INTRACLUSTER LIGHT AT THE FRONTIER: A2744

A substantial fraction of stars in clusters are not gravitationally bound to any particular galaxy. These stars constitute the so-called intracluster light (ICL). The ICL is distributed around the central galaxy of the cluster and extends to several hundred kpc away from the cluster center (e.g., Murante et al. 2004; Zibetti et al. 2005). This diffuse light is thought to form primarily by the tidal stripping of stars from galaxies that interact and merge during the hierarchical accretion history of the cluster (e.g., Gregg & West. 1998; Rudick et al. 2006; Conroy et al. 2007; Contini et al. 2014). Therefore, the characterization of the ICL provides a direct way of determining the assembly mechanisms occurring inside galaxy clusters. In this sense, the ICL is the signature of how violent the assembly of the cluster has been through its cosmic history. For that reason, it is absolutely key to determine how and when the ICL formed. However, the identification of this light observationally remains difficult and uncertain. Indeed, the typical surface brightness of the ICL is μ_V ≳ 26.5 mag arcsec⁻² (e.g., Mihos et al. 2005; Zibetti et al. 2005; Rudick et al. 2006) and it is contaminated by foreground and background galaxies. Moreover, it is difficult to dissociate between the ICL and the brightest central galaxy surface brightness profile (e.g., Gonzalez et al. 2005; Krick et al. 2007).

Although the properties of the stellar populations of the ICL provide a vital tool to understand its formation, little is known about their characteristics. The ICL is reported to be metal-poor in some studies (e.g., Durrell et al. 2002; Williams et al. 2007) while other studies show super-solar metallicities (e.g., Krick et al. 2006). Simulations predict that the ICL forms very late (z < 1; e.g., Murante et al. 2007; Contini et al. 2014) and that the bulk of the ICL is produced by the most massive (M_⋆ ∼ 10^10–11 M_☉) satellites as they fall into the cluster core (e.g., Murante et al. 2007; Purcell et al. 2007; Martel et al. 2012; Contini et al. 2014). If this scenario is correct, the ages and metallicities of the intracluster population will be highly dependent on the progenitor galaxies from which they were stripped. Therefore, the ICL is expected to have a metallicity that is similar to that of these massive satellites. The goal of this paper is to explore this question in detail and characterize, for the first time, the age and metallicity of the ICL of a massive cluster at radial distances R > 50 kpc with unprecedented accuracy.

We achieve our goal by taking advantage of the deepest optical and near-infrared images ever taken by the Hubble Space Telescope (HST) of the Abell Cluster 2744. The Hubble Frontier Fields³ (HFF) project represents the largest investment of HST time for deep observations of galaxy clusters. A2744 is the first cluster observed as part of the HFF program. This structure, at z = 0.3064 (Owers et al. 2011), is a rich cluster (virial mass of ∼7 × 10¹⁵ M_☉ within R < 3.7 Mpc) undergoing a major merger as evidenced by its complex internal structure (Boschin et al. 2006). Therefore, this cluster represents an excellent target for studying the formation of the ICL. The ages and metallicities of the ICL are studied in detail, taking advantage of the inclusion of very deep near-infrared (NIR) data to break the age–metallicity degeneracy (e.g., Anders et al. 2004).

Throughout this work, we adopt a standard cosmological model with the following parameters: H₀ = 70 km s⁻¹ Mpc⁻¹, Ω_m = 0.3, and $\Omega _\Lambda$ = 0.7. At z = 0.3064; this corresponds to a spatial scale of 4.52 kpc arcsec⁻¹.

The data used in this work are based on the first release of the HFF program and include all the NIR WFC3 data (ID13495, PI: J. Lotz and ID13386, PI: S. Rodney) of A2744. The Advanced Camera for Surveys (ACS) images were taken from the HST archive as part of the program ID11689 (PI: R. Dupke) and consist of six orbits in F435W and five orbits each in F606W and F814W. NIR observations include imaging in the four filters F105W, F125W, F140W, and F160W based on 24, 12, 10, and 24 orbits, respectively. The data were directly retrieved from the archive.⁴ The images were reduced by the HFF team following standard HST procedures both for the ACS and the WFC3 data.⁵ For both cameras, flat fields are claimed to be accurate to better than 1% across the detector. A few of the HFF observations in the IR exhibit a time-variable sky background signal due to time variable atmospheric emission. This exposures were corrected from this variable emission and included into the final mosaics (A. M. Koekemoer 2014, private communication).

