THE X-RAY ZURICH ENVIRONMENTAL STUDY (X-ZENS). I. CHANDRA AND XMM-NEWTON OBSERVATIONS OF ACTIVE GALACTIC NUCLEI IN GALAXIES IN NEARBY GROUPS

We carried out Chandra/ACIS-I observations of 12 ZENS galaxy groups in Cycle 11 (PI: J. Silverman; proposal # 11700688; 120 ks). The observations were executed between 2009 September and 2010 October. Each target is observed for close to 10 ks in order to detect AGNs at z ∼ 0.05 down to a limiting luminosity of L_{(0.5 − 8 keV)} ∼ 3 × 10⁴⁰ erg s⁻¹ for detections with at least 4 counts in the broad energy band 0.5–8 keV. The 169 × 169 field-of-view (FOV) of CCDs $\#$0-3 of ACIS-I is sufficient to cover the sky area of the target galaxy groups. In total, there are 115 galaxies within the 12 ZENS groups that fall within the ACIS-I FOV. The aim-points of Chandra were chosen to maximize the number of galaxies that fall within the ACIS-I FOV. This resulted in offsets as given in Table 2 from the group centers listed in Table 1. We also tried to avoid having galaxies falling within or near chip gaps; this was accomplished by adjusting the pointing location once the planned observation date was set, and thus the roll angle was known. In other cases, this was needlessly achieved by splitting the observation into smaller time intervals and applying small offsets (<1') to the aim point. In Table 2, we provide details on the individual exposures.

We use CIAO tools (Version 4.3 and CALDB version 4.4.6) to perform the data analysis. In cases where multiple exposures were taken, we use the task $merge\_all$ to combine the individual frames that generate a summed event file (level 2). Source counts are measured in three energy bands (Broad (B): 0.5–8.0 keV, Soft (S): 0.5–2 keV, Hard (H): 2–8 keV) by placing circular apertures at the location of galaxies within our groups. The extraction radius for each galaxy is set to include close to 100% of the flux while the background contribution is estimated in an annulus centered on each individual galaxy with an area greater than the source region and free of any other nearby X-ray sources. We consider a positive detection as any source with greater than or equal to four net counts in any of the three energy bands. This is the same significance threshold employed by the Bootes survey (Kenter et al. 2005) that has demonstrated that there is a low false-positive detection rate even at these low count levels. We use the CIAO tool "aprates" to estimate count rates and associated 1σ errors. Exposure maps, required for conversion of count rates to physical fluxes (i.e., photons cm⁻² s⁻¹), are generated in the three bands that correct for off-axis effects such as vignetting. Because of the fact that $merge\_all$ does not properly deal with individual frames having a pointing offset, an exposure map was created for each individual obs_id (mkinstmap, mkexpmap), reprojected ($reproject\_image$) to a common astrometric frame and then coadded ($dm\_merge$). These broadband maps are based on a spectral weighting scheme that assumes a power-law spectral energy distribution (SED) with a photon index of 1.7. Broadband fluxes are then determined by multiplying the photon flux by the mean energy of a powerlaw distribution of photons with the spectral index as given above. In Table 3, the measurements are provided for the individual Chandra detections. The magnitude of the error on the measured counts can be used to estimate uncertainties on flux and luminosity. We note that there will be an additional error associated with the conversion of counts to energy units since we do not have full knowledge of the spectral shape of the individual detections due to the small number of counts.

Table 3. Chandra Photometry of ZENS Galaxies

Name	R.A. (X-Ray)	Decl. (X-Ray)	Off-axis	Counts			fXa	LXb	Class
	(J2000)	(J2000)	Angle (')	B	S	H
s1571_10	02:36:55.95	−25:28:39.4	4.99	$6.1_{-2.4}^{+3.1}$	$3.8_{-1.1}^{+2.4}$	$2.3_{-1.5}^{+2.2}$	−14.2	40.7	AGN
s1671_4	22:23:54.06	−30:00:46.3	2.41	$10.2_{-3.9}^{+3.9}$	$8.8_{-3.2}^{+3.2}$	$1.4_{-1.4}^{+2.3}$	−13.9	40.9	Galaxy
s1752_4	22:21:10.59	−26:00:24.6	1.40	$11.5_{-3.9}^{+3.9}$	$9.8_{-3.1}^{+3.6}$	$1.8_{-1.6}^{+2.1}$	−13.9	41.0	AGN
n1347_3c	09:59:44.04	−05:22:04.9	3.46	$12.0_{-4.0}^{+4.6}$	$5.3_{-2.4}^{+3.0}$	$6.8_{-3.0}^{+3.7}$	−13.8	41.0	Galaxy
n1381_3	14:28:06.04	−02:31:27.32	2.86	$4.8_{-2.8}^{+3.3}$	$2.8_{-1.8}^{+2.3}$	$2.0_{-1.8}^{+2.2}$	−14.4	40.6	AGN
n1598_5	14:36:16.89	−01:23:13.65	8.15	$9.1_{-3.2}^{+3.8}$	$5.8_{-2.4}^{+3.0}$	$3.2_{-2.0}^{+2.6}$	−13.9	40.9	AGN
n1702_3d			4.72	$4.6_{-3.3}^{+2.7}$	$0.7_{-0.7}^{+1.8}$	$3.9_{-2.4}^{+2.9}$	−14.2	40.5	AGN
n1381_add	14:28:08.90	−02:31:24.68	3.18	$80.1_{-9.0}^{+9.1}$	$15.3_{-3.7}^{+4.3}$	$64.8_{-8.1}^{+8.2}$	−13.0	41.8	AGN

