CNO abundances in the Sun and Solar Twins Jorge Meléndez Departamento de Astronomia, IAG, Universidade de São Paulo Sunset in Paracas, Peru (c) www.flickr.com/photos/rodrigocampos/
Why are C, N, O (and Ne) important? After H (X) and He (Y), they are the most abundant elements, so they have a crucial impact on the mass fraction of metals Z (elements heavier than H, He) H He O C Ne F Mg N Al Na Si Fe Asplund et al. 2009 X + Y + Z = 1. B Li Be
The metallicity Z is important for the morphology of evolutionary tracks and isochrones VandenBerg et al. 2007 VandenBerg et al. 2007
Solar models vs. helioseismology E = log(n E /N H ) + 12 O, C, N (dex) + meteoritic 8.86, 7.92, 8.52 8.69, 7.83, 8.43 8.69, 7.83, 8.43 + photospheric 8.66, 7.78, 8.39 GS98: Grevesse & Sauval 1998 AGS05 : Asplund et al. 2005 AGSS09 : Asplund et al. 2009 Serenelli et al. 2009
Solar Oxygen Abundance What is going on with the solar O abundance? Anders & Grevesse 1989 8.93±0.04 8.76±0.07 Caffau et al. 2008 8.66±0.05 Asplund et al. 2004 8.69±0.05 Asplund et al. 2009
Solar carbon abundance What about carbon? Anders & Grevesse 1989 Grevesse & Sauval 1998 Caffau et al. 2010 Asplund et al. 2009 Asplund et al. 2005 Year
Solar nitrogen abundance What about nitrogen? Anders & Grevesse 1989 Grevesse & Sauval 1998 Caffau et al. 2009 Asplund et al. 2009 Asplund et al. 2005 Year
Why are the Asplund, Grevesse & Sauval (2005) solar C, N, O abundances very low? 1) Improvements in the measurements on the solar spectra (blends with other species). Forbidden [OI] line at 6300 A Allende Prieto, Lambert, Asplund (2001) A(O)=8.69±0.05 3D
( dex BLEND with NiI (-0.13 TRANSITION PROBABILITIES FOR THE [OI] 6300 A line : Old log gf = -9.78 New log gf = -9.72 Difference in abundance: -0.06 dex 3D-1D: -0.04 Why are the Asplund et al. (2005) solar C, N, O abundances very low? 2) Improvements in the atomic data (transition probabilities) 3) Improvements in the model atmospheres (3D hydro) After blends + new gf-value + 3D : A(O) = 8,69 dex But ignoring 3D+blend+better data, A(O)=8,92 dex -Allende Prieto, Lambert & Asplund 2001, ApJ, 556, L63 -Asplund et al. 2004, A&A, 417, 751
Why Asplund et al. (2005) C, N, O are low? 4) Departures from local thermodynamic equilibrium accounted for (mostly). A(O)= 8.88 [LTE] Permitted lines: 777 nm OI triplet A(O)=8.66, 8.72 [NLTE] S H =0, 1 NLTE corrections ~ -0.2 dex 8.88 [LTE] Center / limb variation of OI 8.66 [NLTE] Allende Prieto, Asplund & Fabiani ( 2004 ) Bendicho 8.72 [NLTE] ( 1968 ) NLTE effects on OI 777nm already suggested by Altrock
Revision of Asplund et al. (2005): new solar CNO with improved 3D models (Asplund et al. 2009) The temperature gradient of the Asplund et al. (2005) 3D solar model atmosphere seem to be too step when compared to observations of the center-to-limb variation (Ayres et al. 2006; Koesterke et al. 2008). Disk center The new 3D model by Asplund et al. (2009) has much improved radiative transfer and updated opacities. The new model is in excellent agreement with the center-to-limb variation observed in the Sun. Asplund et al. (2009) Limb
Latest set of C, N, O abundances from Asplund, Grevesse, Sauval & Scott (2009) C = 8.43 N = 7.83 O = 8.69 Excellent agreement between atomic and molecular lines
What about other 3D models? Excellent agreement between Stagger (Asplund et al.) and CO 5 BOLD (Caffau et al.) codes Caffau et al. 2008-10 Asplund et al. 2009
Independent analyses by the Caffau et al. group (Caffau, Ludwig, Steffen, Freytag et al.) using 3D CO 5 BOLD models Element Caffau et al. (2008, 2009, 2010) Asplund et al. (2009) Carbon 8.50+/-0.1 8.43+/-0.05 Nitrogen 7.86+/-0.12 7.83+/-0.05 Oxygen 8.76+/-0.07 8.69+/-0.05 Good agreement on N, but not so good for O and C
Caffau et al. (2009) nitrogen abundance Used equivalent widths from the literature (Grevesse et al.) + CO 5 BOLD 3D model + NLTE Only atomic lines were analyzed by Caffau et al. (Asplund et al. analyzed atomic and molecular lines) Caffau et al. (2009) Asplund et al. (2009) 7.86+/-0.12 7.83+/-0.05 Agreement!
