Simulating the Milky Way formation
- comparing models with real data -
Useful books/notes:
Mo, van den Bosch & White: Galaxy Formation & Evolution Andrea Ferrara’s Saas-Fe lectures:
http://www.sns.it/en/scienze/menunews/docentiscienze/ferraraandrea/lectures/
Chapter 2 of my Phd Thesis: http://www.astro.rug.nl/~salvadori/thesis.pdf
Stefania Salvadori
First stars/galaxies: simple sketch
M ~ 106M @ z ~ 25 Tvir < 104K H2-cooling
tcool << tff H2-cooling
Tc ~ 200K, nc ~104cm−3 Mclump ≈ MJ ≈ 700 M
Maccr ≈ Tc3/2
m* > 10 M
τlifc ≈ few Myr Feedback processes:
LW photons H2 dissociation Ionizing photons HII regions
Metal production/dispersion driven by SN explosions
Low binding energy:
gas/metals ejection
The minimum halo mass able to form stars increases Msf(z)
The metallicity Z of the ISM and IGM
increases
Subsequent generations
M > Msf (z) ?# YES# Z >Zcr=10−5±1 Z?#
YES# NO#
NO#
dark halo no stars
Different evolution, photon production, metal enrichment,
SN energy
Mclump ≤ M
m*=(0.1-100) M m* > 10 M
Mclump≈ 700M
PopIII vs PopII star-formation
Tornatore+2007
PopIII stars (Z < Zcr) are very rare and they disappear at z ~ 2
Stellar lifetimes
z = 25 Age = 0.13 Gyr
τ = 13.5 Gyr
z = 6 Age ~ 1 Gyr τ = 12.7 Gyr
z = 2 Age ~ 3.3 Gyr
τ = 10.3 Gyr
Surviving stars
Initial Mass Function
ϕ(m*) ~ m*−1+x exp(−mcut/m*) x = −1.35
mcut = 0.35 M
Looking for metal-poor stars
If the formation of “normal” low mass popII stars is triggered by the presence of metals and dust exceeding
Zcr =10 −5±1Z#
then the most metal-poor stars, Z ~ Zcr , that survive until today may represent the oldest stellar relics of the early Universe.
Where can we observe the most metal-poor stars?
thick disk
Mbulge ≈ 2. 1010 M
Mgas ≈ 1010 M
Mdisc ≈ 6. 1010 M
Mhalo ≈ 3. 109 M Stellar halo
thin disk
30 kpc
8 kpc
Sun
The structure of the Milky Way
4 kpc
thick disk open clusters
bulge thick disk globulars
young halo globulars
old halo globulars thin disk
thick disk
halo
Freeman&Bland-Hawthorn 2002
Signs of metal enrichment
Milky Way stars
N* = 2756 r < 20 kpc
HE1327-2326
HE0107-5240 HE0557-4840
Metallicity Distribution Function
Galactic halo stars
Beers&Christlieb2005
Caffau+11
N* = 2756 r < 20 kpc
HE1327-2326 HE0107-5240 HE0557-4840
Metallicity Distribution Function
Galactic halo stars
Beers&Christlieb2005
Caffau+11
Log Z/Z Zcr =10 −5±1Z
Dwarf spheroidal galaxies
dSph galaxies satellites of the MW
kpc
kpc Galactic center
Total masses M < 109 M. Gas-free systems. Old and metal poor stars
Outer halo
Metallicity-Luminosity relation
Kirby+08
Milky Way dwarf spheroidal satellites
Kirby+2008
Via Lactea simulation
Diemand+2007/2008
≈ 1,000,000,000 dark matter
particles
mp= 4.100×103M
Aquarius simulation
Springel+2008
Increasing resolution
4,252,607,000 mp = 1.712×103 M 148,285,000
mp = 4.911×104 M 2,316,893
mp = 3.143×106 M
Monte Carlo approach
MW
MMW = 1012 M
Time
z = 0
Redshift
Comparison with N-body Binary scheme
€
ψ = ε
*M
gt
ff€
dMg
dt = −
ψ
+ dRdt + dMinf
dt − dMej dt
€
dM
Zdt = − Z
ISMψ + dY
dt + Z
virdM
infdt − Z
wdM
ejdt
Z wZ w
ZISM
Zvir Zvir
Physical prescriptions/free parameters
εw tinf
Model calibration
Evoli&Ferrara2011
SFR ≈ 1.3 M/yr M* ≈ 6×1010 M
Mg/M* ≈ 0.1
68%
99%
99%
95%
Simplified case: only stars/gas no infall
The free parameters
General rule for semi-analytical models:
the higher is the number of equations (physics) involved the higher is the number of free parameters
the higher is the number of observational constraints needed
Example: if we also want to follow the evolution of metals along the build-up of the Milky Way we have to reproduce
the final metallicity of the gas/stars (~ Z)
along with the observed Z-range of Galactic halo stars
Constraining high-z properties
Once fixed the main free parameters (SF/wind efficiency) we can investigate (and then constrain?)