The mosaics we used consist of drizzled science images with pixel size 006. These mosaics were built using the package Astrodrizzle.⁶ In the case of the WFC3, this pixel size is closer to one-half of an original pixel. While working with subpixel-dithered data has several advantages, such as helping the comic ray rejection step, improving the resolution, and lowering the correlated noise, it also reduces the sensitivity to low surface brightness features. This last problem, however, can be avoided by combining several pixels later on as we have done in our work (see the next section).

The 04 diameter aperture depth at 5σ of each image is: 27.4 (F435W), 28.0 (F606W), 27.1 (F814W), 28.6 (F105W), 28.5 (F125W), 28.7 (F140W), and 28.2 (F160W) mag (Laporte et al. 2014). As we will explain later, the age and metallicity of the ICL are determined using the color information provided by both cameras: ACS and WFC3. Consequently, we will limit our study of the ICL to their common field of view (∼2.0 × 2.2 arcmin²) of the galaxy cluster.

2.1. Surface Brightness Limits

2.2. Optical and NIR Colors

In order to study the stellar populations of the cluster, including its ICL, we constructed two restframe colors at the redshift of the cluster: g − r and i − J. The choice of these colors was made for two reasons: first, to diminish the effects of the different point-spread functions (PSFs) combining images from the same camera (ACS data for g − r and WFC3 data for i − J) and second, to include an NIR filter to constrain the metallicity with better accuracy (e.g., Anders et al. 2004). The colors were built interpolating among the observed filters to obtain the flux in the equivalent restframe bands: g, r, i, and J. We also derived the z band, using the same methodology, to construct the r − z color. This color was used to create a measurement of the stellar mass density of the cluster (see below). We corrected our data of Galactic extinction (E(B − V) = 0.012; Schlafly & Finkbeiner 2011) using a Cardelli et al. (1989) extinction law.

To explore the stellar populations of A2744 and their variation across the cluster structure, we make use of three independent analyses based on different parameters: the restframe (corrected of cosmological dimming) surface brightness in J band, μ_J, the logarithm of the stellar mass density, log (ρ), and the radial distance, R, to the most massive galaxies of the cluster. These three parameters will allow us to compare our results with theoretical expectations.

The μ_J map of the cluster was divided in eight surface brightness bins, from 16 to 25 mag arcsec⁻². To estimate log (ρ), we follow the procedure described in Bakos et al. (2008). First, we link the stellar mass density profile ρ with the surface brightness profile at a given wavelength μ_λ using the mass-to-light (M/L) ratio. This is done with the expression

$$\begin{equation} \log (\rho)=\log (M/L)_{\lambda }-0.4(\mu _{\lambda }-m_{\rm abs\, \odot \lambda })+8.629, \end{equation} \tag{ 1 }$$

where m_abs ☉λ is the absolute magnitude of the Sun at wavelength λ and ρ is measured in M_☉ pc⁻². Second, to evaluate the above expression, we need to obtain the M/L ratio at each radius. Following the prescription of Bell et al. (2003), we have calculated the M/L ratio as a function of color. In this work we have used the restframe color r − z and assumed a Salpeter initial mass function (IMF; Salpeter 1955). The expression we have used to estimate the M/L is

$$\begin{equation} \log (M/L)_{\lambda }=(a_{\lambda }+b_{\lambda }\times {\rm color}), \end{equation} \tag{ 2 }$$

where color = r − z, a_λ = −0.041, and b_λ = 0.463 are applied to determine the M/L in the z band. The log (ρ) map of the cluster was divided also in eight mass density bins, from 10^4.8 to 10^0.8 M_☉ pc⁻².

