Post on 29-Jan-2016
First Light from the Fossil Record:A New Synthesis
Jason Tumlinson
Yale Center for Astronomy and Astrophysics
A slice of the Milky Way at z = 6
(1)A Review of Major Themes in the Study of “First Stars”
(2) A New Approach to Constraints on the
IMF of Primordial (“First”) Stars
(3) The IMF of the “Second” Stars
(4) Predictions and Future Tests
The Big ??: When, What, and Where was “First Light”?
Major Themes of “The First Stars”
Physical Models of Star Formation at Zero and Very Low Metallicity
Stellar Evolution and Nucleosynthesis of the First Stars
Chemical Abundance Studies of Metal-Poor Pop II (“Galactic Archaeology”)
At Zcrit ~ 10-5.5 to 10-3.5 Zּס, efficient metal-line cooling may allow fragmentation to low-mass stars (Bromm & Loeb 2003; Santoro & Shull 2006).
But by this time there may also be dust, ionizing radiation, the CMB, cosmic rays, B fields. . so ab initio simulation is too hard.
To cut the knot of theory, we need observations!
Key Concept #2: “The Critical Metallicity” and the “2nd Stars”
Simple recipe for first stars:• CDM • Dark matter “minihalos” of MDM ~ 106-7 M€at z = 20 - 40. • primordial composition (H,He,H2)• the absence of other (in)famously complicating factors (dust, B)
2J
H
GM MkT
R m
3122
3 310 10J
n TM M
cm K
H2 cools primordial gas to Tmin ~ 200 K, for MJ ~ 100 - 1000 Mּס
(Bromm, Coppi, & Larson 1999; 2002, Abel, Bryan, & Norman 2002)
30 – 300 Mּס accretes in a Kelvin-Helmholz time (O’Shea & Norman 2007).
Key Concept #1: “Warmer (Primordial) Gas Forms Heavier Stars”
Major Themes of “The First Stars”Physical Models of Star Formation at Zero and Very Low Metallicity:
Approach: Hydrosims of gas physics in early cosmological halos
Key Results: High mass range (~30 - 300) for limiting Z = 0 case.
Formation of first low-mass stars depends on prior ionization
and/or metal enrichment metals, dust, CMB, other factors (?)
How did the first and second stars form, and what was their IMF?
Chemical Abundance Studies of Metal-Poor Pop II (“The Second Stars”)
Stellar Evolution and Nucleosynthesis of the First Stars:
Approach: numerical stellar evolution and supernova models (1-D)
Key Idea: The chemical signatures of stars vary with initial mass and metallicity in complex but calculable fashion.
Our strategy is to use robust and distinct signatures of stellar mass to diagnose IMF.
Nu
mb
er
per
Ma
ss B
in
M
Major Themes of “The First Stars”Physical Models of Star Formation at Zero and Very Low Metallicity:
Approach: Hydrosims of gas physics in early cosmological halos
Key Results: High mass range (~30 - 300) for limiting Z = 0 case.
Formation of first low-mass stars depends on prior ionization
and/or metal enrichment metals, dust, CMB, other factors (?)
How did the first and second stars form, and what was their IMF?Stellar Evolution and Nucleosynthesis of the First Stars:
Approach: numerical stellar evolution and supernova models (1-D)
Key Results: “Pair Instability SNe” and “Hypernovae” may arise from
the first stars and give distinctive yield patterns.
Big question now is how much rotation alters mass loss and yields.
Given a particular IMF, what are the observational signatures (both radiation and chemical yields)?
Chemical Abundance Studies of Metal-Poor Pop II (“The Second Stars”):
10000 from RAVE
(AAO – now) The Future of “Galactic Archaeology”>100000 from SDSS/SEGUE for halo,
APOGEE for bulge and disk
100000 from LAMOST
(China - 2009)
109 from GAIA (ESA-2011)
Dwarf Abundance and Radial Velocities (DART) @ VLT 106 from WFMOS
@ Subaru (2010?)
Massive spectroscopic multiplexing enables surveys of > 106 stars for studies of
MW structure and formation.
Up to >~ 105 of these stars will have [Fe/H] < -2, so are plausibly from the first few
generations.
About 1% of the abundance data that will exist in 2013 is in hand and analyzed today.
But what information about the first galaxies might these
stars provide?
