1. The Role of Giant LBV Eruptions in the Evolution of Very Massive Stars Nathan Smith CASA, U. Colorado In collaboration with Stan Owocki (U. Delaware) In the near future: Smith & Owocki (2006) ApJ Letters, submitted 120 20 M/M  t=02.5-3 Myr WR LBV MS clumped MS homogeneous

2. CLUMPING IN LINE-DRIVEN WINDS OF HOT STARS Observational mass-loss rates come from H  emission and IR/radio free-free. Both are sensitive to  2. If winds are highly clumped (F C >>1). Then M from H  and free-free is much lower. Examples: Fullerton et al. (2006); factors of 10-20 reduction in Mdot. Bouret et al. (2005); factors of >3. Puls et al. (2006); median of 20, but as much as 100x lower see also Crowther et al. 2003; Hillier et al. 2003; Massa et al. 2003; Evans et al. 2004. see poster by Martins & Schaerer

3. MASS LOSS AND STELLAR EVOLUTION MSLBVWRSN M/M  120  20? Mdot3e-62e-43e-5 T(yr)3e65e45e5 M lost10 15 M/M sun60  12? M dot1e-62e-42e-5 T(yr)5e65e45e5 M lost510 So, the burden of mass loss must fall on post-MS phases. 120 20 M/M  t=02.5-3 Myr WR LBV MS clumped MS homogeneous Smith & Owocki (2006) ApJ Letters, submitted) ?

4. + LBVs on the upper HR Diagram Eta Car RSGs Smith, Vink, & de Koter (2004)  =0.9  =0.5

5. During eruption L Bol =20x10 6 L  T star =8,500K After eruption L Bol =5x10 6 L  T star =30,000K ( ) present-day properties The historical light curve of Eta Car Total mass ejected was >12 M .

6. (Van Dyk et al. 2000) (SN 1954j) (Humphreys, Davidson, & Smith 1999) SUPERNOVA IMPOSTORS (extragalactic Eta Car analogs) Historical Type V supernovae: (Eta Carinae / P Cygni) SN1954j in NGC2403 SN1961v in NCG1058 Recent faint SNe in SN searches: All have been faint with spectral class Type IIn (but not all Type IIn’s are SN impostors) MASS LOSS AND STELLAR EVOLUTION

7. OBSERVED MASSES OF LBV NEBULAE In circumstellar shells around other LBVs and LBV candidates, a mass of ~10 M  is typical for stars with L>10 6 L . In Eta Carinae, at least, we know this is ejecta from a single outburst and not swept-up material. Smith & Owocki (2006) ApJ Letters, submitted

8. SN and GRB Environments Recent observations reveal very massive shells around SN and GRBs: This means that the progenitor stars may have had eruptive mass-loss events shortly before exploding………. SN1988z, nebula = 15 M  (Aretxaga et al. 1999; Williams et al. 2002; Van Dyk et al. 1993; Chugai & Danziger 1994). SN2002hh, nebula =10-15 M  (Barlow et al. 2005). massive shells around SN2001em and SN1994w (Chugai & Chevalier 2006; Chugai et al. 2004). massive shells around GRB021004 and GRB050505 (Mirabel et al. 2003; Berger et al. 2005). circumstellar gas around other GRBs (H.-W. Chen, D. Fox, this meeting)

9. MULTIPLE ERUPTIONS… HST/WFPC2 F502N [O III] F658N [N II] Smith et al. 2005 Each burst will remove a substantial fraction of the star’s mass and will affect its evolution…how many times will this happen? This has happened before: Outside the bipolar Homunculus of Eta Car, there are ionized “outer ejecta” from probably 2 previous eruptions separated by several hundred to 1000 yr….(proper motions: Walborn et al. 1978; Walborn & Blanco 1988). P Cygni also shows multiple separate nebular shells separated by several hundred years (Meaburn 2001; Meaburn et al. 1996, 1999, 2000, 2004; O’Connor et al. 1998).