Notes. ^aThe log of the flux in units erg cm⁻² s⁻¹ in the 0.5–8.0 keV band. ^bThe log of the luminosity in units erg s⁻¹ in the 0.5–8.0 keV band. ^cX-ray emission is slightly extended and X-ray source is not found by wavdetect; given position indicates peak X-ray emission. ^dNo position is given since source is not detected by wavdetect with a significance above our threshold. The optical position was used for count extraction.

Download table as:  ASCIITypeset image

We further run wavdetect on each combined image in a given energy band. This enables us to both validate the significance of the source detections with aperture estimates, as mentioned above, and to construct images devoid of X-ray point sources, required for the detection of diffuse emission (see F. Miniati et al., in preparation for details). A significance threshold of 5 × 10⁻⁶ was set to detect sources with low count statistics (N ∼ 2–5) while minimizing the inclusion of false detections. The search was set to spatial scales of 1, 2, 4, 8 × 196 that are identical to those used by the Chandra survey of the Bootes field (Kenter et al. 2005).

We also acquired X-ray imaging of nine ZENS groups with XMM-Newton (PI: F. Miniati; proposal #065530, #067448, #069374). Of these, seven are of sufficient quality for our study of point-like emission as listed in Table 1 (flaring significantly degraded the quality of the data for the remaining two groups, thus rendering them as ineffective for our analysis). The total exposure times for each observation, even after removing bad time intervals due to flaring, are sufficient to reach depths comparable to the Chandra observations reported above. This is due to the fact that the XMM observations were taken with the aim of providing sufficient count statistics to accurately measure the physical properties of the DIM. While four additional groups have some archival X-ray coverage with XMM-Newton, they are highly offset from the center position of our groups and miss most of the galaxy members. One of them (2PIGG_n1377) suffers from flaring and is not useful even though a single bright X-ray source is present that is associated with a galaxy (2PIGG_n1377_18) belonging to the group. Out of these seven groups, only one (2PIGG_s1571) is in common with the Chandra sample. In fact, this group clearly has diffuse emission detected with both instruments as presented in F. Miniati et al. (in preparation). The XMM-Newton observation further aids in the detection of a X-ray point source associated with the central galaxy in 2PIGG_s1571 that was uncertain from the Chandra observation. This highlights the complementarity of Chandra and XMM.

For point-source detection, we use the 0.5–2 keV and 2–7.5 keV bands while masking those energy intervals impacted by strong instrumental effects, as in Finoguenov et al. (2007). A detailed modeling of the unresolved background, foreground and detector background has been undertaken, following the prescription of Bielby et al. (2010). We use 4'' pixels and run the wavelet image reconstruction on 8'' and 16'' scales, with a 4σ detection threshold. We produce catalogs, based on the detections in two energy bands. Physical flux units are obtained by converting the count rate (adjusted to account the flux losses within the 15'' flux extraction circle due to XMM PSF) using PIMMS with a power law photon index of Γ = 1.7, and no correction for intrinsic absorption. We then cross-matched our X-ray catalog to the positions of the ZENS galaxies. X-ray luminosities in the broadband 0.5–8.0 keV are used throughout. In Table 4, the measurements are provided for XMM-Newton point-source detections. As for the Chandra sources, the errors on flux and luminosity can be propagated from the error on count rate. In total, we have X-ray coverage of 177 unique galaxies that fall within either the Chandra or XMM footprints.

The level of X-ray emission in ZENS galaxies is bordering on the overlap between accretion onto a black hole, stellar remnants and diffuse hot gas. In addition to AGN, point-like X-ray emission may be attributed to the cumulative contribution of low-mass and high-mass X-ray binaries. There is additionally room for a component of thermal emission in early-type galaxies. The high-mass X-ray binary population is likely to contribute significantly only to those galaxies with high star formation rates. There are many studies (e.g., Ranalli et al. 2003; O'Sullivan et al. 2003; Lehmer et al. 2010; Boroson et al. 2011) of X-ray emission from normal galaxies (i.e., without AGN) that enable us to determine whether the level of X-ray emission is characteristically higher than that expected for galaxies of a given luminosity, stellar mass, and star formation rate. While we are mainly concerned with nearby galaxies, such relations between X-ray emission and galaxy type appear to be universal and extend over a wide range of redshift (Lehmer et al. 2007, 2008).

