M. Asplund

Solar chemical composition
Does the Sun have a
subsolar metallicity?
Martin Asplund
Main partners in crime
3D stellar modelling:
Mats Carlsson (Oslo)
Remo Collet (MPA)
Wolfgang Hayek (MPA)
Åke Nordlund (Copenhagen)
Regner Trampedach (ANU)
Solar abundances:
Carlos Allende Prieto (UCL)
Nicolas Grevesse (Liege)
David Lambert (Austin)
Jacques Sauval (Brussels)
Patrick Scott (Stockholm)
Solar abundances
The solar chemical composition is a
fundamental yardstick for almost all astronomy
Big Bang
Some compilations:
Russell (1929)
Unsöld (1948)
Suess & Urey (1956)
Goldsmith et al. (1960)
Anders & Grevesse (1989)
Grevesse & Sauval (1998)
Lodders (2003)
Asplund et al. (2005)
Fe-peak
Neutron
capture
Odd-even
effect
Solar system abundances
Meteorites
Mass spectroscopy
Very high accuracy
Element depletion
Solar atmosphere
Solar spectroscopy
Modelling-dependent
Very little depletion
Elemental fingerprints
Stellar flux
The chemical composition is encoded in stellar
spectra but detailed modelling is required:
• Stellar atmospheres
• Radiative transfer
Wavelength
Mats Carlsson (Oslo)
Solar atmosphere
1D solar atmosphere models
Theoretical models:
• Hydrostatic
• Time-independent
• 1-dimensional
• Convection: mixing length theory
• LTE
• Detailed radiative transfer
• MARCS, Kurucz etc
Semi-empirical models:
• Temperature structure from
observations
• Holweger-Mueller (1974)
3D solar atmosphere models
Ingredients:
• Radiative-hydrodynamical
• Time-dependent
• 3-dimensional
• Simplified radiative transfer
• LTE
Essentially parameter free
Spatially resolved lines
Line profiles vary
tremendously across
the solar surface
3D LTE describes
observations very
well in most cases
Averaged line profiles
1D vs Sun
3D vs Sun
No micro- and macroturbulence needed in 3D!
Solar abundances revisited
Asplund et al. (2000a,b, 2004, 2005a,b, 2006, 2008),
Allende Prieto et al. (2001, 2002), Scott et al. (2006)…
3D hydrodynamical solar model atmosphere
Non-LTE line formation when necessary
Atomic and molecular lines with improved data
Element
Anders &
Asplund
Grevesse (1989) et al. (2005)
Difference
Carbon
8.56+/-0.06
8.39+/-0.05
-0.17 dex
Nitrogen
8.05+/-0.04
7.80+/-0.05
-0.25 dex
Oxygen
8.93+/-0.03
8.66+/-0.05
-0.27 dex
Note: logarithmic scale with H defined to have 12.00
Oxygen diagnostics
Discordant results in 1D: log O~8.6-8.9
Excellent agreement in 3D: log O=8.66+/-0.05
Asplund et al. (2004)
MARCS
HolwegerMueller
3D
[O I]
8.72+/-0.01
8.76+/-0.02
8.68+/-0.01
OI
8.72+/-0.03
8.64+/-0.08
8.64+/-0.02
OH, dv=0
8.83+/-0.03
8.82+/-0.01
8.65+/-0.02
OH, dv=1
8.74+/-0.03
8.87+/-0.03
8.61+/-0.03
OH, dv=2
8.61+/-0.06
8.80+/-0.06
8.57+/-0.06
Lines
[O I] 630nm line
Revised solar O abundance: log O=8.69+/-0.05
Blend with Ni: -0.13 dex
3D-1D model: -0.08 dex
Allende Prieto et al. (2001)
Ni blend (not
noticed in 1D)
O I 777nm lines
Line strength
Revised solar O abundance: log O=8.64+/-0.02
3D non-LTE line formation
Non-LTE - LTE: -0.25 dex
Asplund et al. (2004)
Non-LTE
LTE
