Discovery of Two New Globular Clusters in the Milky Way

For RLGC 1, we use the 2MASS point-source catalog, choosing the profile-fitted magnitudes. On the other hand, for RLGC 2, we use the UKIRT Infrared Deep Sky Survey Galactic Plane Survey (UKIDSS GPS; Lawrence et al. 2007; Lucas et al. 2008) data, which is much deeper than 2MASS. Both data sets are in VEGA magnitudes.

Unfortunately, the point-source catalog of the UKIDSS GPS is significantly incomplete in the central regions of RLGC 2. Therefore, we derived point-source function (PSF) magnitudes of the sources in RLGC 2 from the UKIDSS J and K images, using the Image Reduction and Analysis Facility (IRAF)¹ /DAOPHOT (Stetson 1987). For source detection, we used a 4σ detection threshold. We selected sources with good photometry using the sharpness distribution. These sources are matched with the point sources in the UKIDSS GPS catalog. Using the magnitudes of the matched point sources, we transform the instrumental PSF magnitude to the standard system magnitude. The calibration errors are ∼0.03 mag in both the J and K bands.

We note that K-band magnitudes in the UKIDSS system are known to be very similar to those in the 2MASS system (Hodgkin et al. 2009): ${K}_{\mathrm{UKIDSS}}={K}_{2\mathrm{MASS}}+0.010({J}_{2\mathrm{MASS}}-{K}_{2\mathrm{MASS}})$. For the color range of $0\lt ({J}_{2\mathrm{MASS}}-{K}_{2\mathrm{MASS}})\lt 2$, the difference between the two systems is smaller than 0.02 mag. Therefore, we do not distinguish K-band magnitudes between the 2MASS and UKIDSS systems in the following.

Figures 1(a), (b), and (c) show grayscale maps of the WISE W1 and W3, and 2MASS K_s band images of RLGC 1. Similar images of RLGC 2 are shown in Figures 2(a), (b), and (c). The K-band images of these two clusters show a grouping of resolved stars. The W1-band images show the presence of faint diffuse light in the clusters. However, the W3 band images show little diffuse light features in the cluster regions, indicating that there is no dust associated with the clusters.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We derive the radial number density profile of the RLGC 1 region using the stars with 11 ≤ K_s ≤ 14.5 mag, and that of the RLGC 2 region using the stars with 11 ≤ K ≤ 15 mag. The profiles of RLGC 1 and RLGC 2 are shown in Figures 1(d) and 2(d), respectively.

The two clusters show strong central excesses. In the radial number density profile of RLGC 1, there is an excess at r < 02 and a slight enhancement at 05 < r < 08. This profile becomes almost flat at r > 08. Similarly, in the case of RLGC 2, there is a central excess at r < 05. The radial number density profile of RLGC 2 becomes almost flat at r > 08. These excesses indicate that most of the bright and resolved stars in the central regions of these clusters belong to each cluster.

Figures 1(e) and 2(e) display the K versus $(J-K)$ CMDs of the resolved stars located inside the half-light radius (r_h) of each cluster (which is derived in the following section). We plot the stars located at $r\lt 0.5{r}_{h}$ with red symbols. These stars have a higher probability of cluster membership compared with the stars located in the outer region. For comparison, we plot the CMDs of the background region of each cluster in Figures 1(f) and 2(f). As background regions, we select an annular region at 99'' < r < 104'' for RLGC 1 and at 56'' < r < 63'' for RLGC 2. The area of each background region is the same as that of the cluster region inside the half-light radius.

We subtract background stars from the cluster CMDs using a statistical subtraction process. This subtraction is based on the number difference of stars in the same sub-regions of the CMD between the cluster region and the background region. The details of this statistical background subtraction process are described in Appendix B of Ryu & Lee (2018). In Figures 1(e) and 2(e), we plot the stars that were subtracted in this process with gray symbols.

