Eli Dwek Observational Cosmology Lab
NASA Goddard Space Flight Center
The Origin of Dust in the Early Universe
Isabelle Cherchneff Univ. of Basel - Basel
arXiv: 1011.1303
1Wednesday, November 10, 2010
AGB SNe
Novae WR
accretioncoagulation
chemical reactions
destruction by SN
blast waves
dust sources
molecular clouds diffuse ISM
interstellar dustprocessing
stellar explosions winds
star formationcloud disruption
cycling between ISM phases
The journey of interstellar dust
2Wednesday, November 10, 2010
Chemical EvolutionThe Rate of Change in the Dust Mass
destruction by SN blast waves
added/removedto/from ISM by
infall/outflow
source function
removed from the ISM by
star formation
Equation applicable to any elementMd → MA
τd → ∞τd → τ1/2
stable
radioactive
3Wednesday, November 10, 2010
Chemical EvolutionThe rate of change in the gas mass
SFRMass returned by stars
infall/outflow
instantaneous recycling approximation
4Wednesday, November 10, 2010
Chemical Evolution Dust-to-Gas mass ratio
5Wednesday, November 10, 2010
Chemical Evolution Integral solutions
Integration factor
Solutions
6Wednesday, November 10, 2010
Chemical Evolution A simple differential equation
7Wednesday, November 10, 2010
Chemical Evolution The Source Function
8Wednesday, November 10, 2010
Cygnus Loop detail(HST)
GRAIN DESTRUCTION
Cygnus Loop(ROSAT)
9Wednesday, November 10, 2010
Grain Destruction in the ISM
• 24, 70, 160 µm
MolecularCloud
Bright Eastern Knot (BEK)
Spitzer observations of Puppis AArendt et al. 2010, ApJ in press
10Wednesday, November 10, 2010
~ 1’
Cloud spectraZubko et al dust model
Post shock spectrapre-shock dust predictstoo much NIR emission
Grain Processing in Puppis A
Spitzer• MIPS 24 µm
11Wednesday, November 10, 2010
Grain destruction in Puppis A
pre-shock ZDA grain size distribution
post-shock
About 30% of the dust mass is destroyed in the
~ 500 km/s shock
This is a lower limit!
12Wednesday, November 10, 2010
Grain destruction efficiencies
Mass of dust destroyed by a single SNR
(Jones, Tielens, Hollenbach, & McKee 1994, 1996)
Md = Zd
� vf
v0
fd(vs)
�dMism
dvs
�dvs
13Wednesday, November 10, 2010
Destruction
The dust lifetime
Dusty ISMExpanding SNRs
• For the MW galaxy:
✦ Md ≈ 3x107 Msun RSN ≈ 0.02 yr-1 md ≈ 3 Msun
d ≈ 500 Myr
τd(t) ≡Md
md RSN
The mass, locked up in dust, that is returned to the gas phase by a
single SNR
md
14Wednesday, November 10, 2010
DUSTFORMATION:
Supernovae
15Wednesday, November 10, 2010
Spitzer spectra of Cas A ≈ 0.02–0.04 Msun (Rho et al. 2008)
Akari/BLAST search for warm (~ 35 K) dust
≈ 0.06 Msun (Sibthorpe et al. 2009)
Observations
Dust Formation in Cas A
Herschel (~ 35 K) dust ≈ 0.08 Msun
(Barlow et al. 2009)
total ≈ 0.1 – 0.12 Msun
16Wednesday, November 10, 2010
Theoretical calculations
Chemical kinetic calculations for Pop III stars
(Cherchneff & Dwek 2010)
Dust Formation in SNe
Dust yield in a 20 Msunprimordial CCSNe
≈ 0.15 Msun
17Wednesday, November 10, 2010
Eta Carinae
DUSTFORMATION:
quiescent outflows(AGB, WR stars)
18Wednesday, November 10, 2010
Dust sources: AGB stars
O-richC/O < 1
C ≈ O
C-richC/O > 1
silicates
carbon dustPAHs
MgS
19Wednesday, November 10, 2010
Dust Yield in AGB StarsKarakas & Lattanzio (2007)
20Wednesday, November 10, 2010
AGB stars WR starsSupernovae
AGB stars80%
Supernovae18%
WR stars2%
The relative contribution of the different dust sources
In a steady state
21Wednesday, November 10, 2010
The dust composition MUST evolve !
• SN are the primary sources of silicate dust
✦ they return their processed ejecta “promptly” back to the ISM
• AGB stars (Msun ≈ 1-8 Msun) are the primary sources of carbon dust
✦ they return their processed material a considerable time after their birth
✤ The IMF-averaged mass of a carbon star is ~ 3 Msun
• The dust composition will therefore change over time
✦ extinction, composition, IR emission will therefore depend galactic age
• How can we observationaly test it?
