September 8 2000 Cornelis de Jager (1), Alex Lobel (2), Garik Israelian (3) and Hans Nieuwenhuijzen (1) (1) SRON Laboratory for Space Research, Sorbonnelaan 2, 3584 CA Utrecht, NL (2) CfA, 60 Garden Street, Cambridge MA, 02138 1516, USA (3) IAC, Obs. del Teide, via Lactea, E 38200 La Laguna, Spain WILL HR 8752 BECOME A P CYGNI TYPE STAR? Abstract: As long as the yellow hypergiant HR8752 has been observed spectroscopically it has shown erratic and significant fluctuations in its effective temperature. But an impressive and hitherto never observed rise in its temperature started around 1988. Since that time T_eff rose from 4600 K to 7900 K. Regular further observations are needed to see if and when this rise will stop, and what will happen thereafter. The instability is related to to the fact that in its evolution the star has entered the Yellow Evolutionary Void, this being a region in the Hertzsprung-Russell diagram where blueward evolving supergiants are dynamically unstable. In that region the values of , averaged over the whole of the star is smaller than the critical value of 4/3, and the atmospheric values of g_eff are negative. Already for at least the last 30 years , averaged over the atmosphere of HR8752 is far below 4/3, while recently the atmospheric effective acceleration has decreased to below zero. This star is the first object where stellar dynamic instability can be studied observationally in great detail. We hypothize that HR8752 will traverse the Void and thereafter, entering the "blue" region of dynamic instability, may become a star like P Cygni. 1. The Yellow Evolutionary Void. The Yellow Evolutionary Void (Figure 1) is a part of the Hertzsprung-Russell diagram where evolved supergiants, moving blueward in the HR-diagram become unstable (Nieuwenhuijzen and de Jager, 1995; de Jager, 1998). In this region the atmospheric effective acceleration in the line-forming region is negative. Moreover, in this region - and in a larger area of lower temperatures as well - the atmospheric values of the logarithmic (P,rho) gradient, Gamma_1, fall below the critical value of 4/3. There also exists another, slightly less well defined area of instability, at higher temperatures, where the same two criteria are fullfilled. We called that area the 'Blue Instability Region'. ----------------------------------------------------------- Fig. 1: The upper part of the Hertzsprung-Russell diagram. The central line-bordered area is the Yellow Void; the upper-left bordered area is the Blue Instability Region. Large dots and circles are hypergiants; small dots are supergiants. The temperature excursions of IRC+10420, Rho Cas and HR8752 are shown by horizontal lines. From de Jager (1988). ---------------------------------------------------------- From a determination of masses for well-studied yellow hypergiants, Nieuwenhuijzen and De Jager (2000) found that the hypergiant HR8752 is a post-red blueward evolving supergiant, such in view of its chemical composition and its low mass. The same applies to the related hypergiant IRC10420. There seems little doubt that the same applies to the hypergiant Rho Cas and to the other yellow hypergiants in the same part of the HR-diagram. Independently of our studies, Stothers and Chin (1995) found similar results that agree with ours in broad outline. They derived the value of , averaged over the whole star, for a set of models of evolved supergiants, and found that is smaller than 4/3 in two parts of the HR-diagram. They called these areas the "first and second phase of dynamic instability". We note that their first phase area includes the Yellow Void (actually, the high temperature borders agree fairly well) but that it extends to lower temperatures than the Void. That latter result does not imply a disagreement because the Void was defined after two criteria: atmospheric Gamma_1 < 4/3 and negative g_eff values. The area where the first condition is fulfilled is larger that that where both are satisfied. Stothers (1999) found that the average value should be calculated by weighing the local Gamma_1 values according to the local volume (d(r^3)) and pressure. He stressed that the average should be taken over the whole of the star, particularly when the mean value of Gamma_1 is close to 4/3 (examples in Stothers, 1999). Although this is basically correct, there is the empirical fact (Nieuwenhuijzen and de Jager, 1995, cf. also de Jager, 1998) that those regions where the atmospheric Gamma_1 < 4/3 coincide reasonably well with those where the stellar value (as calculated by Stothers and Chin) does so too. In a study of detailed models of evolved supergiants Stothers and Chin (2000b, in press) confirmed that yellow hypergiants are dynamically unstable post-red supergiants. They also calculated the high-temperature