The CMD of RLGC 1 shows a narrow red giant branch (RGB) feature, although the number of stars that constitute the RGB is small. The brightest part of the RGB is seen at K_s ≈ 11.8 mag and $(J-{K}_{s})\approx 1.4$. On the other hand, the CMD of RLGC 2 shows a much stronger RGB than RLGC 1. The brightest part of the RGB in this cluster is seen at K ≈ 10.8 mag and (J − K) ≈ 1.8.

In summary, based on morphological features, radial number profiles with strong central concentration, and the presence of the RGB, we conclude that RLGC 1 and RLGC 2 are old GCs.

We derive the structural parameters of the new GCs using King profile-fitting on the background-subtracted radial number density profiles of the clusters: $\sigma \,=k[1/\sqrt{1+{(r/{r}_{c})}^{2}}-1/\sqrt{1+{({r}_{t}/{r}_{c})}^{2}}{]}^{2}$ (King 1962). In this equation, r_c and r_t denote the core radius and the tidal radius. The background number densities are averages of the number densities at the outer regions of each cluster, which are 72'' < r < 144'' for RLGC 1 and 48'' < r < 96'' for RLGC 2.

From this fitting, we derive the parameters for RLGC 1: r_c = 018 ± 003, r_t = 093 ± 040, and c = 0.7 ± 0.2. In the same manner, we derive structural parameters of RLGC 2: r_c = 021 ± 001, r_t = 090 ± 010, and c = 0.6 ± 0.1. The errors in these values are fitting errors.

We derive systematic deviations of the structural parameters using the bootstrap method with N_repeat = 3000. The systematic parameter deviations of RLGC 1 are σ_rc = 007, σ_rt = 026, and σ_c = 0.2 for the core radius, tidal radius, and concentration index. These systematic deviations and parameter-fitting errors are similar to each other.

Similarly, we derive the deviations for RLGC 2: ${\sigma }_{{rc}}=0\buildrel{\,\prime}\over{.} 11$, σ_rt = 010, and σ_c = 0.3. The mean value of the core radius during the bootstrap resampling is $\langle {r}_{c}{\rangle }_{\mathrm{boot}}=0.26$, which is consistent with the fitting result. However, the systematic deviation of the core radius is significantly larger than its fitting error. This means the derived core radius value is reliable, but not robust; a different sampling for the radial number density profile can produce a different core radius of RLGC 2 within σ_rc = 011.

Based on the tidal radii of the new GCs, we derive the total magnitude of each cluster using circular apertures with tidal radii. The background levels are estimated using the annular region with 120'' < r_bg < 130'' for both clusters.

The integrated aperture magnitude at each tidal radius is K_total = 7.45 ± 0.02 mag for RLGC 1 and K_total = 5.86 ±0.03 mag for RLGC 2. Finally, we derive half-light radii of ${r}_{h}=32\buildrel{\prime\prime}\over{.} 9\pm 1\buildrel{\prime\prime}\over{.} 9$ for RLGC 1 and r_h = 279 ± 07 for RLGC 2. These half-light radii are plotted with dashed blue line circles in the grayscale images of Figures 1 and 2.

We determine the reddenings, distance moduli, and metallicities of the new GCs from isochrone (PARSEC; Bressan et al. 2012) fitting on the CMDs. Since our photometric data of the new GCs do not reach the main-sequence turnoff points or subgiant branches, we adopt an age of 12.6 Gyr (Log (age) = 10.1) for the new GCs. This age is close to the mean age of metal-poor GCs (12.5 Gyr; VandenBerg et al. 2013). We also adopt [α/Fe] = +0.3 and [Z/H] = [Fe/H] + 0.94[α/Fe] (Thomas et al. 2003). We use the same extinction law as adopted in Bressan et al. (2012): R_V = 3.1, ${A}_{K}=0.366E(B-V)$ (2MASS), ${A}_{K}=0.353E(B-V)$ (UKIDSS), $E(J-K)\,=0.533E(B-V)$ (2MASS), and $E(J-K)=0.524E(B-V)$ (UKIDSS). Finally, we assume the brightest RGB star in each cluster corresponds to the tip of the RGB (TRGB).

