sign in sign up
<<
Home , Dwarf galaxy, Dwarf spheroidal galaxy, Galaxy, Sculptor (constellation)

Reports from Observers

The Dwarf Abundances and Radial-velocities Team (DART) Large Programme – A Close Look at Nearby

Eline Tolstoy1 The dwarf galaxies we have studied are nearby galaxies. The modes of operation, Vanessa Hill 2 the lowest- (and mass) galax- the sensitivity and the field of view are Mike Irwin 3 ies that have ever been found. It is likely an almost perfect match to requirements Amina Helmi1 that these low-mass dwarfs are the most for the study of Galactic dSph galaxies. Giuseppina Battaglia1 common type of galaxy in the , For example, it is now possible to meas- Bruno Letarte1 but because of their extremely low ure the abundance of numerous elements Kim Venn 4 our ability to detect in nearby galaxies for more than 100 Pascale Jablonka 5,6 them diminishes rapidly with increasing over a 25;-diameter field of view in one Matthew Shetrone 7 distance. The only place where we can shot. A vast improvement on previous la- Nobuo Arimoto 8 be reasonably sure to detect a large frac- borious (but valiant) efforts with single-slit Tom Abel 9 tion of these objects is in the Local spectrographs to observe a handful of Francesca Primas10 Group, and even here, ‘complete sam- stars per galaxy (e.g., Tolstoy et al. 2003; Andreas Kaufer10 ples’ are added to each year. Within Shetrone et al. 2003; Geisler et al. 2005). Thomas Szeifert10 250 kpc of our Galaxy there are nine low- Patrick Francois 2 mass galaxies (seven observable from The DART Large Programme has meas- Kozo Sadakane11 the southern hemisphere), including Sag- ured abundances and velocities for sev- ittarius which is in the process of merging eral hundred individual stars in a sample with our Galaxy. We will show that only of four nearby dSph galaxies: , 1 Kapteyn Institute, University of Gronin- the closest galaxies can be observed in , and . We have gen, the Netherlands detail, even with an 8-m telescope. used the VLT/FLAMES facility in the low- 2 Observatoire de Paris, France resolution mode (LR 8, R ~ 8 000) to 3 Institute of , University of The lowest-mass galaxies are dwarf irreg- obtain Ca ii triplet estimates as Cambridge, United Kingdom ular (dI) and dwarf spheroidal (dSph) type well as accurate radial-velocity measure- 4 Department of and Astronomy, galaxies. The only difference between ments over large areas in Sculptor, For- University of Victoria, Canada these low-mass dSphs and dIs seems to nax and Sextans out to the tidal radius 5 Observatoire de Genève, Laboratoire be the presence of gas and current (Tolstoy et al. 2004; Battaglia et al. 2006, d’Astrophysique, Ecole Polytechnique formation in dIs. It has already been not- in prep.), see Figure 1 for the example Fédérale de Lausanne, Switzerland ed that dSphs predominantly lie close of Sculptor. In Figure 1 we show the area 6 on leave from Observatoire de Paris, to our Galaxy (< 250 kpc away), and dIs on the sky around the Sculptor dSph France predominantly further away (> 400 kpc galaxy. The large ellipse is the tidal radius 7 University of Texas, McDonald Observ- away). This suggests that the proximity of of Sculptor as determined by Irwin and atory, Fort Davis, Texas, USA dSph to our Galaxy played a role in the Hatzidimitriou (1995). The positions ob- 8 National Astronomical Observatory, removal of gas from these . How- served with VLT/FLAMES are marked as Tokyo, Japan ever, the range of properties found in circles, where the solid circles repre- 9 KIPAC, Stanford University, Menlo dSph and dI galaxies does not allow a sent the 10 individual FLAMES point- Park, California, USA straightforward explanation, particular- ings analysed so far, and the dashed-line 10 ESO ly not the large variations in star-formation circles are the positions of five addition- 11 Astronomical Institute, Osaka Kyoiku histories and chemical-evolution paths al fields that were observed in November University, Japan that have now been observed in different 2005. Also plotted for the analysed fields systems (e.g., Dolphin et al. 2005). are the positions of the stars that are probable members (small red squares) We review the progress of ESO/WFI Im- It is perhaps not a surprise that there is and the non-members ( crosses). aging and VLT/FLAMES spectroscopy apparently a lower limit to the mass of an Each of the four galaxies has been ob- of large numbers of individual stars in object which is able to form more than served at high resolution (Giraffe settings nearby dwarf spheroidal galaxies by the one generation of stars, which is related HR 10, 13 and 14) in the central region to Abundances and Radial- to the limit below which one obtain detailed abundances for a range of velocities Team (DART). These observa­ will completely destroy a galaxy. Numeri- interesting elements such as Mg, Ca, O, tions have allowed us to show that cal and analytic models tell us that this Ti, Na, Eu to name a few (Hill et al. 2006 6 neither the kinematics nor the abundan­ must be around a few × 10 MA. Globular in prep.; Letarte et al. 2006a in prep.) ces nor the spatial distributions are easy clusters typically have much lower masses for about 100 stars. During the Giraffe HR to explain in a straightforward manner than this limit. observations we were also able to use for these smallest galaxies. The main the fibre feed to the UVES spectrograph result is that dwarf galaxies show com- to obtain greater wavelength coverage plex and highly specific evolutionary Observations and higher resolution for a sample of sev- and metal-enrichment processes, espe- en to 14 stars per galaxy (e.g., Venn et cially at ancient times. This conclusively VLT/FLAMES with fibre-feeds to the Gi- al. 2006 in prep.; Shetrone et al. 2006 in proves that these small galaxies are raffe and UVES spectrographs has made prep.). The comparison we can make not the building blocks of the larger gal- a revolution possible in spectroscopic between results from UVES spectroscopy axies in the . studies of resolved stellar populations in (R ~ 40 000) and Giraffe spectroscopy

