in Table 2.

The star appears to have near-solar abundances for most elements except Si which is overabundant by a factor of 3 (Table 8). With its small radial velocity (+11 km s^-1), it is most likely a young star belonging to the disk population. But it can serve as excellent calibrator of photometric indices in the 7500 K temperature range.

5.7 HD 218753

This star was classified as A5III by Cowley et al. (1969). These and the Sp.T. - Colour - $T_{\rm eff}$ calibration of Schmidt-Kaler (1982) give a temperature of 8091 K.

Figure 3: H$_\gamma$, H$_\delta$, Mg and Si loci for HD 218753

 

 

Table 9: Elemental abundances for HD 218753

Species	$\log \epsilon_{\odot}$	[X/H]	s.d.	N	[X/Fe]
C I	8.55	-0.49	$\pm$0.18	5	-0.30
O I	8.87	-0.14		1	+0.05
Na I	6.32	+0.32	$\pm$0.07	2	+0.51
Mg I	7.58	-0.27	$\pm$0.16	5	-0.08
Mg II	7.58	-0.29	$\pm$0.15	4	-0.10
Si I	7.55	+0.11	$\pm$0.15	3	+0.20
Si II	7.55	+0.00	$\pm$0.22	4	+0.26
S I	7.21	+0.17	$\pm$0.09	2	+0.36
Ca I	6.35	-0.08	$\pm$0.15	15	+0.11
Ca II	6.35	-0.09		1	+0.10
Sc II	3.13	+0.00	$\pm$ 0.27	8	-0.19
Ti I	4.98	-0.27		1	-0.08
Ti II	4.98	-0.25	$\pm$0.17	19	-0.06
Cr I	5.67	-0.01	$\pm$0.38	3	+0.18
Cr II	5.67	-0.16	$\pm$0.13	14	+0.03
Mn I	5.39	-0.23	$\pm$0.06	3	-0.04
Fe I	7.51	-0.20	$\pm$0.16	82
Fe II	7.51	-0.18	$\pm$0.13	30
Ni I	6.25	-0.06	$\pm$0.15	9	+0.13
Zn I	4.60	+0.18		1	+0.37
Y II	2.23	-0.24	$\pm$0.21	7	-0.05
Zr II	2.60	+0.18		1	+0.37
Ba II	2.13	-0.08	$\pm$0.16	2	+0.11

Notes - same as Table 3.

Photometric data in the Strömgren's system can also be used to estimate $T_{\rm eff}$, through the $\beta -(b-y)$ relation of Crawford (1979) (his Table 1) and adopting (b-y) = 0.236 and $\beta = 2.773$ (Hauck & Mermilliod 1998). This leads to (b-y)₀ = 0.176 and E(b-y) = 0.054. While (b-y)₀ leads to (B-V)₀ = 0.20-0.25 (Crawford 1970), this implies $T_{\rm eff} = 7800$ K.

Also, if the above colour excess and photometry are used in combination of Napiwotzki et al. (1993) calibration we find $T_{\rm eff} = 7200$ K. However, it must be emphasized that the calibration has been computed using only stars of luminosity classes V and IV, while HD 218753 could be of luminosity class III or II.

Balmer line fitting can also be used to estimate $T_{\rm eff}$. Theoretical profiles have been extracted from Kurucz's (1993) models for H$_\alpha$, H$_\beta$, H$_\gamma$ and H$_\delta$. The observed profiles were fitted with theoretical profiles of given $T_{\rm eff}$ and $\log g$. The solution is not unique but in fact the best fits define a locus on the $T_{\rm eff}$-$\log g$ plane for each Balmer line. As underlying core emission may be present in some stars, we chose to fit the wings H$_\gamma$ and H$_\delta$. Following the above fitting process we found the loci for H$_\gamma$ and H$_\delta$ profiles shown in Fig. 3.

In hot stars, magnesium and silicon lines can be used as $T_{\rm eff}$ indicators (since Fe I lines are strongly affected by non-LTE effects). Given a pair ( $T_{\rm eff}$, $\log g$) one searches for abundance consistency between neutral and ionized lines of Mg and Si. Again, the solution is not unique but rather a locus on the $T_{\rm eff}$-$\log g$ plane is defined for each element, as illustrated in Fig. 3. The intersections of the Mg and Si loci with the H$_\gamma$ and H$_\delta$ loci point to the proper temperature and gravity. In this fashion we estimated $T_{\rm eff}$ between 7600 and 7900 K and $\log g$ between 1.75 and 2.0.

The turbulent velocity, $\xi_{\rm t}$, was estimated from Fe II lines by requiring abundance to be independent of line strength. We found $\xi_{\rm t} = 3.3$ km s^-1.

