next up previous contents index
Next: HST Spectral Maps Up: StarsStellar Populations, Previous: Is there a

Science with the Hubble Space Telescope -- II
Book Editors: P. Benvenuti, F. D. Macchetto, and E. J. Schreier
Electronic Editor: H. Payne

Spectroscopic Analysis of the Wind Lines of the Symbiotic Nova AG Peg

W. Schmutz
Institute of Astronomy, ETH Zentrum, CH-8092 Zürich



I present stellar parameters of the symbiotic nova AG Peg for the years 1970 and 1994. For both dates the analyses yield approximately the same mass loss rate of Myr and a remarkably similar terminal velocity of the wind of 700kms and 950kms, respectively. The effective temperatures increased from 50,000K to over 100,000K and the radius shrunk from 0.5R to less than 0.06R. In 1994 nitrogen was over-abundant with a N/He number-ratio of 0.01 and carbon was deficient with a C/N ratio of less than 0.01.

Keywords: Novae, Symbiotic stars, Stars: AG Peg


AG Peg is the nova with the longest outburst known. The eruption started in 1850, and today, it is still not finished (Kenyon et al. 1993, Mürset & Nussbaumer 1994). From theoretical calculations of slow novae (Kenyon & Truran 1983, Livio et al. 1989, Shara et al. 1993), it is expected that after the outburst the photosphere is extended and then it contracts on a time scale of years at constant luminosity to a white dwarf radius. At the end of this evolution the effective temperature is higher than 100kK. It appears that AG Peg's temperature and luminosity evolution followed the expected trend (Kenyon et al. 1993, Mürset & Nussbaumer 1994) and that the outburst is now close to its end. It has been predicted that soon, the mass loss from the nova should stop and an accretion phase should start (Zamanov & Tomov 1995). The characteristics of AG Peg make it a key object to study the slow novae phenomenon.


The spectra of AG Peg show three distinct types of line profiles. These are attributed to different emission regions within the symbiotic system (Penston & Allen 1985, Kenyon et al. 1993). Here, I focus on the broad components that are attributed to a fast wind from the symbiotic nova (Nussbaumer et al. 1994). I have used two data sets for a spectroscopic analysis. The first set is based on observations by Hutchings & Redman (1972). The tracings of September 25, 1970, show for the last time a HeI \ P Cygni absorption that was present prominently in earlier spectra (e.g., Merrill 1951). In later spectra all HeI wind features have disappeared and, thus, an analysis based on the helium ionization is not possible for later dates. The tracings of Figs. 2, 5, and 6 of Hutchings & Redman (1972) have been scanned and corrected for light present in addition to that of the nova, i.e., after correction, the wind emission lines are measured relative to the estimated nova continuum. At the wavelengths 3888, 4340, and 4686 contributions of 18%, 38%, and 52% have been subtracted to account for nebular emission and a red giant continuum. These corrections result from a fit of a red giant spectrum to the absolute energy distribution of AG Peg in the red wavelength region, and from the strength of a calculated nebular spectrum. Despite considerable effort to determine these corrections, the values are admittedly rather uncertain. The re-normalized profiles of HeI and of the HI-HeII blend are shown in Fig. 1. It can be seen that in addition of the uncertain level of the nova continuum there is also the difficulty that the wind profiles are blended in the central part by a strong component from the nebula.

Figure: Comparison of the calculated HeI line (left panel) and the HI-HeII blend (right panel) with spectra observed in 1970 (Hutchings & Redman 1972).

In contrast to the optical region, the signatures of the nova wind remained observable in the UV (Penston & Allen 1985). Nussbaumer et al. (1995) detected a P Cygni absorption in the profile of NV . The depth of the absorption reaches almost zero intensity and, therefore, the wind is unequivocally associated with the nova. This absorption is blue shifted by the same velocity that corresponds to the width of other broad emission features. The HST observations have been obtained on June 6, 1994. There are four line profiles that show a broad component from the nova wind: NV , OV, HeII , and NIV. The GHRS observations shown in Figs. 2 to 4 are smoothed with a 10 pixel box filter.

