Collapsed Cores in Globular Clusters

General Relativistic Instability Supernova of a Supermassive Population III Star


The formation of supermassive Population III stars with masses 10,000  in primeval galaxies in strong UV backgrounds at 15 may be the most viable pathway to the formation of supermassive black holes by 7. Most of these stars are expected to live for short times and then directly collapse to black holes, with little or no mass loss over their lives. But we have now discovered that non-rotating primordial stars with masses close to 55,000  can instead die as highly energetic thermonuclear supernovae powered by explosive helium burning, releasing up to 10 erg, or about 10,000 times the energy of a Type Ia supernova. The explosion is triggered by the general relativistic contribution of thermal photons to gravity in the core of the star, which causes the core to contract and explosively burn. The energy release completely unbinds the star, leaving no compact remnant, and about half of the mass of the star is ejected into the early cosmos in the form of heavy elements. The explosion would be visible in the near infrared at 20 to Euclid and the Wide-Field Infrared Survey Telescope (WFIRST), perhaps signaling the birth of supermassive black hole seeds and the first quasars.

cosmology: early universe - theory - galaxies: formation – hydrodynamics – galaxies: high-redshift – stars: early-type – supernovae: general – shocks – quasars: supermassive black holes

1 Introduction

The existence of supermassive black holes (SMBHs) in most massive galaxies today poses significant challenges to the paradigm of hierarchical structure formation (Kormendy & Richstone, 1995; Ferrarese & Merritt, 2000; Ferrarese & Ford, 2005; Gebhardt et al., 2000; Di Matteo et al., 2005; Beifiori et al., 2012; McConnell & Ma, 2013). In particular, it is not known how some SMBHs reached masses of 10  by 7, less than a Gyr after the big bang (Fan et al., 2002, 2006; Mortlock et al., 2011). The two leading contenders for the origins of SMBHs are 100 - 500  Population III (Pop III) stars at 20 and supermassive (10 - 10 ) Pop III stars at 10 - 15 (e.g., Rees, 1984; Madau & Rees, 2001; Volonteri, 2010, 2012; Karlsson et al., 2013).

The main argument against conventional Pop III stars as SMBH seeds is that their BHs must accrete at the Eddington limit continuously down to 7 to become supermassive. How they could sustain such growth is unclear because they form in low-density relic H ii regions that delay initial accretion (Whalen et al., 2004; Johnson & Bromm, 2007). Later, when they do accrete, the relatively shallow dark matter potential wells in which they reside cannot retain their fuel supply because of radiative feedback (Alvarez et al., 2009), so they grow at well below the Eddington limit (Park & Ricotti, 2011, 2012, 2013). Furthermore, many low-mass Pop III black holes are ejected from their halos (and thus their fuel supply) at birth by asymmetries in their explosion engines (Whalen & Fryer, 2012).

For these reasons, there has been growing interest in supermassive stars (SMSs) as candidates for SMBH seeds (e.g., Loeb & Rasio, 1994; Bromm & Loeb, 2003; Begelman et al., 2006; Bromm & Yoshida, 2011; Johnson et al., 2013c). Such stars could form in atomically-cooled halos at 15. In this scenario, a primeval galaxy forms in a strong Lyman-Werner (LW) UV background that sterilizes its constituent halos of H prior to assembly (Agarwal et al., 2012; Johnson et al., 2013a), preventing them from forming stars. When the protogalaxy reaches a mass of 10  and virial temperature of 10 K, atomic cooling triggers the rapid collapse of gas at its center at rates of up to 1  yr (Lodato & Natarajan, 2006; Spaans & Silk, 2006; Wise et al., 2008; Regan & Haehnelt, 2009; Shang et al., 2010; Latif et al., 2013a, b; Schleicher et al., 2013). In this manner, a 10 - 10  Pop III star could be formed in less than a Myr.

