DISCOVERY OF RR LYRAE STARS IN THE NUCLEAR BULGE OF THE MILKY WAY

There are very limited observational tests that can be applied to shed light on the origins of galactic nuclei, using their stars and black holes. The only galactic nucleus where detailed stellar population properties can be derived with sufficiently high resolution and accuracy is that of the Milky Way, making it a fundamental testbed for different formation models. There we can in principle resolve individual stars, probing a wide range of stellar ages and metallicities. There are two main scenarios proposed for the formation of the nuclear bulge of the Milky Way, and of all galactic nuclei in general: merging of globular clusters (Tremaine et al. 1975; Capuzzo-Dolcetta 1993; Gnedin et al. 2014; Guillard et al. 2016), and fast gas accretion and star formation onto the central region (Milosavljević 2004; Schinnerer et al. 2008).

Here we concentrate on testing the first of these two theories. In that scenario, dynamical friction causes orbital decay, dragging globular clusters deep into the potential well, where they merge and form a high-density nuclear bulge with a nuclear star cluster at its center. While merging globular clusters typically bring in old stellar populations, comparison with observations requires some additional in situ star formation (Antonini et al. 2015), or subsequent growth of the newly formed nuclear cluster via wet merger with other clusters that bring with them additional gas reservoirs that contribute younger stars (Guillard et al. 2016). Therefore, young or intermediate-age stellar populations often dominate the total light in galactic nuclei, even in cases where they make a minor contribution to the total stellar mass. It is then very difficult, in an environment like the Galactic center, to establish the presence of the ancient stellar populations, and to estimate their ages and metallicities. Such old populations must be present if the nuclear bulge of the Milky Way was made through the merging of primordial globular clusters (Capuzzo-Dolcetta 1993).

Theoretically, if the nuclear stellar bulge formed through the merging of several globular clusters, the expected extension of the final merger product is about 100 pc (Capuzzo-Dolcetta 1993; Antonini et al. 2012; Antonini 2014; Gnedin et al. 2014). This size appears to be obtained by the simulations regardless of the presence or absence of a central massive black hole (Capuzzo-Dolcetta 1993; Capuzzo-Dolcetta & Miocchi 2008; Capuzzo-Dolcetta & Mastrobuono-Battisti 2009; Antonini et al. 2012; Antonini 2014).

Observationally, the nuclear bulge of the Milky Way is well fit by two components: the nuclear star cluster, a compact component with a half-light radius of 4 pc (2 arcmin) that dominates the inner  30 pc, and the nuclear stellar bulge, a shallower component extending out to about 120 pc (Launhardt et al. 2002). This size is comparable to the sizes of well studied nuclear stellar bulges of other external galaxies (Lotz et al. 2001; Carollo et al. 2002; Hartmann et al. 2011).

As globular clusters are tidally disrupted, they yield their stars, including RR Lyrae, to the field. So far, there has been no search for variable stars that was deep enough to find RR Lyrae in the complex Galactic center region. However, this can now be tested observationally with the VISTA Variables in the Via Lactea (VVV) ESO public survey (Minniti et al. 2010; Saito et al. 2012), which contains deep multi-epoch photometry in the near-IR, allowing us to find faint variable sources.

In the present search we also find numerous bright LPVs/Miras, eclipsing binaries, Cepheids, and microlensing events, which will be reported elsewhere. Here we concentrate  only on the RR Lyrae because in this context they play a crucial role. Their properties make them prime representatives of the primordial stellar populations of the Milky Way because: (a) they have a well known period–luminosity (P–L) relation, and are therefore excellent distance indicators; (b) they have a very narrow range of intrinsic colors, which makes them excellent reddening indicators; and (c) they are old (age >10 Gyr) and metal-poor (i.e., [Fe/H] < −0.5).

The limiting magnitudes (K_s ∼ 18 mag, J ∼ 20 mag) and spatial resolution (∼0.8 arcsec) of the near-IR data provided by the VVV survey enable for the first time a successful search of RR Lyrae throughout the Galactic center region. The point-spread function (PSF)-fitting photometry of the individual VVV images for ∼100 epochs of the tiles b333 and b334 was carried out following the procedure described by Alonso-García et al. (2015) and Minniti et al. (2015). The search for periodic variable stars, phasing of the light curves, and classification of the RR Lyrae type ab were made following the strategies outlined by Gran et al. (2016) and Dékány et al. (2013). We searched for RR Lyrae with magnitudes of 12 < Ks < 17, amplitudes of 0.2 < A < 1.0, and periods of 0.3 < P < 1.0 days.

