Impact of baryonic streaming velocities on the formation of supermassive black holes via direct collapse
aa r X i v : . [ a s t r o - ph . C O ] M a r Mon. Not. R. Astron. Soc. , 1– ?? (2009) Printed 26 June 2018 (MN LaTEX style file v2.2) Impact of baryonic streaming velocities on the formationof supermassive black holes via direct collapse
M. A. Latif, J. C. Niemeyer, D. R. G. Schleicher Institut f¨ur Astrophysik, Georg-August-Universit¨at,
Friedrich-Hund-Platz 1, D-37077 G¨ottingen, Germany
ABSTRACT
Baryonic streaming motions produced prior to the epoch of recombination becamesupersonic during the cosmic dark ages. Various studies suggest that such streamingvelocities change the halo statistics and also influence the formation of PopulationIII stars. In this study, we aim to explore the impact of streaming velocities on theformation of supermassive black holes at z > via the direct collapse scenario. Toaccomplish this goal, we perform cosmological large eddy simulations for two halos ofa few times M ⊙ with initial streaming velocities of 3, 6 and 9 km / s . These massiveprimordial halos illuminated by the strong Lyman Werner flux are the potential cradlesfor the formation of direct collapse seed black holes. To study the evolution for longertimes, we employ sink particles and track the accretion for 10,000 years. Our findingsshow that higher streaming velocities increase the circular velocities from about 14 km / s to 16 km / s . They also delay the collapse of halos for a few million years, but donot have any significant impact on the halo properties such as turbulent energy, radialvelocity, density and accretion rates. Sink particles of about ∼ M ⊙ are formedat the end of our simulations and no clear distribution of sink masses is observed inthe presence of streaming motions. It is further found that the impact of streamingvelocities is less severe in massive halos compared to the minihalos as reported in theprevious studies. Key words: methods: numerical – cosmology: theory – early Universe – galaxies:formation
Baryonic acoustic oscillations prior to the epoch of recombi-nation generated baryonic streaming motions of ∼
30 km / s (Tseliakhovich & Hirata 2010). During recombination, theradiation dominated plasma transformed to neutral gas andthe sound speed dropped from relativistic to thermal veloci-ties of × − c . The latter is smaller than the relative motionof gas and dark matter particles. Consequently, the stream-ing motions of baryons became supersonic with typical Machnumbers of about 5. Such streaming velocities impede smallscale density perturbations, may lead to the suppression ofhalo abundances and also enhance the bias and clustering ofthe halos (Tseliakhovich & Hirata 2010; Dalal et al. 2010;Maio et al. 2011; O’Leary & McQuinn 2012; Fialkov et al.2012). This bulk motion may further move the baryons outof the dark matter potentials. Their impact on the forma-tion of supermassive black holes in massive primordial halosremains unexplored.Population III stars are the first sources of light tobe formed at the end of cosmic dark ages (Abel et al.2000; Clark et al. 2011; Greif et al. 2012; Stacy et al. 2012; Bovino et al. 2013; Latif et al. 2013c; Bovino et al. 2013).The role of streaming velocities in the context of PopulationIII star formation in minihalos of − M ⊙ has been stud-ied by Stacy et al. (2011) and Greif et al. (2011), and theyfound that typical streaming velocities of / s at z = delay the formation of the first stars, and increase the virialmass required for molecular hydrogen cooling to become ef-fective. The enhanced mass for H cooling may influencethe statistics of minihalos and even the presence of higherturbulent energy may reduce the masses of the first starsby inducing fragmentation. In a recent study, Tanaka et al.