1. 1)Observations: where do (massive) stars form? 2)Theory: how do (massive) stars form? 3)Search for disks in high-mass (proto)stars 4)Results: disks in B stars, toroids in O stars 5)Implications: different formation scenarios for B and O stars? Disks, toroids and the formation of massive stars Riccardo Cesaroni O-B star  >10 3 L O  >8 M O  high-mass

2. High-mass star forming regions: Observational problems  IMF  high-mass stars are rare  large distance: >400 pc, typically a few kpc  formation in clusters  confusion  rapid evolution: t acc =20 M O /10 -3 M O yr -1 =2 10 4 yr  parental environment profoundly altered Advantage:  very luminous (cont. & line) and rich (molecules)!

3. Where do massive stars form?

4. Clump UC HII Core HMC

5. Clump UC HII HMC

6. Clump

7. High-mass star forming region 0.5 pc

8. Clumps and hot molecular cores R clump = 10 R HMC M clump = 10 M HMC n clump = 0.01 n HMC D (pc)M (M O )n H 2 (cm -3 )T (K) Clump1100010 5 30 HMC0.110010 7 100

9. n  R -p with p=1.5-2.5  no break at HMC  unstable density profile M clump > M virial  clumps unstable V clump = V HMC  HMCs at rest wrt clumps T  R -q with q=0.4-0.5  clumps centrally heated  Clumps might be collapsing  HMCs are density peaks in clumps  HMCs are T peaks “enlightened’’ by embedded stars

10. n H 2  R -2.6 ClumpHMC Fontani et al. (2002)

11. n  R -p with p=1.5-2.5  no break at HMC  unstable density profile M clump > M virial  clumps unstable V clump = V HMC  HMCs at rest wrt clumps T  R -q with q=0.4-0.5  clumps centrally heated  Clumps might be collapsing  HMCs are density peaks in clumps  HMCs are T peaks “enlightened’’ by embedded stars

12. Fontani et al. (2002) sample of 12 Clumps

13. n H 2  R -p with p=1.5-2.5  no break at HMC  unstable density profile M clump > M virial  clumps unstable V clump = V HMC  HMCs at rest wrt clumps T  R -q with q=0.4-0.5  clumps centrally heated  Clumps might be collapsing  HMCs are density peaks in clumps  HMCs are T peaks “enlightened’’ by embedded stars  HMCs are deeply related to clumps

14. n H 2  R -p with p=1.5-2.5  no break at HMC  unstable density profile M clump > M virial  clumps unstable V clump = V HMC  HMCs at rest wrt clumps T  R -q with q=0.4-0.5  clumps centrally heated  Clumps might be collapsing  HMCs are density peaks in clumps  HMCs are T peaks “enlightened’’ by embedded stars  HMCs are deeply related to clumps

15. How do massive stars form?

16. Low-mass vs High-mass Shu et al. (1987): star formation from inside-out collapse onto protostar Two relevant timescales: accretion  t acc = M * /(dM/dt) contraction  t KH = GM * /R * L * Low  mass (< 8 M O ): t acc < t KH  “birthline’’ High  mass (> 8 M O ): t acc > t KH  accretion on ZAMS

18. Palla & Stahler (1990) dM/dt=10 -5 M O /yr t KH =t acc Zero-age main sequence Sun

19. PROBLEM: High-mass stars “switch on” still accreting   radiation pressure stops accretion (Kahn 1976)   stars > 8 M O cannot form!? SOLUTIONS Yorke (2003): K dust < K crit  M * /L * 1)“Increase’’ M * : large accretion rates 2)“Reduce’’ L * : non-spherical accretion 3)Reduce K dust : large grains (coalescence of lower mass stars)

20. Possible models 1)Large accretion rates: competitive accretion (Bonnell et al. 2004); turbulent cores (McKee & Tan 2002) 2)Non-spherical accretion: disk+outflow focus ram pressure and dilute radiation pressure (Yorke & Sonnhalter 2002; Krumholz et al. 2003) 3)Coalescence: many low-mass stars merge into one massive star (Bonnell et al. 2004)

21. Disk + outflow may be the solution (Yorke & Sonnhalter, Kruhmolz et al.): Outflow  channels stellar photons   lowers radiation pressure Disk  focuses accretion   boosts ram pressure  Detection of accretion disks would support O-B star formation by accretion, otherwise other mechanisms are needed

22. The search for disks in high-mass YSOs

23. Disks seem natural outcome of star formation: collapse + angular momentum conservation   flattening + rotation speed up  disk Disks are likely associated with outflows: outflow detection rate = 40-90% in massive YSOs (luminous IRAS sources, UC HIIs, H 2 O masers,…) (Osterloh et al., Beuther et al., Zhang et al., …)  disks should be widespread! BUT… Where and what to search for…?

24. Where to search for? 0.5 pc disk?

25. outflow Theorist’s definition: Disk = long-lived, flat, rotating structure in centrifugal equilibrium Observer’s definition: Disk = elongated structure with velocity gradient perpendicular to outflow axis core disk What to search for?

