A MATURE GALAXY CLUSTER AT z = 1.58 AROUND THE RADIO GALAXY 7C 1753+6311

The field surrounding 7C 1753+6311 was imaged with Spitzer's Infrared Array Camera (IRAC; Fazio et al. 2004) at 3.6 and 4.5 μm by the CARLA survey (Wylezalek et al. 2013), reaching a 3σ depth of 23.8 mag and 24.4 mag at 3.6 μm and 4.5 μm, respectively. This field was identified as a protocluster candidate with a 4.5σ overdensity by Wylezalek et al. (2013) and was followed up in ${i}^{\prime }$ and J using the William Herschel Telescope in La Palma. The i' band image was taken with the auxiliary-port camera (ACAM), with an exposure time of 6000 s. Full details of the i' data are available in Cooke et al. (2015). The J band image was obtained with the long-slit intermediate resolution infrared spectrograph (LIRIS), with an exposure time of 8160 s, and reduced in the standard way using the publicly available program THELI (Erben et al. 2005; Schirmer 2013). A 3σ depth is reached at ${i}^{\prime }=26.0$ mag and J = 23.6 mag, with seeing of ∼0.76 arcsec for both images.

The IRAC images have a much broader point-spread function (PSF) than the ${i}^{\prime }$ and J images, so selecting sources using the 4.5 μm image is prone to blending and some galaxies may be missed which are distinct in the i' or J bands. Using solely the i' band to detect sources would result in biasing the selection toward intrinsically blue sources, whereas the J image is relatively shallow, so we use a deep F140W image as a detection image. The field around 7C 1753+6311 was imaged with the F140W filter of the Hubble Space Telescope Wide-Field Camera 3 (HST/WFC3) in 2015 July as part of an on-going 40-orbit spectroscopic program (P.I. D. Stern). The HST spectra and photometry will be discussed in a future paper (G. Noirot et al. 2016, in preparation). We retrieved the calibrated, dither-combined (drizzled) image from MAST¹² to use as a detection image. This image has 0.5 ks exposure and is complete¹³ to at least 24 mag.

Source extraction was done with SExtractor (Bertin & Arnouts 1996) in dual-image mode. The HST F140W detection image was used to detect sources, and photometry was obtained on the ${i}^{\prime }$, J, 3.6, and 4.5 μm images in 2 arcsec diameter apertures. We are unable to use large apertures to measure the IRAC fluxes (e.g., 4 arcsec) due to the high spatial density of sources in the cluster core (galaxies are typically ∼2 arcsec apart). So we measure the fluxes in 2 arcsec diameter apertures and correct for the broader PSF of the IRAC data compared to the ground-based data using the ratio of the flux in the J band image to the J image convolved with a Gaussian kernel matching the IRAC PSF, following Hartley et al. (2013). Fluxes were corrected to total fluxes using the growth curves of bright, unsaturated stars in the field of view. This method assumes that the blended sources have the same $J-{\rm{IRAC}}$ color, so may provide inaccurate colors for some sources.

To select sources likely to lie at high redshift, we employ two color cuts. The well-tested IRAC cut of [3.6] − [4.5] > −0.1 (Papovich 2008) selects sources at z > 1.3 due to the 1.6 μm peak of stellar emission moving into the IRAC bands at these redshifts. This cut was adjusted to [3.6] − [4.5] > −0.2 in order to be sure of selecting as complete a cluster sample as possible, although this also allows more lower-redshift sources to contaminate the sample. Most stars have [3.6] − [4.5] ∼ −0.5 and so will be removed by this cut (Galametz et al. 2012). A second cut of ${i}^{\prime }-[3.6]\gt -0.5\times [3.6]+11.4$ (Cooke et al. 2015) was applied to further remove bright low-redshift interlopers and contaminating AGN. To remove faint sources with potentially inaccurate flux measurements, only those with magnitudes brighter than 23.8 mag in 4.5 μm (5σ image depth) and 23.6 mag in J (3σ image depth) were considered. Any sources referred to hereafter are those that match these criteria. To maximise the overdensity of (proto)cluster sources to field contaminants, we only consider sources within 0.9 arcmin of the central RLAGN.

The completeness of our catalog is a function of F140W magnitude. This image is deeper than the ground-based imaging and is 100% complete to the J band limit of 23.6 mag.