Model Calculations

For the spectroscopic analysis I used the non-LTE atmosphere code developed in Kiel for hot stars with a strong stellar wind (Hamann & Schmutz 1987, Wessolowski et al. 1988). For the analysis of the 1970 spectrum the calculations are based an a H-He model atmosphere as described by Hamann et al. (1991). For the analysis of the 1994 HST observations nitrogen and carbon model atoms are included in addition to hydrogen and helium. The non-LTE rate equations include six levels for the hydrogen model atom; 13 levels for helium with five levels for HeI and seven levels for HeII, six levels for carbon with five levels for CIV; and 12 levels for nitrogen with six levels for NIV and five levels for NV. In total there are 37 levels and 53 line transitions are treated explicitly. The line radiation transfer is calculated in the co-moving frame. The final line formation is obtained in the observer's frame and includes frequency redistribution of line photons due to electron scattering.

Figure: Comparison of the calculated HeII line (dashed curve) with the HST observation of 1994 (full drawn line). The left and right panels show the same figure but with different scaling.


Table 1: Calculated stellar parameters for the nova AG Peg.

The analysis of the 1970 spectra yields the stellar parameters given in Table 1. Following the method of Schmutz et al. (1991) the relative strengths of the HeII \ and the HI-HeII line blend is used to derive the helium abundance. Within the uncertainties of the observed line profiles the helium abundance is solar, i.e., 10% by number, but the uncertainty of this value is quite large. I estimate that values between 5% and 20% are still compatible with the observations. Similarly, the stellar temperature is not well constrained. The temperature results from the relative strength of HeI compared to HeII . Allowing for some observational noise in the weak absorption feature of the HeI line, and uncertainty of the level of the nova continuum, the effective temperature is found to be between 40 and 70 kK. The radius of the nova results from the observed continuum in the UV (OAO2, Gallagher et al. 1979), that is assumed to be dominated by the nova continuum, combined with the calculated flux, the distance and the reddening. The astrophysical flux at Å is and for the reddening and distance I adopt and d=650 kpc (Mürset et al. 1991, Vogel & Nussbaumer 1994). The error of the derived luminosity is of the order of a factor three.

The analysis of the 1994 spectra does not yield a unique result. There are no pairs of lines from the same ion with adjacent ionization stages that could be used for a stellar temperature diagnostic.gif Therefore, the only temperature diagnostic comes from the fact that there is a nebular component of HeII . This observation implies that the nova wind is transparent for photons more energetic than 54 eV. Given the strength of the observed wind component of HeII , which implies a mass loss rate, there is a minimal effective temperature that yields a complete ionization of helium through the wind. The model calculations require that kK. The IUE observations of AG Peg show (Fig. 2a of Vogel & Nussbaumer 1994) that a nebular HeII component appears between 1981 and 1986. Thus, the nova has reached the threshold temperature for complete ionization in the mid 80s. I assume that the increase in stellar temperature was steady and I have, therefore, adopted a somewhat larger temperature, kK, for the determination of the nitrogen and carbon abundances (Table 1). The determined nitrogen abundance does only weakly depend on the adopted stellar temperature as long as 100 kK kK. If the effective temperature of AG Peg is higher than 130 kK---which cannot be excluded by the present analysis---then the nitrogen abundance is larger than given in Table 1.

The fits to the profiles of NV and HeII require a velocity law that can be described with , with . The usual assumption of can be excluded because it yields a shape that does not fit the observed profile.

As shown in Figs. 1--4 the synthetic wind lines of HeII and NV fit nicely the broad feet of the observed profiles. The extreme line wings are due to line photons scattered by electrons. The line wing of HeII \ is reproduced very well. In the observed line wing of NV there is an absorption at Å\ of uncertain origin.

Figure: As Fig. 2 but for the NV doublet.

The CIV doublet is not at all reproduced by the model calculations. From the comparison of the calculated profile with the observed profile, I derive an upper limit to the carbon abundance. With a carbon abundance given in Table 1, the calculation still predicts a clear P Cygni absorption. In the observed profile, there is a hint of a weak absorption. However, the interpretation of this feature is uncertain because it is blended by a broad line wing; presumably CIV line photons from the nebula scattered by electrons. It appears safe to assume that an absorption as strong as in the synthetic profile would be detected even if it is filled in by the electron scattering wing and, therefore, the given comparison yields an upper limit for the carbon abundance.