The Messenger 123 – March 2006 33 Reports from Observers Tolstoy E. et al., The DART Large Programme – A Close Look at Nearby Galaxies

Sculptor dSph Figure 1: The area on the sky around (R ~ 20 000) for the same stars is very 1.5 useful in convincing ourselves, and oth- the Sculptor dSph galaxy. The posi- tions observed with VLT/FLAMES are ers, that we are able to obtain reliable marked as circles. Also plotted are results with lower resolution spectra than 1.0 the positions of the stars that are prob­ was previously thought possible or advis- able members (red squares) and the able. This lower-resolution is less of a non-members (black crosses). 0.5 problem at lower metallicity (e.g., Sculp- tor) than at higher metallicity (e.g., Fornax). ] ees 0.0 egr [d η Colour- diagrams –0.5 The first step in a detailed analysis of the resolved of a galaxy –1.0 is an accurate colour-magnitude diagram (e.g., Figure 2), ideally down to the old- –1.5 est main-sequence turnoffs (MV ~ + 3.5). 1.5 1.0 0.5 0.0 –0.5 –1.0 –1.5 In Figure 2 is plotted the colour-magni- ξ [degrees] tude diagram of the Sculptor dSph from ESO/WFI imaging out to the tidal radius. The coloured symbols are the stars we –3 Figure 2: Colour-magnitude diagram observed with VLT/FLAMES. The prob­ WFI/FLAMES Sculptor dSph of the Sculptor dSph from ESO/WFI possible members imaging. The coloured symbols are able members of Sculptor are shown as probable non-members the stars observed with VLT/FLAMES. purple circles and the non-members are The probable members of Sculptor shown as green stars. A careful analysis –2 are shown as purple circles and the leads to the star-formation history all the non-members are shown as green stars. Also shown outlined in blue are way back to the first in the regions used to define the Blue FG the Galaxy. This approach is the most ac- –1 (BHB), the Red Hor- curate for intermediate-age populations, izontal Branch (RHB) and foreground comparison (FG) stellar populations. but for stars older than about 10 Gyr V M the time resolution gets quite poor (and the stars are getting very faint), and it 0 BHB becomes hard to distinguish a 12-Gyr- RHB old star from a 10-Gyr-old star. Here it ­becomes useful to consider the Horizon- 1 tal Branch stars (MV ~ 0) which are the bright He-burning phase of low-mass stars > 10 Gyr old. The ratio of red to blue horizontal-branch stars (see Figure 2) 2 tells us about the age and metallicity vari- 0 0.5 1 1.5 V–I ation. In Figure 2 regions used to define the Blue Horizontal Branch (BHB), the by Cole et al. 2005; their Figure 8). The pockets of of differ- Red Horizontal Branch (RHB) and fore- observed magnitude and colour (e.g., ent ages (enriched by different num- ground comparison (FG) stellar popula- MV, V−I) of a star combined with a meas- bers of processes) conveniently covering tions in Sculptor are outlined in blue ured [Fe/H], allows us to effectively the nuclear burning core of stars. This boxes. The spatial distribution of red and remove the age-metallicity degeneracy hot provides a useful bright blue horizontal-branch stars in Sculp- and determine the age of an RGB star background source to be absorbed in the tor is shown in Figure 3. In this Figure the from an isochrone, and thus to trace stellar allowing very detailed outline of the extent of the ESO/WFI im­ the enrichment patterns of as many ele- studies of the elemental abundances in aging is shown, as is the tidal radius of ments as we can measure with time. these ancient gas samples. Thus, a spec- Sculptor. troscopic analysis of the variation of the abundances of different elements seen in We can also consider the much brighter Detailed abundance analysis absorption in of different