The evolution of supermassive protostars and stars has been studied with both analytical and numerical techniques for over 40 years (Fowler, 1966; Wheeler, 1977; Bond et al., 1984; Carr et al., 1984; Fuller et al., 1986; Bisnovatyi-Kogan, 1998; Baumgarte & Shapiro, 1999; Begelman, 2010; Hosokawa et al., 2013; Schleicher et al., 2013). Most SMSs are thought to become fully convective and then later collapse to massive BHs (e.g., Shapiro & Teukolsky, 1979; Shibata & Shapiro, 2002; Reisswig et al., 2013). However, previous calculations of SMS evolution have been done at low resolution, considered a relatively coarse grid of models in mass, and in some cases ignored first-order, post-Newtonian general relativistic corrections to gravity. The latter could play an important role in the evolution of 10 - 10  stars, especially during their pre-explosion phase.

We have revisited the evolution of Pop III SMSs with high-resolution one-dimensional (1D) simulations with updated nuclear reaction rates (Heger et al., 2001; Heger & Woosley, 2002) and post-Newtonian corrections to gravity. We have found a 55,500  star that explodes as a highly energetic thermonuclear supernova (SN) instead of collapsing to a BH (see also Montero et al., 2012a). Our discovery suggests that there may be a range of high-mass stars that can die as SNe. In Section 2 we describe our stellar evolution model. The explosion of the star is examined in 1D in Section 3 and in two dimensions (2D) in Section 4. We discuss the potential impact of this explosion on its host galaxy and its visibility at high redshift in Section 5.

2 Numerical Model

The SMS is evolved from the beginning of the main sequence to the onset of collapse and then explosion in 1D in KEPLER. To verify these processes in a separate code, and to capture the violent fluid instabilities that can occur during the SN, we also model the collapse and explosion of the SMS in 2D in CASTRO with initial conditions taken from KEPLER. Combining 1D and 2D simulations in this manner allows us to realistically simulate all phases of the SMS within a practical computational budget.

2.1 Kepler

KEPLER (Weaver et al., 1978) is a 1D Lagrangian hydrodynamics and stellar evolution code. It includes nuclear burning and mixing due to convection. For a given stellar mass, the initial profiles of density and temperature of a star are determined by solving the Lane-–Emden equation (Chandrasekhar, 1939). We use the 19-isotope APPROX nuclear reaction network (Weaver et al., 1978; Timmes, 1999), which includes heavy-ion reactions, alpha-chain reactions, hydrogen burning cycles, photo-disintegration of heavy nuclei, and energy loss through thermal neutrinos. Nuclear burning is self-consistently coupled to hydrodynamics, and we also account for energy deposition due to radioactive decay of . KEPLER uses the Helmholtz equation of state (EOS; Timmes & Swesty, 2000), which includes contributions from degenerate and non-degenerate relativistic and non-relativistic electrons, electron-positron pair production, and radiation. The star is partitioned into 1201 zones in mass for an effective resolution of 46.2 . Consistent with the usual convention that massive Pop III stars lose little of their mass over their lives (Kudritzki, 2000; Vink et al., 2001; Baraffe et al., 2001; Krtička & Kubát, 2006; Ekström et al., 2008), we turn off mass loss in our models.

When stars become extremely massive, 1000 , general relativity (GR) must be taken into account in gravity in stellar evolution models. Instead of full GR, we apply first-order post-Newtonian GR corrections to gravity in our calculations. We use the Tolman-Oppenheimer-Volkoff equation for hydrostatic equilibrium in GR to calculate the pressure (Zeldovich & Novikov, 1971; Kippenhahn & Weigert, 1990):


where is the radius of star, is the speed of light, is the gravitational constant, and is the enclosed relativistic mass at , which is the rest mass plus total energy (internal plus gravitational) divided by . The relativistic density is therefore , where and are the rest-mass and total energy densities. When , Equation (1) reduces to Newton’s Law,


The effective gravity is calculated from


The term in is negligible in our models. It only becomes comparable to the rest energy when becomes 10 K (, where is the proton mass). Equation (3) therefore becomes


where is the Newtonian gravity field, which is determined from the gravity equation in KEPLER and from the monopole approximation to gravity for nearly spherical matter distributions in CASTRO.