✦ find proxy for AGB dust
✦ observe galaxies at different ages or metallicities
32 Myr
116 Myr
470 Myr
1.1 Gyr
2.8 Gyr
10 Gyr
22Wednesday, November 10, 2010
AGB
SNe
The delayed injection of AGB dust into the ISM
Dwek (2005)
Galliano et al. (2008)
PAHs are proxies for AGB-condensed dust
23Wednesday, November 10, 2010
The high-z universe
24Wednesday, November 10, 2010
What is the Origin of Dust inthe high redshift quasarSDSS J114816+5251 ?
Detected in a search for i (7481 Å) dropouts in the Sloan Digital Sky Survey
(Fan 2003)Follow-up spectroscopy
z = 6.4
2.5 m
AGN
M(H2) = 1.6× 1010 M⊙
Mdyn = (4− 6)× 1010 M⊙
MBH = 3× 109 M⊙
Mbulge = 5× 1011 M⊙
25Wednesday, November 10, 2010
The Spectral Energy Distribution of J1148
Searches for lensing effects were negative
dust emission
Ltot ≈ 1× 1014 L⊙
LIR ≈ 2× 1013 L⊙
26Wednesday, November 10, 2010
Spectral FITS to theFIR Flux of J1148+5251
LFIR = 2× 1013 L⊙
ΣSFR(M⊙ yr−1 kpc−2) = 2.5× 10−4 Σ1.4g (M⊙ pc−2)
CO mass500± 240
Star formation rate (Msun yr-1)
ψ(M⊙ yr−1) = 1.7× 10−10LFIR(L⊙)IR luminosity
∼ 3400
However the SRF can be much lower if the AGN contributes
significantly to the FIR luminosity
3x108 Msun 4x108 Msun
1x108 Msun
Dust Masses
~(1–4)x108 Msun
Dust Mass in J1148
Milky Way Mdust ≈ 3x107 Msun
27Wednesday, November 10, 2010
The SN Scenario
Can SNe produce the required dust mass?
(Dwek et al. 2007)
28Wednesday, November 10, 2010
Age of the universe at z=6.4: 890 Myr
z = 10 Univ = 490 Myr, Gal age = 400 Myrz = 20 Univ = 190 Myr, Gal age = 700 Myrz = 30 Univ = 100 Myr, Gal age = 790 Myr
32 Myr
116 Myr
470 Myr
1.1 Gyr
2.8 Gyr
Can SN produce the required mass of dust?
NO PROBLEM!
SN rate ≈ 70 per yr
in 400 Myr ≈ 3x1010 SNe
Each SN needs to make 5x108 / 3x1010
≈ 0.02 Msun of dust
These calculations neglect:(1) the effect of grain destruction(2) the finite effective SF time
If galaxy is young enough then AGBs have not yet evolved
off the main sequence
29Wednesday, November 10, 2010
SN Yield Required to Produce an Observed Dust-to-Gas Mass Ratio, Zd
(Dwek, Galliano & Jones 2007, ApJ, 662, 927)
mg(Msun)
Milky Way destruction
No grain destruction
Largest observed SN yield
SN yieldsneeds to be 1 Msun/SN
With grain destruction
SN yieldscan be
0.15 Msun/SN
Without grain destruction
30Wednesday, November 10, 2010
The AGBScenario
Can AGB stars produce the required dust mass?
(Valiante et al. 2009)
31Wednesday, November 10, 2010
Galaxy needs to be older than ≈ 400 Myr
32Wednesday, November 10, 2010
The star formation history of high-z quasars in hierarchical galaxy merger models
(Li et al. 2007) SFR used by Valiante et al. 2009
33Wednesday, November 10, 2010
The Evolution of Dust in J1148+5251
Note the delayed injection of AGB dust
closed box model Schmidt-type SFR-Mgas relation
34Wednesday, November 10, 2010
The Evolution of Stellar Mass
and Luminosity
35Wednesday, November 10, 2010
How unique is the star formation history implied
by the merger scenarioS?
36Wednesday, November 10, 2010
The Dependence of the Dust Abundance on the Star
Formation History
✦ SFR = 1000 Msun/yr✦ 100 Myr duration✦ no grain dstruction
during burst
onset of starburst
first AGB stars evolve off the MS
dust lifetime(Myr)
37Wednesday, November 10, 2010
The Contribution of Succesive Bursts of Star Formation
Approximation the merger history with discrete bursts
The cumulative contribution of the bursts to the dust mass
38Wednesday, November 10, 2010
Can SN form the dust with a yield of
0.15 Msun/event?
39Wednesday, November 10, 2010
Is the SN scenario a viable alternative?
YES!provided that we are
observing a very young starburst
40Wednesday, November 10, 2010
Can we discriminate between the SN and the
AGB models from the galaxy’s spectral energy
distribution (SED)?