Red Giant Branch (RGB, −3 < MV < 0) age stars allows us to trace the detailed stars which have ages > 1 Gyr old, back In the majority of RGB stars it is be- chemical enrichment history of a galaxy to the oldest stars in the Galaxy, however, lieved that the atmosphere of the star with time. Different elements are created the interpretation of the RGB using only remains an unpolluted sample of the in different circumstances and if we are is plagued by the age-metal- interstellar medium out of which it was able to determine the abundance of ele- licity degeneracy (graphically illustrated formed. This means we have small ments known to be created in a particular

34 The Messenger 123 – March 2006 Figure 3: The distribution of horizontal- branch stars from ESO/WFI imaging of the Sculptor dSph selected as shown in Figure 2.

BHB RHB Foreground at RHB 2 2 2

1 1 1 ] ] ] ees ees ees 0 0 0 egr egr egr [d [d [d η η η –1 –1 –1

–2 –2 –2 2 1 0 –1 –2 2 1 0 –1 –2 2 1 0 –1 –2 ξ [degrees] ξ [degrees] ξ [degrees] set of physical conditions, we can assess be an element produced almost exclu- time as the Giraffe spectra (Venn et al. the importance and frequency of these sively by the r-process. The slow capture 2006, in prep.). A representative error conditions during the history of star for- (s-process) is thought to be created by bar is also given in the bottom left hand mation in a galaxy. more quiescent processes such as the of each panel. The Fornax VLT/FLAMES stellar winds common in AGB type stars results are preliminary, as the problems For example, the abundances of light of intermediate age (and mass). Typical with the appropriate stellar models for elements (e.g., O, Na, Mg, Al) are consid- elements which are thought to be created such cool stars as are found in Fornax ered to be tracers of ‘deep mixing’ by the s-process are Ba and La. The have not yet been fully resolved. We can patterns which are found only in globular- ratio between r- and s-process element compare the detailed abundance pat- cluster environments, which gives a limit abundances gives an indication of the terns that we see in dSph galaxies, with to the number of dissolved globular relative importance of these different en­ other galaxies such as the . clusters which can exist in a stellar popu­ richment processes during the history This is shown in Figure 4, where the Ga- lation. These typical (Galactic) globular of star formation in a galaxy. Dwarf sphe- lactic stars shown in both panels as cluster abundance patterns have also roidal galaxies are found to have a strong black dots come from various literature been recently found in the globular clus- evolution from r-process domination to sources (see Venn et al. 2004 for refer- ters of Fornax dSph (Letarte et al. 2006b s-process domination as a function of the ences), and the disc and halo component in prep.), but not (so far) in the field star metallicity of the stars. This may indicate are also labelled. This can also be done populations (e.g., Shetrone et al. 2003). that supernovae products are typically for Saggitarius and the lost to a shallow potential well, or that the (e.g., Venn et al. 2004). These compari­ The creation of a-elements (e.g., O, Mg, slow star formation rate means that sons show that enrichment patterns dif- Si, Ca, Ti) occurs predominately in su- massive stars are not very common (e.g., fer strongly between different types of pernovae type II explosions, i.e. the ex- Tolstoy et al. 2003; Venn et al. 2004). galaxy, making it hard to build one type plosion of massive stars a few 106−107 yrs of galaxy out of another once they have after their formation. The abundance of Our latest results for the a-elements are formed a significant number of stars. the different a-elements is quite sensitive shown in Figure 4. The a-element abun- to the mass of the SNII progenitor so the dance is an average of Ca, Mg and Ti a/Fe ratio traces the mass function of the abundances from high-resolution spec- The Ca ii triplet and kinematics stars which contributed to the creation troscopy for stars in Sculptor and For- of the a-elements, and ratios of different nax dSphs. The VLT/FLAMES measure­ The ideal is to be able to obtain high-re- a-elements themselves can put limits on ments of 92 stars in the centre of Sculp- solution spectra for individual stars the highest-mass star which has enriched tor are shown in the upper panel as in nearby dSph over a large wavelength a galaxy and also the typical mass range purple crosses (Hill et al. 2006 in prep.) range, and make a detailed analysis of (e.g., McWilliam 1997). and the preliminary measurements for a range of different elements along with 55 stars observed in Fornax are shown accurate velocities. However, this is quite Heavy Elements (Z > 30) are a mix of as green crosses in the lower panel time consuming both in telescope time r- and s-process elements. That is to say ­(Letarte et al. 2006a, in prep.). The open (even with VLT/FLAMES) and in analysis. elements which were produced by rap- blue squares, five in Scl and three in Fnx One of the most simple ways to get an id or slow neutron capture which tells us are UVES measurements of individual estimate of the metallicity of RGB stars is about the environment in which enrich- stars (Shetrone et al. 2003). The four with the Ca ii triplet. This is a basic met- ment occurred. Rapid capture (r-proc- open red squares are Geisler et al. (2005) allicity indicator requiring only low or in- ess) is assumed to occur in high-energy measurements of Scl, and the 10 open termediate spectral resolution, based on circumstances, such as supernovae ex- green squares are FLAMES fibre fed three lines around 8 500 Å which have plosions. For example Eu is considered to UVES spectra of Scl, taken at the same been empirically calibrated from obser-