10. MASS LOSS AND STELLAR EVOLUTION MSLBVWRSN M/M  120  20? Mdot3e-62e-43e-5 T(yr)3e65e45e5 M lost10 15 M/M sun60  12? M dot1e-62e-42e-5 T(yr)5e65e45e5 M lost510 Giant LBV outbursts a la Eta Carinae apparently dominate the post-MS mass loss of very massive stars…..but we still don’t know what triggers them. 120 20 M/M  t=02.5-3 Myr WR LBV MS clumped MS homogeneous Smith & Owocki (2006) ApJ Letters, submitted

11. WHAT IF THIS PICTURE IS WRONG? 120 20 M/M  t=02.5-3 Myr WR LBV MS clumped MS homogeneous ? Consider some Alternatives: LBV phase is longer or core He-burning massive stars masquerade as some other type of BSGs? Mass loss by mass transfer/RLOF? Massive stars explode early - end of the LBV phase?

12. MASS LOSS AND STELLAR EVOLUTION The  10 M  ejected in this type of eruption may be enough to fix the mass discrepancy in the post-MS evolution of massive stars. Evolutionary tracks for massive stars depend on adopted mass loss rates (e.g., Maeder & Meynet 1994, 2000, 2003). To reproduce correct ratio of OB/WR stars, WR star M and L, etc., these calculations need to adopt mass-loss rates on MS that are 2 X HIGHER than “standard” mass loss rates. (de Jager et al. 1988; Nieuwenhuijzen & de Jager 1988) Problem: more recent modeling of spectra of O stars winds find much LOWER mass-loss rates due to clumping by about 10x or more. 120 20 M/M  t=02.5-3 Myr WR LBV MS clumped MS homogeneous Angular Momentum?

13. WHAT DRIVES THE EXREME MASS LOSS? So, 12 M  is a safe Lower Limit to the total mass ejected during the 19 th century Great Eruption of Eta Carinae (could be 20-30 M  ). Mass Loss rate during the 20-yr eruption > 0.5 M  yr -1. (Probably much higher, since proper motions indicate a small range of ejection dates.) Lines are saturated! HUGE amount of Kinetic Energy ( ½mv 2 =10 50 ergs. But this is a lower limit...) Luminous energy ~10 49.5 ergs There’s NO WAY these can be line driven winds…either super- Eddington continuum-driven winds or hydrodynamic explosions. (see Owocki et al. 2005; Arnett et al. 2005) Typical M  typical  t for other LBV eruptions implies mass-loss rates of order 10 -2 to ~1 M  /yr.

14. THE FIRST STARS (Pop III) The “first stars” may have been mostly massive (peak ~100 M  ) and luminous, and may have re-ionized the Universe ( Bromm & Larson 2004, Bromm this meeting ). Hot stars have line-driven winds (opacity dominated by Fe), but the “first stars” have no metals…so they don’t have any mass loss…….? …but… Giant eruptions like Eta Carinae and the “supernova impostors” are insensitive to metallicity (electron-scattering opacity, or hydrodynamic explosions). Important for mass loss in the first stars? (Abel et al. 2000) MASS LOSS AND LOW METALLICITY

15. Heger et al. 2003 MASS LOSS AND LOW METALLICITY Using standard mass-loss rates (Nieuwenhuijzen & de Jager 1988).

16. At solar metallicity, MS mass loss is not very different from First stars! 120 20 M/M  t=02.5-3 Myr WR LBV MS clumped MS homogeneous Pop III MASS LOSS AND LOW METALLICITY

17. SUMMARY/CONCLUSIONS MASS LOSS - Main Implications Because of clumping, mass-loss rates for line-driven winds have been revised downward by an order of magnitude. Thus, line-driven winds on the main-sequence are vastly insufficient to remove the H envelope and produce WR stars (M WR  20 M  )…best alternative is that LBV eruptions dominate the mass loss of the most massive stars. The mass loss of giant LBV eruptions is insensitive to metallicity --- their extreme mass-loss rates cannot arise from line-driven winds. The possibility that the mass loss of massive stars even at solar metallicity may be dominated by a metallicity-independent mechanism should at least raise caution signs for the notion that Pop III stars did not suffer mass loss. //////////// ROAD BLOCK: we don’t know what triggers LBV eruptions \\\\\\\\\\\\\\\\\ Pop III stars were massive -- could they shed mass through LBV eruptions? Proving this wrong will tell us a great deal about stellar evolution.

