### Some Relations between Wolf-Rayet and P Cygni-Type Stars [chapter]

Mart De Groot
1973 Wolf-Rayet and High-Temperature Stars
DISCUSSION Thomas: Could you sketch how Ca II looks? De Groot: The normal interstellar components (Figure 1 ) lie at -10 and -23 km s" 1 ; they are quite strong. Sometimes, on the shortward flank of the component at -10 km s -1 , another can be seen at -16 km s _1 . Then there are two much weaker components at + 17 and + 26 km s -1 . Finally there are the two weak components at -36 and -49 km s _1 . available at https://www.cambridge.org/core/terms. https://doi. Morton: Which ones are permanent
more » ... ones are permanent and which ones appear sporadically? De Groot: The components at -10 and -23 km s -1 are always there. The components at 0, 4-17 and + 26 km s _1 are quite permanent, that means visible roughly on one plate out of three. The components at -36 and at -49 km s _1 1 have seen only once. Van Blerkom: Do you see the same sort of behaviour with the sodium lines? De Groot: That I did not check yet. The one at -36 km s -1 is sometimes showing up in the Na I D lines as well, but not at the corresponding positive velocity. However, at this wavelength the number of plates is only 5. Kuhi: I am just wondering how you can so definitely rule out the interstellar component? De Groot: That was based on the fact that when I found these components, 1 was more or less convinced that they were really there, and then the fact they just fit was such a nice coincidence that I wanted to explain it in this way. Paczynski: I still do not understand what the coincidence could possibly mean. De Groot: The mean velocity of two particular components is the stellar velocity. It means that you have a stream going out and one coming back at the same velocity with respect to the star. Paczynski: So you have matter ejected from the star, and at the same time other matter falling back. I am trying to visualize the picture. It seems difficult. Thomas: No, if you are willing to accept an outward decelerated field, you can accelerate it for a while and then decelerate it for a while and then throw it back. I am not suggesting anything. Paczynski: You have two components always present. So you have a more or less stationary situation. Thomas: It depends on where the excitation is. If, for example, I throw it out, its excitation changes and if it comes back in again its excitation changes. You pick the locus of where Ca is predominantly Ca II. It is a way to make a model. Underhill: I think it is interesting that these inward components are seen against the emission because that emission gives you a background against which you see some absorption. After all, in stars like available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0074180900098697 Downloaded from https://www.cambridge.org/core. IP address: 207.241.231.81, on 27 Jul 2018 at 16:19:06, subject to the Cambridge Core terms of use, the B star at one side of the K star but you can see plus and minus components at the same time. It is a satisfactory description. Kuhi: Could I also ask whether you have looked at any other O and B stars in the vicinity, to see what the interstellar component out there looks like. De Groot: I remember from the work of Adams that there is another star very close to P Cygni. It has the same four components at the same velocities. Bappu: But one finds a temporal change. That is the striking feature. It is not a permanent feature. One of those components comes and goes. De Groot: It has happened only once. I did not see all of them always, but quite often. The real interstellar ones are always there. Thomas: So when you average the -49 and the +17, which you call interstellar, what you are seeing is really a shell associated with the star. How many shells associated are interstellar? De Groot: No, it is what you would call a circumstellar line. Thomas: So this is something which is partly going out and partly coming back. Is that what you suggest? De Groot: I suggest, and it seems to be quite controversial, that you see here the sum of the inter action between the expanding envelope and the interstellar medium. Paczynski: Is there any observational evidence that could disprove the suggestion that P Cygni behaves, as far as the light curve is concerned, like a W UMa system with a period of half a day and an amplitude of 0.1 mag. De Groot: Fernie calculates this case. From the mass-luminosity relation, he finds a mass of about one hundred solar masses. If you take one particle and let it go round this star which has a large radius as well, then the velocity is something like 10 4 kilometers per second, and that is not possible. Paczynski: I have asked you about the observations. According to the results of Appenzeller, a main sequence star of about 100 solar masses may become vibrational^ unstable and may increase its radius by a factor of four, and decrease its effective temperature by a factor of two, and look like P Cygni itself. The pulsation period for the main sequence star is about 10 hours when the star expands by a factor of four. Therefore, there is no theoretical reason at present to deny a half-day period for the luminosity variation of P Cygni. De Groot: I think there are photoelectric observations of Alexander and Wallerstein (Publ. Astron. Soc. Pacific 79, 500,1967), who have tried to check the half-day period, and have found no variation. I would not reject at first the half-day period, but reject at least the W Ursae Majoris system idea, and