border of the Yellow Void; it appears to be situated at about 9000 K, with small uncertainty dependent on the choise of the convection parameter. The approach of Stothers and Chin has been criticized by Glatzel and Kiriakidis (1998; references to other authors in Stothers, 1999)). The latter remark that a nonadiabatic correction has to be applied to sigma, the eigenfrequency of the slowest mode of oscillation. To this, Stothers (1999; cf. also Stothers and Chin, 1993) replies that the remark may apply to pulsational instability, but not to dynamic. Dynamic instability is a strictly adiabatic phenomenon, as was shown already by Jeans (1929) and by Baker (1966). As to the calculation of the local value of Gamma_1 one of us (Lobel, 1997, p. 97) remarked that its value should be calculated including microscopic processes such as thermal and photo-ionisation, radiation pressure, non-LTE excitation and ionization. This remark naturally applies particularly to atmospheric regions, where such effects are more important than in the interior. A paper on this more complete calculation of Gamma_1 is presently in preparation (Lobel, 2001). We note that the atmospheric Gamma_1 values derived by us in the present paper are already based on the still unpublished extended expression. 2. HR8752, a hypergiant running into instability HR8752 is a yellow hypergiant. The hypergiant classification (spectral luminosity type Ia+) signifies that supergiant criteria such as mass loss and strong turbulence are spectrally more pronounced than is the case for 'normal' supergiants (for a discussion of the hypergiant criterion and earlier references cf. de Jager, 1980, p. 18ff). The effective temperature of HR8752 has been determined from spectral data since the years '60. Table 1, based on Nieuwenhuijzen and de Jager, 2000) lists those T_eff data that were derived from studies of high-resolution spectra; we consider these the most reliable ones. --------------------------------------------------------- Table 1: Input data for HR8752. Here, g_K is the Kurucz g-value, deduced from a comparison of spectral line profiles with data calculated for Kurucz models; zeta_t is the microturbulence (km/s). References: ILS = Israelian et al. (1999); NdJ: Nieuwenhuijzen and de Jager (2000). yr and mo of obs. T_eff log g_K zeta_t ref _____________________________________________________ 1969/09 5250 - 0.5 10 ILS 1973/08 4930 - 1.8 5.3 NdJ 1978/08 5540 0.15 11.3 NdJ 1984/07 4570 - 6 ?? 4.9 NdJ 1995/04 7170 - 0.18 13.2 NdJ 1998/08 7900 1.1 11 ILS ______________________________________________________ Figure 2 presents the T_eff data against time. The Figure includes also temperatures estimated from color observations (open circles). The hatched areas along the abscissa are periods for which Smolinski (1989) noted excessive mass loss. These data present a fairly complete picture of the life history of this star during the past half century, although one would wish that more observations had been taken during that period. It leads to a scenario in which T_eff decreased around 1970, after a first period of large mass loss, which was followed by stellar contraction as one deduces from the higher T_eff. After the periods of large mass loss around 1980 T_eff decreased again but since 1987 it has been rising steadily. The consequent radius of the star (assuming constant bolometric luminosity) decreased in this period from 900 to 335 solar radii, which means an average shrinking spreed of 0.8 km/s. This behaviour shows great similarity to one of the theoretical models derived by Stothers (1999). His Figure 2 presents the shrinking of of a dynamically unstable and pulsationally stable star. His model has Log (L/Lo) = 6 and log T_eff = 4. For that model = 1.330. (But note that the time scale of the model's shrinking is about one year, ten times smaller that the observed values). ------------------------------------------------------------ Fig. 2: Effective temperatures of HR8752 during the past half century. Filled dots are values derived from high-resolution spectra and open circles are derived from colors. The hatched areas along the abscissa are period of enhanced mass loss. ------------------------------------------------------------ One wonders how this process will continue. In any case, it has been agreed during the Armagh Workshop to henceforth secure one spectrum monthly, in order to obtain detailed information about the future behaviour of this remarkable object. 