SSTELLAR SPECTROSCOPY OF INDIVIDUAL ININ WITH THE VLT: KINEMAINEMATICS AND CALCIUM TRIPLET ABUNDANCES

E. TOLSTOY1, M. IRWIN2, A. COLE1, F. FRATERNALI3,4, T. SZEIFERT5, G. MARCONI5

1KAPTEYN ASTRONOMICAL INSTITUTE, UNIVERSITY OF GRONINGEN, THE NETHERLANDS 2INSTITUTE OF , UNIVERSITY OF CAMBRIDGE, UNITED KINGDOM 3ASTRON, DWINGELOO, THE NETHERLANDS 4DEPT OF THEORETICAL PHYSICS, UNIVERSITY OF OXFORD, UNITED KINGDOM 5EUROPEAN SOUTHERN OBSERVATORY

THE LARGE COLLECTING AREA AND HIGH-THROUGHPUT MULTI-OBJECT INSTRUMENTS ON THE VLT MAKE IT POS- SIBLE TO CARRY OUT DETAILED STUDIES OF STELLAR PROPERTIES AND DISTRIBUTIONS IN ENVIRONMENTS WELL BEYOND OUR .

HE HIGH RESOLUTION SPECTRO- However, FLAMES is not always the three absorption lines near 8500 Å. The graph, UVES, has provided best instrument for every stellar project. equivalent widths of these lines (ie. the line outstanding data for the study Due to light loss from fibre transmission, flux with respect to the continuum luminosi- of stellar abundances in extra- and restrictions on choice of resolution and ty of the ) have been shown to correlate galactic environments. It is on placing fibres close to each other, very well with high resolution [Fe/H] abun- TpossibleT to observe branch (RGB) FLAMES is well suited for abundance stud- dance (see Armandroff & Da Costa 1991; stars at high resolution in galaxies as far ies of stars in uncrowded regions of our Rutledge 1997). We have also made a com- away as I (250 kpc) with UVES nearest neighbour galaxies (within parison of our own, not matched on individ- (Shetrone et al. 2002; Tolstoy et al. 2002), ~220 kpc). If we want to look at more ual stars, but using preliminary results from and to look at super-giants up to 1.5 Mpc crowded regions (e.g., the bar of the LMC), the HR FLAMES data we are collecting for away at the boundary of the Local Group or if we want to use the VLT to push beyond 100+ stars in the galaxy (Hill et al., (e.g., Venn et al. 2003; Kaufer et al. 2004). the galaxies in and around the halo of our in prep.). We compare the distribution of These studies have been, of necessity, Galaxy to look at a completely separate and [Fe/H] abundances found with FLAMES painstakingly long integrations of individual independent environments, we have to use with those measured on the basis of the CaT stars down to the faint limit of the instrument the slit spectrographs, FORS1 and FORS2. from a previous FORS1 study of Sculptor (I ~ 19.5). These instruments also have enhanced mul- (Tolstoy et al. 2001). As can be seen in The multi-fibre instrument, FLAMES, is tiplexing capabilities such as slit masks, and Figure 1, there is good agreement between providing a dramatic increase in multi-plex- with a blocking filter, FORS2 can match or the two samples, although we note that the ing with a 25 arcmin diameter field of view, even exceed the capabilities of FLAMES. FLAMES results are preliminary. and 130 fibres in Medusa mode (mag. limit Of course this means fewer lines from a sin- The FORS2 spectrograph, combined I ~ 18.5), over selectable wavelength ranges gle element at lower resolution, but the con- with the red-optimised MIT/LL CCDs, is a (per setting), at resolutions useful for stellar verse is that we can look at significantly uniquely powerful instrument for measuring spectroscopy (R ~ 20000), and 8 fibres feed- fainter stars, probing a more distant and var- the of RGB stars as faint as I = ing UVES over a longer wavelength range, ied set of environ- with a typical UVES resolution (R ~ 40000). ments. This field of view and density of fibres is a A commonly good match to the size of the central regions used abundance of nearby dwarf spheroidal galaxies, and the indicator at inter- number of RGB stars at magnitudes bright mediate resolution enough to obtain good quality high resolu- is the Ca II triplet tion spectra. (CaT), a set of

