The$ﬁrstgalaxies$– radiave $feedback$ from$the$ﬁrststars · 2018. 4. 20. · Assemblyunder...

The first galaxies – radia0ve feedback from the first stars

Andreas Pawlik (MPA)

Reioniza0on in the Red Center, Uluru, July 2013

Joop Schaye (Leiden) Ali Rahma0 (Leiden)

Milan Raicevic (Leiden) Claudio Dalla Vecchia (MPE)

Milos Milosavljevic (UT Aus0n) Volker Bromm (UT Aus0n) Jacob Hummel (UT Aus0n)

Myoungwon Jeon (UT Aus0n)

Outline

•  The first galaxies: assembly under radia0ve feedback from the first stars •  Tracing the first galaxies – stellar radia0on – pair instability supernovae – miniquasars/HMXBs

•  Outlook: large-‐scale reioniza0on


first atomic coolers

The First Galaxies: current simula0ons*

Abel et al. ‘99

Greif et al. ’08 Wise & Abel ‘08

Yoshida et al. ’06

Reagan & Haehnelt ‘09

106 107 10

8 10

9 >10

10

virial m

ass [M

sun]

redshij 20 15 10 6

Mashchenko et al. ’08

Wise et al. ‘11

minihalos

JWST (dwarf galaxies)

too faint for JWST (unless lensed, e.g., Zackrisson et al. ’12, Johnson et al. ’09)

*a representa0ve selec0on, not to scale


Assembly under radia0ve feedback: numerical methods

•  SPH/Gadget-‐3 zoomed simula0ons (Schaye et al. 2010; Springel 2005)

•  Mvir ~ 109 Msun at z = 10, mgas ≈ 500 Msun •  Non-‐equilibrium H2/HD chemistry

(Johnson & Bromm 2006; Greif et al. 2010) •  Star forma0on above nH = 500 cm-‐3

(metal-‐free, top-‐heavy IMF, Schaerer 2003)

•  LW in the op0cally thin limit + selfshielding correc0on (e.g., Wise & Abel 2008, Greif et al. 2011)

•  Ionizing radia0ve transfer (TRAPHIC; AP & Schaye 2008, 2010)


Radia0on-‐hydrodynamical galaxy assembly: gas densi0es

-‐3 -‐2 -‐1 0 1 2 Log10 nH [cm-‐3]

circle = virial radius zoom

AP, Milosavljevic, Bromm, ApJ, 2013


The First Galaxies: gas morphology*

Abel et al. ‘99

AP, Milosavljevic, Bromm 2011

Prieto et al. 2013

Wise & Abel ‘08

Yoshida et al. ’06

Reagan & Haehnelt ‘09

106 107 10

8 10

9 >10

10

virial m

ass [M

sun]

redshij 20 15 10 6

Mashchenko et al. ’08

Wise et al. ‘11


disks no disks

18 Prieto, Jimenez & Haiman

Figure 19. The logarithm of the projected mass–weighted density. Despite significant variations in their details, all haloes are charac-terised by a dense core, and filamentary accretion. Two of the most massive haloes (#16 and #18) develop compact over-dense blobs,here seen to be located near the haloes’ outskirts.

connection between the strength of the filamentary accretionand the nature of the object formed at the center of the halo.Namely, that stronger streams lead to stronger shocks and amore turbulent environment, which can produce COBs. Bycontrast, the less collimated, and weaker accretion proceedsin a more orderly fashion, which allows the specific angularmomentum of the gas to be conserved, and a rotationallysupported core (RSC) to develop.

Next, we examine the merger histories of the haloes. InTable 2, for each halo, we list the total number of merg-

ers it has experienced, the number of minor and majormergers (defined as having mass ratios below and aboveM1/M2 = 1/3, respectively), the redshift and mass ratioof each merger, and the first (highest) redshift at which thegas in the halo could cool. For our available hydro outputs,zcool was computed as the z when the gas temperature insideRvir reaches the T = 104K. Because we do not have all thehydro outputs at high z for every simulation, for the z⇤ casesin the last column of Table 2 we used the DM-only simula-

c� 0000 RAS, MNRAS 000, 000–000

18 Prieto, Jimenez & Haiman

Figure 19. The logarithm of the projected mass–weighted density. Despite significant variations in their details, all haloes are charac-terised by a dense core, and filamentary accretion. Two of the most massive haloes (#16 and #18) develop compact over-dense blobs,here seen to be located near the haloes’ outskirts.

connection between the strength of the filamentary accretionand the nature of the object formed at the center of the halo.Namely, that stronger streams lead to stronger shocks and amore turbulent environment, which can produce COBs. Bycontrast, the less collimated, and weaker accretion proceedsin a more orderly fashion, which allows the specific angularmomentum of the gas to be conserved, and a rotationallysupported core (RSC) to develop.

