Abstracttest the abstract



The BRAVA-RR survey: mapping the bulge with RR Lyrae stars.




LARGE te \AbstractRR Lyrae stars are representatives of the oldest, most metal poor stellar population in the inner Galaxy. A pilot program has revealed that their kinematics are starkly distinct from those of the more metal-rich bulge giants and trace a slow (or non-) rotating, high velocity dispersion component that may be identified as the old classical spheroid of the Milky Way, a metal-rich inner halo, or a different stellar population. Futher, the RR Lyrae population has been confirmed as being unexpectedly metal rich with a wide range in metal abundance, and showing strong hints of complex kinematic substructure. We propose to define as completely as possible the structural properties of this component and determine accurate Ca infrared triplet metallicities via a spectroscopic survey with FLAMES. We will probe \(\sim\)2100 RR Lyrae stars spanning \(-6<b<-2\) and \(-6<l<+6\). The project is ideally suited for mediocre weather conditions, as it requires only moderate seeing and can be carried out in grey time. However, we must visit each RR Lyrae star at least 3 times to probe different phases and give a precise fit to the star’s systemic radial velocity.


A99FLAMES216hjung1.4THNs \ObservingRunB101FLAMES2hjung1.4THNs




Andrea Kunder


G.Bono1345 \CoIR.De Propris9155 \CoIC.I.Johnson1355 \CoIA.Koch12254 \CoID.Nataf1208 \CoIR.M.Rich1938 \CoIJ.Storm1135 \CoIA.Walker1011 \CoIJ.Wojno1135


The bulge of our Galaxy is the only spheroidal system that can be resolved into stars below the main sequence turnoff. It is therefore an essential data point to understand the formation of spiral galaxies. It is now known that our bulge is a composite system, containing a long bar (Blitz & Spergel 1991; Dwek et al. 1995). Its appearance on the sky resembles the class of bulges known as boxy/peanut bulges. The stellar populations of the bulge appear to be largely old and metal rich (Ness & Freeman 2016; Clarkson et al. 2011; see also review by Rich 2013), with evidence of \(\alpha\) enhancement and metal abundance gradients (Hill et al. 2008, Babusiaux et al. 2010). In many respects the bulge resembles a low mass version of an elliptical galaxy,

So, it was with some surprise that the first extensive radial velocity surveys of the bulge stars by our group (Rich et al. 2007; Shen et al. 2010; Kunder et al. 2012), and subsequently by the ARGOS (Ness et al. 2013) and GIBS (Zoccali et al. 2014) teams, showed that the bulge appears to follow pure cylindrical rotation and that there is very little, if any, scope for a traditional pressure-supported component (Fig. 1). Our bulge appears to have formed from the secular evolution of a massive early disk, and the lack of a classical bulge implies that no significant mergers may have taken place since the epoch of disk formation, \(z\sim 3\) (e.g., Di Matteo 2015). Simulations show that this is an unlikely event for a halo resembling the Milky Way (Agertz et al. 2011) , but observations imply that the majority of disk galaxies within the 11 Mpc sphere are either bulgeless or contain a boxy/peanut bulge (Fisher & Drory 2011). Most of these seem to be in cylindrical rotation, i.e. they are bars as well. This may represent a significant challenge to the current paradigm for bulge and spiral galaxy formation. Can we find a classical component and how important is it?

The only bulge stellar population for which age and distance can be precisely determined are RR Lyrae variables (RRLs). These stars are necessarily old (\(\sim 11\) Gyr) and they are accurate standard candles. In general, RRLs have metallicities [Fe/H] \(<-1\), consistent with their old ages. If there is a classical bulge, it may be best identified with these objects. Indeed, simulation by Tumlinson (2010) predict that the oldest stellar population in the galaxy may reside in the bulge, even if not part of the bulge itself. Because of their clean age stamp, these stars plausibly relics of the earliest epochs of star formation within the Milky Way galaxy.