The Messenger 123 – March 2006 35 Reports from Observers Tolstoy E. et al., The DART Large Programme – A Close Look at Nearby Galaxies

0.6 within the central, r < 0.5 degree region Halo (solid black line); and the 229 stars 0.4 Disc beyond r > 0.5 (dashed red line). Clearly e] 0.2 the distributions are very different. Most /F α

[ 0 noticeably Fornax has a substantial –0.2 ‘metal-rich’ population. The lowest metal- Scl licity star in our combined sample of –0.4 more than 850 stars for both galaxies is –4 –3 –2 –1 0 [Fe/H] = −2.7. We find a similar lack of [FeI/H] low-metallicity stars in Sextans and Ca- rina. Although it is difficult to make an ac- 0.6 curate comparison with the ­surveys, where the completeness can be 0.4 Disc hard to quantify, there appears to be a e] 0.2 significantly different distribution between /F α

[ 0 all the dSph and the (metal-poor) halo of –0.2 the Milky Way. It can be seen that there Fnx is a clear difference in the distribution –0.4 of the metal-rich and metal-poor stars in –4 –3 –2 –1 0 both Fornax and Sculptor. [FeI/H]

Figure 4: The a-element abundances from high- (Shetrone et al. 2003; Geisler et al. 2005; Venn et Future work ­resolution spectroscopy for stars in the Sculp- al. 2006, in prep.). Galactic stars coming from various tor dSph (upper panel, Hill et al. 2006 in prep) and ­literature sources (see Venn et al. 2004 for refer- also preliminary results for the Fornax dSph (lower ences) are shown for comparison in both panels as There are indications that the presence panel, Letarte et al. 2006a, in prep.). The open black dots and labelled as disc and halo compo- of two distinct populations is a common squares are UVES measurements of individual stars nents. feature of dSph galaxies. Our prelim- inary analysis of Horizontal Branch stars, vations of stars in globular clusters with satisfying S/N > 10. Likely Sculptor mem- vhel and [Fe/H] measurements in the high-resolution abundances. Assum- ­bers are clearly seen clustered around other galaxies in our sample (Fornax and ing sufficient signal-to-noise spectra the systemic velocity of 110 km/s. The Sextans dSph; Battaglia et al. 2006, in (S/N > 10) it provides [Fe/H] within typical stars which are potential members are prep.) also show similar characteristics (internal) errors of ± 0.1 dex, and also the plotted as red stars ([Fe/H] > −1.7) and to Sculptor, especially in the most metal- of each star with ± 2 km/s blue circles ([Fe/H] < −1.7), while the poor component. Pure radial-velocity accuracy. These accuracies are well suit- green crosses are assumed to be non- studies (e.g., Wilkinson et al. 2004) have ed for ‘quick look’ surveys of the resolved members. There appears to be a metal- also considered the possibility that kin- stellar population of a galaxy. In the DART rich, −0.9 > [Fe/H] > −1.7, and a metal- ematically distinct components exist project these Ca ii triplet measurements poor, −1.7 > [Fe/H] > −2.8, component. in , and Sextans dSph are complementary to the high-resolution The metal-rich component is more cen­ galaxies. Interestingly, the Carina dSph observations made in the centre of each trally concentrated than the metal-poor, appears to go counter to this trend, and dSph. In the low-resolution observa- and on average appears to have a lower another VLT/FLAMES study finds no ob- tions a much larger area is surveyed and , smetal−rich = 7 ± 1 km/s, ­vious evidence for more than one com- we can assess how representative the whereas smetal−poor = 11 ± 1 km/s. A sim- ponent, or even a gradient within Carina detailed study is of the stellar population ilar effect is seen in Fornax, where the dSph (Koch et al. 2006). of the whole galaxy. metal-rich stars are centrally concentrat- ed, and the metal-poor stars appear What mechanism could create two or Our first VLT/FLAMES results (Tolstoy et more uniformly and diffusely distributed. more distinct ancient stellar components al. 2004), were based upon Ca ii triplet in a small ? A measurements, which clearly showed It is clear from the histogram of [Fe/H] simple possibility is that the formation of that Sculptor dSph contains two distinct measurements that both Sculptor and these dSph galaxies began with an initial stellar components with different spatial, Fornax lack a low metallicity tail (see Fig- burst of star formation, resulting in a stel- kinematic and abundance properties (see ure 6). In Figure 6 are plotted in the left- lar population with a mean [Fe/H] ≤ −2. Figure 5). The upper panel shows the hand histogram distributions of [Fe/H] Subsequent supernova explosions from VLT/FLAMES spectroscopic measure- for Sculptor dSph: the 91 stars within the this initial episode could have been suf- ments of [Fe/H] for 307 probable velocity central, r < 0.2 degree region (solid black ficient to cause gas (and metal) loss such members of Sculptor (with S/N > 10). line); and the 216 stars beyond r > 0.2 de- that star formation was inhibited until We see a clear trend of metallicity with grees (dashed red line). In Figure 6 in the the remaining gas could sink deeper into radius. The lower panel shows vhel as right-hand histogram is the [Fe/H] distribu­ the centre. Thus the subsequent gen­ a function of elliptical radius for all stars tion for the Fornax dSph: the 332 stars eration(s) of stars would form in a region