Beers & Christlieb (2005) ARA&A HE
RE
S S
urvey - Barklem
et al. (2005) – 15 elements in 253 stars
“Primary”
“Hydrostatic”
“Explosive”
“neutron capture”
[X/F
e]
[Fe/H]
“Information Overload” from Chemodynamical Probes of Galactic Evolution
Measured proper motion, radial velocity, and position trace galactic components – disk, bulge, or halo.
Color, luminosity, Teff, and metallicity select old, low-mass stars with [Fe/H] < -2 that most likely trace the first generations.
Expand this ~30-D “data space” by at least four orders of magnitude and you begin to get the idea.
VLT data - Cayrel et al. (2004) and Barklem et al. (2005)
HERES Survey - Barklem et al. (2005)
[
Ba/
Fe]
≥ 82% at [Fe/H] ≤ -2.5 show r-process enrichment
HE1327-2326
HE0107-5240
100%100%
after Komiya et al. (2007)
Carbon-Enhanced Metal-Poor Stars (CEMPs):
CEMP = [C/Fe] > 1@ [Fe/H] < -2Beers & Christlieb (2005)
“HMPs”
“C-normal “ ~ solar
Major Themes of “The First Stars”Physical Models of Star Formation at Zero and Very Low Metallicity:
Approach: Hydrosims of gas physics in early cosmological halos
Key Results: High mass range (~30 - 300) for limiting Z = 0 case.
Formation of first low-mass stars depends on prior ionization
and/or metal enrichment metals, dust, CMB, other factors (?)
How did the first and second stars form, and what was their IMF?Stellar Evolution and Nucleosynthesis of the First Stars:
Approach: numerical stellar evolution and supernova models (1-D)
Key Results: “Pair Instability SNe” and “Hypernovae” may arise from
the first stars and give distinctive yield patterns.
Big question now is how much rotation alters mass loss and yields.
Given a particular IMF, what are the observational signatures (both radiation and chemical yields)?
Chemical Abundance Studies of Metal-Poor Pop II (“The Second Stars”):
Approach: massive surveys to discover stars at [Fe/H] < -2,
followed by high-res spectra to obtain abundance patterns.
Key results: discovery of HMPs with [Fe/H] <~ -5 and widespread
strong enhancement of Carbon, the CEMPs.
Where are the oldest low-mass stars, and what do they tell us about star
and galaxy formation during the Epoch of First Light?
So, how can we use Galactic Archaeology to study the first stars?
(1) A Quick Review of Major Themes in the Study of “First Stars”
First Major Conclusion:
The theory of “First Light” is developed to the point of having some testable predictions,
which can be addressed in the near term with rapidly growing data from “Galactic
Archaeology” and in the long term with JWST.
As the sample sizes and dimensionality of the data explode, the theoretical challenge is to:
- make sense of all this data- come to grips with the awesome statistics
- define what “information” is present- place the observations in the proper context of high redshift
- properly translate physical theory into the data space.
The Challenge to Theory
+ + =
Star Formation Theory Nucleosynthesis Structure Formation Observations!
. . . In short, to create a “Virtual Galaxy” that will synthesize all this data, in the high redshift context.
A New Synthesis of Chemical Evolution & Structure Formation
HIERARCHICAL: Halo merger trees allow for chemical evolution calculations much faster than full hydro simulations, much
more realistic than “classical” GCE.
STOCHASTIC: Within each node, gas budget is tracked and new star formation
samples the IMF “one-star-at-a-time”. New star formation is assigned a metallicity
based on random sampling of “enrichment zones” from prior generations.
UNIFIED: Best of all, these “nodes” can be modeled as individual galaxies for direct
comparisons to data at high redshift – this is also the core of a galaxy formation code.
25
20
15
10
5
0
z
Tumlinson 2006, ApJ 641, 1
Pop III Halos
“Milky Way”
mc
= -2.35
“Very Massive Stars”
“Log-normal”
Z < Zcr
“Salpeter”
Z ≥ Zcr
Nu
mb
er
per
Ma
ss B
in
Discrete, Stochastic Chemical Evolution, “One Star at a Time”
Pure Z = 0
progenitors!
Tum
linson 2
006
Fo ≤ 1/N(<2.5)
≤ 0.0019
Zcrit = 10-4
“Pop II” [Fe/H] “Pop II” [Fe/H]
Tumlinson, Venkatesan, & Shull (2004) Yields: Heger+Woosley - Data: McWilliam95, Carretta02, Cayrel04
PISNe yields are characterized by big “Odd Even Effect” and no neutron capture nucleosynthesis.