(2013) have found that streaming motions significantly re-duce the number density of stellar seed black hole at z > ,while their overall impact on the formation of high-redshiftBHs is negligible. They have also proposed that streamingvelocities of 2-3 times the root mean square value could de-lay the formation of stars in minihalos until the halo massreaches the threshold value for atomic cooling and may fur-ther facilitate the formation of supermassive black holes bythe direct collapse (Tanaka & Li 2013). It is expected that c (cid:13) Latif et al. streaming motions may enhance the turbulent accretion andlead to higher seed black hole masses.The existence of ∼ M ⊙ supermassive black holeshas been revealed from the observations of quasarsat z > (Fan et al. 2003, 2006; Willott et al. 2010;Mortlock et al. 2011; Venemans et al. 2013). How suchmassive objects are assembled in the infant Universeand what their potential progenitors are remains anunfathomable conundrum. Numerous theoretical modelspropose various pathways such as accretion and merg-ing of stellar mass black holes (Haiman & Loeb 2001;Haiman 2004; Tanaka & Haiman 2009; Whalen & Fryer2012), run-away collapse of dense stellar cluster dueto the relativistic instability (Portegies Zwart et al.2004; Omukai et al. 2008; Devecchi & Volonteri 2009)and the direct collapse of a protogalactic gas cloud(Oh & Haiman 2002; Bromm & Loeb 2003; Spaans & Silk2006; Begelman et al. 2006; Lodato & Natarajan 2006;Dijkstra et al. 2008; Djorgovski et al. 2008; Shang et al.2010; Johnson et al. 2010; Schleicher et al. 2010; Latif et al.2011; Prieto et al. 2013; Latif et al. 2013a,e; Whalen et al.2013; Aykutalp et al. 2013; Spaans 2013; Latif et al. 2013).The growth of stellar mass black holes is extremely chal-lenging as they have to accrete at the Eddington limitalmost all the time to reach the observed masses. On theother hand, the direct collapse model provides massiveseeds of − M ⊙ which may grow at relatively moderateaccretion rates to form billion solar mass black holes.Massive primordial halos of − M ⊙ formed at z = and irradiated by the strong Lyman-Werner fluxare the potential cradles for the birth of supermassive blackholes forming via the direct collapse scenario. Such condi-tions can be achieved in the early Universe and their oc-currence is frequent enough to produce the observed num-ber density of black holes (Dijkstra et al. 2008; Shang et al.2010; Agarwal et al. 2012), see also Inayoshi & Omukai(2012); Van Borm & Spaans (2013). Numerical simulationsstudying the collapse of such halos show that fragmen-tation remains suppressed in the presence of a strongphoto-dissociating background flux and massive objects arelikely to be formed (Bromm & Loeb 2003; Wise et al. 2008;Regan & Haehnelt 2009; Latif et al. 2011, 2013a). In a re-cent study, Latif et al. (2013e) have evolved the simulationsfor × years after the initial collapse and have shown thatseed black holes of about M ⊙ are formed. This study fur-ther demonstrates the feasibility of the direct collapse sce-nario.In this article, we explore the impact of baryonic stream-ing velocities on the formation of seed black holes via thedirect collapse mechanism. To achieve this goal, we per-form large eddy simulations (LES) with initial streamingvelocities of × (z / / s . To further investigate therole of large streaming velocities, we perform comparisonruns with streaming velocities of × (z / / s and × (z / / s . We make use of sink particles to fol-low the accretion for longer times and employ a fixed Jeansresolution of 32 cells during the entire course of the simula-tions. This study allows us to investigate the role of baryonicstreaming motions in the assembling of seed black holes.This paper is organized in the following way. In thesecond section, we briefly summarize the simulations setupand numerical methods employed. We present our results in the third section and confer our conclusions in the fourthsection. The simulations presented here are performed using the opensource cosmological simulations code ENZO (O’Shea et al.2004; The Enzo Collaboration et al. 2013) which is a paral-lel, Eulerian, grid based, an adaptive mesh refinement code.Our simulation setup is exactly the same as described in anumber of previous studies Latif et al. (2013e,a,b,d). Here,we provide a brief summary of the simulations setup andrefer to the above mentioned articles for details. The sim-ulations are started at z = with cosmological initialconditions and two nested refinement levels in addition tothe top grid are employed each with a resolution of cells. Our computational domain has a periodic box sizeof / h . We employ 15 dynamical refinement levels inthe central 62 kpc region during the course of simulationsand use fixed Jeans resolution of 32 cells. To simulate theevolution of dark matter, we use 5767168 particles. Afterreaching the maximum refinement level, we insert sink par-ticles to follow the evolution for 10, 000 years. The detailsof the sinks algorithm can be found in Wang et al. (2010).We use standard parameters from the WMAP 7-year data( Ω m = . , Ω b = . , Ω λ = .