2.2 Castro

CASTRO is a multidimensional adaptive mesh refinement (AMR) hydrodynamics code. It has a second-order unsplit Godunov hydro scheme and block-structured AMR (Almgren et al., 2010; Zhang et al., 2011). We use the same Helmholtz EOS, APPROX reaction network, energy deposition due to  and GR corrections to pressure and gravity in CASTRO as in KEPLER. As in KEPLER, this reaction network is sufficient to capture the key nucleosynthetic pathways and energetics of SMS evolution and explosion. CASTRO evolves mass fractions for each isotope with its own advection equation. The monopole approximation is used for self-gravity, in which a 1D gravitational potential is constructed from the radial average of the density and then applied to gravitational force updates everywhere in the AMR hierarchy. This approximation is well-suited to the nearly spherical symmetry of the star and is quite efficient.

We port the star from KEPLER onto a 2D cylindrical coordinate grid in and in CASTRO with the conservative mapping scheme of Chen et al. (2013). This is done at the onset of collapse, 600 seconds before the core of the star reaches maximum compression and the most violent burning begins. We simulate the half star in 2D, so the mesh is 2 10 cm in and 4 10 cm in , which just accommodates the entire star. The coarse grid has 256 zones in and 512 zones in , with up to two levels of AMR refinement for an additional factor of up to 16 in resolution. The inner core, where most of the explosive burning occurs, is always at the highest resolution; additional criteria for mesh refinement depend on gradients of the density, velocity, and pressure. The highest spatial resolution of our simulation is about 5 10cm to resolve the length scale of nuclear burning. Reflecting and outflow boundary conditions are set on the inner and outer boundaries in both and , respectively. The simulation is halted when the SN ejecta become frozen in mass coordinates (homologous expansion).

3 The 1D Kepler Explosion

The SMS has a radius of , an effective surface temperature of about 7.19 10 K, and a luminosity of during its pre-collapse phase. The star evolves about 1.69 Myr before beginning to collapse. When it begins helium burning, the central density and temperature are 10 and 2.0 10 K, respectively. Radiation dominates the pressure in the core, and the adiabatic index, , is slightly above 4/3. At this density and temperature, even the most energetic photons in the Maxwellian distribution cannot create electron-positron pairs and cause to fall below 4/3. Pair production therefore does not trigger collapse in the SMS. Instead, the energy density of the thermal photons in the high temperature and very low density of the core begins to affect the gravitational field by contributing to the source term in Einstein’s field equation, in effect becoming an additional source of gravity. The ratio of the radiative pressure to the relativistic density becomes 1-3 10. This causes in the core to fall below 4/3, and it begins to contract.

Although the temperature and density in the core now rise rapidly and accelerate nuclear burning, they are not high enough to ignite carbon burning, which is the next stage after helium burning. Helium instead begins to burn explosively in the core, burning to carbon first through the triple reaction and then to , , and through capture reactions. These reactions release enough energy to reverse collapse. We show the evolution of radial velocities in the star in Figure 1. The onset and reversal of collapse is evident, together with the formation of the outgoing shock, which propagates outward at several thousand kilometers per second. The shock breaks through the surface of the star 11,880 seconds after core bounce.

The binding energy of the star prior to collapse is 5.76 10 erg, but explosive helium burning releases 6.52 10 erg. We show the evolution of density and temperature at the center of the star in Figure 2. They are initially 20 and 3.6 10 K, and then rise to peak values of 360 and 8.26 10 K at core bounce before beginning to fall. The relative magnitudes of the binding and explosion energies together with the rapidly falling density at the center of the star strongly suggest that the SN completely unbinds the star, with no BH formation.