the properties of the first stars/galaxies
and/or the efficiency of feedback processes acting at high-redshifts
• What is the efficiency of star formation in H2-cooling haloes?
• Are H2-cooling haloes a “suicide” population?
• What is the evolution of the minimum halo mass to form stars?
• What is the value of the critical metallicity?
• What is the efficiency of mechanical feedback at high-z?
Questions we can try to address:
Number of DM haloes
Madau+08
The missing satellites problem
If all the haloes are able to form stars with a fixed efficiency
The number of predicted luminous satellites exceeds by several orders of magnitude the one observed.
The higher is the resolution of the simulation
the higher is the expected number of luminous satellites at z = 0
Radiative feedback processes are expected to gradually
reduce the SF in minihaloes and increase the minimum mass of haloes that are able to form stars. This is really a problem?
The SF efficiency of minihaloes
105 104
103
Ltot/L
Observations
Simulations: different SF efficiencies
Madau+08
The SF efficiency of mini -haloes has to decrease at decreasing mass in order to reproduce the observed
luminosity function of dwarf satellites
Imprints of radiative feedback
Munoz+09 105
104
103 106 107 108
Ltot/L
The number of luminous satellite galaxies predicted at z = 0 strongly
depends on the evolution of Msf(z)
Imprints of chemical feedback
Varying the critical metallicity
The predicted Metallicity Distribution Function of Galactic halo stars strongly depends on the assumed critical metallicity. The existence of the Caffau star implies Zcr < 10 −4 Z
Salvadori+07
Second generation stars
Second generation stars are extremely rare. The expected number of second generation stars in the currently limited Galactic halo sample is < 1-2.
Salvadori+07
Zcr = 10 – 4 Z Zcr = 10 – 6 Z Zcr = 0
2nd generation vs all generations
The most metal-deficient star
M ~ 0.7 M SDSS J102915+172927
Z = 5 × 10−5 Z
Can we finally see the imprint of the first stars?
Caffau+11
Just a “common” metal-poor star
The chemical abundance patter of SDSS J102915+172927 is consistent with that of stars with − 4 ≤ [Fe/H] ≤ − 3 no characteristic features produced by massive first stars !
Not surprising: second generation stars are extremely rare !
This observation does not implies that the first stars were not massive
Cayrel+04 Mean [X/Mg] value for 35 stars with [Mg/H] < − 3
The most iron-poor stars
Second generation stars ?
1. If the total metallicity reflects that of the ISM from which these stars form ZISM > 10 −3Z >Zcr. What kind of stars are
responsible for such a chemical enrichment?
2. If the iron abundance reflects the metallicity of the ISM from which they form ZISM ≈ 10 −5Z ≈ Zcr . dust is needed.
But CNO have to be accreted from a companion star
Caveat: for these stars [Fe/H] is not a good metallicity indicator!
Even if [Fe/H] < −4.8 the total metallicity is Z > 10 −3Z
What we learnt?
• Semi-analytical models are “cosmological bridges” that connect the physical processes acting at high-z with the Local observations.
• They are powerful tools to investigate the feedback imprints left in the Local Universe and the properties of the first stars/galaxies.
• If you want to build up a good semi-analytical model you have to compare your results with most of the available observations
• They have several free parameters (physical unknowns) that are fixed in order to reproduce the observed properties of the analyzed system.
• There are still many puzzling questions about the first cosmic objects that can be solved using these methods and the new observations!!