2.3.1. Control Field

2.3.2. Subtraction of Field Contaminants

Figure 3 shows a density map of the field around 7C 1753+6311 and ClG 0218.3−0510, a well-studied cluster at z = 1.62 (Papovich et al. 2010; Tanaka et al. 2010). These maps were produced by measuring the number density of sources (selected using the color and magnitude criteria described in Section 2) within 30 arcsec radius apertures around each 5 arcsec pixel. The UDS was mapped in the same way, and the mean and standard deviation of densities in the UDS were used to convert each number density value to a significance above the expected field density. The pixels in Figure 3 are therefore correlated as each 5 arcsec pixel indicates the overdensity within a 30 arcsec radius aperture. Using the selection criteria defined in Section 2, the peak overdensity around 7C 1753+6311 is an 8.9σ significance of galaxies within a 30 arcsec aperture, centered 16 arcsec (136 kpc) from the radio galaxy. These galaxies appear to be highly clustered around the central RLAGN (Figure 1). The source density is so high in the central 0.9 arcmin region that five pairs of sources were blended in the 4.5 μm image, and their true nature was only discovered in the higher resolution ground-based data. It is possible that many of the other CARLA cluster candidates which have extremely high galaxy overdensities also suffer from blending.

Zoom In Zoom Out Reset image size

Follow-up near-infrared Keck spectroscopy of 7C 1753+6311 revealed five galaxies, including the RLAGN, with spectroscopic redshifts between 1.578 < z < 1.587 within a projected diameter of 2 Mpc (A. E. Rettura et al. 2016, in preparation). This structure therefore satisfies the criteria set out by Eisenhardt et al. (2008) for a spectroscopically confirmed z > 1 (proto)cluster, and so we refer to the structure as CARLA J1753+6311 from now on.

Besides the RLAGN, there are 29 ± 6 excess galaxies within 0.9 arcmin of 7C 1753+6311 that are selected with the above color and magnitude criteria.¹⁵ This level of clustering and overdensity is slightly greater than that of the ClG 0218.3−0530 protocluster at z = 1.62, which has a galaxy excess of 22 ± 6 using the same criteria. The cluster ClG 0218.3−0530 is a well-studied structure with a tentative 4.5σ X-ray detection potentially indicating a collapsed core (Tanaka et al. 2010). The comparably high galaxy overdensity surrounding 7C 1753+6311 (Figure 3) suggests that this RLAGN is surrounded by a protocluster consisting of a dominant main halo that is already a relatively high-mass group.

The approximate mass of CARLA J1753+6311 can be estimated from the galaxy richness. Andreon & Congdon (2014a) reported that galaxy richness was a good proxy for cluster mass, with little dependence on redshift. ClG 0218.3−0530 has an X-ray determined mass of 4–8 × 10¹³M_⊙ (Tanaka et al. 2010; Pierre et al. 2012), which is consistent with the galaxy velocity dispersion (Tran et al. 2015). Since CARLA J1753+6311 is richer than ClG 0218.3−0530, its mass is likely to be slightly greater. Using Equation (3) from Andreon & Congdon (2014a), and the calculated value of ${m}_{4.5\;\mu {\rm{m}}}^{*}+1=21.2$ from Wylezalek et al. (2014) we measure a cluster richness of 10 ± 1 galaxies¹⁶ with $[4.5]\leqslant 21.2$, and estimate the mass within 500 kpc of 7C 1753+6311 to be (9.2 ± 4.5) × 10¹³${M}_{\odot }$, consistent with this structure being a slightly more massive group than ClG 0218.3−0530.

One of the signs of a mature cluster is the presence of a red sequence. Ubiquitous in clusters at z < 1, red sequences persist in galaxy clusters out to at least z = 1.4 (Stanford et al. 2005; Snyder et al. 2012), and have been found in some dense (proto)clusters at even higher redshifts (e.g., Kodama et al. 2007; Stanford et al. 2012; Newman et al. 2014).

In Figure 4 we show the ${i}^{\prime }-J$ color–magnitude diagram of sources within 0.9 arcmin of the RLAGN (∼500 kpc at this redshift).¹⁷ Larger squares indicate sources that are within 30 arcsec of the radio galaxy. The histogram in the top panel of Figure 4 shows the excess number of galaxies in CARLA J1753+6311, compared to the UDS control field. There is a significant overdensity in the field around 7C 1753+6311 at all magnitudes, increasing at the faint end. Although we expect contamination from fore- and background field sources in the color–magnitude diagram, the majority of the red data points are likely to be cluster members, and there is a clear, strong red sequence at J < 23 mag, with hints of the sequence continuing to fainter magnitudes.

