A HIGH-VELOCITY BULGE RR LYRAE VARIABLE ON A HALO-LIKE ORBIT

It is well known that within the Galactic bulge at longitudes $| l|$ < 10°, there is a bar-like structure with a bar angle in the range 20°–30°, traced by numerous old stellar population probes (e.g., red clump giants, RCGs; Stanek et al. 1994; Groenewegen & Blommaert 2005). This bar/bulge is a rotating Box/Peanut (B/P) structure with an X-shape protrusion, made up largely of old and metal-rich stars (∼10 Gyr, $[\mathrm{Fe}/{\rm{H}}]$ falling between −0.5 and +0.5 dex; e.g., Gonzalez et al. 2012; Johnson et al. 2013; Wegg & Gerhard 2013).

There is speculation that the Milky Way also has an older, more spheroidal bulge population, and perhaps the greatest possibility of uncovering such a component would be within the most metal-poor bulge stars. One example of a metal-poor bulge population is the RR Lyrae stars (RRLs), and over the past few years, ∼38,000 RRLs toward the bulge have been identified from photometric surveys (e.g., Soszyński et al. 2014). These RRLs are thought to exhibit a small metallicity spread and are centered around $[\mathrm{Fe}/{\rm{H}}]$ = −1 dex (Walker & Terndrup 1991; Kunder & Chaboyer 2008; Soszyński et al. 2014). The small $[\mathrm{Fe}/{\rm{H}}]$ spread may suggest that RRLs trace a more ancient stellar population than the majority of bulge red giant branch and red clump stars, which are more metal-rich, on average, and likely do not evolve to become RRLs (e.g., Walker & Terndrup 1991; Lee 1992).

Recent photometric studies have reached different conclusions regarding the relationship of the bulge RRLs and the bar/bulge. Dékány et al. (2013) combined optical and infrared photometry of ∼8000 ss OGLE-III discovered RRLs to find that unlike the RCGs, RRLs do not trace a strong bar. Instead, they have a more spheroidal, centrally concentrated distribution, indicating that the RRLs belong to a classical bulge that has co-evolved with the bar (e.g., Saha & Gerhard 2013).

In contrast, using 28,000 RRLs from the more spatially extended OGLE-IV bulge sample, Pietrukowicz et al. (2014) assert that the RRLs trace closely the barred structure formed of RCGs, and hence that the bulge RRLs are in the same gravitational potential together with the more massive Galactic bar.

Missing still are the kinematics of the bulge RRLs, which can resolve this discrepancy and provide an understanding of the origin of the old, metal-poor bulge component. The last published paper on bulge RRLs radial velocities, Gratton (1987), used a sample of 17 RRLs to conclude that the kinematic properties of RRLs in Baade's Window are similar to that of the Miras, M-giants, K-giants, OH/IR sources, and planetary nebulae of the Galactic bulge.

In this paper, we report on a high-velocity RRL found serendipitously in the Bulge RR Lyrae Radial Velocity Assay (BRAVA-RR) to have v_r = −372 km s⁻¹, which is well above the typical speed of the stars one might expect to find in the bulge. High-velocity stars are intriguing in part because they can provide insight to the mechanisms that produce their velocities. The origin of high-velocity stars can also provide useful information about the environments from which they are produced. Here, we investigate the cause for the high-velocity of MACHO 176.18833.411 to discern whether or not it is consistent with stars in the Galactic bulge and what this suggests about the formation of the Galaxy.

MACHO 176.18833.411 was originally cataloged by the MACHO survey as a fundamental mode RRL (Kunder et al. 2008).¹⁷ The star was one of the ∼100 RRLs surveyed spectroscopically as part of BRAVA-RR (NOAO PropID: 2014A-0143; PI: A. Kunder) in a field at (l, b) = (3, −2.5) using the AAOmega multifiber spectrograph on the Anglo-Australian Telescope. We observed this star twice on 2014 June 21, separated in time by 7 hr, and the observations were taken in dual beam mode centered on 8600 Å, with the 580 V and 1700D gratings to probe the Calcium Triplet. This covers the optical window from about 8300 to 8800 Å at a resolution of R ∼ 10,000.

The data were reduced using the automated pipeline supplied by AAOmega, 2DFDR, and the spectra were cross-correlated using the IRAF cross-correlation routine, xcsoa. Four Bulge RAdial Velocity Assay (BRAVA) stars (Kunder et al. 2012) observed with the same setup were selected as radial velocity standard stars. Due to less than optimal weather conditions, the signal-to-noise ratio is ∼10 (see the spectrum in Figure 1), and the consistency of our velocity result is 8 km ${{\rm{s}}}^{-1}$, in agreement with the errors reported by xcsao for each individual measurement.

