The first galaxy cluster discovered by the VISTA Variables in the Vía Láctea Survey

In order to detect and characterize the extragalactic sources in the tile d015, we applied the photometric methodology developed in Baravalle et al. (2018), which uses the combination of ${\tt SExtractor\ v2.19.1}$ and ${\tt PSFEx\ v3.17}$. The sources were selected using the stellar index CLASS_STAR, the SPREAD_MODEL ($\Phi$) parameter, the radius that contains 50% of the total flux of an object (R_1/2), the concentration index (C) and the near-IR colors. The adopted criteria consider: ${\tt CLASS\_STAR}<0.3$; 1.0 $<$ R${}_{1/2}<5.0~arcsec$; 2.1 $<$ C $<$ 5; and $\Phi>0.002$; and the colors 0.5 $<$ (J - K_s) $<$ 2.0 mag; 0.0 $<$ (J - H) $<$ 1.0 mag; and 0.0 $<$ (H - K_s) $<$ 2.0 mag with the color restriction (J - H) + 0.9 (H - K_s) $>$ 0.44 mag to minimize false detections. Also, the sources with Z and Y detections should satisfy -0.3 $<$ (Y - J) $<$ 1.0 mag and -0.3 $<$ (Z - Y) $<$ 1.0 mag.

Following these criteria, we detected 933 extragalactic sources in the d015 VVV MW disk tile with 24 located in the region of 30$\times$ 30 $arcmin^{2}$ centered at VVV-J144321.05-611753.9. Three objects were discarded because they are low surface brightness sources with strong strong stellar contamination that could not be effectively corrected. The objects VVV-J144243.0-611529.8, VVV-J144331.6-612133.1 and VVV-J144426.4-611101.3 have contamination by nearby stars and they were corrected using the procedure described in Baravalle et al. (2018). Additionally, four sources that satisfied the above criteria but had slightly smaller R_1/2 than 1 arcsec, R_1/2, between 0.78 and 0.95 $arcsec$), were also included after a visual inspection. Our final sample consists of 25 extragalactic sources, all of them visually confirmed to be galaxies. They were all detected in the J, H and K_s passbands and only four sources were also detected in the Z and Y VVV passbands. Figure 1 shows a cutout made from a composite (J, H and K_s) color image of the field showing the concentration of galaxies, centered at the brightest galaxy (VVV-J144321.05-611753.9). The 25 galaxies of the concentration are marked by small circles. The left panel corresponds to a 30$\times$30 $arcmin^{2}$ box identifying with dashed lines, two circles with diameters of 3 and 7 $arcmin$. The right panel shows the 3$\times$3 $arcmin^{2}$ central parts of the cluster candidate.

Table 1 shows the photometric properties of the 25 galaxies. Column (1) gives the VVV identification; columns (2) and (3), RA and Dec in J2000 coordinates; columns (4) to (8) show the PSF Z, Y, J, H and K_s magnitudes; columns (9) to (13) show the aperture magnitudes within a 2 $arcsec$ diameter, respectively; finally, columns (14) to (17) list the morphological parameters based on K_s-band photometry: the half-light radius; concentration index; the ellipticity; and the Sersic index. The PSF magnitudes are estimates of the total K_s magnitudes and they range between 13.70 and 16.00 mag. All magnitudes were corrected for galactic extinction using the maps of Schlafly & Finkbeiner (2011) and the relative extinctions of Catelan et al. (2011). The completeness was estimated for the tiles d010 and d115 in Baravalle et al. (2018, Figure 3) showing that 80 percent completeness is reached for sources fainter than 17 to 17.5 K_s magnitudes. We are reaching these magnitude limits in the tile d015, with a median K_s extinction value A_Ks = 0.73 $\pm$ 0.01 mag. In the studied region of 30$\times$30 $arcmin^{2}$, the extinction was not constant with median A_Ks = 1.07 $\pm$ 0.04 mag: 1.13 $\pm$ 0.05 mag in the central parts (radius smaller than 1.5 $arcmin$); 1.12 $\pm$ 0.07 mag for radius between 1.5 to 3.5 $arcmin$; and 0.93 $\pm$ 0.09 mag in the most external parts.

The galaxies in the concentration are clearly small and red when compared against those in the whole tile. Extinction corrected median values of the photometric properties are: PSF magnitudes, J = 15.50 $\pm$ 0.13 mag; H = 15.09 $\pm$ 0.14 mag; and K_s = 14.64 $\pm$ 0.14 mag and colors, (H - K_s) = 0.34 $\pm$ 0.05 mag; (J - H) = 0.57 $\pm$ 0.08 mag; and (J - K_s) = 0.87 $\pm$ 0.06 mag. The median morphological parameters are: R_1/2 = 1.59 $\pm$ 0.16 $arcsec$; C = 3.01 $\pm$ 0.08; $\epsilon$ = 0.30 $\pm$ 0.03 and Sersic index, n = 4.63 $\pm$ 0.39.

