The Spectrum of Fg Sagittae After Photometric Recovery in 2000

Baltic Astronomy, vol. 10, 393-401, 2001.

THE SPECTRUM OF FG SAGITTAE AFTER PHOTOMETRIC RECOVERY IN 2000

Tonu Kipper1 and Valentina G. Klochkova2

1 Tartu Observatory, Toravere, 61602, Estonia 2 Special Astrophysical Observatory, Nizhnij Arkhyz, 357147, Karachaevo-Cherkessia, Russian Federation Received November 15, 2000.

Abstract. The high resolution spectra of FG Sge obtained in 2000, after the stars photometric recovery from a deep brightness drop, are analysed. The elemental abundances are derived using the spectral synthesis method. The results show that the abundances are close to the ones before the fading which means that the increase in carbon and 5 — process elements abundances has levelled out. Key words: stars: atmospheres, late-type, individual: FG Sge

1. INTRODUCTION

The remarkable variable star FG Sge was photometrically very active in recent years. Starting in 1992, it shows R CrB-like brightness drops. At the same time the very broad blue-shifted absorption components of the Nal D doublet profiles indicate intensive mass loss (Kipper et al. 1995). These phenomena have been suggested to follow the last helium flash in the shell of the star. Also recorded are the variations of the C2 band strengths and the appearance of emission lines and the C2 emission bands during the light minima (Smith et al. 1996, Kipper & Klochkova 1999). The story of FG Sge until 1995 was presented by Kipper (1996). The latest thorough reviews of FG Sge studies were given by Gonzalez et al. (1998) and by Jurcsik & Montesinos (1999), where the observations until the end of 1996 were discussed. 394 T. Kipper, V. G. Klochkova

Soon afterwards FG Sge recovered almost to the level which it had before the onset of R CrB-like photometric behavior in 1992. Later FG Sge has again suffered two very deep brightness drops. During the deep minimum in 1998 we again observed rich emission line spectrum of the star (Kipper & Klochkova 1999) which was observed by Smith et al. (1996) for the first time. After the relative light maximum (V m 10.5) in 1999 the brightness dropped again at the end of 1999 to a very low level of V « 16. The star almost recovered from that deep fading by the summer of 2000.

2. OBSERVATIONS

We obtained some high resolution (R « 20000) spectra with the echelle spectrograph "Lynx" of the 6 m telescope of the Special Astrophysical Observatory RAN (Panchuk et al. 1999) on 25/26 June and 17/18 July 2000. The exposures were 2 times 3200 s, and the S/N varies considerably with the echelle orders. The highest S/N values reach 80 in July and 50 in June. The spectra cover 4725- 6320 A without any gaps. A somewhat lower resolution (R sa 10000) spectrum was obtained with the prime focus spectrometer (Panchuk et al. 1998) on 8/9 July 2000. This spectrum covers 4300-7850 A, without gaps up to 7000 A. This lower resolution spectrum was used as a supplementary spectrum for the continuum placing and for the Ha line region fit. The spectra were reduced using the NOAO astronomical data analysis facility IRAF. The continuum was placed by fitting low order spline functions through the high flux points in every order. The pro- cedure was verified using the overlapping regions of adjacent echelle orders. Visual inspection of these spectra shows no notable changes (except in the Nal D and Ha profiles) compared with the spectra obtained before the R CrB episode in 1992 and during the light maximum in 1997. The spectra obtained on two dates in 2000 practically coincide in details.

3. ELEMENT ABUNDANCES

The only abundance analysis of the star after the onset of R CrB-type activity is that of Gonzalez et al. (1998), who found The spectrum of FG Sagittae 395 about 1.2 dex higher relative abundances of rare-earths ([Me/Fe]« 3) than we derived from our 1992 spectra before the dimming episodes started (Kipper & Kipper 1993). Therefore it was decided to per- form a new abundance analysis despite the heavy blending in the observed spectral region.

