5th Russbach Workshop, March 3 – 7, 2008
Russbach, Austria
Chemical evolution
of galactic stellar populations
from observations of cool stars
Lyudmila Mashonkina
Institute of Astronomy, Russian Academy of Sciences, Moscow
Outline
• Basics of stellar atmosphere and
line formation modelling.
• Accuracy of stellar abundances.
• Element abundance trends:
- thin disk vs. thick disk,
- thick disk vs. halo.
• Stellar isotope abundances.
I. Basics of stellar atmosphere and
line formation modelling
Astrophysical experiment is measurement of stellar fluxes.
Stellar atmosphere: thin layer on top of the stellar core from
which radiation can escape into space.
• Optically thick (τν > 1), geometrically thin (H ~ 10-3 R).
• Energy is not generated. Energy transport by radiation and
convection.
Stellar parameters and chemical abundances
are obtained from fitting theoretical spectra to
the observed spectral data.
Assumptions and restrictions in modelling the atmosphere.
Atmosphere is ‐ composed of homogeneous layers, plane‐parallel or spherically‐symmetric, 1D‐geometry,
‐ static,
‐ in Local Thermodynamic Equilibrium (LTE).
For given
effective temperature Teff,
surface gravity log g,
chemical composition, {εi}
unique temperature- and pressure-stratification is computed.
is model atmosphere.
T(z), p(z), pe(z)
Calculations of theoretical spectrum
Radiation transfer (RT) equation:
z
dI (z)
μ ν = − χ ν (z)I ν (z, μ) + ην (z)
dz
Opacity
c
l
χ ν =χ ν +χ ν
c
l
emissivity ε ν =ε ν +ε ν
c
θ
hν
μ = cos θ
Continuum opacity χ ν : photoionization of H, He, metals,
f-f absorption, Thomson scattering, etc.
Line absorption:
χνl = χνl (nA,r,i , fij ,γR ,γ4 ,γ6 , HFS,IS)
number density of absorbers
Atomic parameters
nA,r,i
how to compute?
• LTE: nA,r,i = f(T, p)
from the Saha‐Boltzmann equations.
• Is LTE valid in line formation layers?
– mean free-path of photons is not small,
– radiation field is far from TE.
non-LTE: nA,r,i are determined from balance among radiative and collisional population and
de‐population processes.
Statistical equilibrium (SE) equations:
∑ n (R
j≠i
j
ji
+ C ji )= n i ∑ (R ij + C ij )
j≠i
ni = f(n1 ,…, nNL , J 1 ,…, J NF )
dn i / dt=0
i = 1,... NL
Emissivity: how to compute?
• LTE: for thermal processes
• non-LTE:
3
l
l 2hν ij
ε ν / χ ν= 2
c
εν / χν = Bν (T)
1
ni g j
−1
n j gi
Line formation problem is reduced to simultaneous solution
of huge system of RT and SE equations.
Stellar abundances are not observed, but calculated!
II. Can we reliably extract the abundances
encoded in stellar spectra? • observational errors:
noise, identification of the local stellar continuum flux,
• stellar parameter (Teff, log g, ξt) errors,
• atomic parameter uncertainties,
• simplified stellar atmosphere and line formation
modelling: LTE vs. non-LTE, 1D vs. 3D
Observational errors:
For weak line with Wλ = 10 mA:
R ~ 40000, S/N ~ 100
The bias in a continuum fit is
Wλ ~10 mÅ
shown by dotted curves.
Solid curve is theoretical
spectrum.
Full measurement error is σ(Wλ) ~ 10 %
Stellar parameter errors
Teff, log g are determined using different methods based on - color indexes, e.g. V-K, b-y, c1
σ(Teff) = 100 K
- line profiles, e.g. Hα, Mg Ib
- Wλ ratios,
e.g. W(FeI)/W(FeII)
For example:
Teff from b-y,
g from c1
Teff from Hα, Hβ, g from Mg Ib
σ(log g) = 0.1
(Edvardsson et al. 1993)
(Fuhrmann 1998, 2000)
Fuhrmann 1998, 2000
Atomic parameter uncertainties
Effect on derived abundances can be avoided (minimised)
- fij in differential analysis: [X/H] = log(ε gifij)∗ – log(ε gifij)Sun
- γ4, γ6 using weak lines.
• HFS of Mn I lines: effect on derived Mn abundances.
outdated
modern HFS data
[Mn/Fe]
Nissen et al. 2000
Important! for constraining nucleosynthesis and the Galaxy chemical evolution models.