734 H =
71 kms − Mpc − )with a value of σ = . (Jarosik et al. 2011). To follow thethermal evolution, we solve the rate equations of the species H , H + , He , He + , He ++ , e − , H − , H , H + self-consistently alongwith the cosmological simulations. We further presume thata strong photo-dissociating background flux of strength in the units of J = − erg cm − s − Hz − sr − is produced bya star forming galaxy in the vicinity of the halo with a stel-lar radiation temperature of K. Indeed, such estimatesof the UV field strength are in accordance with previousstudies of Dijkstra et al. (2008) and Agarwal et al. (2012).The effect of self-shielding is ignored in these calculationswhich may raise the strength of the critical flux even further.We use the subgrid scale turbulence model of Schmidt et al.(2006) to include unresolved turbulent fluctuations. Theadaptively refined large eddy simulations (LES) approachis used to implement subgrid scale (SGS) turbulence modelin AMR cosmological simulations (Maier et al. 2009). A de-tailed discussion on this topic can be found in dedicatedstudies (Schmidt et al. 2006, 2009; Latif et al. 2013a). Ourapproach of implementing the streaming motions is the sameas in Greif et al. (2011). Additional constant streaming ve-locities of 3, 6 and 9 km/s were added to each grid cell inthe x-direction at the start of our simulations (z=100).
In total, eight cosmological large eddy simulations are per-formed for initial baryonic streaming velocities of 0, 3, 6and 9 km / s for two distinct halos. The properties of the sim-ulated halos such as masses, collapse redshifts and spins arelisted in table 1. They have typical masses of a few times M ⊙ and collapse redshifts of 10.8 and 13.5, respectively.We have computed the circular velocity (also called cool-ing threshold velocity) which is found to be a good measure c (cid:13) , 1– ?? lack hole formation in the early universe V s [km/s] V c [ k m / s ] V s [km/s] ∆ z Figure 1.
The circular velocities and delay in the collapse redshifts for halo A and B are shown in this figure. The circular velocity ofthe halo (in the left panel) and the difference is the collapse redshift are shown for the streaming velocities of 0, 3, 6 and 9 km / s . Blackline represents halo A while green line stands for halo B. Radius [AU] -4 -2 ˙ M [ M ⊙ / y r ] -2 -1 V r o t / V c S0S1S2S3 M a ss [ M ⊙ ] Radius [AU] −20−15−10−50 V r a d [ k m / s ] E t u r b [ e r g/g ] -27 -24 -21 -18 -15 ρ [ g/ c m ] Figure 2.
The properties of the halo A at its collapse redshift are shown here. The top two panel show the profiles of enclosed massand density. Turbulent energy and the ratio of rotational to circular velocity are depicted in the middle panel. The bottom panels showthe profiles of mass accretion rates and radial infall velocity. S0,S1,S2 and S3 represent the simulations with initial streaming velocitiesof 0, 3, 6 and 9 km / s .c (cid:13) , 1– ?? Latif et al.
Table 1.
Properties of the simulated halos are listed here.Model Collapse Mass Spin parameter Redshift Streaming velocity at z = Circular velocity Sink mass M ⊙ λ z [ km / s ] [ km / s ] M ⊙ A (s0) . × . × A (s1) . × . × A (s2) . × . × A (s3) . × . × B (s0) . × . × B (s1) . × . × B (s2) . × . × B (s3) . × . ×
50 PC 50 PC 50 PC 50 PC -24 -23 -22 -21 -20 -19 -18 D e n s i t y [ g c m − ] -24 -23 -22 -21 -20 -19 -18 D e n s i t y [ g c m − ] Figure 3.
