Does feedback from low mass stars affect fragmentation in massive protoclusters?

1. Does feedback from low mass stars affect fragmentation in massive protoclusters? Steve Longmore Harvard-Smithsonian Center for Astrophysics Thushara Pillai (CalTech), Eric Keto (CfA), Qizhou Zhang (CfA), Keping Qiu (MPIfR) Jill Rathborne (CSIRO), Jim Jackson (BU)

2. Fragmentation of molecular gas within massive protoclusters Observational aim • Select massive protoclusters with similar global properties but at different evolutionary stages • Get complete census of fragments within each region Requires sufficient • Resolution to resolve smallest fragment separation • Sensitivity to detect fragments with weakest emission • Dynamic range to detect both brightest and weakest fragments

3. Time Fragments more massive and closer together with time?

4. What causes differences in fragment mass distribution? • Global properties of G8.68-0.36 (IRAS 18032-2137) • Distance = 4.8kpc • Mass = 1500Msun • L = 2x104 Lsun • dM/dt = 10-4 Msun/yr • Numerical simulations (Krumholz & McKee, 2008, Nature, 451, 1082) show above threshold of N > 1 g cm-2 & Lsun/Msun >10 • Accretion luminosity from low-mass stars becomes trapped • Gas equation of state changes from isothermal to stiffer than isothermal • Mass of fragments increase • Can not determine E.O.S. observationally (so direct test of KM08 not possible) • Higher sensitivity/resolution observations to search for evidence of population of low mass stars

5. G8.68-0.36 (IRAS 18032-2137) Fragment mass • MMM1 = 14Msun • MMM2 = 7Msun • MMM3 = 5Msun Fragment separation • ~0.02pc

6. Implications • No fragments large enough to form O-star through direct collapse • Any feedback not efficient enough or not happened quickly enough to halt fragmentation to this extent 2. Where are the fragments M<5Msun? • Top-heavy mass function? • Missing population of low-mass stars/fragments? • Don’t see them in optical/IR  too embedded • Interferometer only sensitive to density contrasts  missing population of uniformly distributed low mass stars? • Searching for evidence of population of low mass stars • Method • Use present day conditions to infer gas properties at time of fragmentation • See what population of low mass stars would be required to create these conditions

7. ρ ~ 104 cm-3 T ~ 10-20 K ρ = ρfrag T = Tfrag ρ = ρnow T = Tnow Simple model of fragmentation Feedback from low mass stars changes cluster-scale gas properties Initial conditions Gas fragments to form population of low mass stars t = t0 t = t1 t = t2 Cluster-scale gas Use present-day mass and separation of fragments to infer conditions at time of fragmentation Change in cluster-scale gas properties allows more massive fragments to form Feedback from low and high mass stars alter cluster-scale gas properties t = t3 t = t4 t0 tfrag tnow

8. Inferring gas properties at tfrag • Start with simplest scenario. Assume: • 1. ρfrag = ρnow • 2. Mass/location of MM1, MM2 and MM3 not changed Upper/lower limit to mass of MM1 to MM3 Using Jeans analysis to determine what combination of temperature and density are required to reproduce the observed separation and mass of MM1 to MM3. Upper/lower limit to separation of MM1 to MM3

9. r100K What population of low-mass stars needed to heat gas to 100K? • Assuming accretion luminosity of 10Lsun and main-sequence stellar temperature requires: • 2.9×106 0.8 Msun stars • 1.25×105 1.2 Msun stars • 2.9×103 1.8 Msun stars 0.05pc

10. Implications for the formation of massive stars • If the cluster-scale gas density and mass/position of MM1 to MM3 have not changed since fragmentation, a population of low mass stars would not have been able to create the conditions required for thermal fragmentation. • Number of low mass stars required drops considerably if relax previous assumptions: • ρnow > ρfrag and/or • Separations of MM1, MM2 and MM3 were larger in the past Fragments part of global infall? Future work required to test with larger sample/more realistic support mechanisms

11. Conclusions Massive stars do not form from in the collapse of individual fragments, but rather in smaller fragments that themselves continue to gain mass by accretion from larger scales 1. Despite large (1500Msun) reservoir, no fragments with enough mass to form O-star through direct collapse • Any feedback not efficient enough or did not happen quickly enough to halt collapse to this extent 2. Several lines of evidence suggest the observed cores are currently fed via infall from the very massive reservoir (1500M) of gas in the larger pc scale cloud around the star-forming cores.

13. Region selection: G8.68-0.36 (IRAS 18032-2137) G8.68-0.36 Properties • Distance=4.8kpc • Mass = 1500Msun • L = 2x104 Lsun • F8GHz < 1 mJy (MichelePestalozzi priv com) • Prior to UCHII region formation • Signs of star formation • EGO/Green Fuzzy • Methanol maser • Single-dish dust emission • Centrally peaked • Symmetric • Easy to model! • Relatively isolated • Easier to interpret observations Criteria • N > 1 g cm-2 • Lsun/Msun >10 • Enough raw material to expect O-star to form • Young enough that fragments still exists • Old enough for feedback from low-mass stars to have affected gas properties

14. Determining gas temperature and density 1 2 3 NH3(1,1) Modelling • Ammonia • NH3(1,1), (4,4), (5,5) emission from Longmore et al 2007 • Eric Keto’s 3D radiative transfer code “Mollie” • Model with 5 parameters • ρ(r)=ρ0r –κ1 • T(r)=T0r –κ2 • ΔVNT • Χ2 minimisation fitting of spectra • CH3CN (12-11) • Emission unresolved in SMA compact config • Solve for optical depth assuming LTE • Χ2 minimisation fitting of spectra • 1.2mm continuum • Fit measured visibility amplitude vs uv-distance assuming power law density and temperature 1 NH3(4,4) 2 3 NH3(5,5) Result from 3 methods consistent ρ(r) = 9.5x104 r –2.08 [cm-3] T(r) = 42 r –0.35 [K] 1

