Giuseppina Battaglia ESO

1. The Chemo-Dynamical Structure of Galaxies: intermediate resolution spectroscopy of resolved stellar populations out to Virgo Giuseppina Battaglia ESO Simulations as part of the Design Reference Mission for E-ELT

2. Motivation:insights on galaxy formation and evolution • Low mass stars can have lifetimes comparable to the age of the Universe -> record of changing galaxy properties • By deriving ages, chemistry and kinematics of LARGE samples of individual stars we can gain insights on how galaxy properties were built up in time These observations are needed for a variety of galaxy types (e.g. dwarfs, spirals, ellipticals) and in a variety of environments (e.g. groups, clusters)=> beyond the Local Group

3. CaII triplet at  8600 Å Sculptor dSph: 93 stars * Fornax dSph: 36 stars ESO/WFI Colour-Magnitude Diagram of the Fornax dSph (Battaglia et al. 2006) [Fe/H] from CaT EW [Fe/H]HR from about 60 Fe lines The NIR CaII triplet (CaT) • Well tested method for obtaining large numbers of metallicities and l.o.s. velocities of individual stars (currently used in studies of Local Group dwarf galaxies, M31, RAVE, GAIA in the future…) • Red Giant Branch stars as targets (usually): -> bright in the NIR -> cover wide age range (> 1 Gyr old) • The CaT is a strong feature in the NIR spectra of late-type stars -> need only moderate resolution (R  6000) to obtain accurate l.o.s. vel and EW • For individual RGB stars an empirical relation holds between [Fe/H] and CaT EW -> extensive literature for stellar clusters -> in composite stellar population tested in range -2.5 < [Fe/H] < -0.5 Battaglia et al. 2008, MNRAS, 383, 183

4. Example from VLT/FLAMES intermediate resolution (R=6500) studies of dSphs from DART team The Sculptor dSph (distance = 80 kpc) With 1h exposure time per pointing S/N=10/Å @ V=19.5. At this S/N: -> l.o.s. velocities accurate ± 5 km/s -> [Fe/H] accurate to ± 0.25 dex 20 pointings -> 670 individual members coverage of Sculptor with VLT/FLAMES GIRAFFE

5. Looking farther out Can we perform similar studies out to the distance of the Virgo cluster with the E-ELT? => Can we derive accurate [Fe/H] and line-of-sight velocities from the CaT lines for large numbers of individual RGB stars in a “reasonable” observing time? Main points to address: • Larger distances • Crowding: effect of stellar background in the spatial resolution element (spaxel) on the properties of target RGB star -> to which extent will the CaT [Fe/H] and velocity derived from the integrated spectrum resemble the ones of the target RGB star?

6. Setting up the simulations

7. 1Re 2Re 5Re Stellar population code developed by J.Liske & E.Tolstoy Simulations: creating the stellar catalogue The stellar catalogue contains all the stars that contribute a fraction of their light to the spaxel (target + stellar background) • Stellar population: constant SF (14-12 Gyr ago); [Fe/H] = -1.8 (MP) or -1 (MR); solar [alpha/Fe] • Target: RGB stars of different magnitude • Stellar background: - fix the distance and the surface brightness of the object to observe - explore different projected radii (different crowding) - at each radius, simulate a stellar catalogue over a large area (5” x 5”) - include PSF effects (we use LTAO) - find which stars contribute to the light in a 50mas spaxel, where the RGB star target is

8. Creating the simulated spectrum • For each star in the catalogue find the appropriate synthetic spectrum (log g, Teff, [M/H], [alpha/Fe]) in the Munari et al. (2005) library at R=20˙000 • Rescale each spectrum according to the light contributed by the star to the spaxel • Redshift individual spectra according to stellar velocities (we assume a Gaussian velocity distribution) • Produce the integrated spectrum (R=20000) in the spaxel • take into account effect of atmosphere, telescope, instrument, site, mirror coating • Convolve to desired resolution (R=6000) and add noise • Measure the line-of-sight velocity and CaT EWs from the integrated spectrum & compare to the ones of the target RGB star

9. We perform 300 random realizations to account for varying stellar background and noise • We derive the distribution of (V_los_simulated - V_los_input) • and (∑W_simulated - ∑W_input) where ∑W = EW2 + EW3 • -> the systematics are quantified by the mean of the distributions • -> the errors are quantified by the scatter of the distributions Figures of merit (accuracies) We consider that the simulations produce accurate enough velocities and ∑W if they agree with the input values within 0.7 Å -> 0.3 dex ( [Fe/H] = 0.44 x ∑W ) 30 km/s (CenA & M87); 5 km/s (NGC205) (this is similar to accuracies of current studies)

10. Explored parameters SCIENTIFIC: • STELLAR POPULATION: constant SFH (14-12 Gyr ago), [alpha/Fe]=0, [Fe/H] = -1.8 (“metal poor”) and [Fe/H]= -1.0 (“metal rich”) • DISTANCES: Local Group (800 kpc), Centaurus A (4 Mpc), Virgo (17 Mpc) • GALAXIES: NGC205 (dwarf, internal dispersion used = 35 km/s), CenA, M87 (internal dispersion used = 100 km/s). Surface brightness profiles from Mateo 1998, van den Bergh 1976, goldmine web database, respectively • RADII: 1, 2, 3.5, 5 Re (depending on the galaxy)