Extreme crowding and extinction variations are clearly evident in near-IR images taken by the VVV survey (Figure 1). Searching for RR Lyrae in the most crowded and reddened region of the Milky Way is therefore a daunting task, in which several problems need to be faced and sorted out. Specifically, the completeness depends on the position in the field, as these are near-IR mosaics. The stellar density is very high, and the presence of numerous saturated stars in the field that obliterate their surroundings is a limiting factor in the photometry. In addition, the large and differential reddening, highly variable even on small spatial scale, affects the photometric completeness, although the effect is less severe than the crowding. Indeed, the VVV photometry is generally deep enough to reach well below the RR Lyrae region of the color–magnitude diagram even in the most reddened region at the distance of the Galactic center (Figure 3). The photometric completeness in this region measured from red clump giants at 14.3 < K_s < 15.9, the magnitude range of the spanned by the observed RR Lyrae, has an average value of 80% (Valenti et al. 2016). However, the variable seeing and uneven sampling of the observed epochs contribute to further reducing the completeness of our sample. The faint magnitudes of the targets also prevent us from finding/classifying RR Lyrae that have very small amplitudes (<0.2 mag).

Zoom In Zoom Out Reset image size

For all these reasons, we do not claim full completeness for the detection of RR Lyrae, but conversely we expect many more to be found in dedicated high-resolution deep searches that might enable us to establish the total RR Lyrae density number in the Galactic center region.

The shape of the light curves is also an important limiting factor, with contamination from eclipsing binaries being a serious problem for the sinusoidal light curves, and therefore limiting us to mostly select RRab with asymmetric light curves. The total number of epochs (i.e., points in the light curves) is ∼100 epochs, generally sufficient to select RR Lyrae with confidence. However, sampling is a problem, with many good candidates that need to be discarded as aliases. We therefore have many more RR Lyrae candidates for which additional observations are needed in order to measure their periods accurately. These observations will be acquired during the next three years as part of the VVV extended survey (VVVX).

Here we report the discovery of a dozen RR Lyrae type ab (fundamental mode pulsators) stars within 36 arcmin (84 pc) from the Galactic center (Figure 1), plus a couple of c-type RR Lyrae (pulsating in their first overtone). We measure accurate positions, projected distances from the Galactic center, mean IR magnitudes and colors, periods, and amplitudes for all of our targets (Table 1). The clear variability signature of RR Lyrae, including their characteristic saw-tooth light curve shape (Figure 2), and their measured amplitudes and periods, are the unambiguous signatures that we are detecting individual ancient and faint RR Lyrae, and not stellar blends or other artifacts.

Zoom In Zoom Out Reset image size

Table 1.  Photometric Observations^a

ID	R.A.	Decl.	R	Ks	J	$(J-{K}_{s})$	P	A
	(J2000)	(J2000)	arcmin				days
40405	266.17164326	−28.86342987	15.5	15.78	20.05	4.27	0.780597	0.32
65743	266.40425763	−29.31187570	18.3	14.27	18.17	3.90	0.594484	0.22
55278	266.79651367	−29.08202592	20.4	15.60	19.07	3.47	0.624883	0.35
37068	266.28880383	−28.68220478	20.7	15.26	18.34	2.98	0.618563	0.30
80042	266.62151461	−28.71032145	20.8	15.44	18.63	3.19	0.408153	0.29
84844	266.06131561	−28.83296846	21.4	15.87	19.69	3.82	0.549641	0.36
58214	266.11652879	−28.65279182	26.5	15.79	18.87	3.08	0.403153	0.26
55144	266.10448919	−28.65918014	26.6	14.64	18.00	3.36	0.508399	0.39
89691	266.24177968	−28.58633606	26.9	15.74	19.33	3.59	0.376027	0.25
33007	266.92238653	−28.83231159	28.6	15.27	18.74	3.47	0.621276	0.24
42332	266.10220084	−28.61729278	28.7	15.34	17.93	2.59	0.520099	0.30
65271	266.26347962	−28.47343076	33.1	14.81	17.75	2.94	0.369534	0.30
8444	266.09247811	−28.51255523	34.3	15.52	18.60	3.08	0.487035	0.44
33289	266.20795127	−28.43858475	35.9	15.58	18.61	3.03	0.480476	0.33

Note.

^aTypical photometric errors are σ_Ks = 0.01 mag, and σ_J = 0.06 mag. Periods are good to 10⁻⁵ days, and K_s-band amplitude errors are σ_A = 0.02 mag.

Download table as:  ASCIITypeset image

We also found a candidate type II Cepheid at a projected distance of 45pc (20 arcmin) from the Galactic center. This type II Cepheid with P = 1.809 days, mean K_s = 14.66, and $(J-{K}_{s})=4.02$, is also representative of an old and metal-poor population present in the vicinity of the nuclear star cluster. A fundamental implication about the old age of the variable stars found here is that the nuclear stellar bulge must have been in place since the origins of the Milky Way. In addition, we discovered several more bonafide RR Lyrae over a wider area, within 36–50 arcmin (84–109 pc) of the Galactic center, just outside of the nuclear bulge.