Zoom In Zoom Out Reset image size

Figure 1 (middle) shows the pulsation curve using the radial velocity template and scaled by its V-amplitude as outlined from Liu (1991). The two radial velocity measurements were folded by the known period to find the radial velocity as a function of phase. Both measurements fit the radial velocity template well for an RRL with a line of sight radial velocity, V_los of −372 km s⁻¹. It is noteworthy that we did not adjust the template in phase, which indicates that the OGLE time of maximum brightness is reliable for this star.

The OGLE-IV catalog of RRLs provides V- and I-band light curves of their stars, with the photometric observations spanning from 2010 March to 2013 October (Soszyński et al. 2014). The optical light curve for MACHO 176.18833.411 from the OGLE-IV observations is shown in Figure 1 (left), and its most important properties are summarized in Table 1.

Table 1.  Properties of MACHO 176.18833.411

α (J2000)	18:00:13.08
δ(J2000)	−27:15:39.1
l	3.0795
b	−1.9209
$\langle V\rangle$	17.586 mag
$\langle I\rangle$	15.969 mag
V-amp	1.27 ± 0.05 mag
Period	0.51521996 ± 0.00000005 days
Time of max brightness	2,456,000.21386
$[\mathrm{Fe}/{\rm{H}}]$	−1.62 ± 0.2 (from light curve)
${(V-I)}_{\mathrm{min}}$	1.79 ± 0.01
$E{(V-I)}_{\mathrm{min}}$	1.21 ± 0.03
AI	1.49
${(m-M)}_{0}$	14.29 ± 0.13 mag
Heliocentric distance	7300 ± 600 pc
Galactocentric distance	850 pc
Vlos	−372 ± 8 km s−1
VGRV	−350 ± 8 km s−1
${\mu }_{\alpha *\mathrm{cos}(\delta )}$	8.08 ± 0.20 ± 0.40 mas yr−1
${\mu }_{\delta }$	4.41 ± 0.20 ± 0.40 mas yr−1
X	−1.04 ± 0.56 kpc
Y	0.39 ± 0.03 kpc
Z	−0.243 ± 0.02 kpc
U	−377 ± 20 km s−1
V	−262 ± 26 km s−1
W	−147 ± 20 km s−1

Download table as:  ASCIITypeset image

To find the distance to the RRL, we first use the mean-flux magnitude as listed by OGLE-IV. Next, the $E(V-I)$ color excess along the stars line of sight is calculated from the observed V – I color at minimum light, ${(V-I)}_{\mathrm{min},\mathrm{obs}}$ (see, e.g., Guldenschuh et al. 2005), by carrying out a Fourier fit to the V- and I-band light curves. The extinction, A_I, can then be derived using

$$\begin{eqnarray}&&{A}_{I}=0.7465E(V-I)+1.37E(J-K),\end{eqnarray} \tag{ 1 }$$

as introduced in Nataf et al. (2013). Here, $E(J-K)$ was taken from the bulge reddening maps of Gonzalez et al. (2012). Finally, it has been shown that RRL V-band light curves and the phase difference of their Fourier decomposition can be used to estimate photometric metallicities to an accuracy of 0.2 dex(Jurcsik & Kovács 1996), although with high-quality light curves, the fitting accuracy can be ∼0.12 dex (Kovács 2005). This relation is well-calibrated over the metallicity range from ∼ −2.0 to ∼0 dex. Applying this method to the OGLE V-band light curve, the metallicity ($[\mathrm{Fe}/{\rm{H}}]$) of MACHO 176.18833.411 is estimated and placed on the Carretta et al. (2009) metallicity scale. Using the recalibration of the RRL luminosity scale by Catelan & Cortés (2008), MACHO 176.18833.411 has an absolute magnitude of M_V = 0.58 ± 0.13 mag. Similarly, using a quadratic relation between RRL absolute magnitude and metallicity from Bono et al. (2007), we find M_V = 0.61 ± 0.08 mag, where 0.08 is a reasonable error in the RRL absolute magnitude—$[\mathrm{Fe}/{\rm{H}}]$ zero-point calibration. The Benedict et al. (2011) ${M}_{V}$–$[\mathrm{Fe}/{\rm{H}}]$ relation, however, results in M_V = 0.43 ± 0.07, indicating that the level of agreement between independent M_V measurements is as large as ∼0.2 mag. A systematic uncertainty of 0.2 mag M_V is factored into the uncertainty in the estimated distance, and does not change the results significantly. The apparent magnitude, reddening, and absolute magnitude of MACHO 176.18833.411 lead to a distance of ${(m-M)}_{0}$ = 14.29 ± 0.13 mag.