The spectroscopic analysis of the nova AG Peg is hampered by the scarceness of observed wind features and in addition, by the blends with other light sources from the system. In particular, the determination of the effective temperature turns out to be rather uncertain. Mürset & Nussbaumer (1994) used the nebular lines in order to determine the temperature. The largest uncertainty of their method is the assumption of a nebula with a simple geometry and symmetry. A comparison of their effective temperature with the present determinations does not show any significant discrepancies. Since the two methods are based on different assumptions the agreement supports the results.

Figure: As Fig. 2 but for the CIV doublet.

The high nitrogen abundance and the strong carbon deficiency supports the commonly accepted view that on novae there is CN processing. The fact that we see the CN anomaly implies that there is sufficient mixing that the processed matter is visible.

The most striking result is that for 1970, as well as for 1994, the analyses yield identical mass loss rates and terminal velocities that increased only marginally from 700kms to 900kms. From the point of view of radiation-driven winds, this result cannot be understood. The theory predicts the terminal velocity to be proportional to the escape velocity (Castor et al. 1975, Kudritzki et al. 1989). Since the photospheric radius changed by about a factor of 10 between 1970 and 1994, it is expected that the terminal velocity should have changed by a factor of 3. Thus, it is unlikely that the mass loss of AG Peg can be explained by radiation pressure alone. However, there are no other plausible mechanisms than radiation pressure for the acceleration, at least in the optically thin part of the wind. Therefore, the wind momentum should be proportional to the luminosity, as predicted by the theory of radiation driven winds if gravitation is neglected (Abbott 1980, Eq. 88). If true, then the wind momentum is a very sensitive tool to measure the nova luminosity. The wind parameters of AG Peg then indicate that the luminosity of AG Peg did not change between 1970 and 1994. This is not in contradiction with the results of Table 1 because the errors allow an unchanged luminosity between 1970 and 1994. It is, therefore, possible that AG Peg is still in the evolutionary phase of shrinking radius at constant luminosity and the phase of declining luminosity has not yet started.


I appreciate stimulating discussions and valuable suggestions by my colleagues U. Mürset, H. Nussbaumer, H. Schild, and R. Walder.


Abbott, D.C. 1980, ApJ 242, 1183

Castor, J., Abbott, D.C., & Klein, R. 1975, ApJ 195, 157

Hamann, W.-R. & Schmutz, W. 1987, A&A 174, 173

Hamann, W.-R., Dünnebeil, G., Koesterke, L., Schmutz, W., & Wessolowski, U. 1991, A&A 249, 443

Hutchings, J.B. & Redman, R.O. 1972, PASP 84, 240

Kenyon, S.J. & Truran, J.W. 1983, ApJ 273, 280

Kenyon, S.J., Mikolajewska, J., Mikolajewski, M., Polidan, R.S., & Slovak, M.H. 1993, AJ 106, 1573

Kudritzki, R.P., Pauldrach, A., Puls, J., & Abbott, D.C. 1989, A&A 219, 205

Livio, M., Prialnik, D., & Regev, O. 1989, ApJ 341, 299

Merrill, P.W. 1951, ApJ 119, 338

Mürset, U., Nussbaumer, H., Schmid, H.M., & Vogel, M. 1991, A&A 248, 458

Mürset, U. & Nussbaumer, H. 1994, A&A 282, 586

Nussbaumer, H., Schmutz, W., & Vogel, M. 1995, A&A 293, L13

Penston, M.V. & Allen, D.A. 1985, MNRAS 212, 939

Schmutz, W., Leitherer, C, Hubeny, I., Vogel, M., Hamann, W.-R., & Wessolowski, U. 1991, ApJ 372, 664

Shara, M.M., Prialnik, D., & Kovetz, A. 1993, ApJ 406, 220

Vogel, M. & Nussbaumer, H. 1994, A&A 284, 145

Wessolowski, U., Schmutz, W., & Hamann, W.-R. 1988, A&A 194, 160

Zamanov, R.K. & Tomov, N.A. 1995, Observatory 115, 185

next up previous contents index
Next: HST Spectral Maps Up: StarsStellar Populations, Previous: Is there a