Next, we examine the merger histories of the haloes. InTable 2, for each halo, we list the total number of merg-

ers it has experienced, the number of minor and majormergers (defined as having mass ratios below and aboveM1/M2 = 1/3, respectively), the redshift and mass ratioof each merger, and the first (highest) redshift at which thegas in the halo could cool. For our available hydro outputs,zcool was computed as the z when the gas temperature insideRvir reaches the T = 104K. Because we do not have all thehydro outputs at high z for every simulation, for the z⇤ casesin the last column of Table 2 we used the DM-only simula-

c� 0000 RAS, MNRAS 000, 000–000

Greif et al. ’08

*a representa0ve selec0on, not to scale

Detec0on with JWST: UV con0nuum flux

Z=0, top-‐heavy IMF Z=0, normal IMF Z=5 x 10-‐4 Zsun, normal IMF

SFR

Mstar


UV1500

AP, Milosavljevic, Bromm, ApJ, 2013; see also, e.g., Zackrisson et al. 2012

Pair instability supernovae Jacob Hummel, AP, Milosavljevic, Bromm, ApJ, 2012

THE FIRST SUPERNOVAE 5

Fig. 2.— a) npisn in the upper limit of no feedback (blue), withchemical feedback (green), LW feedback (red) and the resultingPISN rate for the conservative (chemical plus LW) feedback case(black). b) Same as (a), but for enhanced massive star formation.

enriched beyond a critical metallicity of Zcrit ⇠ 10�4 Z�will no longer form Pop III stars (Bromm et al. 2001;Schneider et al. 2002; Bromm & Loeb 2003), and henceno PISNe. Chemical feedback can thus be representedas the fraction of halos forming from pristine gas at agiven redshift. Realistic three-dimensional simulations ofthis process starting from cosmological initial conditionshave become possible in the past decade, showing thatenrichment by Pop III SNe, if they are highly energetic,proceeds very inhomogeneously, enriching the IGM be-fore penetrating into denser regions (Scannapieco et al.2005; Greif et al. 2007; Tornatore et al. 2007; Wise &Abel 2008; Maio et al. 2010).

In modeling ⌘chem, we use the results of Furlanetto &Loeb (2005). Their semi-analytic treatment of SN windsutilizes the Sedov (1959) solution for an explosion ex-panding into a uniform medium and yields a probabilityfunction Ppristine(z) that the gas in a newly formed halois pristine. This is plotted in Figure 2 of their paper forvarious strengths of chemical feedback. We identify thisquantity as the fraction of newly collapsed halos thathave been polluted with metals, ⌘chem. Given the re-cent detection of pristine gas at z = 3 by Fumagalliet al. (2011), we choose the weakest feedback scenariopresented by Furlanetto & Loeb (2005) among the sce-narios that incorporate a clustering of sources. The re-sulting PISN rate is given by the green line in Figure 2.

2.3. Enhanced Massive Star Formation

Gas cooling and subsequent star formation in halos af-fected by LW feedback can be delayed until nearly an or-der of magnitude more gas is available for star formation(Figure 1). This increases the likelihood that multiplemassive stars form per halo, o↵seting the negative e↵ectsof LW radiation considered above. We quantify this bypositing that the number of PISNe produced per haloat redshift z is given by the ratio of the critical mass in

5 10 15 20 25 30z

10�6

10�5

10�4

10�3

10�2

10�1

100

101

SNra

te[y

r�1

(10

arcm

in�

2 )]

Miralda-Escude & Rees 97

Mackey et al. 03

Weinmann & Lilly 05

Wise & Abel 05

Fig. 3.— The observable PISN rates in number per year perJWST field of view above a given redshift in the upper limit of nofeedback (blue line), in the conservative feedback case (solid redline), and the enhanced star formation case (dashed red line). Therates calculated by Miralda-Escude & Rees (1997), Mackey et al.(2003), Weinmann & Lilly (2005) and Wise & Abel (2005) are alsoshown for reference. Red points account for feedback; blue pointsdo not.

the presence of LW feedback Mcrit,lw to the critical massin the no-feedback case Mcrit. For example, at z = 17,Mcrit,lw/Mcrit ⇡ 1.4, so for every 10 pristine halos thatform, 14 PISNe are produced. In this case the PISN rateis modified such that

npisn(z) =Mcrit,lw(z)

Mcrit(z)⌘chem(z) ⌘rad(z) n+(z). (10)

The resulting enhanced PISN rate can be seen inFigure 2b. In contrast to the conservative feedback case,the net e↵ect of LW feedback is much less significant here,with chemical feedback controlling the final PISN rate.

