«ABSTRACT The Galactic bulge is dominated by an old, metal rich stellar population. The possible presence and the amount of a young (a few Gyr old) ...»
The age of the young bulge-like population in the stellar system Terzan5:
linking the Galactic bulge to the high-z Universe1
F. R. Ferraro2, D. Massari3,4, E. Dalessandro2, B. Lanzoni2, L. Origlia3, R. M. Rich5, A.
2 Dipartimento di Fisica e Astronomia, Universit` degli Studi di Bologna, Viale Berti Pichat 6/2,
I–40127 Bologna, Italy
3 Kapteyn Astronomical Institute, University of Gr¨ningen, Kapteyn Astron Institute, NL-9747 o AD Gr¨ningen, Netherlands o 4 INAF- Osservatorio Astronomico di Bologna, Via Ranzani, 1, 40127 Bologna, Italy 5 Department of Physics and Astronomy, 430 Portola Plaza, Box 951547, Los Angeles, CA 90095-1547, USA 25 June 2016 ABSTRACT The Galactic bulge is dominated by an old, metal rich stellar population. The possible presence and the amount of a young (a few Gyr old) minor component is one of the major issues debated in the literature. Recently, the bulge stellar system Terzan 5 was found to harbor three sub-populations with iron content varying by more than one order of magnitude (from 0.2 up to 2 times the solar value), with chemical abundance patterns strikingly similar to those observed in bulge ﬁeld stars. Here we report on the detection of two distinct main sequence turn-oﬀ points in Terzan 5, providing the age of the two main stellar populations: 12 Gyr for the (dominant) sub-solar component and 4.5 Gyr for the component at super-solar metallicity. This discovery classiﬁes Terzan 5 as a site in the Galactic bulge where multiple bursts of star formation occurred, thus suggesting a quite massive progenitor possibly resembling the giant clumps observed in star forming galaxies at high redshifts. This connection opens a new route of investigation into the formation process and evolution of spheroids and their stellar content.
Subject headings: Galaxy: bulge - Globular Clusters: Individual (Terzan 5) - Technique:
photometry Based on data obtained with (1) the ESA/NASA HST, under programs GO-14061, GO-12933, GO-10845, (2) the Very Large Telescope of the European Southern Observatory during the Science Veriﬁcation of the camera MAD;
(3) the W.M. Keck Observatory, which is operated as a scientiﬁc partnership among the California Institute of Technology, the University of C
The picture of galaxy bulges formation is still highly debated in the literature (for a review of the Milky Way bulge, see, e.g., Rich 2013; Origlia 2014). Among the proposed scenarios, three main channels can be grossly distinguished: dissipative collapse (e.g., Ballero et al. 2007; McWilliam et al.
2008), with possibly an additional component formed with a time delay of a few Gyr (e.g. Tsujimoto & Bekki 2012; Grieco et al. 2012), dynamical secular evolution of a massive disk that buckles into a bar (e.g., Combes & Sanders 1981; Raha et al. 1991; Saha & Gerhard 2013), and merging of substructures either of primordial galaxies embedded in a dark matter halo (e.g., Scannapieco & Tissera 2003; Hopkins et al. 2010), or massive clumps generated by early disk fragmentation (e.g.
Immeli et al. 2004; Carollo et al. 2007; Elmegreen et al. 2008). In the merging scenarios most of the early fragments rapidly dissolved/merged together to form the bulge. However, a few of them could have survived the total disruption (e.g. Immeli et al. 2004; Carollo et al. 2007; Elmegreen et al. 2008) and it is very possible that such fossil relics are still observable somewhere in the host galaxy, grossly appearing as normal globular clusters (GCs). Because of their original large mass, these clumps should have been able to retain the iron-enriched ejecta and the stellar remnants of the supernova (SN) explosions, and they likely experienced more than one burst of star formation.
As a consequence, we expect these fossil relics to host multi-iron sub-populations and, possibly, also a large number of neutron stars. Clearly, ﬁnding stellar systems with these properties would provide crucial observational support to the ’merging” scenario for bulge formation.