We obtained follow-up spectroscopic data using the near-infrared imaging spectrograph, FLAMINGOS-2 at Gemini South, in the Fast Turnaround (Program ID: GS-2016A-FT-18) observing mode. The observations were made during the nights of July 30 and August 02 of 2016. We collected long-slit spectra with the JH/HK grism and 0.72 arcsec slit. This resulted in a peak resolution R$\sim$600, a dispersion of 6.55 Å/pix and a total spectral coverage from 0.98 to 1.80 $\mu$m. The projected on-sky slit length was 4.48 $arcmin$. We observed the 8 brightest galaxies in the central part of the concentration. We set each pointing to include two objects per slit, constraining the angular separation between each galaxy pair to be lower than 3 $arcmin$. In addition, we optimized the slit position angle configuration or slit rotation to avoid bright stars for each setup. The objects have K_s magnitudes between 13.7 to 14.8 mag and the exposure time was 8 $\times$ 120s to reach a signal-to-noise ratio $S/N>3$ for each combined galaxy spectrum. Calibration frames were acquired together with telluric standard star spectra to remove telluric absorption features from the science data.

The spectra were reduced using the IRAF/Gemini tasks and the Gemini FLAMINGOS-2 PyRAF data reduction package. We performed the basic data reduction and combination of the 2-dimensional data with nsreduce and nscombine routines, respectively. The arc-line identification and wavelength calibration steps were carried out with the nsfitcoords and nstransform routines, respectively. After obtaining the two-dimensional spectra, the one-dimensional spectra were extracted via the nsextract routine, to then be telluric-corrected following the Maiolino et al. (1996) procedure. Finally, the spectra were also flux calibrated using the photometric data available for the telluric standard star observed.

We performed an automatic clustering analysis in order to gain insight about the nature of the galaxy clustering. This analysis aims to split the data into sets of associated objects with a joint conditional probability distribution. In our case, a mixture of simple Gaussian densities would separate the data into different groups by varying the number of free parameters, such as the number and shape of the Gaussians. The best model was obtained using the Bayesian Information Criteria, which penalizes those models with high number of free parameters.

We used the public code Mclust (Scrucca et al. 2016) in the catalog of extragalactic sources found in the tile d015, applying the mixture of Gaussians to the sky coordinates, Right Ascension and Declination. In order to use galaxies with higher probability of belonging to a galaxy cluster, they should satisfy that: R${}_{1/2}<$ 3 $arcsec$; n $>$ 2; and K_s magnitudes fainter than 15.7 mag. We found that the best model has 4 spherically symmetric Gaussians associated to 4 groups with different circular angular areas. The group 1 identified with the concentration of galaxies VVV-J144321-611754 has the smallest angular radius of the sample, 0.138 degrees, whereas the other three groups have angular radii greater than 0.230 degrees. Galaxies within this radius have a higher probability of belonging to the group. Figure 2 shows the distribution of the extragalactic sources of the tile d015 in galactic coordinates. The 25 galaxies of VVV-J144321-611754 are represented by small dots; the objects identified with the groups 1 to 4, with open squares, circles, crosses and triangles, respectively. The group angular sizes are shown in Figure 2. We also took into account some photometric properties of the extragalactic sources, such as the extinction corrected colors, C and n to characterize the groups. The median values for group 1 that contains the concentration of galaxies are (J - K_s) = 0.84 $\pm$ 0.05 mag; (H - K_s) = 0.30 $\pm$ 0.04 mag; R_1/2 = 1.27 $\pm$ 0.09 $[arsec]$; C = 3.01 $\pm$ 0.06; $\epsilon$ = 0.301 $\pm$ 0.03; n = 4.3 $\pm$ 0.3. The other three groups include bluer galaxies with C and n higher than 3.03 and 4.75, respectively. Group 1 is formed with 24 extragalactic objects that were visually inspected; some of them are fainter galaxies with nearby bright stars or located at the outer parts with lower group probability. This result suggests that the VVV-J144321-611754 is a real concentration of galaxies formed by early-type objects.

EzGal (Mancone & Gonzalez 2012) is a tool to model the evolution of the spectral energy distribution (SED) of a stellar population that gives some derived properties such as the magnitude evolution as a function of redshift for different passbands. This tool uses different stellar population synthesis (SPS) models including Bruzual & Charlot (2003), the Maraston model (Maraston 2005), the BaSTI model (Percival et al. 2009), and the Flexible Stellar Population Synthesis (FSPS, Conroy, Gunn & White 2009). The Bruzual & Charlot (2003) model is the most commonly used, which computes the spectral evolution of stellar populations for different ages and a wide range of metallicities. The Maraston model includes a detailed treatment of thermally pulsating asymptotic giant branch stars that dominate the infrared light of a young stellar population. The BaSTI models include a wide range of metallicities for both solar-scaled and $\alpha$-enhanced metallicities. The FSPS models are unique in dealing with important SPS inputs, such as, the initial mass function (IMF) or stellar evolution phases, treating them as free parameters, allowing the uncertainties introduced by various SPS inputs to be quantitatively measured.