We visually match the RGB feature of each cluster and 12.6 Gyr isochrones with varying metallicities. In particular, we tried to match the position of the TRGB and the slope of the RGB of each cluster with isochrones. The errors of the parameters are estimated by eye, considering the uncertainties in visual fitting.

Finally, we derive the values of the reddening, distance modulus, and metallicity: $E(B-V)=1.3\pm 0.1$, $E(J-K)\,=0.7\pm 0.1$, ${(m-M)}_{0}=17.3\pm 0.3$, and [Fe/H] = −2.2 ± 0.2 for RLGC 1, and $E(B-V)=1.9\pm 0.2$, $E(J-K)\,=1.0\pm 0.1$, ${(m-M)}_{0}=16.0\pm 0.3$, and [Fe/H] = −2.1 ± 0.3 for RLGC 2. The distance moduli correspond to the metric distances of d = 28.8 ± 4.3 kpc and d = 15.8 ± 2.4 kpc for RLGC 1 and RLGC 2. These results, in particular metallicities, show that these two clusters are genuine metal-poor GCs. We list all determined parameters of the new GCs in Table 1.

Table 1.  Fundamental Parameters of New GCs

	RLGC 1	RLGC 2
αJ2000 (hh:mm:ss)	16:17:08.41	18:45:28.17
δJ2000 (dd:mm:ss)	−44:35:38.6	−05:11:33.3
l (degree)	336.8696	27.6309
b (degree)	4.3031	−1.0421
${(m-M)}_{0}$	17.3 ± 0.3	16.0 ± 0.3
d (kpc)	28.8 ± 4.3	15.8 ± 2.4
$E(B-V)$	1.3 ± 0.1	1.9 ± 0.2
$[\mathrm{Fe}/{\rm{H}}]$	−2.2 ± 0.2	−2.1 ± 0.3
rc (arcmin)	0.18 ± 0.03	0.21 ± 0.01
rc (pc)	1.51 ± 0.34	0.97 ± 0.15
rt (arcmin)	0.93 ± 0.40	0.90 ± 0.10
rt (pc)	7.79 ± 3.55	4.14 ± 0.78
c	0.7 ± 0.2	0.6 ± 0.1
rh (arcmin)	0.55 ± 0.03	0.47 ± 0.01
rh (pc)	4.59 ± 0.74	2.14 ± 0.33
Ktotal (mag)	−10.35 ± 0.30	−10.84 ± 0.31
K50'' (mag)	−10.28 ± 0.30	−10.10 ± 0.31
V50'' a (mag)	−8.25 ± 0.31	−8.03 ± 0.33

Note.

^aV_50'' magnitude is converted from K_50'' using an equation $V-{K}_{s}=2.93\,+0.409$[Fe/H] (Cohen et al. 2007).

Download table as:  ASCIITypeset image

In Figure 3 we plot the spatial location of the new GCs in comparison with other known GCs listed in Harris (1996), Minniti et al. (2017a), and Camargo (2018), in both the face-on view (Figure 3(a)) and the edge-on view (Figure 3(b)). We adopted the distances to Minniti 01–22 given by Minniti et al. (2017a; see Piatti 2018 for other distance estimations). The two GCs are located at the far-half region of the Milky Way. Their distances from the closest neighbor GCs are 10.2 kpc (NGC 5824) and 4.5 kpc (Pal 11) for RLGC 1 and RLGC 2; practically, no neighbor GCs are found in the vicinity of the new GCs.