41Wednesday, November 10, 2010
The SED of J1148+5251
The SN model
The AGB model
All UV-optical is produced by stars
All UV-optical is produced by the
AGN
42Wednesday, November 10, 2010
Could the UV-NIR SED of J1148+5251 be purely AGN?
The probability of the photometric spectral index
43Wednesday, November 10, 2010
Can we discriminate between the SN and the AGB scenarios from the
frequency of J1148-type
objects?
44Wednesday, November 10, 2010
21 QSOs@ z ≈ 6
Comovingnumber density≈ 10–9 Mpc–3
High redshift (z ≈ 6) QSOs(Jiang et al. 2010)
dust free dust free
dust free?
45Wednesday, November 10, 2010
How rare are J1148-type objects?The AGB Scenario
Press-Schechter (PS) formalism
7x1011 Msun objects must have formed
at z≈8.5
Mhalo ≈ 4x1012 Msun
✦ PS formalism predicts that the number density of
such objects is ≈ 10–12 Mpc-3
✦ But the comoving number density of z ≈ 6 QSOs is ≈ 10–9 Mpc-3
comoving number density of collapsed halos (Mpc-3)
46Wednesday, November 10, 2010
How rare are J1148-type objects?
PS formalism may underestimates the
number of collapsed halos
✦ So AGB scenario may be consistent with observations if PS formalism is wrong by factor of 1,000 !
✦ All QSOs could be hyperluminous IR objects,✦ But J1148-type objects are more probably very rare among z~6 QSOs
The AGB Scenario
47Wednesday, November 10, 2010
How rare are J1148-type objects?The SN scenario
1x1011 Msun objects must have formed
at z≈7
PS formalism predicts that the number density of
such objects is ≈ 10–5 Mpc-3
The GOODS survey (Stark 2009)
Rough agreement with the comoving number density of objects with stellar masses of ~ 1011 Msun
48Wednesday, November 10, 2010
How rare are J1148-type objects?The SN scenario
Comoving luminosity density vs redshift (Franceschini 2010)
If all M*≈1011 Msun galaxies had LIR ≈2x1013 Lsun
2x108 Lsun Mpc-3
Comoving luminosity density of SF galaxies
(Franceschini 2010)
log10(L) = 13.2~10–7 Mpc-3
If J1148-type objects are powered by SNe, they should be
“common”
but veryluminous
objects arerare
49Wednesday, November 10, 2010
Is grain growth in molecular clouds
important in J1148+5251?
50Wednesday, November 10, 2010
• Accretion time must be less that the cloud lifetime
• The condition is fulfilled in J1148+5251, so grain growth in MCs can be important in this object
• However, grain growth in MCs is not required to explain the origin of the dust in this object
Conditions for net grain growth in Molecular clouds
τacc ≡�
1
ρgr
dρgrdt
�−1
∼ a
nc T 1/2≈ 3× 106 yr
τc =Mc
ψ∼ 2× 1010
3000≈ 7× 106 yr
τacc < τc
51Wednesday, November 10, 2010
Is grain growth in molecular clouds important in the
Milky Way?
52Wednesday, November 10, 2010
14
Condensation in the Sources(Field 1974) Accretion in the
ISM (Snow 1975)
Destruction in the ISM (Dwek & Scalo 1980)
What is the Origin of the Interstellar Depletion Pattern?
TcondU0
Iion
53Wednesday, November 10, 2010
54Wednesday, November 10, 2010
The Evolution of Dust in the Local Neighborhood
(VERY preliminary result)
Only ~ 10% of the interstellar dust in the MW was formed in stars!
MdustMgas/100
5× 106 M⊙
Hernandez et al. (2000)
stochastic SF in the solar neighborhood
55Wednesday, November 10, 2010
Do AGN form dust?
56Wednesday, November 10, 2010
Dust formation in AGN winds(Elvis+ 2002; Maiolino+ 2006)
(Kartje & Konigl 2006)
{n, T} conditions similar to those in AGB outflows
BUT: is the dust in the BELC (broad emission line clouds) newly-formed or “just” reformed
Conditions are conducive to dust formation
57Wednesday, November 10, 2010
• Dust evolution models are a useful tool for describing the evolution of dust mass - a new formalism
• Local universe - PAH vs metallicity
• Early universe
✦ AGB likely scenario, but:
✤ may require a massive galaxy - objects rare✤ however SFH not unique
✦ SN are not likely to be the source of dust
✤ SN yields too low, grain destruction too efficient✤ predicts too many hyperluminous IR galaxies
✤ requires contrived star formation history
✤ More common objects than the AGB scenario
✦ Grain growth in clouds can be important but is not necessary
Conclusions
58Wednesday, November 10, 2010
The Origin of Dust at High RedshiftSUMMARY
SF history@ z~6 growth
of BH mass
frequencyof J1148-
type objects
dustformationSNe, AGB
ISM processing
dynamical-stellar mass
59Wednesday, November 10, 2010
END60Wednesday, November 10, 2010