4 The 2D Castro Explosion

We initialize CASTRO with the KEPLER profile of the collapsing star 600 seconds prior to maximum core compression. In the beginning of the CASTRO run, the core temperature and density are 8.21 10 K and and 356 , respectively. Element masses as a function of radius are shown in Figure 3. The mass of the helium core is 31,000, and it also contains , , ,  and . As in KEPLER, the collapse is reversed by nuclear energy release, but violent fluid instabilities now arise in the core during the explosion. Helium burning creates pressure gradients that are opposite to the gradients of density and mass fraction, which in turn gives rise to Rayleigh-Taylor instabilities. These instabilities are more violent than those in non-reactive flows because they mix the fuel with hot ash that then quickly burns the fuel, which in turn enhances the pressure gradient and drives more mixing. This explosion released 8.82 10 erg, slightly more than in the 1D model because of the additional burning due to mixing.

Burning continues until the shock reaches the hydrogen envelope. Table 1 shows elemental abundances before and after the explosion. Energy release in this SN is primarily due to the burning of helium, oxygen and neon, which consumes 757 , 1116 , and 291 , respectively. Burning in turn yields 1195  of and 970  of . Few isotopes beyond are synthesized, with 1  of being formed, not enough to be detected. Since the hydrogen envelope of the SMS is not as extended as in red supergiants, the shock does not plow up much mass when it crashes into it. The reverse shock is therefore weak and does not drive further mixing. We show the degree to which , , , have mixed by the time the shock has broken out of the surface in Figure 4. Most mixing has now ceased. The major drivers of mixing in this model are fluid instabilities that emerge during burning. Mixing can cause heavy elements to be dredged up from deeper layers and appear in the SN spectra at early times.

In KEPLER models that exclude GR, the star collapses to a BH instead of exploding, because its core continues evolving through helium burning then encounters the pair instability before it ignites carbon burning. Once the collapse of core triggered by pair instabilities happens, the nuclear burning is not able to halt the collapse at this time and the entire star eventually dies as a black hole. Likewise, if the star is 56,000  it collapses to a BH, even when GR is included. This suggests that there is a narrow window in mass around 55,500  for the progenitor masses of general relativistic supernovae (GSNe). More masses need to be tested, and more extensive nuclear reaction networks may be needed to capture all the nucleosynthetic pathways in GSNe. Our KEPLER and CASTRO simulations show that GSNe are only weakly dependent on dimensionality.

GSNe are different from pair-instability SNe (PSNe; Heger & Woosley, 2002; Scannapieco et al., 2005; Chen et al., 2014) in every aspect: what triggers the collapse, what drives the explosion, the degree of mixing, and production. We compare properties of PSNe and GSNe in Table 2. In GSNe, only a trace amount of ( 1 ) is produced at the edge of the oxygen-burning shell by capture. PSNe can synthesize up to 50  of through explosive burning. The light curves of GSNe are mainly powered by the thermal emission by hot ejecta and the conversion of kinetic energy into entropy by the shock, not decay. Large masses of elements with atomic masses between and are synthesized during the explosion, and they are all dispersed into the surrounding medium.

5 Discussion and Conclusion

We have discovered that Pop III stars with masses of 55,500  may explode as SNe instead of collapsing to BHs. GR effects, rather than the pair instability, trigger the explosion, which at 9 10 erg is the most energetic thermonuclear SN known. Mixing in 2D enhances burning and thermonuclear yields during the explosion, which completely unbinds the star and leaves no compact remnant. Energy release is primarily due to explosive helium burning after the onset of central helium burning. The explosion yields mostly elements between carbon and silicon, with almost no iron group elements. Besides enhancing burning, mixing during the SN can dredge up heavy elements from deeper layers and cause them to appear at earlier times in spectra. Montero et al. (2012b) recently studied the relativistic collapse of SMS and discovered that explosions of SMS occurred even for stellar masses up to , which are powered by the hydrogen burning due to hot CNO cycle. In their simulations, they use simplified stellar models, with a single fluid of fixed hydrogen, helium, and CNO metal abundance. More importantly, all of their exploded models are associated non zero metallicity, however, there is no metal inside the Pop III stars when they are born. Hence the hot CNO cycle may not occur for the Pop III SMS at early on. Instead, our SMS model evolves from the main sequence through collapse to explosion with the detailed stellar physics which self-consistently produces CNO from the primordial elements. This provides a more realistic pre-supernova model. The differences in pre-supernova models as well as other microphysics used can significantly alter the explosion mechanics that may explain the discrepancies between their findings and ours.