Finally, we constructed a radial distance indicator using the centers of the most massive galaxies in the cluster as the starting points. These massive galaxies were identified as those galaxies with central surface brightness μ_J < 17 mag arcsec⁻². This surface brightness translates to a density of 10⁴ M_☉ pc⁻² equivalent to the stellar mass densities in the centers of massive ellipticals (e.g., Hopkins et al. 2009). Then, the distance to their centers was calculated as the elliptical distance defined by the morphological parameters of these galaxies, derived by sextractor. The less massive galaxies of the cluster were masked to reduce the contamination of their different stellar populations at a given radius. This radial distance was logarithmic spaced in eight bins, from 0 to 120 kpc. We illustrate the different bins in Figure 1.

Zoom In Zoom Out Reset image size

The reason to choose the above parameters, and their binning, is to characterize the stellar populations of the cluster properly, i.e., averaging zones with similar properties. This allows us to describe the cluster from its inner parts (brighter magnitudes, higher densities) to its outer parts (fainter magnitudes, lower densities) in a consistent way. In doing this, the comparison of the ICL properties with those from the cluster galaxies is direct and homogeneous. Consequently, we will infer the properties of the ICL and cluster galaxies based on their relative differences.

In Figure 1, the three color–color diagrams based on the μ_J, log (ρ), and R maps of the cluster are shown (left panels). A grid of Bruzual & Charlot (2003) models based on a Salpeter IMF for single stellar populations is also plotted. Each of the points of the grid is flagged with its corresponding age and metallicity. For illustration, in the right panels, the color-coded bins of the three parameters are drawn over an image of the cluster in the F160W band.

3.1. Color–Color Diagrams

3.2. Age and Metallicity Gradients

To quantify the changes of the stellar populations across the galaxy cluster, we have computed the age and metallicity of the stellar populations based on their restframe g − r and i − J colors. Within each spatial bin, we measure the age and the metallicity of each pixel corresponding to their g − r and i − J colors. Then, we estimate the corresponding mean age and metallicity. The errors in age and metallicity were also drawn from the bootstrapping simulations used to derive the color errors. The gradients of age (upper panels) and metallicity (bottom panels) for each of the parameters are shown in Figure 2. The colors, ages, and metallicities for the three parameters are listed in Table 2.

Zoom In Zoom Out Reset image size

Table 2. Colors, Ages, and Metallicities with Their Corresponding Errors of the Cluster A2744 as a Function of: μ_J, log (ρ), and Radial Distance

Bin	g − r	i − J	Age (Gyr)	Z
μJ (mag arcsec−2)
16–18	0.840 ± 0.011	0.759 ± 0.017	13.401 ± 1.102	0.028 ± 0.002
18–19	0.825 ± 0.008	0.760 ± 0.012	11.631 ± 1.835	0.030 ± 0.003
19–20	0.800 ± 0.005	0.733 ± 0.010	9.579 ± 1.339	0.029 ± 0.003
20–21	0.776 ± 0.006	0.695 ± 0.007	8.703 ± 1.341	0.026 ± 0.002
21–22	0.764 ± 0.009	0.660 ± 0.006	9.846 ± 1.847	0.021 ± 0.002
22–23	0.735 ± 0.016	0.624 ± 0.004	7.574 ± 2.238	0.021 ± 0.003
23–24	0.728 ± 0.034	0.590 ± 0.009	8.605 ± 3.850	0.018 ± 0.004
24–25	0.711 ± 0.051	0.562 ± 0.017	7.866 ± 3.884	0.018 ± 0.005
log (ρ(M☉ pc−2))
4.8–3.6	0.841 ± 0.010	0.765 ± 0.016	13.281 ± 1.242	0.029 ± 0.003
3.6–3.2	0.822 ± 0.008	0.758 ± 0.010	11.262 ± 1.812	0.030 ± 0.003
3.2–2.8	0.799 ± 0.005	0.729 ± 0.007	9.661 ± 1.187	0.028 ± 0.002
2.8–2.4	0.776 ± 0.006	0.692 ± 0.006	9.052 ± 1.400	0.025 ± 0.002
2.4–2.0	0.764 ± 0.009	0.656 ± 0.006	10.060 ± 1.746	0.021 ± 0.002
2.0–1.6	0.733 ± 0.016	0.622 ± 0.005	7.499 ± 2.210	0.021 ± 0.003
1.6–1.2	0.721 ± 0.034	0.584 ± 0.012	7.959 ± 3.654	0.018 ± 0.004
1.2–0.8	0.721 ± 0.034	0.575 ± 0.017	8.118 ± 3.318	0.018 ± 0.004
Radial Distance (kpc)
0–3	0.813 ± 0.018	0.746 ± 0.022	10.528 ± 2.882	0.030 ± 0.005
3–7	0.773 ± 0.009	0.713 ± 0.007	6.571 ± 1.458	0.030 ± 0.003
7–15	0.793 ± 0.008	0.689 ± 0.006	13.500 ± 0.740	0.021 ± 0.001
15–25	0.776 ± 0.008	0.655 ± 0.007	12.488 ± 1.522	0.019 ± 0.001
25–35	0.760 ± 0.013	0.634 ± 0.006	10.784 ± 2.161	0.019 ± 0.002
35–55	0.734 ± 0.019	0.600 ± 0.005	8.674 ± 2.421	0.018 ± 0.002
55–85	0.691 ± 0.036	0.561 ± 0.011	5.864 ± 2.922	0.019 ± 0.004
85–120	0.678 ± 0.052	0.569 ± 0.023	5.298 ± 3.536	0.023 ± 0.008