Uncertainties of modelling: non-LTE vs. LTE
Departures from LTE grow towards lower metallicity due to ‐ decreasing collision rates,
‐ increasing photoionization rates for important levels. One example:
C abundance from C I ~9100 A.
LTE:
[C/Fe] = 0.3 – 0.5 at [Fe/H] < -0.7.
Carbon is α–process element?
Non-LTE: [C/Fe] ≈ 0
Common origin of C and Fe?
[C/Fe] vs. [Fe/H]
Non-LTE (filled squares) and
LTE (open squares)
(Fabbian et al. 2006)
Uncertainties of modelling: 3D vs. 1D Atmospheres of late type stars: developed convection.
Homogeneity ‐ ?
• Effect of 3D modelling is maximal for spectral lines sensitive to T variations. n(ОН):
3D
1D, HM
strong dependence on T
IR: OH rotational lines
log O/H = -3.35 to -3.25
~ -3.17
UV: OH vibration- rotation lines
~ -3.39
~ -3.13
Sun:
HM – 3D = 0.08 – 0.26 dex
(Asplund et al. 2004, A&A 417, 751)
Galactic abundance trends as important tools of
the Galaxy chemical evolution study
For example,
Cayrel et al. 2003
[Fe/H]
9 Observations: [Mg/Fe] ≈ 0.4 at [Fe/H] < ‐1 →
common origin of Mg and Fe.
Theory: astrophysical site is SNeII.
9 Observations: [Mg/Fe] decline at [Fe/H] > ‐1 →
increasing contribution to Fe from the lower mass stars.
Theory: Fe production in SNeIa. First explosions 0.5‐1 Gyr after the beginning of protogalactic collapse. Stellar populations of the Galaxy:
◊ thin disk (Population I),
W
-- H ≈ 300 pc, 93% of stars,
U
-- small dispersion of Vspace,
-- young,
V
-0.5 < [Fe/H] < +0.3.
◊ halo (Population II),
-- 0.6% of Galactic stars,
-- eccentric, arbitrary inclined orbits,
-- old, [Fe/H] < -1.
◊ thick disk,
-- H ≈ 1.5 kpc, 7 % of stars,
-1.5 < [Fe/H] < 0(?).
◊ bulge.
corona
(dark halo)
subdwarf halo
(globular clusters)
thin disk
Sun
bulge
thick disk
halo
thick
disk
thin disk
from Reddy et al. (2006)
Observations: thick disk and thin disk stellar populations have different chemical history. • Thick disk;
ο Thin disk
Thick disk stars are more enhanced
in α–elements (O, Mg, Si, Ca, Ti):
Gratton et al. (1996, 2000), ...,
Reddy et al. (2003, 2006).
• Step‐like (!) change in α/Fe.
• Decline of O/Fe.
Fuhrman 2004
∗ Halo, • Thick disk, ο Thin disk
Thick disk stars are more enhanced in r–process elements (Eu, Nd): Mashonkina et al. (2000, 2003, 2004), ...,
Reddy et al. (2006).
• [Eu/Fe] – [Fe/H] resembles
[α/Fe] – [Fe/H].
(Arlandini et al. 1999)
• [Eu/Ba]: step‐like (!) change. ⇒ Ba in thin disk:
s:r is close to solar (81:19, Arlandini et al. 1999),
⇒ Ba in thick disk: r–process contributes 50‐70%.
Halo
Cayrel et al. 2004
[Mg/Fe]=0.27±0.13
• Enhanced in α–elements with small intrinsic scatter of α/Fe.
(Cayrel et al. 2004, Cohen et al. 2004,
Barklem et al. 2005).
∗ halo,
• thick disk
Nissen&Schuster 1997
⇒ Common astrophysical site for α–elements and Fe in the early Galaxy. • α-poor stars:
‐ accreted halo?
? Halo is non‐homogeneous in origin?
Halo, n‐capture elements
• n‐capture/Fe, significant scatter
⇒ different sites for production of n‐capture elements and Fe.
Sneden&Cowan 2003
• Sr,Y,Zr/Ba grow toward lower Ba abundance.
? What process yielded Sr, Y, Zr with little contribution to Ba?
Barklem et al. 2005
r-process
(Arlandini et al.1999)
• Eu/Ba ≈ constant. ⇒ r‐process synthesis of Ba and Eu.
Mashonkina et al.
(2007)
Isotopic abundances in stars. Why are they important?
For n‐capture elements, isotopic abundances directly confront s‐ and r‐process predictions.