Two recent studies based on photometric identification of bulge RRLs have reached contradictory conclusions. Pietrukowitz et al. (2015) from the OGLE survey data, claim that the spatial distribution of RRLs is consistent with their being part of the bar. On the other hand, Dekany et al. (2013) from the VVV survey data claims that RRLs follow a spheroidal distribution which is not associated with any bar or subcomponent and therefore trace the old classical bulge. It is also possible that the RRLs are actually part of the inner halo instead of the bulge (Pérez-Villegas et al. 2016). It is clear a clear measurement of the internal kinematics remains now as the best path forward for discriminating between these geometries.

We have started a survey of RRLs kinematics using data from the AA\(\Omega\) instrument on the AAT. We targeted fundamental mode RRLs from the OGLE survey in four 3 deg\({}^{2}\) fields. Each field typically has \(\sim 250\) RRLs per field and between 2 and 15 epochs of observations. The multiple epochs of observations are needed to fit radial velocities templates to find the velocity of the center-of-mass with RV precision of \(\sim\)5 km s\({}^{-1}\). Multiple epochs also increase the likelihood that a reliable metal abundance can be obtained, as abundances can only be derived when the RRL atmosphere is in a quiescent state and of course stars in the AAOmega fields are all observed at different phases each time. Our sample so far consists of 947 RRLs (Kunder et al. 2016). While valuable for kinematics, the AAOmega data have to date yielded only \(\sim 200\) spectra that may be of high enough quality that abundances can be measured for them. Only with an 8m class telescope is there the possibility of measuring abundances for most of the sample.

The velocity dispersion profile we have derived is shown in Fig. 2. It is immediately evident that these RRLs do not follow the cylindrical rotation of the long bar and are better fitted with a dynamically hot component. At face value our results are consistent with the presence of an old, classical component, consisting of largely metal poor old stars and showing no significant rotation. We have estimated its mass to be about 1% of the bulge mass. Ness et al. (2013) also find slowly rotating metal-poor stars, and it is unclear if these two populations share a common origin or if their red clump giants represent contamination from the disk and halo. The bulge RRL are without any doubt within the inner \(\sim 1\) kpc of the bulge and are unquestionably old, making them especially suitable to probe the inner Galaxy.

We have also been able to derive the metal abundance of these stars from the Calcium Triplet line at 8498 Å (Wallerstein et al. 2012). These range from [Fe/H] \(\sim-2.5\) to a fraction that have super-solar metallicities. The distribution peaks at [Fe/H] \(\sim-1\), about 1/10\({}^{th}\) the typical metallicity of red giants in the long bar. However, they are more metal rich than the stellar halo (An et al. 2013) although the metallicity gradient observed in the inner halo may account for this (Suntzeff et al. 1991). The large metallicity spread is suggestive of the presence of multiple populations in the bulge and it also appears that more metal rich RRLs have smaller velocity dispersion, again suggesting that there may be several formation mechanisms at hand (see Fig. 4).

Multiple populations in the bulge RRLs have recently been traced out in two distinct sequences in the period-amplitude diagram (Pietrukowicz et al 2015), which can be explained as the manifestation of two major old bulge populations co-existing in the bulge. Roughly 25% of bulge RRL can be found in these two sequences. Schiavon et al. (2016), using APGEE giants, have also confirmed chemical diversity for bulge stars in the range \(\rm-2.0<[Fe/H]<-1.00\). About 10% of those stars have abundances of second-generation globular cluster stars. These subgroups are not meaningless, and need to be placed in proper context with the formation of the inner Galaxy. In our sample, there will be \(\sim\)2100 RRLs measured, the majority having \(\rm[Fe/H]<-1.00\), and these will span the complete period-amplitude space. Therefore, hundreds of the RRLs will belong to the various subgroups already identified, with superior kinematic information both individually and collectively.

