1. The circumstellar environment of evolved stars as seen by VLTI / MIDI Keiichi Ohnaka Max-Planck-Institut für Radioastronomie, Infrared Interferometry Group kohnaka@mpifr-bonn.mpg.de

2. Asymptotic Giant Branch (AGB) Post-AGB Red Rectangle Teff ~ 3000K L ~ 10 4 L  Mass loss ~10 -8 —10 -5 M  /yr Mass loss mechanism The structure of the outer atmosphere: Molecule & dust formation Morphology from AGB to Planetary Nebulae Carbon star, IRC+10216 AGB, AFGL2290 PN, Cat’s Eye Nebula

3. Outer atmosphere Molecular layers Near-infrared Expanding dust shell What can interferometry do to study AGB stars? Mid-infrared Dust formation Mira variables: Large variability amplitude ~ 9 mag (in V) MIDI AMBER

4. Infrared long-baseline interferometry Ｂ Spatial resolution = /B p N band B p = 200m  10mas What’s observed: Visibility, not an image! MIDI AMBER Visibility = Amplitude of the (complex) Fourier transform of the object’s intensity = Fringe contrast Is it useful? Yes, especially with spectral resolution! It “contains” information on the angular size and shape Point source  V = 1 Extended source  V < 1 Larger size  lower V ＢpＢp

5. MIDI: first fringe in December 2002 Open to the community since April 2004 VLTI Interferometric Laboratory N band: 8 – 13  m (dust features, molecular bands) Spectral resolution: 30 / 230 Sensitivity @  m: 1 Jy / 10 Jy

6. UT1 UT3 102m MIDI observation of the Mira variable RR Sco 2003 June, Science Demonstration Time Unit Telescopes 1 & 3 (8m) Projected baseline = 74—100m Angular resolution @ 10  m = ~20mas RR Sco (Phase = 0.6 in 2003 June) P = 284 days, d = 320 pc (Hipparcos) Dust emission not strong  Good for studying the molecular layers

7. 7.5  m 13.3  m MIDI observation of RR Sco: spectrally dispersed fringes

8. Observed N-band visibility of RR Sco Visibility increases from 8 to 10  m, constant > 10  m  UD diameter constant between 8 and 10  m (~18 mas), UD diameter increases  > 10  m (~25 mas @ 13  m) N-band UD diameter (MIDI) twice as large as that in the K band (VINCI, 3 weeks later) Why?

9. Observed N-band visibility of RR Sco

10. Expanding dust shell Optically thick emission from H 2 O (pure rotation) + SiO (fundamental) gas + Dust emission N band (8—13  m) H 2 O + SiO gas K band (2—2.4  m) No dust emission H 2 O + CO bands Not optically thick  Angular size smaller Modeling H 2 O + SiO layer (constant temperature, column densities, radius) Optically thin dust shell (silicate+corundum) (Inner radius, optical depth) Basic idea  Ohnaka et al. 2005, A&A, 429, 1067  Angular size larger

11. H 2 O+SiO emission Dust emission (silicate 20% + corundum 80%) T = 1400 K N(H 2 O) = 3 x 10 21 cm -2 N(SiO) = 1 x 10 20 cm -2 R = 2.3 Rstar Tin ~ 700 K, Rin = 7--8 Rstar  = 0.2 – 0.3 (V band), 0.025 (10  m)  Comparison with pulsation models is ongoing

12. MIDI observation of the silicate carbon star Hen 38 Silicate carbon star : carbon-rich photosphere, oxygen-rich circumstellar dust Usually… carbon star, carbon-rich circumstellar dust (amorphous carbon, SiC) M giants (O-rich), oxygen-rich circumstellar dust (silicate, Al 2 O 3 : corundum) How can silicate (O-bearing dust) exist around a carbon star?

13. Oxygen-rich dust (silicate) reservoir Mass loss AGB, primary star: oxygen-rich, mass loss  Circumbinary disk is formed Primary star becomes a carbon star. Oxygen-rich dust is stored in the disk  Silicate carbon star High-resolution observation in the silicate emission feature is the most direct approach  VLTI/MIDI AGB star + main sequence star

14. Compact silicate disk + extended corundum disk Silicate Corundum (Al 2 O 3 ) Compact disk Only present in the outer region  Increase of the angular size  m

15. Silicate torus (ring) dominant Corundum dominant Silicate + corundum disk model 15 – 35 Rstar,   m ) = 1.5 > 35 Rstar,   m ) = 0.4

16. Concluding remarks First “spectro-interferometric” observation of RR Sco  Wavelength dependence of the angular size Angular size constant between 8 and 10  m, increases longward of 10  m, More than twice as large as in the K-band Observed N-band visibilities and spectra can be explained by optically thick emission from H 2 O + SiO gas & dust emission  Consistent with ISO and previous results Potential to probe the circumstellar environment (molecule and dust) with spatial and spectral information disentangled  Totally new picture of the circumstellar environment Upcoming: more Miras (oxygen-rich, carbon-rich, S-type), Circumstellar dust disks around (post-)AGB stars, Symbiotic stars (Mira + hot companion), etc…