26. TRACERPROsCONTRAs Maser lines High angular & spectral resolution Unclear geometry & kinematics Continuum Sensitivity (and resolution) No velocity info Confusion with free- free and/or envelope Thermal lines Kinematics and geometry of outflow and disk Limited angular resolution and sensitivity (but see SMA and ALMA)

27. Results of disk search Two types of objects found: Toroids M > 100 M O R ~ 10000 AU L > 10 5 L O (O stars) (dM/dt) star > 10 -3 M O /yr t rot ~ 10 5 yr t acc ~ M/(dM/dt) star ~ 10 4 yr  t acc << t rot  non-equilibrium, circum- cluster structures Disks M < 10 M O R ~ 1000 AU L ~ 10 4 L O (B stars) (dM/dt) star ~ 10 -4 M O /yr t rot ~ 10 4 yr t acc ~ M/(dM/dt) star ~ 10 5 yr  t acc >> t rot  equilibrium, circumstellar structures

28. Examples of rotating toroids:

29. Furuya et al. (2002) Beltran et al. (2004) Beltran et al. (2005)

30. Furuya et al. (2002) Beltran et al. (2004) Beltran et al. (2005)

31. Furuya et al. (2002) Beltran et al. (2004) Beltran et al. (2005)

32. Furuya et al. (2002) Beltran et al. (2004) Beltran et al. (2005)

33. Furuya et al. (2002) Beltran et al. (2004,2005) Moscadelli et al. (2007) M dyn = 19 M O M dyn = 55 M O

34. First result : velocity gradient perpendicular to bipolar outflow  rotating toroid conservation of angular momentum from 2” (15000 AU) to 0.5” (4000 AU)  possible formation of circumstellar disk?

35. outflow axis absorption HC HII hypercompact HII + dust O9.5 (20 M O ) + 130 M O

36. outflow axis Beltran et al. (2006, Nature)

37. Second result : Red-shifted absorption in molecular line towards HII region  infall towards star  accretion onto star?

38. Hypercompact HII region Moscadelli et al. (2007) Beltran et al. (2007) 7mm free-free & H 2 O masers 500 AU

39. Hypercompact HII region Moscadelli et al. (2007) Beltran et al. (2007) 7mm free-free & H 2 O masers 30 km/s

40. Third result : H 2 O masers along HII region border have proper motions away from star  expansion of shell HII region with t HII = 500 AU/50 km/s = 50 yr !!! note that this is distance independent  hyperyoung HII region

41. Final scenario: G24 A1 is a massive toroid, rotating about a bipolar outflow and infalling towards an O star with very young expanding HII region  a 20 M O star has been formed through accretion (now finished…?)

42. Example of rotating disk:

43. IRAS 20126+4104 Cesaroni et al. Hofner et al. Moscadelli et al. Keplerian rotation: M * =7 M O Moscadelli et al. (2005)

44. Results of disk search Two types of objects found: Toroids M > 100 M O R ~ 10000 AU L > 10 5 L O (O stars) (dM/dt) star > 10 -3 M O /yr t rot ~ 10 5 yr t acc ~ M/(dM/dt) star ~ 10 4 yr  t acc << t rot  non-equilibrium, circum- cluster structures Disks M < 10 M O R ~ 1000 AU L ~ 10 4 L O (B stars) (dM/dt) star ~ 10 -4 M O /yr t rot ~ 10 4 yr t acc ~ M/(dM/dt) star ~ 10 5 yr  t acc >> t rot  equilibrium, circumstellar structures

45. Are there disks in O stars? In L star ~ 10 4 L O (B stars) true disks found In L star > 10 5 L O (O stars) no true disk (only toroids) found - but distance is large (few kpc) Orion I (450 pc) does have disk, but luminosity is unclear (< 10 5 L O ???)  Difficult to detect disks in O (proto)stars. Why? Observational bias or physical explanation?

46. Observational bias?

47. Assumptions: HPBW = R disk /4 FWHM line = V rot (R disk ) M disk  M star same in all disks T B > 20 K obs. freq. = 230 GHz 5 hours ON-source spec. res. = 0.2 km/s S/N = 20 edge-on i = 35°

48. Assumptions: HPBW = R disk /4 FWHM line = V rot (R disk ) M disk  M star same in all disks T B > 20 K obs. freq. = 230 GHz 5 hours ON-source spec. res. = 0.2 km/s S/N = 20 no stars edge-on i = 35°

49. Physical explanation?

50. O-star disks “hidden” inside toroids O-star disk lifetime too short, i.e. less than rotation period:  photo-evaporation by O star (Hollenbach et al. 1994)  tidal destruction by stellar companions (Hollenbach et al. 2000) In both cases we assume M disk =M star /2 and disk surface density ~ R -1, i.e. M disk  R disk :

51. tidal destruction rotational period photo-evaporation Cesaroni, Galli, Lodato, Walmsley, Zhang (2007)

52. Disks in O (proto)stars might be shorter lived, and/or more deeply embedded than those detected in B (proto)stars Photoionosation: inefficient disk destruction mechanism, for all spectral types (if M disk comparable to M star ) Tidal interaction with the stellar companions: more effective to destroy outer regions of disks in O stars than in B-stars

53. Conclusions Disks found in B (proto)stars  star formation by accretion as in low-mass stars No disk found yet (only massive, rotating toroids) in O (proto)stars  –observational bias (confusion, distance, rarity,…) –disks hidden inside toroids and/or truncated by tidal interactions with stellar companions –disks do not exist; alternative formation scenarios for O stars needed: coalescence of lower mass stars, competitive accretion (see Bonnell, Bate et al.)

55. Furuya et al. (2002) Beltran et al. (2004) Beltran et al. (2005)

56. 10 pc G9.62+0.19 NIR J+H+K

57. 2 pc

58. 0.5 pc G9.62+0.19 350 micron Hunter et al. (2000)

59. Testi et al. Cesaroni et al. 3.6cm

60. Furuya et al. (2002) Beltran et al. (2004) Beltran et al. (2005) 1200 AU

61. Beltran et al. (2005); Hofner et al. (in prep.) NH 3 red-shiftedNH 3 blue-shiftedNH 3 bulk CH 3 CN(12-11)

62. IRAS 20126+4104 Edris et al. (2005) Sridharan et al. (2005) disk NIR & OH masers