In the OGLE-III catalog, Soszyński et al. (2011) marked ∼400 stars as possessing proper motions relative to mean motion of the bulge that are large enough to be easily detected. MACHO 176.18833.411 is included in this list. To determine its proper motion, the centroids of stars were measured on each OGLE-IV (Udalski et al. 2015) image separately. All measured centroids were transformed to the common grid using positions of bright red giants, i.e., stars belonging to the bulge, not the disk population. The transformed positions were fitted with a model that takes into account proper motion as well as differential refraction effects (see Poleski et al. 2013 for details).

With our proper motion, the space velocity and position vector can be resolved, and the 6D position and velocity information for MACHO 176.18833.411 is given in Table 1. For this convention, the Sun's orbital velocity vector ${v}_{\odot }$ =[${U}_{\odot }$,${V}_{\odot }$, ${W}_{\odot }$] = [14.0, 12.24, 7.25] km s⁻¹, V_LSR = 220 km s⁻¹, and position = [8.28, 0, 0] kpc. The derived true space velocity is −482 ± 22 km s⁻¹ relative to the Galactic rest frame.

4.1. Possible Explanations for the High-velocity Star

BRAVA-RR currently has surveyed only 94 RRLs, including MACHO 176.18833.411. Is such a bulge star therefore really anomalous? To give us some idea if this RRL in fact belongs to the bulge, we integrated its orbit through an assumed Galactic potential, which is a sum of the potential of a logarithmic halo, Miyamoto–Nagai disk, and a Hernquist bulge, as in Hawkins et al. (2015). Uncertainties in the orbital integrations were estimated by a Monte Carlo approach, where the initial conditions were varied to within their uncertainties over 100 orbital integrations (see Hawkins et al. 2015 for details).

The orbital integration over the past 1 Gyr is shown in Figure 2, as is the distribution of the maximum distance from the Galactic plane, ${Z}_{\mathrm{max}}$, and the minimum and maximum distance from the Galactic center for 100 orbital draws. Although MACHO 176.18833.411 is currently close to or in the bulge, its orbit clearly suggests it is not confined to the bulge. The escape velocity at the radius of the bulge is ∼650 km s${}^{-1}$; this means that a star with the high velocity of MACHO 176.18833.411 can still be inside the bulge. However, it is clear that MACHO 176.18833.411 spends most of its time well outside the radius of the formal bulge/bar structure.

Zoom In Zoom Out Reset image size

We next address how likely it is for a bulge star to have such a negative radial velocity. The Galactocentic radial velocity, V_GRV, distribution of 229 giants toward the bulge at (l, b) = (4, −2) and 320 RCGs toward the bulge at (l, b) = (2.4, −2.2) is shown in Figure 3. The velocities of the giants are taken from APOGEE (the Apache Point Observatory Galactic Evolution Experiment), which is part of Sloan Digital Sky Survey III, and we analyze data from data release 12 (DR12; Alam et al. 2015). The velocities of the RCGs are taken from the GIRAFFE Inner Bulge Survey (GIBS; Zoccali et al. 2014). The mean velocity and velocity dispersion of both the APOGEE giant sample and the GIBS RCGs are within 1σ of the mean velocity and velocity dispersion for the RRLs in the (l, b) = (3, −2.5) BRAVA-RR field (A. Kunder et al. 2016, in preparation). Assuming the radial velocity distribution for the giants, red clump, and RRLs is Gaussian, a star with V_GRV = −350 km s⁻¹ is a 4σ outlier in velocity space.

Zoom In Zoom Out Reset image size

The ∼9000 BRAVA giant sample (Kunder et al. 2012), ∼17,000 ARGOS RCG sample (Freeman et al. 2013), and the ∼1200 Gaia-ESO Survey bulge RCG sample (Rojas-Arriagada et al. 2014) probe further from the Galactic plane than where MACHO 176.18833.411 resides, so more contamination from, e.g., the disk and halo would be expected, but these large radial velocity surveys can also give an indication of how unusual it is for a star toward the bulge to have such a negative velocity. Within the Gaia-ESO Survey bulge sample, there is no star with V_GRV less than −350 km s⁻¹ and <0.1% of stars have V_GRV less than −300 km s⁻¹. For both the BRAVA and ARGOS samples, ∼0.02% of stars have V_GRV less than −350 km s⁻¹ and ∼0.1% of stars have V_GRV less than −300 km s⁻¹.

4.2. A Halo Star

There are a few possibilities that can explain the presence of MACHO 176.18833.411. The most likely is that this star is a halo interloper that happens to be at the same distance and location as where bulge RRLs reside. This is supported by our orbital solution of MACHO 176.18833.411, which shows its ${R}_{\mathrm{max}}$ is mostly larger than 8 kpc, strongly suggesting this RRL is not confined to the bulge. The time spent in the bulge is 145 ± 100 Myr, which corresponds to less than 15% of the integration. The Z_max of ∼8 kpc suggests this star is not part of the disk, which is also consistent with its high velocity. This star is on an elliptical orbit (orbital ellipticity is 0.95 with a period of ∼140 Myr) and resembles a star in the halo.