Our simulations are approximate for several reasons. First, we do not model the evolution of the protostar from much lower masses. Instead, the star is initialized at the beginning of the main sequence in KEPLER. Next, we did not evolve the star under the heavy ongoing accretion that gave birth to the star. We also did not consider stellar rotation, which could lower the mass at which the star explodes (e.g., Chatzopoulos & Wheeler, 2012; Chatzopoulos et al., 2013). There is growing evidence that many Pop III stars are born with high rates of spin (Stacy et al., 2011; Greif et al., 2012). Finally, we did not include UV feedback from the star, which could regulate its accretion rates. However, in some cases luminosity from the star is thought to terminate accretion (Johnson et al., 2012), so our assumption that the star has a constant mass is plausible. Efforts are now underway to survey Pop III protostellar evolution under a variety of accretion rates and SMS evolution with much larger and finer grids in mass.

What effect would such an explosion have on the protogalaxy that gave birth to the star? The answer depends on the ambient density of the SN. If the SMS grows by accretion through a disk its UV radiation will break free of the disk and ionize the surrounding envelope. In these circumstances the star would explode in low densities, and Johnson et al. (2013b) show that the SN would drive all the gas from the protogalaxy, perhaps engulfing nearby protogalaxies with metals. Much of this material would later fall back to the halo on timescales of 50 - 100 Myr and form stars. If accretion is spherical, then the star will not ionize its envelope and it will explode in very high densities (Johnson et al., 2012; Hosokawa et al., 2013; Schleicher et al., 2013). In this scenario, Whalen et al. (2013a, b) find that much of the energy of the explosion is promptly radiated away by bremsstrahlung X-rays and inverse Compton scattering of cosmic microwave background (CMB) photons, but that the SN remnant still expands to roughly the virial radius of the halo, 1 kpc, and then collapses back into the halo. Collapse thoroughly mixes baryons in the halo with metals and may trigger a starburst that would easily distinguish this protogalaxy from its dimmer and less rapidly evolving neighbors.

Radiation hydrodynamical simulations show that the GSN would be visible at 20 to future all-sky NIR surveys by Euclid, the Wide-Field Infrared Survey Telescope (WFIRST) and the Wide-field Imaging Surveyor for High-Redshift (WISH) (Whalen et al., 2013c). The wide survey fields and high sensitivities of these missions would enable them to detect such an event anywhere in the universe, even if their numbers are small. Furthermore, a single GSN can create 100 times the chemical yield of a PSN. But unlike PSNe, which synthesize more iron group elements, GSNe would mainly enrich primordial gas with elements from to . Traces of GSNe might therefore be found in early galaxies that are deficient but enhanced with and (Keller et al., 2014). Whether in future NIR campaigns or in the fossil abundance record, the detection of GSNe in the early universe may soon signal the birth of SMBH seeds and the first quasars.