Observed Fe-peak, eg. [Zn/Fe], require ≤ ½ of Fe from PISNe.
PISNe have no r-process, so cannot give 82% of EMPs with Ba.
Constraints on the Primordial IMF
Tumlinson (2006)
A
B
C
Too many “True” Pop III stars.
Too much Fe from PISNe
Too little r-process
Convergence on the First Stars IMF?
= -2.35
Tumlinson 2006a, ApJ, 641, 1N
um
be
r pe
r M
ass
Bin
“Theory”
Theory is still missing feedback of young star on final mass?
A
B
C
“Data”
Q: How can we study the IMF at Z > 0, i.e. for most stars during the
Epoch of “First Light”?
(2) A New Approach to Constraints on the
IMF of Primordial (“First”) Stars
Second Major Conclusion:
Using a new synthesis of theory that tracks stochastic early chemical evolution in the
proper high-z, hierarchical context, we can show that the first stars were predominantly massive stars, but find hints that additional feedback might be needed in simulations to
resolve remaining discrepancy.
The Answer: CEMP stars are born as low-mass partner in a binary system.
80% are CEMP-s that are rich in s-process elements (indicating AGB).
CEMP-s consistent with 100% binarity (Lucatello et al. ’05).
HE1327-2326
HE0107-5240
100% 40% 20% 10%
after Kom
iya et al. (2007)
A: The CEMPs!
CEMP = [C/Fe] > 1@ [Fe/H] < -2Beers & Christlieb (2005)
CEMP
Primary
From CEMPs to the IMF
LMS
IMS
1.5 80.5 40 1.5 80.5 40
IMS
LMS
Estimate from early CEMP studies: Mc > 0.8 Mּס (Lucatello+05).
There are no C-normal stars at [Fe/H] = -5.5, so Mc = 1.5 - 6 Mּס (Tumlinson07).
Komiya+2007 find mc ~ 10 Mּס to match s-element patterns of CEMPs.
The ratio of C-rich to C-normal stars in a population measures the ratio of intermediate to low-mass stars in the IMF!
1.5 - 8 Mּס
M ~ 0.8 Mּס
0.8 + 0.8 binaries are
favored.
“Low fCEMP”
0.8 + IMS binaries are
favored.
High fCEMP.
Zcrit
C-rich Pop II stars“CEMP”
Why would the IMF form more IMS, if Z ~ 10Why would the IMF form more IMS, if Z ~ 10-3-3ZZּס is high is high
enough to cool efficiently (enough to cool efficiently (Bromm+Loeb03, Schneider+02Bromm+Loeb03, Schneider+02)?)?
Tumlinson (2007a)
MW
H2 cools primordial gas to Tmin ~ 200 K, for MJ ~ 100 - 1000 Mּס
(Bromm, Coppi, & Larson 1999; 2002, Abel, Bryan, & Norman 2002)
30 – 300 Mּס accretes in a Kelvin-Helmholz time (O’Shea & Norman 2007).
2J
H
GM MkT
R m
3122
3 310 10J
n TM M
cm K
Key Concept #1: “Warmer (Primordial) Gas Forms Heavier Stars”
Studies of local star formation (Larson ‘98,’05; Jappsen et al. ’05) suggest that the characteristic mass of stars responds to the minimum T at which gas
becomes optically thick to cooling radiation and thermally coupled to dust.
At low redshift, Z = Zmin = 10 K is set by metal and dust cooling.
But at high z, the CMB at T = 2.73(1+z) K sets the minimum gas temperature!
Thus stars formed early in MW history, at z > 5, should be affected!
MC ≈ 0.9 Mּס [TCMB/10K]1.70-3.35
z = 5, 10, 20 TCMB = 16, 30, 57 K MC = 2, 6, 17 Mּס
Q: How can we test the CMB-IMF hypothesis?
A: Look for agreement between what we see as old in the nearby Universe and what we see
as young in the distant Universe.
(3) CEMPs and the IMF of the “Second” Stars
Third Major Conclusion:
IMF diagnostics in the most metal-poor stars, interpreted by a new hierarchical, stochastic theoretical framework, show evidence for a top-heavy IMF at high redshift that may be
physically independent of metallicity.
Low-z Test #1: Variation of CEMP Fraction with Metallicity
Stochastic, local phenomenon of chemical evolution implies that, on average, more metal-
poor stars form earlier, so fCEMP should increase with
declining [Fe/H].