For the 1986 spectra we estimated Teff = 5500 K and log <7 = 1.0 mainly on the ground of the spectral type and photometric data given by Taranova (1987). Montesinos et al. (1990), for the period 1982- 1989, have estimated the effective temperature within 6000-6500 K from the fit of synthetic Kurucz spectra and the observed continuum in the UV spectral region, as well as from the IR color diagram (J- H) versus (J-K). Thus, some temperature rise after 1980 was noted. However, we have found no significant changes in the spectra for most of the eighties. Photometry by van Genderen et al. (1995) also does not indicate any temperature increase during 1980-1992.6. Jurcsik & Montesinos (1999), for the epochs 1992.9 and 1996.3, after the first 4-magnitude drop in 1992, found the temperature of the central star to be ~5500 K from the fits of broadband fluxes with Kurucz models. As we noted, the overall appearance of the spectrum after several dimmings has not considerably changed in comparison with the spectrum obtained before 1992 dimming. This was noted also for the first dimming episode by Stone et al. (1993), who estimated the effective temperature of the star at that time to be ~5600 K. Therefore, in the following analysis we use the same atmospheric model 5500/1.0 from the Kurucz (1979) set which was used for the 1992 spectra (Kipper &: Kipper 1993). We adopt the microturbulent velocity ift = 5 km s-1 following Gonzalez et al. (1998). Their choice was based on the fact that most supergiants have £t in the range of 3-4 km s_1. Due to severe line crowding, the spectrum synthesis method was used for estimating element abundances. The blending and the absence of many heavy-element lines in the used line-list (Bell 1976) prevent obtaining a high precision. Also, only quite strong lines could be used. We estimate the random uncertainties to be around 0.5 dex without taking into account the errors in effective temperature, continuum position and microturbulent velocity which may add 0.3 dex to the abundance error; hopefully, these errors give a smaller effect if [El/Fe] ratios are considered. The atmospheric model used corresponds to solar hydrogen abundance. Gonzalez et al. (1998) were the first to suspect the 396 T. Kipper, V. G. Klochkova

H-deficiency of FG Sge by about 2-3 dex from the analysis of the Ha profile and the relative strength of the CH and C2 bands. If the H/? profile is calculated with our model, no serious H-deficiency is needed, however. The situation with the Ha profile is somewhat more complicated. In Fig. 1 we plot observed and calculated spectra around the Ha line. The hydrogen-deficient model used in this plot was calculated with the code MARCS (Gustafsson et al. 1975) which was modified in order to include molecular and metallic line opacity in the OS ap- proximation. The 3.0 dex hydrogen deficiency was assumed together with C/He = 0.01 (by number) and with abundances of other elements corresponding to the majority of R CrB stars (Rao & Lambert 1996). From this plot one could derive a H-deficiency up to 3 dex for the latest spectrum. This result probably does not correspond to reality, however. In the next plot (Fig. 2) this latest spectrum and the spectra obtained during the brightness rise from 1996 minimum are presented. The Ha profile obtained on 20 July 1997 corresponds to normal H abundance. The results of abundance determinations are given in Table 1 together with Gonzalez et al. (1998) data for 1994. Some of.the oscillator strengths in the line-list were changed to the values given by Thevenin (1989, 1990). Also, some heavy element lines from the lists by Cowley et al. (2000) and by Kurucz (1993) were added. The sources of log gf-s are also indicated in Table 1. If several sources are listed, all log g values were reduced into the scale of the source listed first, using the coinciding lines in the 4500-6500 A spectral region. The carbon abundance log e(C) = —2.5 was derived from the fit of the calculated and observed profiles of the C2 Swan band heads and several CI lines. These results are quite consistent. The oxygen abundance loge(O) = —2.9 was estimated from a single [01] line at 6300.22 A. The carbon-to-oxygen abundance ratio is then C/0= 2.5. We again find considerably smaller Fe abundance, [Fe/H]= —1.0, than a value of —0.1 found by Gonzalez et al. (1998). The probable reason of this difference lies in the much higher effective temperature Teff = 6500 K used by Gonzalez et al. (1998). We have experimented with the Kurucz model 6500/1.0 and in this case [Fe/H]= —0.1 in- deed follows from our spectra too. In that case, however, the photo- spheric C2 Swan bands should not be observable unless the carbon The spectrum of FG Sagittae 397 abundance is increased more than tenfold which we consider to be unacceptable. Taking into account the large errors due to enormous blending and possible differences in the used oscillator strength sys- tems our results for other elements are quite close to the data by Gonzalez et al.