2.4. The Observable Rate

The observed PISN rate per unit time per unit redshiftper unit solid angle is given by

dN

dtobs dz d⌦=

dN

dtobs dV

dV

dzd⌦

=1

(1 + z)

dN

dtem dVr2

dr

dz.

(11)

Cosmological time dilation between tobs and tem is ac-counted for by the (1 + z) in the denominator; dV isthe comoving volume element and r(z) is the comovingdistance to redshift z given by

r(z) =c

H0

Z z

0

dz0p⌦m(1 + z0)3 + ⌦⇤

, (12)

where c/H0 is the Hubble distance. With the assump-tions outlined above, we estimate the PISN rate in eventsper year per comoving Mpc3 in the source rest frame:

dN

dtem dV= npisn(z). (13)

These results—shown in Figure 3—are in reasonableagreement with previous work; our no-feedback limit of

THE FIRST SUPERNOVAE 7

0 5 10 15 20 25 30tobs [years]

10�19

10�18

10�17

10�16

Flu

x[e

rgss�

1cm

�2]

z = 5z = 10z = 15z = 20z = 25z = 30

5 10 15 20 25 30 35z

0

5

10

15

20

�t v

is[y

ears

]

F444WF356WF277WF200WF150WF115WF090WF070W

0 5 10 15 20 25 30tobs [years]

10�22

10�21

10�20

10�19

10�18

10�17

Flu

x[e

rgss�

1cm

�2]

z = 5z = 10z = 15z = 20z = 25z = 30

2 4 6 8 10 12z

0

1

2

3

4

5

6

7

8

�t v

is[y

ears

]

F444WF356WF277WF200WF150WF115WF090WF070W

Fig. 5.— Left: Lightcurves for the Kasen et al. (2011) R250 (top) and B200 (bottom) models as they would be observed by JWST’sF444W NIRCam filter at z = 5, 10, 15, 20, 25 and 30. The flux limits for a 106 s (dashed line) and 104 s (dotted line) exposure are shownfor reference. Right: The visibility time �tvis in years for R250 (top) and B200 (bottom) as a function of redshift for each of the NIRcamwide filters. Note that the axes are scaled independently. Similar plots for models He100 and R175 are included in the appendix.

3.2. Visibility

The NIRCam instrument on the JWST will observethe early universe through a number of narrow, medium-width, and wide filters5. The widest, longest-wavelengthfilter, F444W, will observe from 3.3 to 5.6 µm with asensitivity limit of 24.5 nJy required for a 10� detectionin 104 seconds (Gardner et al. 2006). Shown in the left-hand column of Figure 5 is the observable flux as it wouldappear in the F444W NIRCam filter at various redshiftsfor the most and least easily observable models, R250and B200, respectively. See Figure 7 for why these twowere chosen; models He100 and R175 can be found in theappendix. The flux limits for the filter of 4.4⇥ 10�19 ergs�1 cm�2 for a 106 s exposure and 4.4 ⇥ 10�18 erg s�1

cm�2 for a 104 s exposure are also shown for reference.We see that the brightest explosions (R250) would bevisible to beyond z ⇠ 25, but are never so bright as tobe detectable with current generation telescopes. This isconsistent with the non-detection by Frost et al. (2009)in a search of the Spitzer/IRAC Dark Field for possiblePop III PISN candidates.

To account for absorption of flux by neutral hydrogenalong the line of sight we implement a simple model of

5 http://www.stsci.edu/jwst/instruments/nircam/instrument-design/filters

instant reionization at z = 10. For sources above thisredshift, we assume no flux is observed shortward of therest frame Ly↵ line. This is not relevant for the F444WNIRcam filter as Ly↵ does not redshift into the filter untilz ⇠ 40, when the lightcurve is already far below even the106 s sensitivity limit. It does however have an e↵ect,albeit a small one, on the F115W and F090W filters.

At low redshifts the duration of the lightcurve pre-sented in Kasen et al. (2011) is not quite long enough forthe observed flux to reach the sensitivity limit; we ex-tend it to the limit by extrapolating assuming a power-law scaling. The visible time �tvis is then simply givenby the time the lightcurve is above the filter sensitivitylimit. Shown in Figure 5 are the visibility times as afunction of redshift for each of the NIRcam filters.