Tumlinson (2006)
stochastic MW
(Tumlinson 2007b, ApJL, 664, L63)
With a CMB-IMF, fCEMP is high at low [Fe/H], and
declines with increasing [Fe/H] as the typical
formation redshift at a given metallicity declines.
HMPs
Key Idea for Prediction 2: The Halo is Built from the Inside Out. . .
Inside-out construction the halo causes extended epoch of star formation at fixed [Fe/H], so
fCEMP should increase in “older” regions of the Galaxy and decrease in “younger” regions, at fixed metallicity.
Low-z Prediction #2: Variation of CEMPs with Galactic Location
The CMB-dominated mass scale at ~ 10 kpc is 2 -10 Mּס.
At a given metallicity, stars in the inner halo are older, and this gradient gives a gradient of C-rich/C-normal fraction.
NCEMP
NCEMP+NC-normal
UPDATE
Also:
Faint end of WD luminosity function? (JWST)
fCEMP in dwarf spheroidals (GSMT)?
>100000 from SDSS/SEGUE for halo,
APOGEE for bulge and disk
SEGUE2: 105 more halo and thick disk stars w/ current SDSS spectrograph.
APOGEE: H-band spectroscopic survey of 105 giants in inner disk and bulge.
with the ARCHES spectrograph (PI Majewski at UVa).
“Virtual Galaxy” will be important to comparing the results of the two surveys for chemical and kinematic substructure in the ancient MW.
SDSS-III =
SEGUE-II (2008)
+ APOGEE (2011)
High-z Test: Mass-to-light ratios in High-z Galaxies
When estimates of dynamical mass / light ratios of “first-light” galaxies become possible with JWST and GSMTs, expect to see M/L decline with redshift, 2 - 5 times lower
than for a normal IMF.
van Dokkum (2008)technique from Tinsley (1980)
The Morals of the Story
1. Because many of the first galaxies are still with us, “Galactic Archaeology” with growing stellar surveys can uncover unique insights into the history of star and galaxy formation during First Light.
2. With this rich dataset and a new synthesis of theory, we can directly address some of the most pressing questions about the galaxies of “First Light” – such as how metallicity, redshift, and environment interact in shaping the IMF.
3. Early indications are that the Pop III and early Pop II IMFs during the epoch of reionization preferred intermediate and massive stars, with major implications for observable features of galaxies by JWST.
4. A new synthesis of theory is being developed to take advantage of this wealth of data, and connect it explicitly to high-z, as a perfect partner and complement to JWST. In the JWST era, we can test and extend these models to uncover a deep, unified view of First Light.
A Three-fold Vision for the Future
Theoretical: Complete the N-body theoretical framework, including many MW realizations, sharpened predictions for tests of the CMB-IMF hypothesis, and a systematic study of dSph abundances. Begin building framework for high-z. Observational: Collaborate (join?) with observers to test predictions and develop new ideas. Sloan SEGUE (current) > 20000 @ [Fe/H] < -2Radial Velocity Experiment (RAVE, current), 10000+
SDSS3 = SEGUE2 (Halo) + APOGEE (Bulge) 2008 – probably most criticalWFMOS: Wide Field Multi-Object Spectrograph (?) and others later
The challenge: to integrate the results and make optimal use of all information.
Unification: The goal is a full realization (gas included) that follows both a high-resolution MW to z=0 and a cosmological volume at high redshift. This model will allow us to test the same galaxy formation physics with both JWST and Galactic Archaeology data.
•Extra slides follow
H2 cools primordial gas to Tmin ~ 200 K, for MJ ~ 100 - 1000 Mּס
(Bromm, Coppi, & Larson 1999; 2002, Abel, Bryan, & Norman 2002)
30 – 300 Mּס accretes in a Kelvin-Helmholz time (O’Shea & Norman 2007).
2J
H
GM MkT
R m
3122
3 310 10J
n TM M
cm K
Key Concept #1: “Warmer (Primordial) Gas Forms Heavier Stars”
At Zcrit ~ 10-5.5 to 10-3.5 Zּס, efficient metal-line cooling may allow fragmentation to low-mass stars (Bromm & Loeb 2003; Santoro & Shull 2006).
But by this time there may also be dust, ionizing radiation, the CMB, cosmic rays, B fields. . so ab initio simulation is too hard.
To cut the knot of theory, we need observations!