The near-IR color–magnitude diagram obtained from PSF fitting photometry shows the high extinction of the field where these RR Lyrae have been discovered (Figure 3). The target RR Lyrae are fainter and bluer than the bulge red clump giants. When taking into account the large extinction differences across the bulge, the comparison of the color–magnitude diagram of these RR Lyrae and those in other regions of the bulge (Minniti et al. 2010; Saito et al. 2012) is consistent with them being RR Lyrae located in the region of the Galactic center.

Zoom In Zoom Out Reset image size

3.1. Extinction Corrections

Extinction corrections to the measured near-IR magnitudes are a mandatory step in order to assess the location of the sample RR Lyrae within the Galactic nucleus. The RR Lyrae lie in the instability strip, which is a narrow band in the color–magnitude diagram, and their intrinsic colors can be assumed to be ${(J-K)}_{0}=0.15\pm 0.05$ mag. Although we only have K_s-band light curves and a single (or a few) J-band epoch, the color corrections for de-reddened RR Lyrae due to the single J-band observation is negligible (Dékány et al. 2013; Gran et al. 2016). In fact, the color variation along the light curves is typically small in the near-IR (${\rm{\Delta }}(J-{K}_{s})\lt 0.05$ mag). When computing the reddening for a specific target, the most important systematic error is the uncertain slope of the reddening law (Majaess et al. 2016; Nataf et al. 2016; Gonzalez et al. 2012). For example, comparing ${A}_{k}=0.528E(J-K)$ given by Nishiyama et al. (2009) with ${A}_{k}=0.72E(J-K)$ from Cardelli et al. (1989), and ${A}_{k}=0.435E(J-K)$ from Alonso-García et al. (2015), the corresponding differences for the typical extinction values of the Galactic center region ($E(J-K)\sim 3.0\mbox{--}4.0)$ are significant. Adopting the most recent value given by Alonso-García et al. (2015) that applies to the VVV data, and which also agrees with the slope of the reddened red giant clump seen in the color–magnitude diagram (Figure 3), we find for each candidate the reddening $E(J-{K}_{s})$ and extinction A_k listed in Table 2.

Table 2.  Measured Stellar Parameters

ID	E(J − Ks)a	AK	${(m-M)}_{0}$ b	D⊙b	${(m-M)}_{0}$ c	D⊙c	[Fe/H]a	Comments
				(kpc)		(kpc)	(dex)
40405	4.12	1.79	14.67	8.6	14.92	9.6	−2.2	RRab
65743	3.75	1.63	13.02	4.0	13.27	4.5	−1.5	RRab foreground?
55278	3.32	1.44	14.59	8.3	14.85	9.3	−1.6	RRab
37068	2.83	1.23	14.45	7.8	14.67	8.6	−1.6	RRab
80042	3.04	1.32	14.09	6.6	14.32	7.3	−0.6	RRab
84844	3.67	1.60	14.56	8.2	14.82	9.2	−1.3	RRab
58214	2.93	1.27	14.47	7.8	14.71	8.8	−0.6	RRab
55144	3.21	1.40	13.45	4.9	13.70	5.5	−1.2	RRab foreground?
89691	3.44	1.50	14.12	6.7	14.36	7.4	−0.5	RRc
33007	3.32	1.44	14.26	7.1	14.50	7.9	−1.6	RRab
42332	2.44	1.06	14.51	8.0	14.75	8.9	−1.2	RRab
65271	2.79	1.21	13.54	5.1	13.69	5.8	−0.4	RRc
8444	2.93	1.27	14.41	7.6	14.66	8.6	−1.1	RRab
33289	2.88	1.25	14.47	7.8	14.72	8.8	−1.0	RRab

Notes.

^aTypical reddening errors are ${\sigma }_{E(J-{Ks})}=0.1$ mag, and metallicity errors are σ_[Fe/H] = 0.3 dex. ^bCase 1: found using the P–L relation given by Equation (14) from Muraveva et al. (2015) and the extinctions listed in Table 1. ^cCase 2: found using the P–L–Z relation of Alonso-García et al. (2015) and assuming a mean [Fe/H] = −1 dex.