3.3. The Observable Number

With this estimate for �tvis, we may finally calculatethe observable number of PISNe on the sky, given by theproduct of the PISN rate at z, as seen in the observerframe, and the time a PISN at z is visible, �tvis. Thisyields an estimate for the number of PISNe visible onthe sky at any given time per unit redshift per unit solidangle:

dN

dz d⌦'

dN

dtobs dz d⌦�tvis. (17)

JWST 106 s

JWST 104 s

light curves based on Kasen et al. 2011

explosion rate (per JWST field of view)

PISNe: bright but rare


Pair instability supernovae Jacob Hummel, AP, Milosavljevic, Bromm, ApJ, 2012

10 HUMMEL ET AL.

102 103 104 105

JWST Fields of View

10�1

100

101

102

103

104

Num

ber

Vis

ible

106 s

102 103 104 105

JWST Fields of View

107 s

102 103 104 105

JWST Fields of View

108 s

Fig. 8.— The total number of PISNe observable with a campaign of 106, 107 and 108 s (from left to right) as a function of survey area forthe R250 PISN model. In each case, the total campaign time is apportioned equally over the total survey area to determine the exposuretime for individual pointings. The blue region represents all PISNe, the red only PISNe from z > 15. Upper boundaries correspond to theno-feedback upper limit to the PISN rate and lower boundaries to the conservative feedback case. For reference we mark the case of onlyone PISN visible (dashed line).

of PISNe that will be observable with the JWST in ob-serving campaigns totalling 106, 107 and 108 s for theR250 PISN model. The exposure time for each point-ing varies with the total area covered by the survey inorder to keep the total observing time constant. Up-per boundaries correspond to the number visible in theno-feedback case, lower boundaries to the conservativefeedback case. As in Figure 7, the blue region shows theobservable number from all redshifts, the red region onlythose from z > 15. We see that the observable numberincreases until the resulting exposure time is no longersu�cient to detect PISNe. The optimal search strategythen will be to cover as large an area as possible, going

only as deep as necessary, possibly in a similar manner tothe ongoing Brightest of Reionizing Galaxies survey withthe Hubble Space Telescope (Trenti et al. 2011; Bradleyet al. 2012).

V.B. and M.M. acknowledge support from NSF grantsAST-0708795 and AST-1009928 and NASA ATFP grantNNX09AJ33G. V.B. thanks the Max-Planck-Institut furAstrophysik for its hospitality during part of the workon this paper. The simulations were carried out at theTexas Advanced Computing Center (TACC).

REFERENCES

Abel, T., Bryan, G. L., & Norman, M. L. 2002, Science, 295, 93Alvarez, M. A., Bromm, V., & Shapiro, P. R. 2006, ApJ, 639, 621Alvarez, M. A., Wise, J. H., & Abel, T. 2009, ApJ, 701, L133Barkana, R., & Loeb, A. 2001, Phys. Rep., 349, 125Barkat, Z., Rakavy, G., & Sack, N. 1967, Phys. Rev. Lett., 18, 379Bradley, L. D., et al. 2012, ApJ, submitted (arXiv:1204.3641)Bromm, V., Coppi, P. S., & Larson, R. B. 1999, ApJ, 527, L5—. 2002, ApJ, 564, 23Bromm, V., Kudritzki, R. P., & Loeb, A. 2001, ApJ, 552, 464Bromm, V., & Larson, R. B. 2004, ARA&A, 42, 79Bromm, V., & Loeb, A. 2003, Nature, 425, 812Bromm, V., Yoshida, N., & Hernquist, L. 2003, ApJ, 596, L135Bromm, V., Yoshida, N., Hernquist, L., & McKee, C. F. 2009,

Nature, 459, 49Chatzopoulos, E., & Wheeler, J. C. 2012, ApJ, 748, 42Chen, K.-J., Heger, A., & Almgren, A. S. 2011, Comp. Phys.

Communications, 182, 254Chiappini, C., Frischknecht, U., Meynet, G., Hirschi, R., Barbuy,

B., Pignatari, M., Decressin, T., & Maeder, A. 2011, Nature,472, 454

Ciardi, B., Ferrara, A., & Abel, T. 2000, ApJ, 533, 594Clark, P. C., Glover, S. C. O., Smith, R. J., Greif, T. H., Klessen,

R. S., & Bromm, V. 2011, Science, 331, 1040Couchman, H. M. P., & Rees, M. J. 1986, MNRAS, 221, 53Dahlen, T., & Fransson, C. 1999, A&A, 350, 349Fraley, G. S. 1968, Ap&SS, 2, 96Frost, M. I., Surace, J., Moustakas, L. A., & Krick, J. 2009, ApJ,