Until recently, no empirical probes of such fossil clumps in galaxy bulges were available. The situation changed in 2009, when Ferraro et al. (2009) discovered two stellar components with very diﬀerent iron abundances in Terzan 5, a stellar system in the Milky Way bulge previously catalogued as a GC, with the only peculiarity of hosting the largest population of millisecond pulsars in the Galaxy (Ransom et al. 2005). The two components of Terzan 5 appear well distinct at the level of the red clump (RC) and red giant branch (RGB) in the combined NIR-optical color-magnitude diagram (CMD; Ferraro et al. 2009; Massari et al. 2012). Moreover, they display signiﬁcantly diﬀerent iron and α-element abundances, the metal-poor population having [Fe/H]= −0.25 dex and [α/Fe]= +0.34, the metal-rich one showing [Fe/H]= +0.27 dex and [α/Fe]= +0.03 (Origlia et al. 2011; a minor component at [Fe/H]= −0.8 has been also detected, extending the internal metallicity range of Terzan 5 over more than 1 dex; see Origlia et al. 2013; Massari et al. 2014).
Given these abundance patterns, the observed RC split could be explained either in terms of an age diﬀerence of a few Gyr, or in terms of a diﬀerent helium content in two nearly coeval sub-populations (D’Antona et al. 2010; Lee et al. 2015).
The chemical abundance patterns measured in Terzan 5 are strikingly similar to those observed toward the Galactic bulge, while no other stellar system within the Milky Way outer disk and halo or in the Local Group show analogous properties (Chiappini et al. 1999; Matteucci & Chiappini 2005; Lemasle et al. 2012). This opens the intriguing possibility that Terzan 5 is a fossil relic of one of the structures that contributed to generate the Galactic bulge. The bulge is known to be dominated by an old ( 10 Gyr) stellar population with solar-like metallicity (Clarkson et al. 2008;
Here we present the accurate determination of the main sequence turn-oﬀ (MS-TO) region in Terzan 5, from which we determined the age of its two main stellar populations. The paper is organized as follows. The data-sets used and the data reduction procedures are presented and discussed in Section 2. The analysis of the CMDs, and the measure of the ages and radial distributions of the two populations are presented in Section 3. Section 4 is devoted to discussion and conclusions.
2. Observations and Data Analysis
For this study we used a wide photometric database collected with diﬀerent instruments and telescopes: (i) ultra-deep images in the F606W and F814W ﬁlters acquired with the Advanced Camera for Surveys (ACS) on board the Hubble Space Telescope (HST), in three diﬀerent epochs
spanning more than 11 years (GO: 9799, PI: Rich; GO: 12933, PI: Ferraro and GO: 14061, PI:
Ferraro); (ii) a set of images in the F110W and F160W ﬁlters secured with the IR channel of the Wide Field Camera 3 (WFC3) on board the HST (GO: 12933, PI: Ferraro); (iii) a set of high-resolution images in the J and K bands acquired with the Multi-conjugate Adaptive Optic
Demonstrator (MAD) at the ESO Very Large Telescope (Science Demonstration Proposal, PI:
Ferraro); (iv) a set of K-band images obtained at the Keck Telescope by using the camera NIRC2 assisted with laser adaptive optics (U156N2L, PI: Rich). Most of the dataset used in this work was already presented and discussed in previous papers (see Ferraro et al. 2009; Lanzoni et al. 2010;
Massari et al. 2014, 2015; Ferraro et al. 2015). Hence detailed information on the observations, data quality and data analysis can be found in those papers, while here we report only a schematic summary of the datasets and the reduction processes.
The optical HST dataset acquired with the Wide Field Channel (WFC) of the ACS camera is described in detail in Massari et al. (2015) and Ferraro et al. (2015). Here we used exclusively the FLC images, already ﬂat ﬁelded and corrected for charge transfer eﬃciency (CTE) losses with the pixel-based correction in the pipeline (Anderson & Bedin 2010; Ubeda & Anderson 2012). The data reduction was performed using the software DAOPHOT-II (Stetson 1987, 1994), following the standard procedure described e.g. in Dalessandro et al. (2015). Brieﬂy, for each exposure we determined the best PSF model using a few hundreds of bright, isolated stars. The model was then applied to all the sources detected (using ALLSTAR) above a 3σ threshold from the background.
Since the images secured with the F814W ﬁlter (being less aﬀected by the strong extinction present in the cluster direction) turned out to be signiﬁcantly deeper than those acquired in the ﬁlter F606W, we created a master list of stars composed by sources detected in at least three frames.
This master list was used as input to force the PSF ﬁtting with DAOPHOTII/ALLFRAME. The –4– star magnitudes obtained in each frame secured in the same ﬁlter were then homogenized, and their weighted mean and standard deviation were ﬁnally adopted as magnitude and photometric error of each source. The stellar instrumental magnitudes and positions were ﬁnally calibrated onto the Johnson-Cousins system and reported onto the 2MASS equatorial reference frame, respectively, as described in Lanzoni et al. (2010).