Density projections of the simulations at their collapse redshifts are shown for the runs with various streaming velocities.Top panels show the morphology of halo A for the streaming velocities of 0-9 km / s (from left to right) and bottom panels show themorphologies of halo B for the streaming velocities of 0-9 km / s (from left to right). of the halo’s virial temperature (Fialkov et al. 2012) and isplotted against the strength of streaming velocity in figure 1.It is found that the cooling velocity increases with enhancingthe magnitude of initial streaming velocity. This trend is ob-served for both halos and typical circular velocities are about . − . / s . Streaming motions delay the infall of gasinto the dark matter potential and enhance the halo masswhich results in a higher circular velocity. We also computedthe delay in halo collapse redshift which is depicted in figure1. We note that the difference in the collapse redshift forzero and extreme streaming velocity cases is 0.1 and 0.5 forour simulated halos. This delay is less significant comparedto the values reported for minihalos (Greif et al. 2011). Thiscomes from the fact that the gravitational potential of ourhalos is much deeper and the halos retain their gas with ashort time delay of few million years.The properties of the halo A as a representative caseat its collapse redshift are shown in figure 2 and comparedfor different initial streaming velocities. The density de-creases with radius and follows an R − behavior as expected for an isothermal collapse. The maximum density is about − g / cm . The specific turbulent energy is about erg / g and the radial velocity is about −
10 km / s which shows in-fall of gas into the center of the halo. The average accretionrate is about . ⊙ / yr at larger radii and decreases down to − M ⊙ / yr in the Jeans length. The ratio of rotational to cir-cular velocity is 0.3. The mass profile increases linearly withradius as expected from an isothermal collapse and declinessharply within the Jeans radius, where the density profilebecomes almost flat. No significant differences are observedin the above mentioned quantities with and without stream-ing motions.The state of the simulations at their collapse redshifts isrepresented by the density projections and is shown in figure3. The maximum density is − g / cm . The morphology ofthe halo is slightly different for each case of streaming mo-tions due to the different turbulence realizations in the halo.We noticed that a monolithic collapse occurs and fragmenta-tion remains suppressed. As mentioned in the previous sec-tion, we evolved the simulations for 10,000 years by employ- c (cid:13) , 1– ?? lack hole formation in the early universe
50 PC 50 PC 50 PC 50 PC -24 -23 -22 -21 -20 -19 -18 D e n s i t y [ g c m − ] -24 -23 -22 -21 -20 -19 -18 D e n s i t y [ g c m − ] Figure 4.
Same as figure 3 but at the final stage of simulations. Sink particles of the masses listed in the table 1 are overplotted on thedensity projections. Radius [AU] -4 -2 ˙ M [ M ⊙ / y r ] -2 -1 V r o t / V c S0S1S2S3 M a ss [ M ⊙ ] Radius [AU] −20−15−10−50 V r a d [ k m / s ] E t u r b [ e r g/g ] -27 -24 -21 -18 -15 ρ [ g/ c m ] Figure 5.
The properties of halo A at its final state are shown here. The top two panel show the profiles of enclosed mass and density.Turbulent energy and the ratio of rotational to circular velocity are depicted in the middle panel. The bottom panels show the profiles ofmass accretion rates and radial infall velocity. S0,S1,S2 and S3 represent the simulations with initial streaming velocities of 0, 3, 6 and9 km / s .c (cid:13) , 1– ?? Latif et al. Radius [AU] -3 -2 -1 Λ /4 π G ρ δ Λ therm +Λ therm −Λ turb +Λ turb −Λ SGS +Λ SGS − 10 Radius [AU] -3 -2 -1 Λ /4 π G ρ δ Λ therm +Λ therm −Λ turb +Λ turb −Λ SGS +Λ SGS − Figure 6.
The contribution of support terms for halo A with streaming velocities of zero (left) and 9 km/s (right). The local supportof thermal, turbulent and SGS is shown in this figure. The solid lines show the positive support of the quantities while dashed linesrepresent the negative support. For the definitions of support terms see Schmidt et al. 2013. ing sink particles after reaching the maximum refinementlevel. Sink particles are overplotted on the density projec-tions and are shown in figure 4. It is found that massive sinksof about M ⊙ are formed and their masses are listed intable 1 for the various initial streaming velocities. No cleartrend is observed in the masses of sinks for the cases withstreaming motions.We also show the central properties