 

 

Table 10: Elemental abundances for HD 331319

Species	$\log \epsilon_{\odot}$	[X/H]	s.d.	N	[X/Fe]
C I	8.55	-0.33	$\pm$0.30	5	-0.09
O I	8.87	+0.06	$\pm$0.02	2	+0.30
Na I	6.32	-0.03		1	+0.21
Mg I	7.58	-0.32	$\pm$0.21	3	-0.08
Mg II	7.58	-0.18	$\pm$0.12	3	+0.06
Al I	6.47	+0.07		1	+0.31
Si II	7.55	-0.18	$\pm$0.20	2	+0.06
S I	7.21	+0.29	$\pm$0.18	5	+0.53
Ca I	6.35	-0.29	$\pm$0.18	10	-0.05
Sc II	3.13	-0.19	$\pm$ 0.22	5	+0.05
Ti II	4.98	-0.33	$\pm$0.22	22	-0.09
V II	4.01	+0.19	$\pm$0.16	3	+0.43
Cr I	5.67	-0.01	$\pm$0.29	3	+0.23
Cr II	5.67	+0.02	$\pm$0.14	18	+0.22
Mn I	5.39	-0.17		1	+0.07
Fe I	7.51	-0.27	$\pm$0.16	48
Fe II	7.51	-0.20	$\pm$0.15	21
Ni I	6.25	+0.00	$\pm$0.22	4	-0.24
Ni II	6.25	-0.13		1	+0.11
Sr II	2.90	+0.01		1	+0.25
Y II	2.23	-0.58	$\pm$0.09	3	-0.34
Ba II	2.13	-0.51	$\pm$0.16	3	-0.27

Notes - same as Table 3.

For HD 218753 we finally adopted $T_{\rm eff}$ = 8000 K, $\log g$ = 2.0 dex and $\xi_{\rm t}$ = 3.3 km s^-1. With these parameters the abundances for the rest of the detected species were calculated and the results are given in Table 9. For this star our spectrum has S/N = 67 and we could measure a large number of clean weak lines for most important elements. HD 218753 shows a significant carbon deficiency most likely caused by CNO processing. There is also a significant enrichment of sodium that has been found in many A-F supergiants as discussed before for HD 725 (Takeda & Takada-Hidai 1994). Among $\alpha$-capture elements, only S shows enrichment above detection limit. These are indications of the star having experienced the first dredge-up (e.g. [C/Fe] = -0.3 dex). However its position in the H-R diagram of Fig. 5 is consistent with a $M\sim1.5-2~M_{\odot}$ and an age of about 7.9   10⁸ yr.

5.8 HD 331319

This star is an IRAS source (19475+3119). The IR fluxes are quite large (Table 1), the heliocentric radial velocity is very small (-2.5 km s^-1). With a spectrum of S/N = 98 we could measure a large number of lines and hence derive the atmospheric parameters as well as abundances with good accuracy (Table 10). The star is Fe-poor by about 1.8 times, but shows significant enrichment of sulphur. The derived abundances are based on 5 good lines hence the derived sulphur abundance cannot be ascribed to measurement errors. Similarly, vanadium also shows some enrichment. Sulphur enrichment appears to be a common feature of our sample stars. The derived abundance of carbon for HD 331319 clearly indicates the effect of CN processing. The star is most likely a young massive disk supergiant or bright giant ascending the red giant branch. The IR fluxes are probably caused by material ejected at this evolutionary stage.

5.9 The proto-Planetary Nebula HDE 341617

It is an interesting object that has been showing relatively fast changes in temperature and brightness. It has faded from $m_{\rm v} = 8.8$ reported in BD catalogue to $m_{\rm v} = 11.4$ estimated by Stephenson (1986). Extensive photometry was taken up by Arkhipova et al. (1999), who found rapid light variations with an amplitude of up to 0.3 mag during 1996-1997. These authors also obtained a spectrum and describe the absorption and emission lines present. The light variations of HDE 341617 and spectral appearance confirmed the suggestion of Volk & Kwok (1989), based on IRAS color indices, that it is a candidate post-AGB star. From spectral type A5 given in HDE, it had changed to class Be (1986) as reported by Downes & Keyes (1988). A very extensive spectral investigation by Parthasarathy et al. (2000) has shown the star to have $T_{\rm eff}$ = 22000 K with large deficiency of carbon. These authors derived $T_{\rm e}$ of 10000 K and $N_{\rm e}$ of 2.5 10⁴ cm^-3 for the nebula from the study of emission lines. The spectra used by Parthasarathy et al. were taken in 1993. As this object presents rapid variations, we felt we could examine the spectrum at our epoch (1999) and study the changes. A comparison of the emission features common to Parthasarathy et al. (2000) 1993 spectrum and ours from 1999 is presented in Fig. 4. Figure 4a shows that in 1999 the emission lines were weaker while Fig. 4b shows that the differences in equivalent widths are not a function of wavelength. Unfortunately, we did not observe any flux standard and so were unable to measure the fluxes in the emission lines. The shape of the pseudo continuum used for each echelle order may also be affected by instrumental sensitivity function. The large scatter is due to this fact. Weakening of lines (if it is not caused by resolution difference) might indicate further increase in temperature of the central star. Systematic monitoring of this fast evolving object could be very rewarding.

