Present accuracies in spectroscopic chemical abundances

”These things … were not discovered by philosophy or the arts of reason, but by chance. …. There are still many things of excellent use, stored up in the lap of nature, … lying quite out of the path of imagination.”

Francis Bacon, Novum Organum (1620)

Present accuracies in spectroscopic chemical abundances

• Rarely better than 0.1 dex (even relatively)• Some ~0.01 dex. What do we learn from

those?

Solar composition not normal

for solar-type stars

Meléndez et al. (2009)

≈0.08 dex≈20%

Birth environment? Effects of planets? Not full mixing?

Present accuracies in spectroscopic chemical abundances

• Rarely better than 0.1 dex (even relatively)• ~0.01 dex in special cases only.What are the problems?• Obs. data: Blends, continua• Fundamental parameters: Teff, log g, mass, radius, distance,

extinction, …• Modelling: Atomic data, non-LTE, convection, …• Interpretation: What do atmospheric abundances represent?Solution: Fit realistic and physically consistent 3D NLTE, MHD models to adequate observables. (Any other decent ways??)

The multi-D world of stars

And there are more: , B, d/dt of all of the above.

Stars contain a wealth of information about themselves, their planetary systems and their birth environments

Gilm

ore

et a

l. (2

012)

, The

Mes

seng

er

The many Ds of hi-res instrumentation

• Wavelength coverage• Resolving power• Achievable signal-to-noise ratio• Various calibrational needs• Stability• Multiplexing• Polarimetry• ... Can we have it all

in one instrument?

Wavelength coverage: Spectral lines and continua

GJ 849 (M 3.5 V) Heiter et al. (2012)

YCVn (C 5,4) UUAur (C 5 II) Lambert et al. (1986)

J-band K-band ”Quasi continua”for C stars CO 1st overtone offers diagnostics on

structure and dynamicsNote: unidentified lines!

Near IR useful for cool stars… but many unidentified lines

as yet.

J-bandSatisfactory continua for M stars

Spectral resolution:Line profiles for convection, rotation, magnetic fields, …

Ramírez et al. (2010): HD 122563

Line profiles for convection, rotation,

magnetic fields, …

Gray & Brown (2006):Arcturus

Resolve the spectra fully!

NG

C 63

97 @

[Fe/

H]=-

2.1

Korn

et a

l. (2

006,

200

7)N

ordl

ande

r et a

l. (2

012)

Achievable S/N

NGC 6752 @ [Fe/H]=-1.6: Ca, Sc, Ti and Fe

TOP stars: 30 h with FLAMES-UVESS/N 35 per rebinned pixel

Gruyters et al. (2012)

Blaze correction reliability

More generally: why accept calibrations that are ”astronomical” rather than physical?

cf. Korn (2002)

amplitude ≥5%!

Physical calibration of spectrometer

See Stubbs & Tonry (2012): arXiv1206.6695:Addressing the Photometric Calibration Challenge: Explicit determination of the Instrumental Response and Atmospheric Response Functions, and Trying it All Together.

(Spectro)photometry

Teff to 50K => B-V to 0.01 or V-K to 0.03 mag.or corresponding accuracy in spectrophotometric gradients.

Problem: Variable stars! Simultaneous data needed.

Polarization

With a 1‰ accuray in polarizationmean fields of about 100 Gauss should be measurable => 0.03 dex in abundance(better with IR lines observed!)

See Fabbian et al. (2010)

Our CODEX: No SIMPLE compromises!

E-ELT instrumentation: second to none

So: Do new things!

• Look deeper!• Go for higher resolution and S/N!• Explore new wavelength regions!• Take control of calibrations!• Invest into extra dimensions like polarization …

Time resolution

Cieslinski et al. (2010): Polar (AM Her star) RBS 0324

15 min intervals.Much higher frequency requires special measures.

Hd HeII Hb

Achievable S/N

• S/N per unit timetime domain ever more importantability to study ever shorter phenomena limited by the light collecting power and the read-out time

Barclay et al. (2011)