of halo A for arepresentative case at the final stage of the simulations infigure 5. With the passage of time, turbulent energy is in-creased by an order of magnitude. According to the Kol-mogorov scaling, the turbulent energy should decrease to-wards smaller radii for the homogenous turbulence but it isenhanced in the core of the halo for our case. This is be-cause of the higher turbulence production rate in the centerof halo mainly driven by the gravity. Such behavior reflectsthe decreasing dynamical times and is noted in a number ofprevious studies (Latif et al. 2013b,a,f). The ratio of v rot / v cir is also enhanced to 0.6 which shows that the halos have ahigh degree of rotational support. The latter may furtherdelay the accretion to the sink and may also slow down thecollapse, see Latif et al. (2013e). Overall, no significant dif-ferences are observed in the halo properties after varying thestreaming velocities. We further investigated the contribu-tions of local support terms which are computed by solvingthe differential equation for the rate of compressions of thegas (Schmidt et al. 2013). Support by thermal pressure, re-solved turbulence and SGS turbulence scaled by the grav-itational compression is shown for two representative casesin figure 6. In general, it seems that turbulence support be-comes important in the core of the halo and extends outto radii of 10 pc. Particularly, support of SGS turbulencebecomes important in the core of the halo and yields rele-vant contributions to the support against gravity. Overall,no qualitative difference is found in the support terms forthe cases with and without streaming motions. In all, we have performed eight cosmological large eddy sim-ulations for two distinct halos to study the impact of bary-onic streaming motions on the formation of supermassiveblack holes via the direct collapse scenario. To accomplishthis goal, we introduced the baryonic streaming velocities of3, 6 and 9 km / s in the x-direction at z = , and comparedthe results with and without streaming motions. We added15 dynamical refinement levels and a fixed Jeans resolutionof 32 cells during the entire course of simulations. We furtheremployed sink particles to follow the accretion on them for10,000 years after reaching the maximum refinement level.Our findings show that the threshold circular veloc-ity which defines the minimum cooling mass of the halois about 14-16 km / s and gets enhanced in the presence oflarge streaming velocities. The overall increase in the circu-lar velocity is 1-2 km / s and less significant compared to theminihalos. This is due to the fact that the halos in our simu-lations have masses of a few times M ⊙ , their gravitationalpotentials are much deeper and can retain sufficient gas evenin the presence of baryonic streaming motions. We also no-ticed that such an increase in the circular velocity delaysthe collapse for a few million years. Again, in the minihalosdelay was ∆ z = and in our case it is about ∆ z = . .Streaming velocities decay as + z , and their impact maybecome more significant if such halos are formed at redshiftaround 30. In this case, the energy ratio of streaming tocircular velocity (i.e., v s / v c ) may increase by a factor of 4,making their impact potentially more significant. Such haloswould thus correspond to high-sigma peaks, but the impactis likely still reduced compared to minihalos.No significant differences are found in the general prop-erties of the halos such as density, radial velocity, turbu-lent energy and accretion rates with and without bary-onic motions. Sink particles have typical masses of M ⊙ and are potential candidates for the formation of super-massive stars as an intermediate stage to supermassiveblack holes (Schleicher et al. 2013; Hosokawa et al. 2012;Ball et al. 2011; Johnson et al. 2013; Hosokawa et al. 2013).The observational imprint of such stars can be probed c (cid:13) , 1– ?? lack hole formation in the early universe with upcoming telescopes such as JWST and ATHENA+ (Whalen et al. 2013, 2014). No systematic trend is observedin the masses of sinks for streaming motions. We have ex-plored here even the extreme cases with streaming velocitiesof 6 and 9 km / s and their impact on the halo properties andsink masses is negligible.In a recent study Tanaka & Li (2013) have proposedthat extreme streaming velocities may facilitate the directcollapse scenario by enhancing the critical mass for the col-lapse to commence, suppressing the formation of stars andconsequently avoiding the metal enrichment. This study