Numerical simulations that match the kinematic observations of the bulge well do not predict such high velocity bulge stars in this region of the sky (Shen et al. 2010), although we note that it is possible for significant radial velocity outliers to be present in any population. Instead, high velocity RRLs have been found in the halo (e.g., ∼1.5% of the local RRLs in the Layden 1994 and Kollmeier et al. 2013 samples have V_GRV less than −300 km s⁻¹, which is more than a factor of 10 larger than high velocity giants and RCGs found in the bulge surveys discussed previously). Similarly within the halo globular cluster (GC) sample, 3% have V_GRV less than −350 km s⁻¹.

If this RRL is a halo interloper, we presume additional halo contamination exists in the bulge RRL sample, with velocities indistinguishable from bulge stars. We would then expect at least a few percent of the RRLs located in the direction toward the bulge to be halo constituents. To obtain a rough approximation of the presence of the halo, the relation of the RRL density profile in the halo is extrapolated to the Galactic center (Watkins et al. 2009) as shown in Figure 4. To find the total number of halo RRLs expected in our observations, we use this relation along with the following approximations: (1) the bulge is a cylinder with a radius of 2 kpc and a minor-to-major axial ratio of 0.55, and (2) the bulge extends from l = −10 to l = +10 and b = −10 to b = +10 degrees, and so the AAOmega 3 square degree field covers 3/314 of the Galactic bulge. It follows, then, that ∼85 ± 70 halo RRLs are expected with a Galactocentric distance of 0.9 kpc in a 3 square degree field. The uncertainty in the number of halo RRLs is a function of the exact scale height and length of the bulge, and is illustrated in Figure 4. The total number of OGLE RRLs in our 3 square degree field is ∼1400, and the Galactocentric distance of these stars peaks at 0.8 kpc. Therefore, assuming the OGLE sample of RRLs is relatively complete, ∼6% of the OGLE RRLs stars toward (l, b) = (3, −2.5) are expected to be from the halo, and it is not unexpected for a few percent of halo stars to exist so close, both in spatial and in distance distribution, to the Galactic center. Proper motions of a larger sample of RRLs would be desirable to identify halo RRLs from bulge RRLs.

Zoom In Zoom Out Reset image size

It is worth noting that the orbit of MACHO 176.18833.411 has its largest excursions perpendicular to the plane—this extreme high velocity star is confined (statistically) vertically. If the most extreme halo stars have an almost spheroidal zone of excursion, the (halo) population from where this star originated from is not the most extreme. It may well be that this RRL is a relic from an earlier era, but likely not the earliest.

4.3. A Bulge Star

4.4. A Runaway Star

In this paper, we take a detailed look at a high negative-velocity RRL observed toward the Galactic bulge found in the BRAVA-RR survey. Stars of such high velocity are rare among bulge giants, and the more precise distance of the RRL makes it possible to explore its origin in greater detail, by integrating its orbit. We argue that MACHO 176.18833.411 is most likely a halo interloper, suggesting that contamination from halo stars is relevant when attempting to trace out the metal-poor tail of the bulge's metallicity distribution function (e.g., Howes et al. 2015).

We thank the Australian Astronomical Observatory, which has made these observations possible. This research was supported in part by the National Science Foundation under grant No. NSF PHY11-25915. This work was supported by Sonderforschungsbereich SFB 881 "The Milky Way System" (subprojects A4, A5, A8) of the German Research Foundation (DFG). This work has been partially supported by the Polish Ministry of Science and Higher Education through the program "Ideas Plus" award No. IdP2012 000162 to I.S. R.M.R. acknowledges support from grant AST-1413755 from the National Science Foundation. C.I.J. gratefully acknowledges support from the Clay Fellowship, administered by the Smithsonian Astrophysical Observatory.

J.S. acknowledges support from the 973 Program of China under grant No. 2014CB845700, the National Natural Science Foundation of China under grant Nos. 11333003 and 11322326, and the Strategic Priority Research Program "The Emergence of Cosmological Structures" (No. XDB09000000) of the Chinese Academy of Sciences. Z.Y.L. acknowledges support from the National Natural Science Foundation of China under grant No. 11403072 and Shanghai Sailing Program (No. 14YF1407700). This work made use of the facilities of the Center for High performance Computing at Shanghai Astronomical Observatory. Hospitality at APCTP during the 7th Korean Astrophysics Workshop is kindly acknowledged.