11. Explored parameters TECHNICAL: • TELESCOPE DIAMETER = 42 m • SPAXEL SIZE= 50mas • PSF: LTAO in I band (to minimize the loss of flux from the spaxel) • AIRMASS = 1.0 & SEEING = 0.8” • INSTRUMENT EFFICIENCY: 0.28 • WAVELENGTH RANGE: 8000-9000 Å • RESOLVING POWER: R  6000 (3 pixels per resolution element) • MIRRORS COATING: bare Al; Ag/Al • SITE: Paranal-like; High&Dry • EXPOSURE TIME: from 20min to 50h, depending on the object

12. Results

13. Tip of the RGB 1 mag below the tip * MP MR Velocity EW Results for the Local Group: NGC205 • Down to 1 mag below the tip (I  21.4) • Exp. times: 20m, 30m, 1h, 2h • Stars resolved with a 50 mas spaxel at every radius • 20min, Paranal-like site, bare Al coating • 20m exposure time x pointing => accuracy: ± 2 km/s accuracy: ± 0.2 dex Velocity

14. Centaurus A: example of spectra • Targets: MR and MP RGB at the tip and 0.5 mag below tip (MR: I= 23.9, 24.4; MP: I= 23.7, 24.2) • R/Re = 1, 2, 3.5, 5 -> R  6, 12, 22, 31 kpc (crowding is not a problem at these radii, for the considered spaxel size) • Exp. time: 1h, 2h, 5h MR, Paranal, Ag/Al, 5h, tip of the RGB Example of different crowding Tip of the RGB 0.5 mag below the tip Tip of the RGB 0.5 mag below the tip

15. Centaurus A: velocity (0.5 mag below the tip) 5h 2h 1h High&Dry+ Ag/Al Paranal-like + Ag/Al Paranal-like + Bare Al

16. Centaurus A: EW (0.5 mag below the tip) 5h 2h 1h High&Dry+ Ag/Al Paranal-like + Ag/Al Paranal-like + Bare Al

17. Paranal, Ag/Al, 50h, tip of the RGB MR VRGB= -79km/s MP VRGB= +312 km/s Results for M87 (Virgo): R= 2Re • Targets: MR and MP RGB at the tip ( I = 27.2, I = 27.0) • R/Re = 2 (a lot of crowding), 5 (not so much crowding) -> R  18, 45 kpc • Exp. time: 20h, 50h • Integration on: 1 spaxel, 9 spaxels, 25 spaxels • integrated spectrum on 25 spaxels • integrated spectrum on 1 spaxel • rescaled spectrum of the target RGB star Integrated spectrum on 25 spaxels dominated by stellar background

18. Paranal, Ag/Al, 50h, tip of the RGB MR VRGB= -79km/s MP VRGB= +312 km/s Results for M87 (Virgo): R= 5Re • Targets: MR and MP RGB at the tip ( I = 27.2, I = 27.0) • R/Re = 2 (a lot of crowding), 5 (not so much crowding) -> R  18, 45 kpc • Exp. time: 20h, 50h • Integration on: 1 spaxel, 9 spaxels, 25 spaxels • integrated spectrum on 25 spaxels • integrated spectrum on 1 spaxel • rescaled spectrum of the target RGB star Integrated spectrum on 25 spaxels considerable improvement

19. Results for M87 (Virgo): summary • Integrated spectrum on 1 spaxel has too low a s/n both at 2Re and 5Re, in all of the explored cases => too much flux is lost because of small spaxel size + PSF effect • At 2Re, integrated spectrum on 25 spaxels is dominated by stellar background => at this and smallest projected radii there is too much crowding • At 5Re: - metallicities are usually underestimated of about 1 dex - the Paranal-like + bare Al case yields very discrepant velocities - velocities can be measured within the desired accuracy for very bright targets (at the tip of the RGB, MP) with 20h exposure time when integrating on 25 spaxels (50h for 9 spaxels).

20. Observing time to collect samples of 1000 targets ASSUMPTION: single IFU with 10”x10” field-of-view (spaxel size= 50mas) NGC205 (10” = 40pc): Down to 1 mag below the tip, with 20m exp. time per pointing => 25 pointings (within 2Re), I.e. 8h exposure time on source Centaurus A (10” = 180pc): Down to 0.5 mag below the tip: -5h exp. time per pointing(Paranal-like, bare Al), 25 pointings within 5Re => 130h on source -2h exp. time per pointing(Paranal-like, Ag/Al), 25 pointings within 5Re => 50h on source • M87 (10” = 0.8 kpc): • At the tip of the RGB, in the outer parts (around 5 Re), about 6 pointings would be needed: • 20h exp. time per pointing -> 120h on source • 50h exp. time per pointing -> 300h on source ±

21. Sensitivity to input parameters SITE: negligible influence MIRRORS COATING: Appears to be important for -CenA => reduced exp. time per pointing from 5h to 2h -M87 => when considering 20h exp. time, velocities are not recovered with a bare Al coating FIELD-OF-VIEW: A 10”x10” field-of-view is sufficient in most cases. Given the goal of acquiring large number of targets over large areas, a larger field-of-view (or multi-IFUs) would allow shorter exposure times INSTRUMENT EFFICIENCY: M87 case appears to be “border-line” -> important to have values as realistic as possible