15. What population of low-mass stars needed to heat gas to 100K? • Assume many individual sources heating dust envelope in their immediate vacinity • Use analytic solutions to solve radial dust temperature profile for different central heating sources. • Accretion luminosity (Garay & Lizano 1999) • Main-sequence stars (Lamers & Cassinelli 1999) • Find radius out to which sources heat gas to 100K  r100K • How many sources uniformly separated by r100K required to fill volume encompassing fragments (radius=0.05pc)? • Assuming main sequence stellar temperature and accretion luminosity of 10Lsun requires: • 2.9×106 0.8 Msun stars • 1.25×105 1.2 Msun stars • 2.9×103 1.8 Msun stars • Numbers are unfeasibly large!! • Higher accretion luminosity doesn’t work as bolometric luminosity exceeds observed limit r100K 0.05pc

16. Motivation and project aim • Typical star forming clouds: T=10-20K & ρ~104cm-3 • If star formation proceeds by fragmentation and subsequent gravitational collapse of the fragments, how do O and B stars form when global Jeans mass ~1Msun? • Feedback from low-mass stars increase fragment mass? • Numerical simulations (Krumholz & McKee, 2008, Nature, 451, 1082) show above threshold of N > 1 g cm-2 & Lsun/Msun >10 • Accretion luminosity from low-mass stars becomes trapped • Gas equation of state changes from isothermal to stiffer than isothermal • Mass of fragments increase • Can not determine E.O.S. observationally (so direct test of KM08 not possible) • Project aim: • Use criteria to select regions where fragmentation may have been affected by feedback from low mass stars • Determine mass distribution of fragments • Top heavy? • Massive enough to form O-star through direct collapse? • Determine properties of low-mass stars

17. Region selection: G8.68-0.36 (IRAS 18032-2137) G8.68-0.36 Properties • Distance=4.8kpc • Mass = 1500Msun • L = 2x104 Lsun • F8GHz < 1 mJy (MichelePestalozzi priv com) • Prior to UCHII region formation • Signs of star formation • EGO/Green Fuzzy • Methanol maser • Single-dish dust emission • Centrally peaked • Symmetric • Easy to model! • Relatively isolated • Easier to interpret observations Criteria • N > 1 g cm-2 • Lsun/Msun >10 • Enough raw material to expect O-star to form • Young enough that fragments still exists • Old enough for feedback from low-mass stars to have affected gas properties

18. Results: Part 2 • If the cluster-scale gas density and mass/position of MM1 to MM3 have not changed since fragmentation, a population of low mass stars would not have been able to create the conditions required for fragmentation.

19. Implications for the formation of massive stars • Number of low mass stars required drops considerably if relax previous assumptions: • ρnow > ρfrag and/or • Separations of MM1, MM2 and MM3 were larger in the past • Simple interpretation of Case 1 = global infall • Observational evidence? • Signatures of large-scale infall in HCO+ and HCN spectra of Purcell et al 2006,2009 • Infall rate ~ 10-4 Msun/yr • Predictions for Case 2 • Assume free-fall from rest (Walsh et al 2004). • Proper motions of MM1, MM2 and MM3 too small to detect • Relative velocity of MM1 to MM3 ~1km/s • Future high angular/spectral resolution molecular-line observations planned to look for this. • Will G8.68-0.36 form O-stars as expected from large mass and infall rate? • Current fragments have insufficient mass so must gain mass from elsewhere • 1500Msun reservoir of gas • Several models for how cluster-scale gas coupled to dense cores • Accretion through common “disk” • Krumholz et al 2009 • Peters et al 2009 • Keto et al 2004, 2007 • Un-bound cluster-scale gas • Smith, Longmore, Bonnell 2009 • Future observations required to test these models Massive stars do not form from in the collapse of individual fragments, but rather in smaller fragments that themselves continue to gain mass by accretion from larger scales

20. Conclusions • Through SMA observations we have conducted a census of the mass/spatial distribution of fragments in G8.68-0.36. We find: • 3 fragments • Mass -- 14Msun, 7Msun, 5Msun • Separation -- ~0.03pc • No fragments with enough mass to form O-star through direct collapse • Any feedback not efficient enough or did not happen quickly enough to halt collapse to this extent • No fragments of mass <~ 5Msun • Top-heavy IMF? • An unobserved population of low-mass stars/fragments? • Using a simple model for low-mass star feedback and subsequent fragmentation, we infer the gas conditions at the time of fragmentation • If the current density and mass/location of fragments have not changed since the time of fragmentation, the number of low mass stars required to create these conditions exceeds the limits imposed by the observations. • However, if the fragments formed in lower density gas then subsequently moved together in a global contraction of the cloud, the number of low-mass stars required could have been lower • The observations suggest the observed cores are currently fed via infall from the very massive reservoir (1500M) of gas in the larger pc scale cloud around the star-forming cores. This suggests that massive stars do not form in the collapse of individual massive fragments, but rather in smaller fragments that themselves continue to gain mass by accretion from larger scales.

21. G8.68-0.36 (IRAS 18032-2137) Distance = 4.8kpc Mass = 1500Msun L = 2x104 Lsun dM/dt = 10-4 Msun/yr F8GHz < 1 mJy (Michele Pestalozzi priv com) • Prior to UCHII region formation Signs of star formation • EGO/Green Fuzzy • Methanol maser ρ(r) = 9.5x104 (r/r0) –2.08 [cm-3] T(r) = 42 (r/r0) –0.35 [K] r0 = 0.29pc