698, L68

Fryer, C. L., Woosley, S. E., & Heger, A. 2001, ApJ, 550, 372Fumagalli, M., O’Meara, J. M., & Prochaska, J. X. 2011, Science,

334, 1245Furlanetto, S. R., & Loeb, A. 2003, ApJ, 588, 18—. 2005, ApJ, 634, 1Gal-Yam, A., et al. 2009, Nature, 462, 624Gardner, J. P., et al. 2006, Space Sci. Rev., 123, 485Glover, S. C. O., & Brand, P. W. J. L. 2001, MNRAS, 321, 385Greif, T. H., & Bromm, V. 2006, MNRAS, 373, 128Greif, T. H., Bromm, V., Clark, P. C., Glover, S. C. O., Smith,

R. J., Klessen, R. S., Yoshida, N., & Springel, V. 2012,MNRAS, in press (arXiv:1202.5552)

Greif, T. H., Glover, S. C. O., Bromm, V., & Klessen, R. S. 2010,ApJ, 716, 510

Greif, T. H., Johnson, J. L., Bromm, V., & Klessen, R. S. 2007,ApJ, 670, 1

Greif, T. H., Johnson, J. L., Klessen, R. S., & Bromm, V. 2008,MNRAS, 387, 1021

Greif, T. H., Springel, V., White, S. D. M., Glover, S. C. O.,Clark, P. C., Smith, R. J., Klessen, R. S., & Bromm, V. 2011,ApJ, 737, 75

Haiman, Z., Abel, T., & Rees, M. J. 2000, ApJ, 534, 11Haiman, Z., Rees, M. J., & Loeb, A. 1997, ApJ, 476, 458Haiman, Z., Thoul, A. A., & Loeb, A. 1996, ApJ, 464, 523Heger, A., Fryer, C. L., Woosley, S. E., Langer, N., & Hartmann,

D. H. 2003, ApJ, 591, 288Heger, A., & Woosley, S. E. 2002, ApJ, 567, 532

number controlled by scarcity of PISNe

number controlled by JWST detec0on limit

op0mal number of poin0ngs

all redshijs

z > 15

feedback

Num

ber visib

le with

JWST

10-‐1 10

0 10

1 10

2 10

3 10

4


The first black holes Myoungwon Jeon, AP, et al., ApJ, 2012

w/o X-‐ray feedback w/ X-‐ray feedback

10 kpc

Eddington rate

z = 20

see also Kuhlen & Madau 2005; Wise & Abel 2011 Andreas Pawlik

(MPA)

Outlook: spa0ally adap0ve radia0on-‐hydro simula0ons of reioniza0on with TRAPHIC

AP, Schaye, Rahma0, Raicevic, Dalla Vecchia •  Computa0on 0me independent of the number of sources (vs. propor0onal to the number of sources)

•  Spa0ally adap0ve radia0ve transfer (vs. radia0ve transfer on uniform grids)

•  Radia0on-‐hydrodynamically coupled (vs. post-‐processing of sta0c density fields)

Andreas Pawlik

(MPA)

TRAPHIC in the cosmological galaxy forma0on code GADGET

(version 3-‐OWLS; Springel 2005; Schaye+ 2010)

•  accurate radia0ve transfer coupled to cosmological hydrodynamical simula0ons

•  high spa0ally adap0ve resolu0on (equivalent to ~130003 uniform grid)

25 Mpc/h, 2x5123

Mhalo > 8 x 108 solar


Effect of feedback on stellar mass func0on at z = 6


No feedback RHD only SN only D. Vecchia & Schaye ‘12

Posi0ve feedback

Effect of photohea0ng on IGM clumping factor

(AP, Schaye, & van Scherpenzeel 2009)

See also, e.g., Shull et al. 2012, Finlator et al. 2012

IGM clumping factor


Summary

•  Extended disks can form at redshijs as high as z≈10 in halos as small as ~109 Msun

•  JWST will collect the stellar light from halos with masses as low as ~>109 Msun

•  Halos with masses ~< 109 Msun may be traced by hun0ng for PISNe and accre0ng black holes

•  Spa0ally adap0ve radia0on-‐hydro simula0ons of reioniza0on with TRAPHIC


The$ﬁrstgalaxies$– radiave $feedback$ from$the$ﬁrststars · 2018. 4. 20. · Assemblyunder...

Documents

Transcript of The$ﬁrstgalaxies$– radiave $feedback$ from$the$ﬁrststars · 2018. 4. 20. · Assemblyunder...