explain any variations as pulsational instability. Paczynski: That is why I was asking for the observational evidence. De Groot: I think the observations pointing to the half-day periods are still better than the ones denying it, since they are more numerous, but this is still a controversial point. May be we can summarize by saying that normally the variations are quite irregular. Thomas: Could you tell us what density and radii and velocity you use, that yield a mass loss of 10~4 to 10" 5 M Q per year? De Groot: For stars between 46 and 144 solar masses you have radii between 40 and 80 solar radii; roughly 100 solar masses give a radius of 60 solar radii. Paczynski: What do you mean by a mass-radius relation? De Groot: The luminosity-radius relation. Paczynski: Luminosity-temperature-radius ? De Groot: Yes, the luminosity-temperature-radius. You have the logarithm of the electron densities which range between 12.5 to 13, and you have velocities of one hundred to two hundred kilometers per second. Now if you combine these you come to mass-loss figures in this range. The only thing is that you have to believe this. Hutchings has more extensively studied mass loss in OB supergiants and he finds that it is dependent on both temperature and luminosity. He draws a HR diagram; you have the main sequence, you have the evolution of a thirty and a fifteen solar mass star and the stars showing the highest degree of mass loss are all in the 30 Af 0 region and above; we should go down to find lower masses in order to the stars of less and less mass loss. Thomas: And what happens if you go in the other direction? De Groot: You have more mass loss. This is only for supergiants. In OB supergiants you have mass losses of 10~5 to 10~7 MQ per year ,so a little bit less, and the lines of constant mass loss are more or available at https://www.cambridge.org/core/terms. https://doi. less parallel to the lines of the evolution off the main sequence of the stars of the masses of the OB supergiants. Thomas: And how do I infer the mass loss of the OB supergiants? De Groot: More or less as it has been done for the P Cygni stars. Thomas: For P Cygni I have some kind of a number to give me a number. De Groot: Because you have in the OB supergiants nearly always a velocity at Ha from the P Cygnitype profile, and you can take your velocity from there. Then, you have the observations in the rocket UV. Paczynski: The rocket UV observations as interpreted by Lucy and Solomon indicated the mass loss rates of 10~8 solar masses per year. This is considerably less than the rates you quoted. Morton: We must not forget the uncertainties in both the theoretical and observed rates of mass loss. For the theoretical estimate Lucy and Solomon put in one mechanism and found about 10~8 solar masses per year. This number can be taken as a lower limit, but the observations show that many resonance lines are effective rather than a single dominant line at each temperature predicted by the calculations. On the other hand, the higher rates derived from the spectra often depend on large, poorly established corrections for unseen ions. Thomas: What makes you think that it will disagree? What is the evolutionary time-scale for these kinds of solar masses? Paczynski: It is less than 10 7 years on the main sequence and less than 10 6 years in the supergiant phase. De Groot: It is less interesting if it concerns a mass that comes from a mass of 100 solar masses, because then it may be only one per cent anyhow. Conti: Theoretically we do not know. It may make a whole lot of difference. De Groot: You could ask why we do not see surrounding P Cygni stars, anything like shells or rings or nebulae. We have to conclude that the extended atmospheres and the matter which is lost is so dense that the region cannot be very large in extent and so it escapes discovery. If you build a Stromgren sphere and you fill it and you make it very dense, the ionizing radiation cannot go very far from the star, so the nebulae you will detect are to be small ones. Thomas: You say this is so dense that I cannot see it? De Groot: No, the matter which leaves the star is so dense that the ionizing radiation from the star does not go very far. Thomas: But if I take the implicit suggestion of velocities that can be a thousand kilometers per second, then you will see them in the rocket UV. I think the absence of a shell is very embarrassing. Morton: You have the same problem with Wolf-Rayet stars. Thomas: Oh, sure. But you still have an embarrassment. Would you think that in terms of the observations that if they are there, you should see them? Or would you accept what Anne Underhill said, that may be we have not looked enough? De Groot: I think we did not look at enough areas. In the southern hemisphere, Thackeray has found very few, two or three objects which have real P Cygni characteristics and are associated with visible nebulae. Thomas: Oh, he does find some! De Groot: Yes. I did not say there were none. But the number must be impressively low. Thomas: Do they have any particular characteristics; are they different from other kinds of shells? De Groot: I think that most of them are circular. The star is more or less in the center. Paczynski: Are the central stars hotter than average P Cygni stars? De Groot: Most of Thackeray's were B's. They were not excessively hot stars. Paczynski: The late B's could hardly ionize the nebulae. De Groot: I think they were probably early B's, but I should have to look it up. I do not remember. Smith: