The challenges of

abundance analysis for LPVsThomas Lebzelter / Walter Nowotny / Bernhard Aringer /

Kenneth Hinkle / Susanne Höfner / Ulrike Heiter /

and many more...

Acknowledging support by the Austrian Science Fund FWF

and the ESF

Why abundances from AGB stars?

Surface abundances are changed by third dredge up:

C/O

Isotopic ratios

s-process elements

Important for:

Stellar and Galactic Evolution

Stellar Nucleosynthesis

LPVs are...

Low temperature objects

With an extended atmosphere

And large amplitude variability

A little experiment from 2010

Analysis results for a Tau

Based on R=80 000, S/N=200 spectrum between 4900 and 9750 A

plus V-I, J-K, and V-K colours. Lebzelter et al. 2012, A&A 547, A108

Variability and atmospheric

structure

Animation: Walter Nowotny

Models vs. Observations

Sample of 5 non-variable red giants

and 7 variable AGB stars in 47 Tuc

L, M, [Fe/H] well known

Teff from NIR photometry

Grid of hydrostatic COMARCS spectra and

5 dynamical model atmospheres for C/O=0.48 and 0.25

Investigation of 9 spectral features

Non-variable vs. variable

Results (example)

Lebzelter et al. 2014, A&A in press

Conclusions ( )

Hydrodynamical models fit the observed high-resspectra of LPVs better than hydrostatic models

BUT: none of the nine configurations tested here gave a good fit to the entire observed high-res spectra

A reasonable fit with hydrostatic models is possible forsome of the mildly pulsating stars.

To fit the EWs observed in LPVs with hydrostatic models:Teff (EW)

Results (example)

Lebzelter et al. 2014, A&A in press

Variability and Line strengthsequ.

wid

th [

A]

0

0.2

0.4

0.6

0.8

1

V [mag]4 6 8 10 12 14

Dynamical model

L: 4000 L

C/O: 0.48

Teff: 3500 K period: 200 d

[Fe/H]: -0.7 mass loss: 2E-7 M/yr

Filled dots: Dynamical model

Triangles: R Cas (Mira)

Lebzelter et al. 2010, A&A 517, A6

Lebzelter et al. 2014, A&A in press

A little experiment from 2010

Pre-defined set of stellar parameters:

Teff = 3900 K, log g = 1.3, [Fe/H] = -0.2, x = 2 kms-1, M = 2 M⊙

A little experiment from 2010

High resolution spectra of four cool giant stars (plus some

photometric colours) were provided to various groups

working in the area of spectral synthesis asking them to derive

stellar parameters from the data provided (Teff, log g, [Fe/H]).

6 results using MARCS models

5 results using ATLAS models

1 result from PHOENIX model

1 result from CODEX model

1 result from Tsuji model atmosphere

Various methods, various spectral synthesis codes

Lebzelter et al. 2012, A&A 547, A108