We do not find the very low metallicity population postulated by Tumlinson (2010) and seen by Howes et al. (2015) in the red giants. This may be because those stars are elsewhere on the horizontal branch. It also may be the consequence of the Calcium Triplet calibration for RRLs, which does not extend to stars below [Fe/H] \(<-2.0\) dex. We also have in hand high-resolution (R\(\sim\)28 000) spectra of \(\sim\)150 bulge RR Lyrae stars using M2FS@Magellan which we are using to refine the RRL-Calcium Triplet calibration. Our survey is intended to shed light on whether an ultra-metal poor population of RR Lyraes may be present.

In order to progress we must extend our survey to a larger portion of the bulge, and especially to higher galactic latitudes and longitudes to trace the rotation curve and explore the bulge/halo connection. Perez-Villegas et al. (2016) reproduce our observations with a model where the RRLs are simply the inner extension of the halo. If this is the case, the RRLs should follow the bar structure only in the inner \(5^{\circ}\) (see Fig. 3), and they would be one of the most metal-rich halo populations known. In order to do so we cannot rely on the modest access to the AA\(\Omega\) given by the OPTICON program and a small time exchange program set by NOAO and the Australian counterpart. A proper study of this issue requires a survey program of the radial velocities and metallicities of bulge RRLs. Gaia will not be releasing observations of the bulge RRLs, as they are too faint and crowded.


We propose to carry out such a survey using FLAMES. This is somewhat time critical as there is the possibility of FLAMES being decommissioned in preparation for a future replacement; the bulge is also only truly visible during semesters A during the Chilean winter. Hence our large program extends over P99 and 101 and asks for 1/2 of the observations in each period. However, our targets are comparatively bright (\(I\sim 17\)), FLAMES fibers are \(1.4^{\prime\prime}\) across (and therefore we do not need seeing better than this) and we wish to target the Calcium Triplet region. Hence our program can be carried out in fairly modest conditions as to seeing, transparency and moon illumination and this allows us to exploit relatively poor quality observing conditions during the Chilean winter.

Our approach will be to observe \(12^{\circ}\) long strips at galactic latitudes of \(-2^{\circ}\), \(-4^{\circ}\) and \(-6^{\circ}\), at longitude intervals of 2 degrees (\(-\)6,\(-\)4,\(-\)2,0,+2,+4,+6). This will allow us to fully sample the rotation curves of our Galaxy and the long bar as traced by the BRAVA survey. As each FLAMES field covers 1/3 of a degree, we have a total of 28 fields per longitude strip and 84 fields in total. Each 2 degree longitude intervals (consisting of 4 FLAMES pointings) will contain about 100 RRLs (based on the OGLE survey counts), which in our experience yields reliable values for the mean radial velocity and velocity dispersion.

In each of the fields our primary targets will be the fundamental mode RRLs already identified by the OGLE survey. We will determine their radial velocities and metal abundances from \(R\sim 11,000\) spectra obtained with FLAMES, centered on the Calcium Triplet region at 8498–8662 Å. A single 45 min. exposure is sufficient to measure the radial velocities and abundances, but we will need to observe at least three epochs for each field. The reason for this is twofold. First, an RRL radial velocity can vary by \(\pm\)50 km/s about its’ systemic velocity. Both Jeffery et al.(2007) and Sesar (2012) showed that center-of-mass radial velocities have a typical uncertainty of \(\pm\)1.5 km s\({}^{-1}\) for variables observed at least three times when using the Liu 1991 radial velocity template. We verified this as well in the case for bulge RRLs, using the OGLE derived periods and times of maximum light (Kunder et al. 2016). We need radial velocities precisions to be \(<\)5 km/s to isolate cold substructures within our RRLs and flag them for follow-up observations. Clearly, there is very interesting science in exploring kinematic/metallicity substructure. Second, RRL abundances can only be derived when the RRL atmosphere is in a quiescent state (which occurs \(\sim\)75% of the time, Pancino et al. 2015), and not when the RRL pulsation cycle is e.g., on the rising branch (pha