Viewing angle on solar disk
OH lines
Revised solar O abundance: log O=8.61+/-0.03
Molecular lines extremely sensitive to temperature
No significant trends with line strength or χexc
Asplund et al. (2004)
Holweger-Mueller
3D model
Carbon diagnostics
Discordant results in 1D: log C~8.4-8.7
Excellent agreement in 3D: log C=8.39+/-0.05
Asplund et al. (2005)
[C I]
CI
CH, dv=1
CH, A-X
8.40
8.35+/-0.03
8.42+/-0.04
8.44+/-0.04
HolwegerMueller
8.45
8.39+/-0.03
8.53+/-0.04
8.59+/-0.04
C2, Swan
CO, dv=1
CO, dv=2
8.46+/-0.03
8.55+/-0.02
8.58+/-0.02
8.53+/-0.03
8.60+/-0.01
8.69+/-0.02
Lines
MARCS
3D
8.39
8.36+/-0.03
8.38+/-0.04
8.45+/-0.03
8.44+/-0.03
8.40+/-0.01
8.37+/-0.01
CO vib-rot lines
Weak CO lines: log C=8.39±0.05
O isotopic ratio: 16O/18O=480±30
C isotopic ratio: 12C/13C=87±4
Scott et al. (2006)
= Terrestrial:
16O/18O=498.7±0.1
12C/13C=89.4±0.2
Holweger-Mueller
3D model
Excitation potential
Line strength
Solar system O isotopes
Solar photospheric isotopic ratios:
δ13C=+30±45
δ18O=+40±60
Comparison with solar wind + lunar soil:
Hashizume & Chaussidon 2005
Ireland et al. 2006
Other elements
Very good agreement
with C1 chondrite
meteorites
Published 3D results
for all elements from
Li to Ca, other
elements ongoing
Change in meteoritic
scale due to 3D
analysis of Si
Volatiles
Solar
depletion
Some other recent work
• 3D-based abundance analyses:
Caffau et al. 2007a,b; 2008a,b;
Mucciarelli et al. 2008, etc
N, O, S, P,
Eu, Hf, Th
• 1D non-LTE line formation:
Gehren et al. 2003
Zhang et al. 2006, 2008
Bergemann et al. 2007, 2008, etc
• Improved atomic data:
Lawler et al. 2001-2008
Den Hartog et al. 2005, 2006
Sobeck et al. 2007
Si, K, Sc, Co,
Fe, Mn etc
Cr, Mn, Sr,
rare earths
etc
(Some) Implications
Significantly lower solar metal mass fraction Z
– Z=0.0194 (Anders & Grevesse 1989)
– Z=0.0122 (Asplund et al. 2005)
Alters cosmic yardstick
– [X/H], [X/Fe] etc
Makes Sun normal compared with surroundings
– Young O,B stars in solar neighborhood
– Local interstellar medium
Changes stellar structure and evolution
– Wrecks havoc with helioseismology
Trouble in paradise
Sound speed difference
Bahcall et al. (2004):
New solar
abundances
Convection
zone
Old solar
abundances
Solar radius
Solar interior models with new abundances
are in conflict with helioseismology
Possible solutions
Missing opacity?
− Unlikely
Underestimated element diffusion?
− Unlikely
Accretion of low-Z material?
− Possibly
Internal gravity waves?
− Possibly
Underestimated solar Ne abundance?
− Possibly
Erroneous solar CNO abundances?
− Hopefully not
Combination of some of the above?
− Contrived?
Oxygen
[O I]: Ni blend
Johansson et al. 2003:
gf-value of Ni I blend
measured experimentally
Scott et al., in preparation:
New solar Ni abundance
Δlog O ≈ -0.03 dex
[O I]: another 3D model
Ayres 2008:
New wavelength calibration
of solar atlas and one
snapshot of a different 3D
model (CO5BOLD)
log O = 8.81 ± 0.02
If Ni I blend treated as
free parameter
Is new Ni gf-value wrong
by 0.15 dex?
