Observational Constraints on Massive Star Evolution

1. Observational Constraints on Massive Star Evolution Phil Massey Lowell Observatory IAU Symp 212

2. Feedback Between Data and Theory is Crucial • Allows theory to develop by highlighting where the models work well and where improvements may have to be made. • Caveat: must be sure the observational database is correct! • Allows us observers to know what to test next.

3. This Is a Wonderful Time! • Improved stellar models: • Stellar evolutionary models are increasingly sophisticated, with improved physics. • Highly accessible thanks to the kindness of the Geneva and Padova teams and the existence of the Internet. • Improved observational capabilities: • Mosaic CCD cameras on 4-m telescopes (LG Survey is producing UBVRI of 300 million stars in 9 galaxies). • Multi-object spectroscopic capabilities on 8-m telescopes.

4. Why Study Stars in the Galaxies of the Local Group? • Z varies by a factor of 10 from the SMC to M31. • Evidence is pretty solid that the IMF does not vary. (IMF slope for massive stars the same in the SMC, LMC, and Milky Way.) • SFR does vary but we can design tests so that this comes out in the wash.

5. Two Types of Tests Possible: • Mixed-age populations (i.e., B/R supergiant ratio, WC/WN ratio, RSG/WR ratio, etc.). • Assumption is that we are looking over enough area of a galaxy to have a totally heterogeneous mix of ages. • Coeval populations containing evolved stars (i.e., young clusters containing WRs or LBVs). • Classical method that allows us to determine progenitor masses from the turn-off ages if the region is sufficiently coeval.

6. Good and Bad Observational Quantities • Good (if careful to avoid selection effects!) • Distribution of stars in HRD in MCs • WC/WN ratio • RSG/WR ratio • B/R Supergiant ratio if carefully defined: • (B-V)o<0, Mbol<-7.5 [includes B supergiants] • (V-R)o>0.6, Mbol<-7.5 [includes K-type RSGs but avoids AGBs] • Turn-off mass in a cluster (if can prove it’s coeval).

7. Good and Bad Observational Quantities (continued) • Bad: • Number of O-type stars relative to anything. Why?

8. The BC! ZAMS 85Mo star will be 15x fainter than a 25Mo A-type supergiant! From Massey et al (1995)

9. Completeness in the MCs  Missing for a reason! From Massey et al. (1995)

10. So What Do We Know, and What Don’t We Know? • Main-sequence evolution • Evolved stars (a) RSGs (b) WRs (c) LBVs

11. Main-sequence Evolution Magellanic Clouds and the Milky Way Image by Roger Smith/NOAO/AURA/NSF

12. Main-sequence Evolution • Meylan & Maeder (1982, A&A 108, 148) described the main-sequence widening problem. • Mixing helps (Langer 1991 A&A, 252, 669) • Blue loops (necessitated by SN1987A) • Higher RSG mass-loss rates extends the blue loops, possibly eliminating the problem (Salasnich et al 1999, A&A, 342, 131) • Fitzpatrick & Garmany (1990, ApJ 363, 119) LMC study: • Established the need for “blue loops” models. • Emphasized the incompleteness problems. • Massey, Lang, DeGioia-Eastwood, & Garmany (1995 ApJ 438, 188) study of the LMC and SMC: • Established good agreement with the number of stars and the Geneva normal mass-loss isochrones after adding many spectra to the database.

13. Things are in pretty good shape for the main-sequence! Furthermore, we are making great strides in improving the data-base in the Magellanic Clouds and other galaxies of the Local Group.

14. The LMC CMD in 1990: LMC From Fitzpatrick & Garmany (1990)

15. The LMC CMD in 2002: From Massey (2002, ApJS, in press)

16. The LMC CMD in 2002: Foreground disk giants disk dwarfs Blue supergiants RSGs From Massey (2002, ApJS, in press)

17. What Can We Do With these New Data? Spectra are needed to take full advantage of the new photometry but in the meanwhile, we do have a reasonable estimate of the number of blue stars for the LMC and SMC. Number of “blue stars” with Mbol < -7.5 well determined in these fields---probably better than in the nearby Milky Way!

18. Evolved Massive Stars • Red Supergiants • Wolf-Rayet Stars • Luminous Blue Variables

19. Red Supergiants • A long-standing problem has been confusion between foreground dwarfs and bona-fide RSGs in these galaxies. BVR two-color diagrams partially resolve this, although spectra are still needed to get the statistics right.

20. Separating RSGs from foreground dwarfs REDDENING VECTOR SUPERGIANTS B-V FOREGROUND DWARFS From Massey (1998 ApJ, 501, 153) V-R

21. Separating RSGs from foreground dwarfs red stars towards M33 SUPERGIANTS B-V - FOREGROUND DWARFS From Massey (1998 ApJ, 501, 153) V-R

22. Do High Mass Stars Become RSGs? Massey (1998) investigated the RSG content of three Local Group galaxies: NGC 6822 (log O/H + 12 = 8.2) M33 (log O/H + 12 = 8.4) M31 (log O/H + 12 = 9.0)

23. Number of RSGs as a f(luminosity) low Z med Z high Z Mbol MV

24. The RSGs in these 3 galaxies show: • As the metallicity goes up the fraction of high luminosity RSGs goes down. • There is not a sharp cut-off in MV or Mbol. Moderately high mass stars become RSGs even at high Z; what changes is that the RSG phase is shorter at high Z (Maeder, Lequeux, & Azzopardi 1980).