Figure 4: A comparison of equivalent widths of the emission spectrum of HDE 341617 as observed in 1993 and in 1999. a) The weakening of the emission spectrum is evident despite the fact that the 1999 spectra have not been flux calibrated. b) The equivalent widths are not a function of wavelength. However, the large scatter is probably due to the lack of calibration. See text for discussion

According to Arkhipova et al. (1999) HDE 341617 has a mass of 0.7 $M_{\odot }$ with an envelope of $\sim$10^-3 $M_{\odot }$ undergoing rapid evolution towards the PN phase which, according to the predictions from Blöcker's (1995) models, should be reached within 100 years. This and the rapid photometric variations reported by Arkhipova et al. (1999), most likely caused by variations in the stellar wind, make the star a very interesting target for continuous monitoring.

We have two relatively low S/N spectra that were used for identification and measurement of absortion and emission lines. We present in Tables 11 and 12 the line strengths and velocities for absorption and emission lines for this object.

Several absortion lines of He I, O II, Si III and Fe II are detected. The radial velocity, for each individual line unambiguously identified is given in Table 11. For He I two groups of lines are identified at average velocities of $52.6\pm8.3$ km s^-1(4 lines) and $75.0 \pm 3.7$ km s^-1 (4 lines). This suggests some stratification in the atmosphere. The rest of the species however average $73.3\pm2.9$ km s^-1 (15 lines). The low dispersion in the radial velocity suggests that all these lines are formed in the same region of the stellar atmosphere and no stratification is evident.

The emission spectrum, formed in the outer envelope and/or in the nebulosity, consists of lines of Fe II, [Fe II], Fe III, [S II], O I, [N II] and Si II. The radial velocities of emission lines are all very consistent and average $54.3\pm3.4$ km s^-1. The difference of radial velocities between the absorption and emission lines indicates that the nebulosity expands at about 19 km s^-1. The Balmer lines show emission on top of the stellar absorption. The emission is shifted relative to the absorption and is consistent with the radial velocities of other emission features, showing that it is also produced in the same region.

Though the number of absorption lines were relatively small, we have done an abundance analysis for a few elements. The O II lines were used for fixing the microturbulence velocity. Since the line data was not adequate to do a detailed excitation equilibrium, we chose to use the temperature and gravity estimated by Parthasarathy et al. (2000) as starting value and tried models both hotter and cooler than their estimate. We got more consistent values for $T_{\rm eff} = 23\,000$ K, $\log g = 3.0$ dex and $\xi_{\rm t} = 15.0$ km s^-1 and hence these parameters were adopted although available lines were not particularly sensitive to the temperature. Temperature errors of $\pm$500 K or more are possible. With these parameters the atmospheric chemical abundances for the central star are those reported in Table 13.

 

 

Table 13: Elemental abundances for the central star of the PPN HDE 341617

Species	$\log \epsilon_{\odot}$	[X/H]	s.d.	N
C II	8.55	-1.43		1
N II	7.97	-0.50	$\pm$0.30	3
O II	8.87	-0.51	$\pm$0.21	14
Mg II	7.58	-1.13		1
Si III	7.55	-0.63	$\pm$0.06	2

Notes - same as Table 3.

Figure 5: H-R diagram showing the positions of sample stars along evolutionary tracks (continuous lines) and isochrones (dashed lines). $T_{\rm eff}$ and log $(L/L_{\odot })$ are those in Table 2. The evolutionary tracks are from Schaller et al. (1992) and all isochrones from Bertelli et al. (1994). a) The star HD 218753 has $M \sim 1.5-2.0\, M_{\odot }$ and age of 7.9   108 yr. b) The two isochrones shown correspond to 3.16  107 and 2.0  107 yr. See text for discussion. c) The post-AGB models are from Blöcker 1995 for ZAMS mass and core mass combinations ( $M_{\rm zams}, M_{\rm H}$); A: = (7$M_{\odot }$, 0.940$M_{\odot }$), B: = (3$M_{\odot }$, 0.836$M_{\odot }$), C: = (5$M_{\odot }$, 0.836$M_{\odot }$), D: = (4$M_{\odot }$, 0.696$M_{\odot }$), E: = (3$M_{\odot }$, 0.625$M_{\odot }$), F: = (3$M_{\odot }$, 0.605$M_{\odot }$). The isochrone shown corresponds to 2.5  108 yr. See text for discussion