Zoom In Zoom Out Reset image size

The red sequence is fit by the line ${i}^{\prime }-J\ =7.688{\rm{\mbox{--}}}0.232\times J$, calculated by iteratively clipping sources more than 1.5σ from the best-fit line, allowing both the slope and normalization to freely vary, until convergence was reached. The fit is shown by the red dotted line in Figure 4. The colors of sources on this red sequence suggest they formed at redshifts of $2\lt {z}_{{\rm{f}}}\lt 3$, assuming the galaxies formed their stars in single bursts. A cluster formation model in which the member galaxies formed over the course of 2–3 Gyr, with galaxy formation peaking at z = 3 predicts an average red sequence color of ${i}^{\prime }-J=2.7$ mag, consistent with the data (Cooke et al. 2015).

We define "red" galaxies as those that lie within 0.5 mag of (or redder than) the red sequence (shown by the lower, gray, dotted line in Figure 4); this cleanly divides the red sequence from the blue cloud. We calculate the red fraction for CARLA J1753+6311, statistically removing the expected number of field contaminants as:

$$\begin{eqnarray}&&{f}_{{\rm{red}}}={\langle \displaystyle \frac{({N}_{{\rm{red}}}^{{\rm{7C1753}}}-{N}_{{\rm{red}}}^{{\rm{field}}})}{({N}_{{\rm{total}}}^{{\rm{7C1753}}}-{N}_{{\rm{total}}}^{{\rm{field}}})}\rangle }_{{\rm{median}}}\end{eqnarray} \tag{ 1 }$$

where ${N}_{{\rm{red}}}^{{\rm{field}}}$ and ${N}_{{\rm{total}}}^{{\rm{field}}}$ are the measured number of "red" and total galaxies that satisfy our color criteria in ∼400 random 0.9 arcmin field regions from the UDS. The uncertainty is the 1σ standard deviation in the calculated red fractions for CARLA J1753+6311.

The fraction of red galaxies in the protocluster is significantly larger than in the blank field. The red fraction of CARLA J1753+6311, after statistically removing field contaminants, is ${f}_{{\rm{red}}}=0.66\pm 0.13$, compared to the average fraction of 1.5 < z < 1.7 galaxies in the UDS control field, which is f_red = 0.27 ± 0.01 (see Table 1). CARLA J1753+6311 has a similar red fraction to the z = 1.62 protocluster ClG 0218.3−0510, which has ${f}_{{\rm{red}}}=0.48\pm 0.15$. So the enhanced red fraction in CARLA J1753+6311 seems typical for mature protoclusters. The dense environment of the protoclusters appears to have a strong impact on the colors of their member galaxies.

Table 1.  Fractions of Quiescent Galaxies in CARLA J1753+6311, ClG 0218.3−0510 and in the Control Field UDS at $1.5\lt z\lt 1.7$

	CARLA J1753+6311	ClG0218.3−0510	UDS
fred	0.66 ± 0.13	0.48 ± 0.15	0.27 ± 0.01
fQ ($J\leqslant 23.6$)	0.50 ± 0.09	0.30 ± 0.08	0.16 ± 0.01
fQ (red galaxies)	0.80 ± 0.06	0.67 ± 0.11	0.61 ± 0.03
fQ (${M}_{*}\geqslant {10}^{10}$M⊙)	0.76 ± 0.13	0.44 ± 0.14	0.28 ± 0.01
fQ (${M}_{*}\geqslant {10}^{10.5}$M⊙)	0.91 ± 0.09	0.38 ± 0.16	0.36 ± 0.02

Note. ${f}_{{\rm{Q}}}$ gives the quiescent galaxy fraction and f_red gives the fraction of galaxies with red colors in each sample.