Figure 1: The solid line in the (normalised) histogram of [Fe/H] is for a sample of RGB stars observed in the Sculptor dwarf spheroidal in the CaT region (by Tolstoy et al. 2001) with FORS1. The dotted his- togram comes from the preliminary results of the FLAMES LP and GTO time which are based on high resolution spectra for ~100 RGB stars in Sculptor. The samples are not overlapping, but both are ran- domly chosen on the RGB. The FLAMES samples are necessarily nearer the tip of the RGB. The FORS1 sample was chosen lower down the RGB, at somewhat fainter magnitudes. 32 The Messenger 115 Figure 2: A FORS1 R-band image of taken in August 1999. The green outline shows the region covered by the MXU slit mask, and thus the region where spectra were obtained. The CCD image on the right is the raw spectra on one of the two FORS2 CCDs (the “master” CCD) for our spectra of stars in Tucana.

22. This brings every Local Group galaxy With this reasonably large sample of green crosses are at systemic velocity. There outside the Galactic into stars across the Tucana galaxy we can see if is intriguing evidence for one side of the range for kinematic and studies there are any obvious kinematic patterns. galaxy appearing to recede and the other to of intermediate-age and old stars. FORS1/2 Figure 4 shows the velocities of the stars be approaching. This could be interpreted as can thus be used to efficiently look at sam- observed in Tucana with their positions in rotation, or possibly some indication of a ples of stars in galaxies which are too distant the galaxy. The blue crosses have velocities tidal disturbance, although Tucana’s isolated to allow high resolution abundances or even less than systemic, and red crosses have position in the Local Group mitigates against fibre spectroscopy (e.g., Tucana, Fraternali greater than systemic velocities. The two the latter possibility. Although the number et al. 2004, in prep.). The field of view of the FORS1/2 spectrograph and the number of slits are a good match to the density of RGB stars in low surface brightness dwarf galaxies out to the edge of the Local Group. In Figure 2 we can see the raw data obtained for the spheroidal galaxy, and the field of view covered by FORS2/MXU on the galaxy. Out of 47 tar- gets placed on the slits 25 could be success- fully extracted. The rest were too faint for a Figure 3: Histogram successful extraction. Of these successfully of the radial veloci- extracted stars 18 were considered of good ties measured from enough quality to determine velocities, and the CaT region for 18 stars in the Tucana of these 3 were quite distinct from the aver- dwaf spheroidal age velocity of the majority of stars. These galaxy. The gaussian cluster around the systemic velocity of fit gives a central, Tucana, determined now for the first time to systemic, optical be 182 km/s. In Figure 3 we show the his- velocity of Tucana of 182 km/s, with a dis- togram of these velocities. persion on 12 km/s. © ESO - March 2004 33 Figure 4: A 20n by (Irwin & Tolstoy 2002). In the case 14n field of view cen- of the Tucana dwarf spheroidal, it has always tred on Tucana, been held to be an unusual object because of taken from DSS its extreme distance from the Galaxy and plates. The blue crosses represent M31, unlike all other Local Group dwarf those stars with spheroidals (at least prior to the discovery of velocities less than ). v (182 km/s) and sys In the case of the Phoenix and the red crosses greater than sys- dwarf galaxies it is possible for the first time temic. The green to make clear the association of the optical crosses are those galaxy with the HI gas seen near these stars at systemic objects. Without a reliable optical velocity velocity. North is up this association remained uncertain. These and East to the left. kinematic studies therefore have far reaching implications for our understanding of evolu- tionary processes in nearby dwarf galaxies. There are currently two large pro- grammes at ESO to use FLAMES to increase the number of stars with velocity and abundance measurements to improve statistics are still very small, interestingly, if functions of galaxies are not well con- detailed modelling in several nearby