Key Concept #2: “The Critical Metallicity” and the “2nd Stars”ORIGINAL
To understand the stars in “First Light” galaxies, we can apply some canonical diagnostic tests in the high-redshift Universe:
- blue colors and unusual emission lines (He II) with JWST and 30-m
- color and luminosity evolution in evolved populations
- GP effect and other tracers of reionization (CMB, 21 cm, LAEs)
However. . .
. . . these tests require facilities that are some years away (2013+), and
. . . they detect direct/reprocessed emission of massive stars, so are insensitive to the bulk of the stellar mass (in a normal IMF), and
provide poor tests of star formation physics at very low metallicity.
Both of these problems can be avoided if we look instead in the low-redshift Universe!
Paths to Star Formation during “First Light”
First Stars: The Hows and Whys
Simple recipe for first stars:• CDM • Dark matter “minihalos” of MDM ~ 106-7 M€at z = 20 - 40. • primordial composition (H,He,H2)• the absence of other (in)famously complicating factors (dust, B)
Red = Bound at z = 10
2J
H
GM MkT
R m
3122
3 310 10J
n TM M
cm K
H2 cools primordial gas to Tmin ~ 200 K, for MJ ~ 100 - 1000 Mּס
(Bromm, Coppi, & Larson 1999; 2002, Abel, Bryan, & Norman 2002)
30 – 300 Mּס accretes in a Kelvin-Helmholz time (O’Shea & Norman 2007).
Key Concept #1: “Warmer (Primordial) Gas Forms Heavier Stars”At Zcrit ~ 10-5.5 to 10-3.5 Zּס, efficient metal-line cooling may allow
fragmentation to low-mass stars (Bromm & Loeb 2003; Santoro & Shull 2006).
But by this time there may also be dust, ionizing radiation, the CMB, cosmic rays, B fields. . so ab initio simulation is too hard.
To cut the knot of theory, we need observations!
Key Concept #2: “The Critical Metallicity” and the “2nd Stars”
ORIGINAL
Major Themes of “The First Stars”Physics of Star Formation at Zero and Very Low Metallicity:
Approach: Hydrosims of gas physics in early cosmological halos
Key Results: High mass range (~30 - 300) for limiting Z = 0 case.
Formation of first low-mass stars depends on prior ionization
and/or metal enrichment metals, dust, CMB, other factors (?)
How did the first and second stars form, and what was their IMF?Stellar Evolution and Nucleosynthesis of the First Stars:
Approach: numerical stellar evolution and supernova models (1-D)
Key Results: “Pair Instability SNe” and “Hypernovae” may arise from
the first stars and give distinctive yield patterns.
Big question now is how much rotation alters mass loss and yields.
Given a particular IMF, what are the observational signatures (both radiation and chemical yields)?
Chemical Abundance Studies of Metal-Poor Pop II (“The Second Stars”):
Approach: massive surveys to discover stars at [Fe/H] < -2,
followed by high-res spectra to obtain abundance patterns.
Key results: discovery of HMPs with [Fe/H] <~ -5 and widespread
strong enhancement of Carbon, the CEMPs.
Where are the oldest low-mass stars, and what do they tell us about star
and galaxy formation during the Epoch of First Light?
ORIGINAL
“Low-z” Predictions of the CMB-IMF Hypothesis
(1)
Stochastic, local phenomenon of chemical evolution implies that, on average, more metal-poor stars form earlier, so fCEMP should increase with declining [Fe/H].
(2)Inside-out construction the halo causes extended
epoch of star formation at fixed [Fe/H], so fCEMP should increase in “older” regions of the Galaxy
and decrease in “younger” regions, at fixed metallicity.
(Tumlinson 2007a, ApJL, 664, 63)
ORIGINAL
Discrete, Stochastic Chemical Evolution, “One Star at a Time”
Pure Z = 0
progenitors!
Tum
linson 2
006
Fo ≤ 1/N(<2.5)
≤ 0.0019
Zcrit = 10-4
“Pop II” [Fe/H] “Pop II” [Fe/H]
ORIGINAL
IMFReionization
Metal enrichment
Colors
Kinetic Feedback
Compact ObjectsSpectral Features
Theme 1: Theory of Star Formation in Early Universe
Theme 1: Theory of Star Formation in Early Universe
IMFHeating (adiabatic,CMB)
Cooling (Metals)
Turbulence
Structure Formation
Magnetic FieldsFeedback