Download table as:  ASCIITypeset image

In low-redshift clusters the galaxies that lie on the red sequence are predominantly passively evolving, old galaxies. However, dusty star-forming galaxies (with ${A}_{V}\sim 1\mbox{--}3$) exhibit colors similar to those expected from quenched, passively evolving (i.e., quiescent) galaxies, and these galaxies make up approximately half of the red infrared-selected galaxy population at higher redshifts (Kriek et al. 2008). Furthermore, recent literature has also shown that high-redshift clusters and protoclusters do contain dusty star-forming galaxies (e.g., Brodwin et al. 2013; Dannerbauer et al. 2014; Smail et al. 2014; Santos et al. 2015). The enhanced red fraction of sources in CARLA J1753+6311 could therefore be ascribed to an excess of dusty star-forming galaxies and/or quenched, passively evolving galaxies. Here we use the rest-frame U, B, J colors (observed i', J, [3.6]) to separate these two populations.

Using the method outlined in Williams et al. (2009), Papovich et al. (2012) used the observed-frame z', J and 3.6 μm bands (rest-frame U, B, J) to separate galaxies in the z = 1.62 cluster ClG 0218.3−0510 into quiescent and star-forming populations. Using the full SED fits to the ClG 0218.3−0510 cluster members, we have converted the Papovich et al. (2012) selection criteria to use our i', J, and 3.6 μm bands. Quiescent galaxies are those which satisfy the following (slightly stricter) criteria:

$$\begin{eqnarray}&&{i}^{\prime }-J\geqslant 2.0\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&J-[3.6]\leqslant 1.7\end{eqnarray} \tag{ 3 }$$

$$\begin{eqnarray}&&{i}^{\prime }-J\geqslant 0.375+1.25\times (J-[3.6]).\end{eqnarray} \tag{ 4 }$$

We caution the reader that these equations were derived specifically for the eighth data release of the UDS and our data. The 3.6 μm magnitudes may be systematically offset by up to 0.5 mag due to the method by which they were determined and so these criteria may change for different datasets.

Figure 5 shows the distribution of all sources selected in CARLA J1753+6311 in i' − J versus J − [3.6] (rest-frame U − B versus B − J) color–color space. The grayscale shows the expected distribution of the control field. The lines show the i', J, [3.6] criteria used to select quiescent galaxies, which lie in the upper-left quadrant. The full cluster membership of CARLA J1753+6311 is not known, so interlopers were statistically removed in i'J[3.6] color–color space using the UDS as the control field. To do this we use ∼400 random 0.9 arcmin radius regions in the UDS, classifying sources as "quiescent" or "star-forming" using the above criteria. Sources in the 7C 1753+6311 field are then classified as "quiescent" or "star-forming," and the quiescent fraction calculated as:

$$\begin{eqnarray}&&{f}_{{\rm{Q}}}={\langle \displaystyle \frac{({N}_{{\rm{Q}}}^{{\rm{7C1753}}}-{N}_{{\rm{Q}}}^{{\rm{field}}})}{({N}_{{\rm{total}}}^{{\rm{7C1753}}}-{N}_{{\rm{total}}}^{{\rm{field}}})}\rangle }_{{\rm{median}}}\end{eqnarray} \tag{ 5 }$$

where ${N}_{{\rm{Q}}}^{{\rm{field}}}$ is the measured number of rest-frame UBJ-selected quiescent galaxies in ∼400 random 0.9 arcmin field regions. The uncertainty is the 1σ standard deviation in the calculated quiescent fractions for CARLA J1753+6311.

Zoom In Zoom Out Reset image size

Without far-IR data, we are unable to locate extremely dust-obscured systems, so these will not be found in either the 7C 1753+6311 field or UDS and would be missing from Figure 5. These extremely dusty galaxies are rare, but could be an important population in protoclusters (e.g., Brodwin et al. 2013). In addition, some galaxies (of order ∼10%) may be misclassified due to very dusty regions within them. Further analysis with submillimeter data would be required to examine the extremely dusty populations in these fields. This means that we cannot analyze the extremely dust-obscured populations in any of the fields considered here, but we are able to do a robust comparison between them as the dusty populations are undetected in all of these fields: CARLA J1753+6311, ClG 0218.3−0510 and the UDS control field.

Half of the detected galaxies in CARLA J1753+6311 are quiescent, with a quiescent fraction (f_Q) for sources with $J\leqslant 23.6$ of ${f}_{{\rm{Q}}}=0.50\pm 0.09$ (see Table 1). Of the "red" galaxies, 80 ± 6% are quiescent, so the vast majority of these objects are not dust-obscured star-forming galaxies, but are already quenched and evolving passively.