Caffau et al. (2010) carbon abundance Used their own measurements of equivalent widths + CO 5 BOLD 3D model + NLTE Only atomic lines were analyzed by Caffau et al. (Asplund et al. used atomic and molecular lines) Caffau et al. (2010) Asplund et al. (2009) 8.5+/-0.1 8.43+/-0.05 disagreement! Main reason for discrepancy is probably the different set of equivalent width measurements used by both groups.
Caffau et al. (2008) oxygen abundance Used their own measurements of equivalent widths + CO 5 BOLD 3D model + NLTE Only atomic lines were analyzed by Caffau et al. (Asplund et al. used atomic and molecular lines) Caffau et al. (2008) Asplund et al. (2009) 8.76+/-0.07 8.69+/-0.05 Caffau et al. like disagreement! H.-G. Ludwig, 2012 Main reason for discrepancy is the different set of equivalent width measurements used by both groups.
Good agreement for N, but who is more likely to be right (if any) regarding C and O? Element Caffau et al. (2008, 2009, 2010) Asplund et al. (2009) Carbon 8.5+/-0.1 8.43+/-0.05 Nitrogen 7.86+/-0.12 7.83+/-0.05 Oxygen 8.76+/-0.07 8.69+/-0.05 We urgently need a new independent study of the photospheric CNO solar abundances
An independent study of C, N, O in the Sun (Meléndez, in preparation, 2012?) - Selection of only best lines (not always possible) - Detailed evaluation of blends (mainly CN) - Careful measurements - Most recent atomic data - Using only telluric-corrected solar spectra (red, IR) - New optical-near IR flux atlas by Wallace et al. 2011 - Complemented in the IR by disk center atlas by Wallace et al. (1993)
Relative flux An independent study of C, N, O in the Sun (Meléndez, in prep., 2012?) Good knowledge of molecular line formation is needed to take into account blends. CN is ubiquitous in the solar spectrum 1.00 0.95 0.90 0.85 5000 5500 6000 6500 Wavelength (A) Meléndez, in preparation
Relative flux Relative flux CN is ubiquitous in the solar spectrum 1.00 (Meléndez, in prep., 2012?) 0.95 0.90 0.85 1.00 7000 7500 8000 8500 0.95 0.90 0.85 9000 9500 10000 10050 11000 11050 12000 Wavelength (A) CN line list from Meléndez & Barbuy (1999)
CN blends for O I triplet (Meléndez, in preparation) CN blends must be taken into account, otherwise the equivalent widths may be overestimated or underestimated CN CN CN CN Other blend CN CN
Comparison of equivalent widths for O I triplet Line (A) Asplund E.W. (A) Asplund et al. A(O) Caffau E.W.(A) Caffau A(O) Melendez E.W. Melendez A(O) 7771 71.2 8.68? 81.4 8.75 74.3 ± 2.6 8.742 7774 61.8 8.69? 68.6 8.74 63.9 ± 0.5 8.743 7776 48.8 8.70? 54.2 8.76 50.4 ± 0.5 8.741 average 8.69 dex? 8.75 dex 8.74 dex (Meléndez, in preparation) Asplund et al. E.W. may be underestimated by 3,5% Caffau et al. equivalent widths may be overestimated by 8,2% Caution: final result depends on adopted NLTE corrections (uncertain at least ~ 0,03 dex)
What about the forbidden [OI] lines? 6300 A line: badly blended with Ni I. Uncertain correction. 6363 A line: blended with CN. Very preliminary result, A(O) ~ 8.7 5577 A line: blended with C 2. Very preliminary result, A(O) ~ 8.7
5577 A [OI] line! Blend by C 2 easily constrained by red side of the feature. O and C abundances determined at the same time! revised Melendez & Asplund (2008). To be
VERY PRELIMINARY RECOMMENDED OXYGEN ABUNDANCE (molecular OH lines will be included later) Oxygen triplet at 777nm : 8,74 dex Forbidden [OI] lines 6363, 5577 A : 8,7 dex Suggested solar O abundance : Meléndez (in prep.) : 8,72 dex Caffau et al. 2008: 8,76 dex Asplund et al. 2009 : 8,69 dex
What about carbon? Also, we must be careful about blends with CN lines Example : C I lines at 7111,4 and 7113,2 A CN CN CN CN CN CN CN CN CN CN C I C I