Download table as:  ASCIITypeset image

3.2. Metallicities

In order to explore the properties of globular clusters that could have initially formed the Galactic nuclear bulge, we examine the properties of the Oosterhoff type I and II globular cluster populations (Oosterhoff 1939; Catelan 2009). A way to distinguish between these two populations is by measuring the average periods of their RR Lyrae, which are shorter in the mean for type I ($\langle P\rangle =0.55$ days) than for type II Oosterhoff cluster populations ($\langle P\rangle =0.65$ days) (Catelan 2009). We find that the distribution of periods of the nuclear bulge RR Lyrae has a mean of P = 0.55 days, resembling an Oosterhoff type I population (more metal-rich than [Fe/H] = −1.6 dex).

Alternatively, the mean metallicities for RR Lyrae type ab can be estimated using their period-amplitude-metallicity relations. After discarding the RR Lyrae type c, we obtain mean metallicities of $\langle [\mathrm{Fe}/{\rm{H}}]\rangle$ = −1.0, −1.4, and −1.3 dex for the sample RR Lyrae using the calibrations from Alcock et al. (2000), Yang et al. (2010), and Feast et al. (2010), respectively.

The Bailey diagram is shown in Figure 4, in comparison to the bulge RR Lyrae. The resulting individual metallicities using the Galactic RR Lyrae calibration from Feast et al. (2010, which should only be taken as indicative until spectroscopic measurements become available), are listed in Table 2. Therefore, we suggest that most of the merged clusters were Oosterhoff type I globulars (with −1.6 < [Fe/H] < −0.9 dex).Even though we cannot deny that a few of the primordial globular clusters that merged into the nuclear stellar bulge might have been very metal-poor (with [Fe/H] < −2 dex likeVVV-RRL-40405), they do not appear to be the dominant population. Interestingly, the star VVV-RRL-40405 that is located only 16 arcmin away from the Galactic center is the most metal-poor star of this sample, with [Fe/H] ∼ −2.2 dex.This very metal-poor object is the first RR Lyrae that is a likely member of the nuclear star cluster. Knowing its intrinsic color, distance, and extinction (Table 2), we can estimate its visual magnitude, V = 29.5 mag, much too faint and out of reach for current optical instruments.

Zoom In Zoom Out Reset image size

3.3. Distances

For the first time, we find that there are RR Lyrae in the region well within the nuclear stellar bulge of our Galaxy, suggesting that they could be the remains of the primordial globular clusters that built up the nuclear bulge. The dozen RR Lyrae stars presented here give a limit to the age and metallicity of the nuclear bulge, and thus provide valuable clues about its origin. While there is ample evidence that the stellar population of the nuclear star cluster is composite, containing a mixture of young, intermediate, and old stellar populations (Genzel et al. 2010; Schödel et al. 2014; Chatzopoulos et al. 2015), the RR Lyrae stars we found suggest that the nuclear bulge is very old (>10 Gyr), perhaps as old as the Milky Way itself.

Are these RR Lyrae special in any way? How do they compare with the RR Lyrae previously found in the Milky Way bulge? RR Lyrae are numerous in globular clusters and in the Milky Way halo, and are taken as prime tracers of old (>10 Gyr), and metal-poor stellar populations. However, not all globular cluster RR Lyrae are similar. For example, there are two populations of globular clusters (Oosterhoff 1939; Catelan 2009): the Oosterhoff type I clusters, which are more metal-rich (−0.9 < [Fe/H] < −1.6 dex),and the Oosterhoff type II clusters, which are more metal-poor ([Fe/H] < −1.6 dex).With a mean period of $\langle P\rangle =0.55$ days, most of the RR Lyrae discovered here are representative of an Oosterhoff type I population (Figure 4).

Overall, the properties of the present sample are consistent with the bulge RR Lyrae population (Gran et al. 2016), being more concentrated to the Galactic center. The evidence supports the scenario where the nuclear stellar bulge was originally made out of a few globular clusters that merged through dynamical friction (Capuzzo-Dolcetta 1993; Guillard et al. 2016), and as such it could well be the most massive and oldest surviving star cluster of our Galaxy.

We gratefully acknowledge the use of data from the VVV ESO Public Survey program ID 179.B-2002 taken with the VISTA telescope, and data products from the Cambridge Astronomical Survey Unit (CASU). The VVV Survey data are made public at the ESO Archive. Support for the authors is provided by the BASAL Center for Astrophysics and Associated Technologies (CATA) through grant PFB-06, and the Ministry for the Economy, Development, and Tourism, Programa Iniciativa Cientifica Milenio through grant IC120009, awarded to the Millennium Institute of Astrophysics (M.A.S.). We acknowledge support from FONDECYT Regular grants No. 1130196 (D.M.), and No. 1150345 (M.Z. and F.G.). We are also grateful to the Aspen Center for Physics, where our work was supported by National Science Foundation grant PHY-1066293, and by a grant from the Simons Foundation (D.M. and M.Z.).

Facilities: ESO@VISTA - .