ClG 0218.3−0510 contains fewer passively evolving galaxies (${f}_{{\rm{Q}}}=0.30\pm 0.08$), with 67 ± 11% of the red galaxies classified as quiescent. These fractions were calculated using the same criteria as in Section 2 and within a 0.9 arcmin aperture of the cluster core. CARLA J1753+6311 has a similar fraction of red, quiescent galaxies at a 1σ level.

Both of these protoclusters contain a significantly higher quiescent fraction than the average field, which is ${f}_{{\rm{Q}}}=0.16\pm 0.01$. This means that the star formation rates of many cluster members are greatly suppressed relative to the field.

The quiescent fraction is a strong function of J band magnitude. As shown in the bottom panel of Figure 4, the quiescent fraction gradually rises with decreasing magnitude. At J < 22.5 mag the fraction of quiescent galaxies in the protocluster rises to >80%, double the field fraction.

The [4.5] flux provides a better correlation with stellar mass than the J flux, and is nearly independent of galaxy type. Using galaxies with known stellar masses (from full SED-fitting) in the UDS (Mortlock et al. 2013), we convert the [4.5] magnitudes to stellar mass using $\mathrm{log}({M}_{*}/{M}_{\odot })\ =22.53{\rm{\mbox{--}}}0.57\times [4.5]$. This equation was derived empirically by fitting a line to the stellar masses and [4.5] magnitudes from the UDS, which was resampled to have the same $J\leqslant 23.6$ quiescent fraction as CARLA J1753+6311, i.e., 50%, to remove the slight dependence of this relation on galaxy type. This line has an intrinsic scatter of 0.2 dex. We use this equation to calculate the masses for CARLA J1753+6311. We use a similar equation but for the full UDS with no resampling, to calculate the masses for the field. This simply corresponds to a slight shift in the bin center in Figure 6. Using these equations, we recalculate the quiescent fractions as a function of stellar mass (Figure 6). These fractions were calculated as in Section 4.3 for sources in [4.5] magnitude bins (corresponding to stellar mass).

Zoom In Zoom Out Reset image size

There is also a strong correlation between galaxy mass and passivity, with a higher fraction of the massive sources being quiescent. Figure 6 shows that this trend is steeper for the protocluster galaxies than for the field galaxies, and there is a divide at ${M}_{*}\sim {10}^{10.5}$ ${M}_{\odot }$. Only 20%−30% of galaxies with stellar masses ${M}_{*}\lt {10}^{10.5}$ ${M}_{\odot }$ are quiescent in both environments, whereas 80%–100% of ${M}_{*}\gt {10}^{10.5}$M_⊙ galaxies are quiescent in the protocluster, compared to only ∼40% in the field (Table 1).

We find that the fraction of quiescent galaxies is dependent on environment: CARLA J1753+6311 contains double the quiescent fraction of the control field at z ∼ 1.6. However, this environmental effect is also mass dependent: only the population of high-mass galaxies has an enhanced quenched fraction relative to the control field. These results are consistent with recent literature on red galaxies in clusters at z > 1.5. Rudnick et al. (2012) and Fassbender et al. (2014) found a strong excess of bright, red galaxies in two z ∼ 1.6 clusters, but a corresponding lack of faint, red galaxies. However, in contrast to these results, Andreon et al. (2014b) found a well-populated red sequence down to ∼10¹⁰M_⊙ in a z ∼ 1.8 cluster and Lee et al. (2015) find that there is no difference in the quiescent fraction between cluster and field environments at $z\gt 1$, with a large variation between individual clusters. Therefore the mass dependence of quiescent galaxies needs to be analyzed in a larger sample of protoclusters to draw firm conclusions.

In van der Burg et al. (2013) clusters at z ∼ 1 were also shown to have an increased quiescent fraction compared to the field. The quiescent fractions in CARLA J1753+6311 are similar to the z ∼ 1 fractions at high masses ($M\gt {10}^{10.5}$ M_⊙), which is further evidence that CARLA J1753+6311 is already a very mature structure, more similar to z = 1 clusters than higher redshift protoclusters. The quiescent fraction at lower masses is much higher at z ∼ 1 than in CARLA J1753+6311. This may suggest a build up of the low mass end of the red sequence in clusters from z = 1.6 to z = 1.