fur-ther suggests that under such conditions there is no need ofbackground UV flux to quench H formation as the later canbe dissociated by the collisional shocks at higher densities.This scenario requires additional mechanisms to suppressthe H cooling at higher densities. In a very recent workFernandez et al. (2014) have explored the feasibility of colli-sional dissociation of molecular hydrogen in the presence ofshocks and found that it is very difficult to avoid the forma-tion of H without background UV flux. Therefore, the needof a strong background UV flux seems necessary for the for-mation of massive black holes. Nevertheless, it is desirableto extend investigation of baryonic streaming velocities tomassive halos where H cooling is relevant in future studies. ACKNOWLEDGMENTS
The simulations described in this work were performed us-ing the Enzo code, developed by the Laboratory for Compu-tational Astrophysics at the University of California in SanDiego (http://lca.ucsd.edu). We thank Wolfram Schmidt foruseful discussions on the topic. We acknowledge researchfunding by Deutsche Forschungsgemeinschaft (DFG) un-der grant SFB / (projects A12, A15) and computingtime from HLRN under project nip00029. DRGS thanksthe DFG for funding via the Schwerpunktprogram SPP1573 “Physics of the Interstellar Medium” (grant SCHL / − ). The simulation results are analyzed using thevisualization toolkit for astrophysical data YT (Turk et al.2011). REFERENCES
Abel T., Bryan G. L., Norman M. L., 2000, ApJ, 540, 39Agarwal B., Khochfar S., Johnson J. L., Neistein E., DallaVecchia C., Livio M., 2012, MNRAS, 425, 2854Aykutalp A., Wise J. H., Meijerink R., Spaans M., 2013,ApJ, 771, 50Ball W. H., Tout C. A., ˙Zytkow A. N., Eldridge J. J., 2011,MNRAS, 414, 2751Begelman M. C., Volonteri M., Rees M. J., 2006, MNRAS,370, 289Bovino S., Grassi T., Latif M. A., Schleicher D. R. G., 2013,MNRAS, 434, L36Bovino S., Schleicher D. R. G., Grassi T., 2013, ArXiv e-printsBromm V., Loeb A., 2003, ApJ, 596, 34 http://athena2.irap.omp.eu/ Clark P. C., Glover S. C. O., Smith R. J., Greif T. H.,Klessen R. S., Bromm V., 2011, Science, 331, 1040Dalal N., Pen U.-L., Seljak U., 2010, JCAP, 11, 7Devecchi B., Volonteri M., 2009, ApJ, 694, 302Dijkstra M., Haiman Z., Mesinger A., Wyithe J. S. B., 2008,MNRAS, 391, 1961Djorgovski S. G., Volonteri M., Springel V., Bromm V.,Meylan G., 2008, ArXiv e-prints-0803.2862Fan X., Strauss M. A., Richards G. T., Hennawi J. F.,Becker R. H., White R. L., Diamond-Stanic A. M., 2006,AJ, 131, 1203Fan X., Strauss M. A., Schneider D. P., Becker R. H., WhiteR. L., Haiman Z., Gregg M., 2003, AJ, 125, 1649Fernandez R., Bryan G. L., Haiman Z., Li M., 2014, ArXive-printsFialkov A., Barkana R., Tseliakhovich D., Hirata C. M.,2012, MNRAS, 424, 1335Greif T. H., Bromm V., Clark P. C., Glover S. C. O., SmithR. J., Klessen R. S., Yoshida N., Springel V., 2012, MN-RAS, 424, 399Greif T. H., White S. D. M., Klessen R. S., Springel V.,2011, ApJ, 736, 147Haiman Z., 2004, ApJ, 613, 36Haiman Z., Loeb A., 2001, ApJ, 552, 459Hosokawa T., Omukai K., Yorke H. W., 2012, ApJ, 756, 93Hosokawa T., Yorke H. W., Inayoshi K., Omukai K.,Yoshida N., 2013, ArXiv e-prints:1308.4457Inayoshi K., Omukai K., 2012, MNRAS, 422, 2539Jarosik N., Bennett C. L., Dunkley J., Gold B., GreasonM. R., Halpern M., Hill R. S., Hinshaw G., 2011, ApJS,192, 14Johnson J. L., Khochfar S., Greif T. H., Durier F., 2010,MNRAS, pp 1427–+Johnson J. L., Whalen D. J., Li H., Holz D. E., 2013, ApJ,771, 116Latif M. A., Schleicher D. R. G., Schmidt W., 2013, ArXive-printsLatif M. A., Schleicher D. R. G., Schmidt W., Niemeyer J.,2013a, MNRAS, 433, 1607Latif M. A., Schleicher D. R. G., Schmidt W., Niemeyer J.,2013b, MNRAS, 430, 588Latif M. A., Schleicher D. R. G., Schmidt W., Niemeyer J.,2013c, ApJ, 772, L3Latif M. A., Schleicher D. R. G., Schmidt W., Niemeyer J.,2013d, MNRAS, 432, 668Latif M. A., Schleicher D. R. G., Schmidt W., NiemeyerJ. C., 2013e, MNRASLatif M. A., Schleicher D. R. G., Schmidt W., NiemeyerJ. C., 2013f, MNRAS, 436, 2989Latif M. A., Zaroubi S., Spaans M., 2011, MNRAS, 411,1659Lodato G., Natarajan P., 2006, MNRAS, 371, 1813Maier A., Iapichino L., Schmidt W., Niemeyer J. C., 2009,ApJ, 707, 40Maio U., Koopmans L. V. E., Ciardi B., 2011, MNRAS,412, L40Mortlock D. J., Warren S. J., Venemans B. P., Patel M.,Hewett P. C., McMahon R. G., Simpson C., Theuns T.,Gonz´ales-Solares E. A., Adamson A., Dye S., HamblyN. C., Hirst P., Irwin M. J., Kuiper E., Lawrence A.,R¨ottgering H. J. A., 2011, Nature, 474, 616Oh S. P., Haiman Z., 2002, ApJ, 569, 558 c (cid:13) , 1– ?? Latif et al.