Download table as:  ASCIITypeset image

In Figure 2, tentatively, we marked in orange the region corresponding to the ICL component. Observationally, the exact definition of which part of the cluster light describes the ICL is ill defined. There have been many different attempts to label the ICL region using criteria such as the surface brightness or the radial distance. For instance, Zibetti et al. (2005) defined the ICL as the light with μ_r > 25.0 mag arcsec⁻², whereas Toledo et al. (2011) used R > 50 kpc (see also Gonzalez et al. 2005). For each of the parameters used, we have defined the ICL region as the region with either restframe μ_J > 24 mag arcsec⁻², log (ρ) < 1.2, or R > 50 kpc. The μ_J limit defining the ICL region is based on the Zibetti et al. definition and was calculated using i − J = 0.6 (see Figure 1) and r − i = 0.4 (Vazdekis et al. 2012, for solar metallicity and ∼7 Gyr), while the log (ρ) limit is calculated using the above μ_J limit and the color i − J = 0.6. These values are inserted into Equation (1) in Bakos et al. (2008) with the mass-to-light ratio in the J band given by Vazdekis et al. (2012) models for the corresponding age (∼7 Gyr) and metallicity (Z ∼ 0.018) of the ICL region.

According to Figure 2, the metallicity within the cluster is continuously decreasing from super-solar (Z = 0.028 ± 0.003) in the central part of the main galaxies to solar (Z = 0.018 ± 0.007) in the ICL region. The ICL metallicity is similar to that found in the stellar halo of nearby massive galaxies (Coccato et al. 2010; Roediger et al. 2011; Greene et al. 2012; Montes et al. 2014). The stellar population ages show a slight negative gradient toward the outskirts. This gradient is compatible with what is found for the giant elliptical galaxy M87 (Liu et al. 2005; Montes et al. 2014), the elliptical galaxies in Roediger et al. (2011), and for some galaxies in Greene et al. (2012).

State-of-the-art semianalytical models of galaxy formation (i.e., Contini et al. 2014) suggest that the origin of the ICL is the result of the disruption and tidal stripping of massive (10^10–11 M_☉) satellite galaxies infalling in the cluster potential (see also Purcell et al. 2007; Murante et al. 2007; Martel et al. 2012). The largest contribution of these massive satellites to the ICL is understood due to the stronger effect of dynamical friction on these galaxies orbiting the cluster compared to less massive satellites. If this theoretical scenario is correct, we would expect that the mean metallicity of the ICL would be slightly subsolar Z ∼ 0.009–0.014 (Contini et al. 2014; corresponding to the typical metallicities of the galaxies described above). Here we find that the mean metallicity of the ICL in our cluster is solar (Z = 0.018 ± 0.007), but still in agreement with the theoretical predictions. Using the mass–metallicity relation (Gallazzi et al. 2005), the derived ICL metallicity corresponds to the metallicities of galaxies with M_⋆ ∼ 3 × 10¹⁰ M_☉. This mass is only a factor of two different from that of the Milky Way, and the metallicity is similar (6.43 ± 0.63 × 10¹⁰ M_☉, McMillan 2014; Z ≈ 0.02 = Z_☉, Rix & Bovy 2013). Therefore, the ICL of A2744 can be understood as mainly produced by the disruption of galaxies with similar stellar properties as the Milky Way.