36 The Messenger 123 – March 2006 closer to the centre of the galaxy, and have a higher average metallicity and dif- VLT/FLAMES Sculptor dSph ferent kinematics. Another possible cause –1.0 are external influences, such as minor mergers, of additional gas or –1.5 the kinematic stirring by our Galaxy. It might also be that events surrounding the epoch of reionisation influenced the evo- [Fe/H] –2.0 lution of these small galaxies and re- sulted in the stripping or photo-evapora- tion of the outer layers of gas in the dSph, –2.5 meaning that subsequent more metal- extent of RHB enhanced star formation occurred only in –3.0 the central regions. 0.0 0.5 1.0 1.5 2.0 Elliptical Radius [degrees]

Acknowledgements 200 Eline Tolstoy gratefully acknowledges support from VLT/FLAMES Sculptor dSph a fellowship of the Royal Netherlands Academy of Arts and Sciences. The data presented here were collected under ESO Large Programme 171.B-0588 100 and GTO programme 71.B-0641. l he v References 0 Cole A. A. et al. 2005, AJ 129, 1465 Dolphin A. E. et al. 2005, Resolved Stellar Popula- tions, eds. D. Valls-Gabaud and M. Chavez (astro-ph/0506430) Geisler, D. et al. 2005, AJ 129, 1428 –100 Irwin M. and Hatzidimitriou D. 1995, MNRAS 277, 0.0 0.5 1.0 1.5 2.0 1354 Koch A. et al. 2006, AJ, in press (astro-ph/0511087) Elliptical Radius [degrees] McWilliam A. 1997, &A 35, 503 Shetrone M. D. et al. 2003, AJ 125, 684 Figure 5: The upper panel shows the VLT/FLAMES radius. The stars that are probable members are Tolstoy E. et al. 2003, AJ 125, 707 spectroscopic measurements of [Fe/H] for 307 Red plotted as red stars ([Fe/H] > –1.7) and blue circles Tolstoy E. et al. 2004, ApJL 617, 119 Giant branch stars in Sculptor plotted as a func- ([Fe/H] < –1.7), showing metal-rich and metal- Venn K. et al. 2004, AJ 128, 1177 tion of distance from the centre of the galaxy. The poor stars respectively. The 128 green crosses are

Wilkinson M. et al. 2004, ApJL 611, 21 lower panel shows vhel as a function of elliptical unlikely to be members of Sculptor.

Figure 6: In the left panel is the histogram distribu- tion of [Fe/H] measurements for the Sculptor dSph: inner region, solid black line outer region dashed red line. In the right panel is the same for the Fornax dSph.

Sculptor Fornax 50 60 r < 0.2 r < 0.5 50 40 r > 0.2 r > 0.5

40 30 . . 30 No No 20 20

10 10

0 0 –3.0 –2.5 –2.0 –1.5 –1.0 –0.5 0.0 –3.0 –2.5 –2.0 –1.5 –1.0 –0.5 0.0 [Fe/H] [Fe/H]

The Messenger 123 – March 2006 37