25. RSGs: How Well Do the Evolutionary Tracks Do? Not too well! We can see that according to any of the models we should NEVER have M-type RSGs with Mbol < -8! This is contrary to the observations....

26. Z=0.02 Models, Normal M-dot

27. Do high mass loss tracks help? No!

28. What About A Different Brand? Same problem!

29. Are Things Any Better at Low Z? Nope! Even in the SMC something is awry...

30. What’s the Problem? Either there is a serious problem with the effective temperature scale of M-type supergiants or The evolutionary tracks don’t go far enough to the right at sufficently high masses

31. Why? =1.5 Convection is hard; mixing length matters! =0.3 =1.5 Maeder & Meynet 1987 A&A 182, 243

32. One Last Point about RSGs... Elias et al. (1985 ApJS 57, 91) found that the average spectral type of RSGs depended upon metallicity. New data will allow us to answer how this effect varies with luminosity.

33. B/R Supergiant Ratio Needs to Count Both K-type and M-type B/R Ratio including only M-type in the “R” LMC 53 SMC 165 B/R Ratio including both K-type AND M-type: LMC 26 SMC 20 From Massey 2002, ApJS in press So, whether or not the B/R ratio is a f(Z) depends upon what you count!

34. Wolf-Rayet Stars Small sections of several nearby galaxies have now been surveyed for WRs with sufficient sensitivities to detect even weak-lined WNs. Soon we will have candidates for all of M31, M33, and other LG galaxies. Need followup spectroscopy. But for now, here’s what is known...

35. WC/WN Ratio Depends Upon Z Number Ratio WC/WN log (O/H)+12

36. With the Possible Exception of IC10 Number Ratio WC/WN log (O/H)+12

37. IC10 is a classic starburst

38. What Could This Mean? • IMF inverted for high-mass stars? • Star-formation occurred in lock-step across the entire galaxy (t < 200,000 yrs???)? • Or we didn’t do as good a job as we thought! • New survey has identified many more WR candidates, including all of the Royer et al (2001) except the WC9s The WC/WN ratio in IC 10 may be normal---but with an even more remarkably large number of WRs than previously supposed!

39. How Well Do the Models Do? Number Ratio WC/WN log (O/H)+12

40. WC/WN Ratio (continued) • Normal Mass-Loss Geneva models (Schaller et al. 1992; Schaerer et al. 1993; Charbonnel et al. 1993) and Padova models (Fagotto et al 1994; Bressan et al. 1993) do very well, except at high Z. • Enhanced mass-loss Geneva models do not match the data: they predict far more WCs than actually observed!

41. RSGS/WRs ---a strong f(Z) log (number of RSGS/WRs) log (O/H) + 12

42. How Well the Models Do? log (number of RSGS/WRs) log (O/H) + 12

43. Summary of Mixed-Aged Tests (1) Main-sequence: Normal mass-loss models do well in LMC/SMC. More data are coming! (2) RSGs: None of the models produce sufficiently cool and luminous RSGs. (3) WC/WN as a f(Z): Normal mass-loss models do OK, although they predict too few WCs at the highest Z. Enhanced mass-loss models do not match the data at all. (4) RSG/WR: None of the models match the observations, but not surprising given (2).

44. Progenitor Masses Determined from Coeval Regions • Selected regions that contain WRs and LBVs: • 19 regions in the MCs (Massey, Waterhouse, & DeGioia-Eastwood, 2000, AJ 119, 2214) • 12 regions in the Milky Way (Massey, DeGioia-Eastwood, & Waterhouse 2001 AJ, 121, 1050) • About half of these were sufficiently coeval to use the turn-off masses to measure the progenitor masses.

45. Milky Way LMC What We Found SMC 100 80 70 60 50 Progenitor Mass (Mo) 40 30 20 WNE WNL Ofpe/WN9 WC WCL LBV

46. Eta Car is Right At Home!

47. Coeval Regions---Conclusions • In the SMC (low Z) all WRs in our sample came from very high mass stars (>85Mo). • In the MW (high Z) WRs come from a large range of masses (18 to >120Mo). • Classical LBVs (S Dor, Eta Car,...) have come from only the highest mass stars. • “Ofpe/WN9” stars come from much lower masses (25-40Mo). Not “true” LBVs???

48. What’s Next? • Move beyond the MCs to complete galaxy-wide surveys for WRs, RSGs, and BSGs in M31, M33, N6822, IC10, and other galaxies of the Local Group. • Follow-up spectroscopy with 8-m will allow meaningful HRDs to be constructed, allowing careful tests of models as a function of metallicity.

49. M33-North

50. NGC 6822