I think the WN7 and WN8 spectra are more like the P Cygni spectra than are the WC spectra. De Groot: Yes. Smith: The WN7's and WN8's are still estimated to have average absolute visual magnitudes -6.8 and -6.2 respectively. I think the WN's are the most closely related to the P Cygni's, in visual appearance. De Groot: If you talk about line widths and about excitation, then, the answer is no, but if you take some other characteristics, especially the fact that N may be a bit overabundant in P Cygni, because at least the N lines are very marked, then you come to WN7 and WN8. The P Cygni stars available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0074180900098697 Downloaded from https://www.cambridge.org/core. IP address: 207.241.231.81, on 27 Jul 2018 at 16:19:06, subject to the Cambridge Core terms of use, exchange and, therefore, at the moment I do not have the problem to assume that they are really helium-burning stars at later phases of their main-sequence life and show this pulsational instability. If I had to classify the phenomena on the scheme of Thomas, I would say that they have a corona and an exosphere. Regarding mass and momentum transfer, etc., that is very difficult to say. Morton: What is the mass qf a typical P Cygni star? Do the P Cygni stars have the problem of the Wolf-Rayet star in the sense that the mass is much less than we would estimate from the mass-lumi nosity relation? De Groot: At the moment it is not clear where we stand on that. But to summarize, the one bit of evidence was one star with a mass not more than 4 solar masses, that somebody just mentioned. Walborn: It was HD 152667. This star is a BO supergiant in Sco OBI, and it has an absolute magnitude of -6.7, which is consistent with other stars with similar spectra in the region. The mass derived, according to that analysis, is mot nuch greater than 4.1 M Q . Morton: It just may be that the P Cygni stars have the same problem that the Wolf-Rayet stars have, that they are very over-luminous for their masses. De Groot: Apart from the fact that you then do not really know their masses, you would introduce the problem that you do not understand the cause of the difference either. You have to invent some thing else. Paczytiski: If you feel attached to the vibrational instability of massive stars you should favour lower masses. If these stars are overluminous indeed, it means that their hydrogen content must be low. The critical mass for the vibrational instability decreases with the hydrogen content. At the limit of zero hydrogen, the critical mass is 15 solar masses. It means that you may reduce this critical mass very considerably. If there is indeed anomalous abundance of N, you may speculate along the lines you suggested yesterday: these stars may be mixed. Of course, this is just a speculation. Underhill: What is the period of HD 152667? De Groot: 7 days point something. Underhill: I was wondering whether you are looking for eclipses which are not known. Is it possible in this star that the apparent supergiant character of it could be partly due to gas streams in a reason ably close pair. Is it double-lined, do you know? De Groot: No, it is single-lined, so it gives the mass function. The star which shows the P Cygnitype line profile is the star which is in front during eclipse. Underhill: Does it have a good light curve? De Groot: Yes; The light curve has not been established very completely, but at least one can say that there are eclipses of 0.10 or 0.15 mag. Kuhi: I have several questions. One is in connection with the color temperatures that you quoted earlier. What wavelengths did those refer to? De Groot: The results I mentioned refer to the region between 3500 and 6000 A and in several stars there is a trend of the colour temperature decreasing with increasing wavelength. Kuhi: I raised a question about that very same point, that in OB supergiants, as well as in Wolf-Rayet stars, you do find that the colour temperature that you get is definitely a decreasing function of increasing wavelength. So it is not clear what the colour temperature means at all. And then I have another question. I am more intrigued by the presence of the circumstellar Ca n lines. Is there any way that you can even guess the size of the distance from the star at which these occur, or can you speculate on it? De Groot: The lowest excitation stellar lines, that is the Balmer lines and, I think, Mg u, are formed at 2£ stellar radii. That means that at 2\ stellar radii we have the amount of excitation that produces the hydrogen lines. This would be at about 160 R 0 . This means that you have to go a little bit farther out to find the excitation region for the calcium lines. I cannot say how much farther out but it gives you an idea; it should be between 3 and 10 stellar radii. Paczynski: Can we observe the component of the Ca line that could show a large expansion velocity? In other words, if you extrapolate your picture going out very far from the star, you are going into steadily less excited regions of the envelope, you may approach the region where the Ca line should be visible? Is there anything at a few hundred km sec -1 ? De Groot: I could not say at the moment. I have a list of identifications, and at two angstroms shorter there is an unidentified weak absorption line. But it has connected with it an unidentified weak emission line, which should not be there. It has been seen only in one spectrogram, so I would hesitate to identify that as a stellar Ca n absorption line.