Many theoretical works (e.g., Willman et al. 2004; Monaco et al. 2006; Murante et al. 2007) find that the majority of the ICL formed at z < 1. According to Contini et al. (2014), it is only since z = 0.4–0.5 that 50% of the ICL of present-day clusters is in place. We find that the ICL age of A2744 is younger (6 ± 3 Gyr) than the age of the most massive (M_⋆ ≳ 10¹¹ M_☉) galaxies of the cluster. This is consistent with the idea that the galaxies that mostly contributed to the ICL were producing stars during a larger period of time than the most massive galaxies of the cluster. This could happen if the satellite galaxies orbiting the cluster were forming stars until their star formation ceased due to ram pressure stripping of their gas content (e.g., Boselli et al. 2009; Chung et al. 2009). Later on, with their star formation stopped, the stellar populations of these satellite galaxies could be stripped from their progenitors and started to contribute to the ICL. However, it is not straightforward to say exactly when the ICL in A2744 was produced. Nonetheless, we can speculate on the basis of the difference between the age of the most massive galaxies and the one derived for the ICL.⁸ If we assume that the stellar populations of the most massive galaxies were formed at z ∼2–3 (i.e., ∼10–11 Gyr ago), then the ICL of A2744 is compatible with being assembled at z < 1.

The restframe optical color of the ICL of A2744 (g − r ∼ 0.68 ± 0.04) is consistent with the results of Krick et al. (2007) for the ICL of the same cluster (V − r = 1.0 ± 0.8). We also compare the color of the ICL of A2744 with the ICL colors available for other clusters. As mentioned before, there are few studies regarding the properties of the ICL, even colors. We warn the reader that the clusters studied in the literature have different masses and redshifts making the comparison of such properties not straightforward. The derived color for the ICL of A2744 is in agreement with the colors derived for various tidal features in the Virgo Cluster (B − V ∼ 0.75; Rudick et al. 2010) and the outskirts of M87 (g − r ≈ 0.72; Chen et al. 2010, private communication). Covone et al. (2006) also found blue colors (B − V ∼ 1.0 ± 0.5) for two diffuse features in the cluster A2667 at z = 0.233 compared to the colors of the most massive galaxies (B − V ∼ 1.5), although there is a hint of a redder V − I color for the ICL (V − I ∼ 1.0 ± 0.3 for the ICL, V − I ∼ 0.7 for the galaxies).