galax- this is rotation it would be, as expected, rota- strained, sample size is a crucial aspect of ies (e.g., 171.B-0588, PI Tolstoy), for which tion about the minor axis. any chemical evolution study. Sparse sam- the first results will be available soon. This In Figure 5 we show preliminary results ples, measuring only a few stars, can all too ability to look at individual stars in different from the CaT metallicities of these stars in easily miss minority populations which may galaxies with such accuracy over such a Tucana. The spread in abundances looks be critical for understanding the chemical large range in distance means we can probe quite similar to Sculptor (from Tolstoy et al. (and dynamical) evolution of the host the properties of stars in many different 2001). From this plot, and taking into galaxy. environments - metal poor, metal rich, account the velocity and position of Tucana There are several galaxies in the Local dense, sparse etc. This is an exciting era for in the Local Group it is clear that the proper- Group for which we do not have even the understanding the detailed properties of ties of Tucana are consistent with those of most basic information, such as the optical resolved stellar populations across the Local nearer Galactic satellite dwarf spheroidal radial velocity. This is especially important Group, which naturally will have implica- galaxies, such as Sculptor. for dwarf spheroidal type galaxies which tions for studies of the most distant galaxies. The efficient use of FORS1/2 is not have no (observable) gas. Once the stellar restricted to the faintest stars; the same tech- velocity of the system is known it is possible niques can be successfully applied to much to look more carefully for associated gas and REFERENCES Armandroff & Da Costa 1991, AJ, 101, 1329 nearer galaxies. Here, the efficiency with to analyse possible orbital trajectories to its Irwin & Tolstoy 2002, MNRAS, 336, 643 which it is possible to change the slit config- current location. Kaufer et al. 2004, AJ, in press uration, the ease of target acquisition based We have used the FORS spectrographs Rutledge et al. 1997, PASP, 109, 907 on pre-images and slit configurations created to determine the optical velocities, previous- Shetrone et al. 2003, AJ, 125, 684 Tolstoy & Irwin 2000, MNRAS, 318, 1241 with the FIMS software, together with see- ly unknown, of Antlia (Tolstoy & Irwin Tolstoy et al. 2001, MNRAS, 327, 918 ing of 0.7 arc-seconds have allowed major 2000), and Tucana (Fraternali et al. 2004) Tolstoy et al. 2003, AJ, 125, 707 new studies of crowded regions like the cen- and a reassessment of the optical velocity of Venn et al. 2003, AJ, 126, 1326 ter of the LMC bar (e.g., Cole et al. 2004, in prep.). With a surface brightness of 20.6 mag/arcsec2, RGB stars in the bar were beyond the capabilities of spectrographs on 4-meter class telescopes. Crowding prob- Figure 5: The black lems can be minimized by target selection crosses are the measure- from pre-images taken in good seeing, and ments of the CaT widths of the two strongest lines the difficulties of sky fiber placement can be in our FORS2/MXU completely avoided by using slits that span spectra of stars in many arc-seconds around each target. These Tucana plotted against capabilities ensure a valuable place for the difference in their FORS even after the advent of large fibre magnitude with that of the horizontal branch in systems such as FLAMES. Tucana (VH). The dashed In nearby galaxies, where targets are lines are the relations bright, the speed with which high signal-to- calibrated by Tolstoy et noise spectra can be acquired becomes a al. 2001 corresponding to where stars of differ- tremendous advantage of FORS1/2. ent metallicity should lie. Comparing CaT spectra of RGB stars in the The red crosses are the LMC obtained in a large survey of field CaT measurements stars, UT4/FORS2 yields higher signal-to- taken with the same noise in 1200 second exposures (Cole et al. instrumental set up and processed in the same 2004) than CTIO- in exposures 10 way for stars of known times as long (Smecker-Hane et al. 2004, in metallicty in Galactic prep). Because the metallicity distribution , M15.

34 The Messenger 115