20. DUST MASS (from the ISO spectrum) Total IR luminosity 4.3x10 6 L  100 x M(dust) 400K 200K 140K 0.02 M  1.5 M  11 M  Total = 12.5 M  Total mass (gas+dust) of Homunculus > 10 M  ----- HUGE! Previous estimates from =2-12  m typically gave 2-3 M . Higher mass comes from cool dust emitting at > 12  m. Smith et al. 2003

21. DUST MASS (from the ISO spectrum) Total mass (gas+dust) of Homunculus > 10 M  ----- HUGE! Conservative assumptions… Optically thin emission Large grains (a~1  m silicate): (small grains have poor Q ABS at long IR wavelengths) Gas : Dust mass ratio =100 (Eta’s ejecta are C and O poor, Fe in gas phase in inner shell)

22. 2.122  m H 2 1-0 S(1) 1.644  m [Fe II] Gemini South/Phoenix R=60,000 Thin walls of the H 2 shell… The ejecta expand as a Hubble flow, so if  R/R ~  t/t then  t ≤ 5 yr. (proper motions agree…) This would require a HUGE mass-loss rate during the eruption of several M  /yr. Implies that the 19 th century event was an explosion. Typical M  typical  t for other LBV eruptions implies mass- loss rates of order 10 -2 to ~1 M  /yr. Smith (2006), ApJ, 644 (June 20)

23. Smith & Ferland (2006, almost done) Thin walls of the H 2 shell… CLOUDY models: The survival of H 2 in a thin layer around Eta Car requires a density of n H  10 7 cm -3 in the outer shell. For the volume of the outer H 2 shell, this implies a total gas mass of 20-30 M .

24. + The historical light curve of P Cygni P Cygni: the other nebula from a Galactic giant LBV eruption that was actually observed. Image: Clampin et al. (1997)

25. + LBVs on the upper HR Diagram Eta Car Smith, Vink, & de Koter (2004)  =0.9  =0.5 P Cygni:

26. + P Cygni: bright [Fe II] 1.644  m emission but no H 2, and almost no dust. (less mass/lower optical depth after ejection) Smith & Hartigan (2006) MASS: Can’t use dust, but from [Fe II] lines, M=0.1 M  Mass and KE much less than Eta Car in 1843, but similar to 1980 outburst of Eta Car (Little Homunculus)

27. Outer shell Cool dust 140 K Molecular hydrogen Thin shell Inner Shell Warm dust 200 K [Fe II] emission, etc. Thick shell n e =10 4 cm -3 Hot dust near star Equatorial clumps ( >400 K ) Smith (2006), ApJ, 644 (June 20)

28. Shape of the Homunculus Bipolar Geometry Smith (2006)

29. Bipolar Geometry Almost 75% of total mass and more than 90% of total KE above θ = 45°… Smith (2006)

30. Bipolar Geometry Almost 75% of total mass and more than 90% of total KE above θ = 45°… Important constraints on shaping mechanism: 1.Rules out spherical shell/wind shaped by circumstellar torus (mass at equator)… 2.Rules out deflection by companion star. (KE in polar ejecta > binding energy of orbit) 3. Must have been an inherently bipolar explosion of the star (ROTATION?). Merger??? (Morris & Podsiadlowski 2006) Smith (2006)

31. 2.122  m H 2 1-0 S(1) 1.644  m [Fe II] Gemini South/Phoenix R=60,000 Instabilities (or lack thereof) in the smooth H 2 shell… Structure does not resemble typical Rayleigh-Taylor instabilities. Instead, it looks like a clumpy, fragmented thin shell. Suggests structure is not dominated by gasdynamic effects, but by thermal instabilities/fragmentation of a dense thin shell shortly after ejection.