The authors thank the anonymous referee for reviewing this manuscript and providing insightful comments, and the members of CCSE at LBNL for help with CASTRO. We also thank Volker Bromm, Dan Kasen, Lars Bildsten, John Bell, and Adam Burrows for many useful discussions. K.C. was supported by an IAU-Gruber Fellowship, a Stanwood Johnston Fellowship, and a KITP Graduate Fellowship. A.H. was supported by a Future Fellowship from the Australian Research Council (ARC FT 120100363). D.J.W. was supported by the Baden-Württemberg-Stiftung by contract research under the programme Internationale Spitzenforschung II (grant P-LS-SPII/18). All numerical simulations were performed at the University of Minnesota Supercomputing Institute and the National Energy Research Scientific Computing Center. This work was supported by the DOE grants DE-SC0010676, DE-AC02-05CH11231, DE-GF02-87ER40328, DE-FC02-09ER41618 and by NSF grants AST-1109394 and PHY02-16783. Work at LANL was done under the auspices of the National Nuclear Security Administration of the US Department of Energy at Los Alamos National Laboratory under Contract No. DE-AC52-06NA25396.
Figure 1: Radial velocities in the star from the onset of collapse to explosion. Infall during collapse reaches velocities above 7000 km sec.
Figure 2: Evolution of central temperature and density during collapse and explosion. In the final 10,000 seconds they reach maximum values of 360 and 8.26 10 K, respectively, before dropping rapidly.
Figure 3: Elemental masses at the beginning of the CASTRO simulation. The mass of the helium core was about 31,000 . Helium abundances at the center of the star are fairly large, which implies that central helium burning is still in progress when the star collapses. Hydrogen shell burning is also still going on at the base of the envelope of the star.
Figure 4: Mixing in , , , prior to shock breakout. Mixing on these scales might be manifest in the spectra of the explosion. Few isotopes with atomic numbers larger than , such as , are formed.
Isotope Total
Before 8336 24902 922 7972 5110 7748 515 55505
After 8335 24145 919 6856 4819 8943 1485 55502
-1 -757 -3 -1116 -291 1195 970 -3
Table 1: Elemental masses before and after the explosion.

Note. – The loss of in total is due to the diffusion of the stellar envelope out of the simulation domain, which is only a tiny fraction of overall mass. It does not affect the explosion and nucleosynthesis.

Characteristic Property PSNe GSNe
Progenitor Mass ()
Collapse Trigger Pair Instability GR Instability
Explosive Burning ,
Production ()
Explosion Energy ()
Fluid Instabilities Reverse Shock Burning
Table 2: Comparison between PSNe and GSNe


  1. affiliation: Department of Astronomy & Astrophysics, University of California, Santa Cruz, CA 95064, USA
  2. affiliation: School of Physics and Astronomy, University of Minnesota, Minneapolis, MN 55455, USA
  3. affiliation: IAU Gruber Fellow;
  4. affiliation: Monash Centre for Astrophysics, Monash University, Victoria 3800, Australia
  5. affiliation: Department of Astronomy & Astrophysics, University of California, Santa Cruz, CA 95064, USA
  6. affiliation: Center for Computational Sciences and Engineering, Lawrence Berkeley National Lab, Berkeley, CA 94720, USA
  7. affiliation: T-2, Los Alamos National Laboratory, Los Alamos, NM 87545, USA
  8. affiliation: Universität Heidelberg, Zentrum für Astronomie, Institut für Theoretische Astrophysik, Albert-Ueberle-Str. 2, 69120 Heidelberg, Germany
  9. affiliation: XTD-PRI, Los Alamos National Laboratory, Los Alamos, NM 87545, USA