O’Leary R. M., McQuinn M., 2012, ApJ, 760, 4Omukai K., Schneider R., Haiman Z., 2008, ApJ, 686, 801O’Shea B. W., Bryan G., Bordner J., Norman M. L., AbelT., Harkness R., Kritsuk A., 2004, ArXiv Astrophysicse-prints 0403044Portegies Zwart S. F., Baumgardt H., Hut P., Makino J.,McMillan S. L. W., 2004, Nature, 428, 724Prieto J., Jimenez R., Haiman Z., 2013, ArXiv e-prints-1301.5567Regan J. A., Haehnelt M. G., 2009, MNRAS, 393, 858Schleicher D. R. G., Palla F., Ferrara A., Galli D., LatifM., 2013, A&A, 558, A59Schleicher D. R. G., Spaans M., Glover S. C. O., 2010, ApJ,712, L69Schmidt W., Collins D. C., Kritsuk A. G., 2013, MNRAS,431, 3196Schmidt W., Federrath C., Hupp M., Kern S., NiemeyerJ. C., 2009, A&A, 494, 127Schmidt W., Niemeyer J. C., Hillebrandt W., R¨opke F. K.,2006, A&A, 450, 283Shang C., Bryan G. L., Haiman Z., 2010, MNRAS, 402,1249Spaans M., 2013, ArXiv e-prints:1309.1067Spaans M., Silk J., 2006, ApJ, 652, 902Stacy A., Bromm V., Loeb A., 2011, ApJ, 730, L1Stacy A., Greif T. H., Bromm V., 2012, MNRAS, 422, 290Tanaka T., Haiman Z., 2009, ApJ, 696, 1798Tanaka T. L., Li M., 2013, ArXiv e-printsTanaka T. L., Li M., Haiman Z., 2013, MNRAS, 435, 3559The Enzo Collaboration Bryan G. L., Norman M. L.,O’Shea B. W., Abel T., Wise J. H., Turk M. J., ReynoldsD. R., Collins D. C., Wang P., Skillman S. W., 2013, ArXive-printsTseliakhovich D., Hirata C., 2010, Phys. Rev. D, 82, 083520Turk M. J., Smith B. D., Oishi J. S., Skory S., SkillmanS. W., Abel T., Norman M. L., 2011, ApJS, 192, 9Van Borm C., Spaans M., 2013, A&A, 553, L9Venemans B. P., Findlay J. R., Sutherland W. J., De RosaG., McMahon R. G., Simcoe R., Gonzalez-Solares E. A.,Kuijken K., Lewis J. R., 2013, ArXiv e-prints:1311.3666Wang P., Li Z.-Y., Abel T., Nakamura F., 2010, ApJ, 709,27Whalen D. J., Fryer C. L., 2012, ApJ, 756, L19Whalen D. J., Johnson J. L., Smidt J., Heger A., Even W.,Fryer C. L., 2013, ApJ, 777, 99Whalen D. J., Johnson J. L., Smidt J., Meiksin A., HegerA., Even W., Fryer C. L., 2013, ApJ, 774, 64Whalen D. J., Smidt J., Even W., Woosley S. E., Heger A.,Stiavelli M., Fryer C. L., 2014, ApJ, 781, 106Willott C. J., Delorme P., Reyl´e C., Albert L., Bergeron J.,Crampton D., Delfosse X., Forveille T., Hutchings J. B.,McLure R. J., Omont A., Schade D., 2010, AJ, 139, 906Wise J. H., Turk M. J., Abel T., 2008, ApJ, 682, 745 c (cid:13) , 1–, 1–