Zoom In Zoom Out Reset image size

Based on the distances and Galactic latitudes of the new GCs, we derive their vertical positions from the Galactic plane: Z = 2.2 ± 0.3 kpc for RLGC 1 and Z = −290 ± 40 pc for RLGC 2. RLGC 1 has a low metallicity and is likely a halo GC, located above the thick disk (Li & Zhao 2017, h_z = 0.9 ±0.1 kpc). RLGC 2 is located in the thick disk. However, its low metallicity ([Fe/H] = −2.1 ± 0.3) indicates that it must be a halo GC. Therefore, RLGC 2 may be a halo GC passing through the disk now.

Cohen et al. (2007) provided the integrated K_s magnitude of the known GCs derived with 50'' radius apertures. For comparison, we derive the 50''-integrated K magnitudes of the two GCs. The 50''-integrated magnitudes are K_50'' = −10.28 ± 0.30 mag (RLGC 1) and K_50'' = −10.10 ± 0.31 mag (RLGC 2), while the peak absolute magnitude of the GC luminosity function noted in Cohen et al. (2007) is M_K = −9.7 mag. The magnitudes of the new GCs are 0.4–0.6 mag brighter than the peak magnitude of the known GCs in the Milky Way. Using the relation for the metal-poor GCs in Cohen et al. (2007), $V-{K}_{s}=2.93+0.409$[Fe/H], we estimate the 50''-integrated V magnitudes of the new GCs. The estimated magnitudes are ${V}_{50^{\prime\prime} }=-8.25\pm 0.31$ mag (RLGC 1) and ${V}_{50^{\prime\prime} }=-8.03\pm 0.33$ mag (RLGC 2), while the peak absolute magnitude is M_V = −7.66 ± 0.09 mag (Di Criscienzo et al. 2009).

The core radii of the two GCs are ${r}_{c}=1.51\pm 0.34$ pc and r_c = 0.97 ± 0.15 pc for RLGC 1 and RLGC 2. These values are consistent with the median core radius of the known GCs, which is $\mathrm{median}({r}_{c})=1.04$ pc. The half-light radii of the two GCs are also comparable with the known GCs: r_h = 4.59 ± 0.74 pc for RLGC 1, r_h = 2.14 ± 0.33 pc for RLGC 2, and $\mathrm{median}({r}_{h})=3.03$ pc.

The concentration indices of the new GCs are relatively lower than those of the known GCs: c = 0.7 ± 0.2 for RLGC 1 and c = 0.6 ± 0.1 for RLGC 2 (${cf}.\ \mathrm{median}(c)=1.50$). The core radii of the new GCs are consistent with those of known GCs, hence low concentration indices would be related to tidal radii. Actually, the tidal radii of the two GCs (r_t = 7.79 ± 3.55 pc for RLGC 1 and r_t = 4.14 ± 0.78 pc for RLGC 2) are much smaller than the median tidal radius of the known GCs, $\mathrm{median}({r}_{t})\,=28.86$ pc. This implies that the derived tidal radii of the new GCs might have been underestimated. However, complex backgrounds in the data we used prevent us from recognizing weak enhancements of number densities in outer cluster regions.

Based on their morphologies, radial number density profiles, CMDs, and other determined parameters, RLGC 1 and RLGC 2 are likely genuine metal-poor halo GCs. However, our photometric parameters are based on uncertain assumptions: the magnitude of the brightest RGB star and the selective extinction value R_V. These assumptions are likely the origins of additional distance uncertainties. Deeper NIR photometry reaching the main-sequence turnoff point is needed to measure more accurate distances, ages, and metallicities for the two new GCs. Considering the low metallicity of these clusters, we expect that blue horizontal branch stars might be detected in deep optical CMDs reaching fainter than V ∼ 23 mag.

Including these two new GCs, the current census of the GCs in the Milky Way is N_GC = 216. The current GC numbers at $| Z| \lt 1\,\mathrm{kpc}$ are 58 and 20 in the close-half region and the far-half region, respectively. Therefore, we predict that there are about 30 undiscovered GCs in the far-half and $| Z| \lt 1\,\mathrm{kpc}$ region of the Galactic disk.