  1. Agarwal, B., Khochfar, S., Johnson, J. L., Neistein, E., Dalla Vecchia, C., & Livio, M. 2012, MNRAS, 425, 2854
  2. Almgren, A. S., et al. 2010, ApJ, 715, 1221
  3. Alvarez, M. A., Wise, J. H., & Abel, T. 2009, ApJ, 701, L133
  4. Baraffe, I., Heger, A., & Woosley, S. E. 2001, ApJ, 550, 890
  5. Baumgarte, T. W., & Shapiro, S. L. 1999, ApJ, 526, 941
  6. Begelman, M. C. 2010, MNRAS, 402, 673
  7. Begelman, M. C., Volonteri, M., & Rees, M. J. 2006, MNRAS, 370, 289
  8. Beifiori, A., Courteau, S., Corsini, E. M., & Zhu, Y. 2012, MNRAS, 419, 2497
  9. Bisnovatyi-Kogan, G. S. 1998, ApJ, 497, 559
  10. Bond, J. R., Arnett, W. D., & Carr, B. J. 1984, ApJ, 280, 825
  11. Bromm, V., & Loeb, A. 2003, ApJ, 596, 34
  12. Bromm, V., & Yoshida, N. 2011, ARA&A, 49, 373
  13. Carr, B. J., Bond, J. R., & Arnett, W. D. 1984, ApJ, 277, 445
  14. Chandrasekhar, S. 1939, An introduction to the study of stellar structure
  15. Chatzopoulos, E., & Wheeler, J. C. 2012, ApJ, 748, 42
  16. Chatzopoulos, E., Wheeler, J. C., & Couch, S. M. 2013, ApJ, 776, 129
  17. Chen, K.-J., Heger, A., & Almgren, A. S. 2013, Astronomy and Computing, 3–4, 70
  18. Chen, K.-J., Heger, A., Woosley, S., Almgren, A., & Whalen, D. 2014, ArXiv e-prints
  19. Di Matteo, T., Springel, V., & Hernquist, L. 2005, Nature, 433, 604
  20. Ekström, S., Meynet, G., Chiappini, C., Hirschi, R., & Maeder, A. 2008, A&A, 489, 685
  21. Fan, X., Narayanan, V. K., Strauss, M. A., White, R. L., Becker, R. H., Pentericci, L., & Rix, H.-W. 2002, AJ, 123, 1247
  22. Fan, X., et al. 2006, AJ, 132, 117
  23. Ferrarese, L., & Ford, H. 2005, Space Sci. Rev., 116, 523
  24. Ferrarese, L., & Merritt, D. 2000, ApJ, 539, L9
  25. Fowler, W. A. 1966, ApJ, 144, 180
  26. Fuller, G. M., Woosley, S. E., & Weaver, T. A. 1986, ApJ, 307, 675
  27. Gebhardt, K., et al. 2000, ApJ, 539, L13
  28. Greif, T. H., Bromm, V., Clark, P. C., Glover, S. C. O., Smith, R. J., Klessen, R. S., Yoshida, N., & Springel, V. 2012, MNRAS, 424, 399
  29. Heger, A., & Woosley, S. E. 2002, ApJ, 567, 532
  30. Heger, A., Woosley, S. E., Martínez-Pinedo, G., & Langanke, K. 2001, ApJ, 560, 307
  31. Hosokawa, T., Yorke, H. W., Inayoshi, K., Omukai, K., & Yoshida, N. 2013, ApJ, 778, 178
  32. Inayoshi, K., Hosokawa, T., & Omukai, K. 2013, MNRAS, 431, 3036
  33. Johnson, J. L., & Bromm, V. 2007, MNRAS, 374, 1557
  34. Johnson, J. L., Dalla, V. C., & Khochfar, S. 2013a, MNRAS, 428, 1857
  35. Johnson, J. L., Whalen, D. J., Even, W., Fryer, C. L., Heger, A., Smidt, J., & Chen, K.-J. 2013b, ApJ, 775, 107
  36. Johnson, J. L., Whalen, D. J., Fryer, C. L., & Li, H. 2012, ApJ, 750, 66
  37. Johnson, J. L., Whalen, D. J., Li, H., & Holz, D. E. 2013c, ApJ, 771, 116
  38. Karlsson, T., Bromm, V., & Bland-Hawthorn, J. 2013, Reviews of Modern Physics, 85, 809
  39. Keller, S. C., et al. 2014, ArXiv e-prints