The WFC3 dataset is made of 32 images, each one acquired with an exposure time of 300 s in both the F110W and the F160W ﬁlters. We worked on the images already pre-reduced by the STScI pipeline, following the same reduction procedure described in the previous paragraph. Also in this case, the F814W ACS catalog was used as input for the ALLFRAME analysis of all the exposures. The ﬁnal catalog was built with the same prescriptions used for the ACS one. The instrumental magnitude were calibrated using the aperture correction and the zero-points listed at http://www.stsci.edu/hst/wfc3/phot zp lbn. Stellar positions were brought to the 2MASS astrometric reference using the CataPack suite of software1 The MAD dataset is already described in Ferraro et al. (2009). Since the observations span a temporal interval of several hours, with seeing conditions varying quite signiﬁcantly from one exposure to another, we limited our analysis only to those with a stable Full Width at Half Maximum (FWHM) of ∼ 0.1 across the entire MAD ﬁeld of view. FWHM were measured on bright, non saturated and isolated stars using the standard IRAF tools. Fifteen exposures survived such a selection, and we reduced them as described for the previous data sets. In this case, however, the PSF model was computed by allowing the look-up table of the star proﬁle ﬁtting residuals to vary across the ﬁeld of view as a third order polynomial. This is because, as demonstrated in Saracino et al. (2015) and Massari et al. (2016), the PSF variability due to anisoplanatism eﬀects in MCAO images is too large to be properly accounted for by a uniform PSF model. The magnitude of the ﬁnal MAD catalogue were calibrated as described in Ferraro et al. (2009).
The NIRC2 data set consists of 82 exposures in the K-ﬁlter with integration time of 180 s, taken during two observing nights (see Table 1). NIRC2 uses one Laser Guide Star (LGS) to correct for the blurring eﬀect of the atmosphere. For this reason, the PSF variation within the ﬁeld of view is much stronger than what observed in MCAO data, and the PSF deformation mainly follows a radial pattern around the LGS itself. In the acquired images, not even a third order polynomial allowed a proper modeling of the PSF, and a good photometry was achieved only within the isoplanatic region. Nonetheless, each exposure has been treated as described for the previous sets, and a ﬁnal catalog including only stars found in at least 5 single exposures was built.
The instrumental magnitudes were calibrated using the MAD catalog as reference.
3. Unveiling the double Turn-oﬀ
The gold standard method to measure the age of a stellar population requires the accurate deﬁnition of the MS-TO region in the color-magnitude diagram (CMD). In Terzan 5 this is hampered by three major problems: (i) strong diﬀerential reddening, (ii) heavy contamination from disk and bulge ﬁeld stars and (iii) severe stellar crowding. To overcome these obstacles we took advantage of previous works published by our group. In particular: the high-resolution extinction map obtained in the direction of Terzan 5 (Massari et al. 2012) was used to identify regions where the color excess is the least and the most homogeneous: we excluded from the analysis all the cluster regions (as the North and the North-East portions of the system) where the reddening map shows large variations of the extinction. Once selected the most appropriate region, we also applied diﬀerential reddening corrections (following the prescriptions by Cardelli et al. 1989) to the magnitudes of all the stars detected therein. To remove possible non-member objects, we excluded all the sources with relative proper motion larger than 1.5 mas yr−1, which likely belong to the Galactic disk and bulge populations (Massari et al. 2015).2 In order to remove blends, only sources with a quite symmetric brightness proﬁle (as measured by the PSF sharpness parameter) have been considered.
Moreover, to minimize the scatter, only sources within 3-σ from the median photometric error at diﬀerent magnitude levels, in the K-magnitude versus photometric error diagram, have been taken into account. Finally, we excluded the innermost and most crowded regions of the cluster, depending on the ﬁeld of view and the performances of each instrument: we excluded stars with r 16 in the MAD and NIRC2 samples, those with r 20 in the ACS sample, and those with r 35 in the WFC3 sample.
Figure 1 schematically shows the portions of the cluster ﬁeld covered by each dataset and considered for the analysis. As can be seen, diﬀerent portions of the South and South-West regions of Terzan 5 are sampled. Figure 2 shows the CMD of the TO/SGB region in a few planes, obtained by combining observations in diﬀerent photometric bands, under the criteria described above. Although they refer to diﬀerent regions of the cluster, they all show (at diﬀerent levels of signiﬁcance) the existence of two distinct evolutionary sequences at the MS-TO/sub-giant branch (SGB) level.