3. The photosphere For obtaining some information about the possible reasons for this strong instability we have derived photospheric models, based on Kurucz models and fitted to the observed photospheric parameters as given in Table 1. We took a stellar mass of 18.8 M_o (Nieuwenhuijzen and de Jager, 2000), a luminosity log (LO/L_o) = 5.6, and a rate of mass loss log(M-dot) = -4.7 (de Jager, 1998). For these models we have derived by integrating over the atmosphere, generally between optical depths of 0.001 and 100. We also calculated the various contributions to the acceleration (Newtonian, wind, turbulence and radiation) and summed them up to find g_eff. As a rule the Newtonian and turbulent accelerations are the largest in absolute numbers. The results are given in Table 2. -------------------------------------------------------- Table 2: The atmospheric value of , (delta-R)/R and g_eff at three op-tical deepths, for the six sets of observed photospheric parameters given in Table 1. Here (delta-R)/R is the fraction of the radius over which the integration leading to has been made. ________________________________________________________ g_eff at tau_Ross = ..... ..... ......... .......... data T_eff (delta-R)/R 0.01 0.1 1.5 ----------------------------------------------------------- 69/09 5250 1.153 .15 -.25 -.17 1.82 73/08 4930 1.197 .23 .85 .14 1.34 78/08 5540 1.140 .05 -2.63 -2.26 4.21 87/04 4570 1.151 .14 .10 .11 .98 95/04 7170 1.066 .24 .20 .31 5.81 98/08 7895 1.168 .02 -10.4 -7.76 18.46 ___________________________________________________________ While appears to be far below 4/3 for all dates of observations, the g_eff value in the outer part of the photosphere was negative at three instants. The first occurred in 1969, thereafter in 1978 and again in 1998. It is perhaps not fortuitous that the first two periods of enhanced mass loss occurred in 1970 and in 1979-1983, both times after the periods during which the upper-atmospheric acceleration was negative. Following that line one might speculate that soon after 1998 one might again have met a period of enhanced mass loss, followed by a downfall of T_eff to values below 5000. This has yet to be seen, dependent on spectra taken since 1998. Much has to be hoped for from new spectral observations. 4. Speculations on the possible further evolution of HR8752. What follows are speculations. Sooner or later, after having bounced various times against the border of the Yellow Void, the star may have lost that much of his outer envelope that it will be possible to 'jump over' the Void to arrive in the region of higher temperature, above 9000 K. (We do not yet have reliable estimates how long this 'jumping over' may take). From there on the initial evolution should be fairly quiet until the star enters the 'second phase of dynamic instability' or (in other terms:) the 'Blue Instability Region'. Then it may become a star like P Cyg. We think that P Cyg and the other S Dor stars (alternatively called LBV's) are evolved objects, as follows from the high atmospheric He/H contents of the envelopes of well-studied S Dor's (Stothers and Chin, 2000b; for a further discussion of P Cyg in that connection, cf. de Jager, 2001). during the next dynamically unstable stage HR8752 might therefore be an S Dor star, like P Cyg. Let us note that these ideas are also found in a recent paper by Stothers and Chin (2000b). If this speculation is correct then we may consider the yellow hypergiants precursors of S Dor stars. The evolutionary scenario might then be: main sequence O supergiant (mass about 60 solar) -- red supergiant -- yellow hypergiant star (mass about 20 solar) -- S Dor star -- WR star or supernova. Acknowledgment: Thanks are due to Dr R. Stothers for enlightening discussions and useful information. References: N.H. Baker: 1966, in R.F. Stein and A.G.W. Cameron (eds), 'Stellar Evolution', Plenum, New York, p. 333 C. de Jager: 1980, 'The Brightest Stars', Reidel, Dordrecht C. de Jager: 1998, A&A Rev. 8, 145 C. de Jager: 2001, 'The Atmosphere of P Cygni', these Proceedings W. Glatzel and M. Kiriakidis :1998, MNRAS 295, 251 G. Israelian, A. Lobel and M. Schmidt: 1999, ApJ, 523, L145 J.H. Jeans: 1929, Astronomy and Cosmogony, Cambridge Univ. Press, Cambridge, UK A. Lobel: 1997, 'Pulsation and Atmospheric Instability of Luminous F and G-type stars', Shaker Publishing, Maastricht A. Lobel: 2001, 'On the Dynamic Stability of Cool Supergiant Atmospheres', in preparation H. Nieuwenhuijzen and C. de Jager: 1995, A&A 302, 811 H. Nieuwenhuijzen and C. de Jager: 2000, A&A 353, 163 J. Smolinski, J.L. Climenhaga and J.M. Fletcher: 1989, in K. Davidson, A.F.J.Moffat and H.J.G.L.M. Lamers (eds) 'Physics of Luminous Blue Variables', Kluwer, Dordrecht, p. 131 R. Stothers: 1999, MNRAS 305, 365 R. Stothers and C.-w. Chin: 1993, ApJ 408, L85 R. Stothers and C.-w. Chin: 1995, ApJ. 451, L61 R. Stothers and C.-w. Chin: 2000a, 'A Clue to the Extent of Convective Mixing inside Massive Stars', Ap.J., in press R. Stothers and C.-w. Chin: 2000b, 'Yellow Hypergiants as Dynamically Unstable Post-red Supergiant Stars', preprint