  40. Kippenhahn, R., & Weigert, A. 1990, Stellar Structure and Evolution
  41. Kormendy, J., & Richstone, D. 1995, ARA&A, 33, 581
  42. Krtička, J., & Kubát, J. 2006, A&A, 446, 1039
  43. Kudritzki, R. 2000, in The First Stars, ed. A. Weiss, T. G. Abel, & V. Hill, 127–+
  44. Latif, M. A., Schleicher, D. R. G., Schmidt, W., & Niemeyer, J. 2013a, MNRAS, 433, 1607
  45. —. 2013b, MNRAS, 430, 588
  46. Lodato, G., & Natarajan, P. 2006, MNRAS, 371, 1813
  47. Loeb, A., & Rasio, F. A. 1994, ApJ, 432, 52
  48. Madau, P., & Rees, M. J. 2001, ApJ, 551, L27
  49. McConnell, N. J., & Ma, C.-P. 2013, ApJ, 764, 184
  50. Montero, P. J., Janka, H.-T., & Müller, E. 2012a, ApJ, 749, 37
  51. —. 2012b, ApJ, 749, 37
  52. Mortlock, D. J., et al. 2011, Nature, 474, 616
  53. Park, K., & Ricotti, M. 2011, ApJ, 739, 2
  54. —. 2012, ApJ, 747, 9
  55. —. 2013, ApJ, 767, 163
  56. Rees, M. J. 1984, ARA&A, 22, 471
  57. Regan, J. A., & Haehnelt, M. G. 2009, MNRAS, 396, 343
  58. Reisswig, C., Ott, C. D., Abdikamalov, E., Haas, R., Moesta, P., & Schnetter, E. 2013, arXiv:1304.7787
  59. Scannapieco, E., Madau, P., Woosley, S., Heger, A., & Ferrara, A. 2005, ApJ, 633, 1031
  60. Schleicher, D. R. G., Palla, F., Ferrara, A., Galli, D., & Latif, M. 2013, A&A, 558, A59
  61. Shang, C., Bryan, G. L., & Haiman, Z. 2010, MNRAS, 402, 1249
  62. Shapiro, S. L., & Teukolsky, S. A. 1979, ApJ, 234, L177
  63. Shibata, M., & Shapiro, S. L. 2002, ApJ, 572, L39
  64. Spaans, M., & Silk, J. 2006, ApJ, 652, 902
  65. Stacy, A., Bromm, V., & Loeb, A. 2011, MNRAS, 413, 543
  66. Timmes, F. X. 1999, ApJS, 124, 241
  67. Timmes, F. X., & Swesty, F. D. 2000, ApJS, 126, 501
  68. Vink, J. S., de Koter, A., & Lamers, H. J. G. L. M. 2001, A&A, 369, 574
  69. Volonteri, M. 2010, A&A Rev., 18, 279
  70. —. 2012, Science, 337, 544
  71. Weaver, T. A., Zimmerman, G. B., & Woosley, S. E. 1978, ApJ, 225, 1021
  72. Whalen, D., Abel, T., & Norman, M. L. 2004, ApJ, 610, 14
  73. Whalen, D. J., & Fryer, C. L. 2012, ApJ, 756, L19
  74. Whalen, D. J., Johnson, J. L., Smidt, J., Heger, A., Even, W., & Fryer, C. L. 2013a, ApJ, 777, 99
  75. Whalen, D. J., Johnson, J. L., Smidt, J., Meiksin, A., Heger, A., Even, W., & Fryer, C. L. 2013b, ApJ, 774, 64
  76. Whalen, D. J., et al. 2013c, ApJ, 778, 17
  77. Wheeler, J. C. 1977, Ap&SS, 50, 125
  78. Wise, J. H., Turk, M. J., & Abel, T. 2008, ApJ, 682, 745
  79. Zeldovich, Y. B., & Novikov, I. D. 1971, Relativistic astrophysics. Vol.1: Stars and relativity
  80. Zhang, W., Howell, L., Almgren, A., Burrows, A., & Bell, J. 2011, ApJS, 196, 20
This is a comment super asjknd jkasnjk adsnkj
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters
Comments 0
Request comment
The feedback must be of minumum 40 characters
Add comment
Loading ...

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test description