4. CONCLUSIONS

Langer et al. (1974) discovered that the atmosphere of FG Sge is enriched in s-process elements and that the abundances of these elements had been increased since Herbig & Boyarchuk's (1968) observations. We later found that this enrichment continued until 1980's and then levelled out (Kipper & Kipper 1993). In this work we confirm these very high abundances of s-process elements. The [El/Fe] ratios found here are in accordance with the results of Conzalez et al.(1998), except for iron abundance, which we estimate to be considerably lower. We also confirm the very large scandium abundance. The possible reasons of that overabundance are discussed by Gonzalez et al.(1998). They were able to obtain the observed huge scandium abundance from the relatively simple s-process calculations with small or moderate single neutron exposure. One expects to see that the surface carbon abundance increases simultaneously with that of s-process elements, since both are syn- thesized in the He-burning shell. Langer et al. (1974), however, found no increase of the carbon abundance. However, the presence of the weak C2 Swan bands was suspected in the spectra taken in 1980 by Acker et al. (1982). In 1981 the star was certainly a carbon star (Iijima & Strafella 1993). Due to low resolution no carbon abundances were found in these papers. In the high resolution spectra, we obtained just before 1992 fading started, we found strong C2 Swan bands and estimated the carbon abundance about loge(C) = —2.7. The fact that the C abundance at the surface of FG Sge has been al- tered after that of the s-process elements cannot be explained by cur- rent understanding of the third dredge-up process (Mowlavi 1999). The C abundance found in this paper loge(C) = —2.5 is close to the value of —2.7 found by Gonzalez et al. (1998). We conclude therefore that the C abundance increase has also levelled out. It is interesting to note that, if this value of [C/H] is used with the Jurcsik (1996) relation of the inter-fade periods of R CrB stars as a function of carbon abundance, we get that period ~700 days, 398 T. Kipper, V. G. Klochkova

Wavelength in Angstroms

Fig. 1. The spectra of FG Sge in the Ha line region. Observed spectra: solid line - 8 July 2000, dash-dotted line - 20 July 1997. Calculated Ha profiles: dashed line - the model (5500/1.0) with normal H abundance, dash-dotted line - the model with a H deficiency by 3 dex.

Wavelength in Angstroms

Fig. 2. The observed spectra of FG Sge in the Ha line region for various dates. Solid line - 8 July 2000, dashed line - 20 July 1997, dotted line - 24 February 1997 and dash-dotted line - 13 August 1996. The spectrum of FG Sagittae 399

Table 1. The chemical composition of FG Sge in 1994 and 2000.

z El. Sun1 [El] Réf. [El/Fe] 2000 log gf 2000 1994

6 C -3.45 0.9Í0.2 B3 1.9 0.98 8 O -3.13 0.2 B 1.2 12 Mg -4.42 -2.3Í0.2 T4 -1.3 21 Sc -8.83 0.5Í0.2 B 1.5 1.72 22 Ti -6.98 -0.4Í0.5 T 0.6 23 V -8.00 -0.5Í0.3 T 0.5 24 Cr -6.33 -0.2Í0.4 T 0.8 25 Mn -6.61 -0.6Í0.5 T 0.4 26 Fe -4.50 -l.OÍO.4 T 0.0 28 Ni -5.75 -0.6Í0.4 B 0.4 0.16 38 Sr -9.03 2.0Í0.5 B 3.0 2.18 39 Y -9.76 1.9Í0.4 T 2.9 3.58 40 Zr -9.40 1.6Í0.6 T 2.6 3.20 41 Nb -10.58 2.6 T 3.6 <3.7 42 Mo -10.08 2.7 B 3.7 4.05 44 Ru -10.16 3.0±0.7 B 4.0 56 Ba -9.87 2.3Í0.5 B 3.3 57 La -10.83 2.0Í0.3 T 3.0 3.11 58 Ce -10.42 2.liO.3 B,C5 3.1 3.27 59 Pr -11.29 2.3Í0.6 T,B,C 3.3 2.96 60 Nd -10.50 1.9Í0.4 K6 2.9 3.34 62 Sm -10.99 2.4Í0.4 B,C 3.4 3.35 63 Eu -11.49 2.8Í0.5 B 3.8 2.06 64 Gd -10.88 2.2Í0.4 B 3.2 2.96 66 Dy -10.86 2.0Í0.4 B,K 3.0 3.13 68 Er -11.07 1.9Í0.5 K 2.9 3.16 70 Yb -10.92 2.5Í0.4 B 3.5 2.47 71 Lu -11.24 2.5Í0.7 B 3.5 3.59 72 Hf -11.12 2.0 B 3.0 3.16