[O I]: another line
Melendez & Asplund 2008:
[O I] 557.7nm not
previously considered
log O = 8.73 ± 0.02
(mean of various 3D and
1D models)
How well can the C2
blending be estimated?
[O I]
C2
O I: another 3D analysis
Caffau et al. 2008:
O I lines with CO5BOLD: log O = 8.76 ± 0.05
- Choice of H collisions: Δlog O ≈ +0.04 dex
- Equivalent widths:
Δlog O ≈ +0.09 dex!
777.1nm
mean
777.4nm
Ca
ffa
u
777.5nm
O I: non-LTE and collisions
Electron collisions
New QM calculations
Barklem 2007
Fabbian et al. 2008
Δlog O ≈ -0.02 dex
Hydrogen collisions
Very poorly known
Use Drawin 1968 recipe with scaling factor SH
SH guessed until recently:
Asplund et al. 2004: SH=0 (other elements)
Caffau et al. 2008: SH=1/3 (why not)
O I: center-to-limb variations
Best agreement:
SH=0.3+/-0.2
Asplund et al. 2004:
log O=8.64
Δlog O ≈ +0.04 (H)
Δlog O ≈ -0.02 (e-)
Line strength
Pereira et al., in preparation:
Better observations of O I 777nm plus [O I] 630nm
and O I 616nm plus spatially resolved spectra
Viewing angle
O I: 3D semi-empirical model
Socas-Navarro & Norton 2007:
Observations of Fe I lines to map T(x,y,z) over surface
⇒ 3D semi-empirical model
O I 777nm non-LTE calculations without H collisions:
log O ≈ 8.63 ± 0.08 dex
OH: erroneous T-structure?
Ayres et al. 2006:
Asplund et al. 3D model
has wrong mean
temperature structure
Asplund et al.
Ayres et al. (CO lines)
Higher abundances
from molecular lines
OH: temperature diagnostics
Center-to-limb variation:
Asplund et al. 3D model
has too steep T-gradient…
(but better than any 1D
theoretical model)
Pereira et al., in preparation
…but agrees better
with H lines than
Holweger & Müller
(1974) model
Oxygen: status report
No clear consensus what the real solar O abundance is
Personal guess: log O ≈ 8.70-8.75?
Lines
HolwegerMueller
Asplund et
al. (2004)
Real Sun?
[O I]
8.76+/-0.02
8.68+/-0.01
~8.7-8.75
OI
8.64+/-0.08
8.64+/-0.02
~8.7
OH, dv=0
8.82+/-0.01
8.65+/-0.02
OH, dv=1
8.87+/-0.03
8.61+/-0.03
OH, dv=2
8.80+/-0.06
8.57+/-0.06
~8.7?
Stay tuned for a re-analysis of all elements
with improved 3D solar model
Galactic neighborhood evidence
ISM and OB stars: O and Ne
Good agreement with low solar O abundance
Cunha et al. (2006), Esteban et al. (2004)
OB stars
Neon?
Orion
Asplund et al.
Old solar
Local OB stars
Excellent agreement with low solar abundances
Przybilla et al (2008)
Corrections for past 4.5Gyr:
-0.05 dex (solar diffusion)
+0.05 dex (chemical enrichment)
Solar twins
Melendez & Ramirez 2007:
Stars with same
stellar parameters
(Teff, log g, [Fe/H])
have identical
abundance ratios
Best solar twin known
Radial mixing in
the Galaxy unlikely
for the Sun
Sun enhanced in biogenics?
Melendez et al. 2009
11 solar twins sofar
Very small
differences but…
Sun is enhanced
(+0.03 dex) in the
biogenic elements
(C, N, O, P, S)
Summary
• Solar chemical composition
- Asplund et al. (2005)
- 3D, non-LTE and input data important
• Low solar abundances
- Consistent with ISM and OB stars
- Problematic for helioseismology
- O abundance (slightly) underestimated?
• Sun is normal thin disk star
- [X/Fe] consistent with [Fe/H]