Finally, another aspect of the ICL that can be directly compared with the theoretical predictions is the fraction of stellar mass of the galaxy cluster that is contained within this component. As we mentioned earlier, the precise definition of what is exactly ICL is observationally ill defined. Here we have used up to three independent parameters to quantify the spatial region corresponding to the ICL: 24 < μ_J < 25 mag arcsec⁻², 0.8 < log (ρ) < 1.2, and 50 < R < 120 kpc. The amount of stellar mass in the spatial regions defined by those constraints were derived as follows. For log (ρ), the derivation of the mass was straightforward, while for μ_J and the radial distance, we used Equation (1) and the corresponding mass-to-light ratio in the J band for the age and metallicity of the ICL region (see Figure 2). The total mass of the cluster, including the ICL region, was obtained similarly but using all the spatial region covered by the light of the cluster explored in this work (i.e., ∼400 × 400 kpc around its core). The amount of total stellar mass in this region of the cluster is: 7.5 × 10¹² M_☉ (μ_J < 25 mag arcsec⁻²), 7.9 ×10¹² M_☉ (log ρ > 0.8), and 4.2 × 10¹² M_☉ (R < 120 kpc, using our definition of radial distance to the centers of the most massive galaxies of the cluster). The ICL mass fractions we got are 5.1% for μ_J, 4.0% for log (ρ), and 10.4% for the radial distance. These ICL mass fractions correspond to the mass contained in 4–6 Milky-Way-type galaxies. Note, however, that these fractions are lower limits to the global contribution of the ICL to the total stellar mass of the cluster. This is due to the fact that we are not considering the whole cluster (but only ∼400 × 400 kpc around its core) and we are also limited by the depth of the optical bands (particularly for estimating log (ρ) based on the r − z color). How do our numbers compare with the theoretical expectations? Simulations predict that the fraction of mass in the ICL of present-day galaxy clusters should be around 10%–40% (Contini et al. 2014). At the redshift of our cluster, z = 0.3, the fraction of stellar mass in the ICL is expected to be around 60% of today's value. Consequently, we assume that the fraction of mass in a cluster as A2744 should be between 6% and 24%. Taking into account that our stellar mass estimates are lower limits, our numbers are in good agreement with theoretical expectations. The simulations are also in nice agreement with other stellar mass fractions measured in other clusters (10%–20%; Feldmeier et al. 2004; Covone et al. 2006; Krick et al. 2006, 2007). Over and over, it seems that the theoretical predictions are reproducing well the observations presented in this work and in other previous papers. We think this is an indication that we are starting to understand the general picture of how the ICL of the galaxy clusters form.

The results presented in this work show the extraordinary power of the Frontier Fields survey to address the origin and evolution of the ICL. Once this survey is completed, it will be possible to explore the properties of the ICL in another five clusters in the redshift range 0.3–0.6, a period of time crucial to understand the formation of this elusive component of the galaxy clusters.

We thank the referee for constructive comments that helped to improve the original manuscript and the HST director and FF teams for their work to make these extraordinary data available. We also thank A. M. Koekemoer for his useful comments in the HFF image reduction process. This research has been supported by the Spanish Ministerio de Economía y Competitividad (MINECO; grant AYA2010-21322-C03-02).

In order to estimate a possible contamination of the colors in the outer regions of the cluster due to the scattered light produced by the different band PSFs, we run the following tests. We have created an artificial galaxy with an effective radius of 5 kpc. This value is typical of galaxies with 10¹¹M_☉ (see, e.g., Shen et al. 2003; Cebrián & Trujillo 2014). The mock galaxy follows a de Vaucouleurs r^1/4 profile (de Vaucouleurs 1948). The surface brightness profile of the simulated galaxy was convolved with the different HST PSFs. As we are interested in exploring the effects caused by the PSF wings in the outer region of our profiles, our PSFs were retrieved from the HST PSF modeling software tinytim. We retrieved from the tinytim webpage the largest PSF available for each band ⁹. This allows us to use PSFs that extends at least over 125 (i.e., 55 kpc at the cluster redshift). We also simulated the effect of adding an ICL profile, an exponential profile with scale length of 18 kpc and μ_J = 24.5 mag arcsec⁻² at R = 50 kpc.

The original model profiles and the convolved PSF models are shown in Figures 3 (without ICL) and 4 (with ICL). As input colors for the model profiles we have used g − r = 0.85 and i − J = 0.75, which are typical colors of the inner regions of the galaxies. In Figure 3, we find that the effect of the different PSFs and their wings in the constructed radial colors is less than 0.03 and 0.02 mag in the g − r and i − J colors, respectively, at the region of the ICL (R > 50 kpc). However, in Figure 4 where the ICL is included, we find that the smoother profile of the ICL diminishes the effect of the PSF and the colors differ by less than 0.01 mag from the simulations. We warn the reader that these are the results of a simple simulation. We only consider the effect of one galaxy, and the ICL region could be affected by the scattered light of other galaxies of the cluster. Consequently, the only purpose of these simulations is to provide a hint about how relevant the PSF effect could be on the ICL region.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size