Grevesse et al. (1996), log CEI relative to H 2 Gonzalez et al. (1998) 3 Bell (1976) 4 Thévenin (1989,1990) 5 Cowley et al. (2000) 6 Kurucz (1993) 400 T. Kipper, V. G. Klochkova

which is right the mean time interval between the observed major brightness drops of FG Sge. We confirm the absence of 13C, which is not an unexpected result taking into account the very large enhancements of s-process elements. 13C would then have been destroyed via 13C(a, n)160 re- action producing free neutrons. We could not confirm Gonzalez et al. (1998) finding that the atmosphere of FG Sge is hydrogen-deficient by 2-3 dex. Taking all together we conclude that the absorption spectrum of FG Sge, after several very deep fadings, has returned to the state it had before fading activity started. The same could be said about the element abundances which stopped increasing also before that time. The difficulties with extremely serious line blending show that much higher spectral resolution than used in this work is needed for detailed comparisons with the s-process modeling.

ACKNOWLEDGMENTS. This research was supported by the ESF grant No. 3166 and the grant N 99-02-18339 from the Russian Foundation for Basic Research.

REFERENCES

Acker A., Jaschek M., Gleizes F. 1982, A&AS, 48, 363 Bell R. 1976, private communication Cowley C. R., Ryabchikova T., Kupka F., Bord D. J., Mathys G., Bidelman W.P. 2000, MNRAS, 317, 299 van Genderen A.M., Gautschy A. 1995, A&A, 294, 453 Gonzalez G., Lambert D.L., Wallerstein G., Rao N.K., Smith V. V., Mc- Carthy J.K. 1998, ApJS, 114, 133 Grevesse N., Noels A., Sauval A.J. 1996, in Cosmic Abundances, eds. S.S.Holt & G. Sonneborn, ASP Conf. Ser., 99, 117 Gustafsson B., Bell R. A., Eriksson K., Norlund A. 1975, A&A, 42, 407 Herbig G. H., Boyarchuk A. A. 1968, ApJ, 153, 397 Iijima T., Strafella F. 1993, IBVS, No. 3959 Jurcsik J. 1996, in Hydrogen-Deficient Stars, eds. C. S. Jeffery & U. Heber, ASP Conf. Ser., 96, 96 Jurcsik J., Montesinos B. 1999, New Astronomy Reviews, 43, 415 The spectrum of FG Sagittae 401

Kipper T., Kipper M. 1993, A&A, 276, 389 Kipper T., Kipper M., Klochkova V. G. 1995, A&A, 297, L33 Kipper T. 1996, in Hydrogen-Deficient Stars, eds. C. S. Jeffery & U. Heber, ASP Conf. Ser., 96, 329 Kipper T., Klochkova V.G. 1999, IBVS, No. 4661 Kurucz R.L. 1993, SAO, Cambridge, Kurucz CDROM18 Kurucz R. L. 1979, ApJS, 40, 1 Langer G.E., Kraft R.P., Anderson K.S. 1974, ApJ, 189, 509 Mowlavi N. 1999, New Astronomy Reviews, 43, 389 Montesinos B., Cassatella A., Gonzalez-Riestra R., Fernandez-Castro T., Eiroa C., Jimenez-Furensalida J. 1990, ApJ, 363, 245 Panchuk V. E., Klochkova V. G., Najdenov I.D., Vitrichenko E. A., Vikuliev N.A., Romanenko V. P. 1999, Preprint SAO, No. 139 Panchuk V. E., Najdenov I. D., Klochkova V. G., Ivanchik A. B., Yermakov S.V., Murzin V.A. 1998, Bull. Spec. Astrophys. Obs. 44, 127 Rao N. K., Lambert D. L. 1996, in Hydrogen-Deficient Stars, eds. C. S. Jeffery k U. Heber, ASP Conf. Ser. 96, 43 Taranova O. G. 1987, AZh Lett., 13, 891 Thevenin F. 1989, A&AS, 77, 137 Thevenin F. 1990, A&AS, 82, 179 Smith V.V., Gonzalez G., Rao N.K. 1996, in Hydrogen-Deficient Stars, eds. C.S. Jeffery & U. Heber, ASP Conf. Ser. 96, 345 Stone R. P.S., Kraft R.P., Prosser C. F. 1993, PASP, 105, 755