The K_s model magnitudes were obtained using EzGal for Bruzual & Charlot (2003) models assuming a formation redshift of z = 3; the Chabrier initial mass function; a solar metallicity; and a simple stellar population at different redshifts. We used a standard normalization of K_s absolute magnitude of -25.51 $\pm$ 0.09 (Brough et al. 2002) for redshifts higher than 0.1 assuming the cluster X-ray luminosity L_X(0.3 $--$ 3.5 keV) $<$ 1.9 $\times$ 10⁴⁴ erg s^-1. In the evolution of K_s magnitudes, the brightest galaxy in the concentration has an observed K_s magnitude of 13.77 mag, which corresponds to a photometric redshift of 0.196 $\pm$ 0.025. This value is also in agreement with the Stott et al. (2009) estimate based on the K_s magnitude and color of the brightest cluster galaxy. Therefore, all these results indicate that the photometric redshift of the galaxy concentration VVV-J144321-611754 is 0.196 $\pm$ 0.025.

The brighter galaxies VVV-J144321.06-611753.9 and VVV-J144319.02-611746.1 were observed in a single pointing. The spectra were flux calibrated and then normalized to the photometry aperture fluxes, in order to be able to perform the dust reddening correction. The final spectra had $S/N~$ 9.5 enough to obtain a precise redshift estimation. Figure 3 shows the spectra of the two galaxies. The solid lines represent the observed spectra with the continuum highly affected by the MW disk dust extinction, making the identification of the spectral features difficult. The dotted lines represent the spectra corrected for absorption of Galactic dust. The spectra were corrected by a simple model of dust reddening extinction in order to retrieve an hypothetical spectrum free of dust. We identified several molecular bands and absorption lines, such as the TiO $\lambda$0.85 $\mu$m and ZrO $\lambda$0.93 $\mu$m molecular bands overlapped with the CaII and Si absorption lines, respectively. Similarly, the CN and TiO bands at $\lambda$1.1 $\mu$m were identified, in addition to the blend of C₂, VO and FeH bands at $\lambda$1.2 $\mu$m. Finally, we recognized the slump of CN and H₂O $\lambda$1.4 $\mu$m bands. All these features, marked by vertical lines in the figure, are typical of old stellar populations (Rayner et al. 2009; Mason et al. 2015). The telluric absorption is shown as a shaded region. The estimated redshift for VVV-J144321.06-611753.9 is $z=$ 0.234$\pm$0.022 and for VVV-J144319.02-611746.1 is $z=$ 0.232$\pm$0.019. The relative large uncertainties in these estimates are mainly related to the use of molecular bands. It was impossible to obtain redshifts for the fainter galaxies in the other pointings due to the lower $S/N$ of the spectra.

For the two brightest galaxies, the spectra include features typical of early-type galaxies. Also, the good agreement between the two redshifts indicates that both galaxies likely belong to the same galaxy cluster. Also, for the brightest galaxy, the difference between the photometric and spectroscopic redshift estimates is about 0.044, and smaller than the uncertainties. We use this spectroscopic redshift as a cluster redshift estimate.

In galaxy clusters, the early-type galaxies follow a tight relation in the Color-Magnitude Diagrams called the Red Sequence (Yee, Gladders & López-Cruz 1999, hereafter RS). We applied the K-corrections to the K_s-band fluxes for the changes of the effective rest- frame wavelengths. We adopted the approximation to the computed observed colors of the 2MASS filters in the Vega system. Following the algorithm by Chilingarian et al. (2010), we used our estimated spectroscopic redshift of the galaxy concentration and the observed (J - K_s) color.

Figure 4 shows the rest-frame near-IR (J - K_s) color – K_s magnitude diagram of the 25 galaxies in the 30$\times$30 $arcmin^{2}$ region. The objects with photometric data are represented by open circles; spectroscopic data by filled circles; and the two objects with estimated redshifts are represented by open squares. The cluster RS shows substantial scatter that might be related to the higher K_s extinction values observed in the central parts of the cluster as compared to outer regions (Section $\S$2.1). The best RS fit relation (J - K_s) = -0.02 $\pm$ 0.01 $\times$ K_s + 0.87 $\pm$ 0.05 was obtained with the 22 galaxies that are also found to belong to the group 1 of the clustering analysis. The solid line in the figure represents the best linear fit and the dashed lines, the 1 $\sigma$ dispersion. Compared with Gladders et al. (1998) that found stable results using 3 $\sigma$ iteration, even this conservative RS has about 64% of the galaxies in the central parts of the galaxy concentration. We obtained similar results using bootstrapping analysis, that supports the idea that these galaxies are part of a galaxy cluster. The cluster RS slope is in agreement with observed near-IR RS for the sample of Stott et al. (2009) and the slope evolution model by Bower et al. (2006) at z $\sim$ 0.2.

Finally, we estimated the virial radius of the galaxy cluster at the derived redshift using the method of Merchán & Zandivarez (2005). We found R_vir = 2.1 $\pm$ 0.4 Mpc for the galaxies in the group 1 of the clustering analysis, and R_vir = 1.7 $\pm$ 0.2 Mpc for the galaxies that follow the RS within 1$\sigma$. The estimates agree with the typical virial radii of galaxy clusters and they are comparable within the uncertainties.