VERY PRELIMINARY RECOMMENDED CARBON ABUNDANCE (based on permitted lines and one [C I] line. Molecular lines to be included later) Suggested solar C abundance : Meléndez (in prep.) : 8,40 dex Caffau et al. 2008: 8,50 dex Asplund et al. 2009 : 8,43 dex
VERY PRELIMINARY RECOMMENDED NITROGEN ABUNDANCE (based on permitted lines. Molecular lines to be included later) Suggested solar N abundance : Meléndez (in prep.) : 7,82 dex Caffau et al. 2008: 7,86 dex Asplund et al. 2009 : 7,83 dex
Summary on C, N, O solar abundances Element Caffau et al. 2008-10 Asplund et al. (2009) Meléndez (2012?) GS98 Carbon 8.50+/- 0.10 8.43+/- 0.05 8,40 8,52 Nitrogen 7.86+/- 0.12 7.83+/- 0.05 7,82 7,92 Oxygen 8.76+/- 0.07 8.69+/- 0.05 8,72 8,83 high thetoreturntounlikelyseemsitdatacurrenttheonbased Grevesse & Sauval (1998) C, N, O abundances
Sunset in Paracas, Peru (c) www.flickr.com/photos/rodrigocampos/ Is it correct to assume the solar abundance pattern for other stars? Solar twins can tell us how typical is the Sun
Magellan ultra high precision study of Observations of the solar twin 18 Sco solar twins -Magellan 6.5m Clay Telescope & Mike spectrometer - R = 65,000 - S/N = 450 per pixel - coverage 340 1000 nm - Solar spectrum: Vesta - 3 nights of observations BLUE frame RED frame -Meléndez et al. 2009 32
Example of Magellan spectra (total spectral coverage 3400 A -1um) Small part (597-603nm) of solar twin & Sun s spectra -Meléndez et al. 2009 33
Δabundance: Sun - <twins> vs. atomic number Z Sun typical : Δ = 0 Sun weird : Δ 0 Our solar system is not host by a typical Sun Melendez et al. 2009 34
Sun s anomalies are strongly correlated to condensation temperature (T cond ) of the elements! Correlation is highly significant probability ~10-9 to happen by chance ~ 0.08 dex ~ 20% It s most likely to win the lottery Condensation temperature Melendez et al. 2009 35
Condensation Condensation in the solar nebula Mercury Venus 36
The late accreted gas in the convection zone was deficient in refractories The late-accreted Sun stheingas convection zone was depleted in refractories which were used to form dust, planetesimals & terrestrial planets Iron gradient in the inner solar system 37
Either 1) the Sun is normal in the volatile C, N, O AND depleted in the refractory elements (e.g. Fe, Al, Ni) or 2) the Sun is normal in the refractories but enhanced in the volatiles (CNO) Option (1) is most likely correct ~ 0.08 dex ~ 20% Condensation temperature Melendez et al. 2009
Results on solar anomalies confirmed by others. Our Sun is indeed chemically peculiar, perhaps due to the formation of terrestrial planets The Sun may be more metal-rich in its interior! Meléndez et al. 2012 39
What about the effect of giant planet formation? 16 Cyg pair of solar analogs 16 Cyg A : no planets 16 Cyg B : giant planet 40
16 Cyg B (planet-host) is 0,04 dex more metalpoor in all elements (photospheric abundances)! Was the missing material used to form the giant planet around the solar analog 16 Cyg B? Is 16 Cyg B more metal-rich in its interior? 41
Conclusions Solar C, N, O abundances seem intermediate between the values of Asplund et al (2009) and Caffau et al. (2008, 2009, 2010), although somewhat closer to Asplund et al. Sun may be more metal-rich in its interior, with refractory elements about 0,08 dex more abundant than in its photosphere (terrestrial planet effect) Sun may be more metal-rich in its interior for ALL ELEMENTS by another 0,04 dex (giant planet effect) In summary, Sun may be richer in its interior by 0,12 dex for refractory elements and 0,04 dex for volatile CNO elements. It is urgent to estimate the Sun s abundances in its interior. 42