Ba:
Solar System: fodd = 0.18
r-process:
• calculations: fodd = 0.44
(Kratz et al. 2007)
• r-residuals:
- Classical model
fodd = 1.0, [Eu/Ba]r = 1.06
- Stellar model
fodd = 0.46, [Eu/Ba]r = 0.69
(Arlandini et al. 1999)
!
AG: Anders & Grevesse (1989)
A : Arlandini et al. (1999)
T : Travaglio et al. (1999)
r-residuals = solar abundance – s-contribution
Determination of fodd is possible due to
HFS affecting Ba II λ4554 Å
of the odd isotopes.
138
136
137
134
135
fodd = 0.18
0.50
1.00
-- The larger fodd , the stronger λ4554 is.
Isotopic and HFS components
of Ba II λ4554. Relative
intensities correspond to SS
Ba isotope mixture.
-- HFS is negligible for Ba II λ5853, λ6496.
Idea: Ba abundances derived from subordinate lines and resonance line must be equal.
Fractional abundance of Ba odd isotopes in stars
(Kratz et al. 2007)
∗ HD140283
Mashonkina et al. 2006,2008
Halo:
HD 84937 fodd = 0.43
HD 103095
0.42
HD 122563, fodd = 0.22
HD 140283
0.31
(Lambert & Allende Prieto 2002)
Fractional abundance of Ba odd isotopes in stars
(Kratz et al. 2007)
∗ HD140283
Mashonkina et al. 2006,2008
Halo:
HD 84937 fodd = 0.43, [Eu/Fe] = 0.70, [Eu/Ba] = 0.70
HD 103095
0.42,
0.56,
0.62
HD 122563, fodd = 0.22, [Eu/Fe] = -0.51, [Eu/Ba] = 0.50
HD 140283
0.31,
-0.31,
0.49
(Lambert & Allende Prieto 2002)
Fractional abundance of Ba odd isotopes in stars
(Kratz et al. 2007)
∗ HD140283
Mashonkina et al. 2006,2008
Halo:
HD 84937 fodd = 0.43, [Eu/Fe] = 0.70, [Eu/Ba] = 0.70
HD 103095
0.42,
0.56,
0.62
pure r‐process nucleosynthesis
HD 122563, fodd = 0.22, [Eu/Fe] = -0.51, [Eu/Ba] = 0.50 r‐process poor
HD 140283
0.31,
-0.31,
0.49
(Lambert & Allende Prieto 2002)
What nucleosynthesis?
Fractional abundance of Ba odd isotopes in stars
(Kratz et al. 2007)
∗ HD140283
Mashonkina et al. 2006,2008
fodd grows towards lower Ba abundance in thin disk and thick disk.
Thick disk:
mean fodd = 0.33 ± 0.04
The larger r‐process contribution to Ba compared to SS matter.
Conclusions
♦ Stellar abundances are not observed, but calculated.
♦ Our understanding of how nucleosynthesis proceeds throughout the Galaxy history depends on accuracy of obtained stellar parameters, uncertainty of atomic data, validity of atmosphere and line formation modelling. ♦ Chemical evolution of halo – thick disk – thin disk was not steady continuous.
Thick disk: 9 older than thin disk, 9 chemical enrichment SNeII dominate,
SNeIa and AGB start to work.
• Element abundance trends are tight.
(Nissen&Schuster 1997, ..., Reddy et al. 2006)
9 Lack of cosmic scatter.
Additional slides
The light element abundance ratios
Zr/Y, Zr/Sr, Sr/Y are almost constant in the halo stars.
Mean Zr/Y = 0.77±0.09, Zr/Sr = 0.16±0.11.
- The r-process-rich stars and halo stars have consistent
within error bars Zr/Y and Zr/Sr ratios.
Barklem et al. (2005), for five r-II stars: Zr/Y = 0.78, Zr/Sr = 0.02.
Stellar parameters
(Fuhrmann 1998; 2004; Mashonkina et al. 2003)
Teff
from Balmer line wings;
log g from Hipparcos parallaxes and Mg Ib line wings;
[Fe/H] from Fe II lines
Тeff = 4990 – 6470 K
Δ Тeff = 80 K
log g = 3.12 – 4.66
Δ log g = 0.1
[Fe/H] = (-1.71) – 0.25
Δ [Fe/H] = 0.1
Stellar population
(Klaus Fuhrmann)
(thin disk, thick disk, halo)
based on
star’s kinematics + [Mg/Fe] + [Fe/H] + star’s age
Uncertainty of the odd isotope fraction in stars
Sources of random errors:
• total Ba abundance error,
• uncertainty of stellar parameters, Teff, log g, microturbulence value;
and systematical errors:
• uncertainty of atomic parameters of the Ba II 4554, fij and C6