We first investigate the nature of the X-ray emission by splitting the galaxy samples into either star-forming or quiescent as classified by the presence of prominent emission lines in the optical spectra combined with NUV-NIR colors (Cibinel et al. 2013a). The diversity of galaxy spectral type within the 18 X-ZENS groups can be seen in Figure 3 where we plot the sSFR as a function of their stellar mass M_galaxy as done in Figure 8 of Cibinel et al. (2013a). It is worth pointing out that the ZENS sample does have incompleteness at low masses that cannot be neglected; the lack of galaxies with M_galaxy ∼ 10⁹ M_☉ and sSFR between ∼10⁻¹¹–10⁻¹² yr⁻¹ is clearly evident. In the figure, we indicate with an open circle those galaxies that are detected in X-rays. There is a wide spread in the distribution of sSFR at a constant stellar mass for galaxies with X-ray emission, with X-ray sources associated with galaxies even at the extreme ends, i.e., either strongly star-forming or quiescent. Within the limited statistics, the relative numbers of quiescent versus star-forming galaxies that emit X-rays reflects the relative number density of the underlying galaxy population. While the hosts span a wide range in mass from 10⁹ M_☉ to above 10¹¹ M_☉, there is a preference for massive galaxies that one would expect given the well-established increase of AGN activity as a function of stellar mass (e.g., Kauffmann et al. 2003; Silverman et al. 2009b; Haggard et al. 2010). The significance of a dependence of AGN activity on the stellar mass of their hosts is further addressed in Section 5.3.

Zoom In Zoom Out Reset image size

In Figure 4, the X-ray luminosity is shown as a function of star-formation rate for galaxies classified as either highly or moderately forming stars (panel (a)) and as a function of B band luminosity for the quenched population (panel (b)). For each ZENS galaxy without an X-ray detection, we indicate an upper limit on the broadband X-ray luminosity. For those galaxies covered by Chandra, this is equivalent to our limit on a source detection of four counts at the position on the detector thus accounting for variations in the effective area as a function of off-axis angle. The upper limits on the XMM-Newton non-detections are based on a flux level at 2σ above the background. For reference, we provide the typical scaling relations for normal galaxies (Ranalli et al. 2003; O'Sullivan et al. 2003) between these quantities plus an interval of ±1σ to indicate the typical spread.

Zoom In Zoom Out Reset image size

With respect to 111 star-forming galaxies (Figure 4(a)) falling within the Chandra and XMM-Newton coverage, we find that 5 out of 10 galaxies with X-ray detections are above the relation for normal, star-forming galaxies, by at least 2σ deviation, thus likely indicative of an AGN contribution effectively boosting the X-ray luminosity above our detection limits. From the distribution of SFRs of the X-ray non-detections, it is clear that the majority of the galaxies in ZENS have SFRs low enough that if we do detect an X-ray source, it is most probably due to an AGN. We have detected with XMM–Newton X-ray emission from three galaxies falling within the region (±1σ) of a normal star-forming galaxy; based on their optical emission-line properties (see below), all of these are indeed likely to have a contribution to the X-ray emission from an AGN.

We highlight the fourth such galaxy (2PIGG-n1598_5) falling within the locus of normal star-forming galaxies. It has the highest SFR of all ZENS galaxies observed by Chandra. With a sSFR 2.6 × 10⁻¹⁰ yr⁻¹ at its stellar mass of M_galaxy = 5.9 × 10¹⁰ M_☉, it has been identified as the central galaxy (with evidence for a strong bar) within its group. We can make use of the high resolution imaging of Chandra and quality optical WFI-ESO imaging from ZENS to determine if the X-rays are coming from the nuclear region or the overall extent of the galaxy. In Figure 5, we show the X-ray and optical images of 2PIGG-n1598_5. It is clear that the X-ray emission is co-spatial with the nucleus. The emission line ratios of [N ii]/Hα and [O iii]/Hβ, seen in an SDSS spectrum (Figure 5 (bottom)), place this object right on the border between AGNs and star-forming galaxies as defined by Kewley et al. (2006), although, the [S ii]/Hα ratio might be in principle more typical of H ii regions. We can also use the Hα emission (detected within the SDSS fiber that covers only the very central region of the galaxy due to the fiber diameter of 3'') to make a more accurate assessment on the upper limit to the level of star formation in the nuclear region cospatial with the X-ray emission. Based on a Hα luminosity of 8.4 × 10³⁹ erg s⁻¹, we estimate the star formation rate of the central region of the galaxy to be around 0.07 M_☉ yr⁻¹ (Kennicutt 1998), substantially lower than that of the entire galaxy and not likely to produce the amount of X-rays observed by Chandra. Therefore, we conclude that a low-luminosity AGN with L_X = 8 × 10⁴⁰ erg s⁻¹ is powering the X-ray emission seen in 2PIGG-n1598_5.

Zoom In Zoom Out Reset image size

Out of the 66 quiescent galaxies observed in X-rays, there are 12 detections with 10 of them falling with a region of the L_X − L_B plane spanned by normal galaxies (see Figure 4(b)). These galaxies need not necessarily have an AGN to explain their level of X-ray emission as was primarily the case for the star-forming population. As done above, we can measure the spatial extent of the X-ray emission, based on Chandra imaging and determine whether it is extended beyond that expected based on the point-spread function (PSF) at the given position on the detector since the size of the PSF is a strong function of the off-axis angle. Based on the three objects having Chandra imaging, we find that two (2PIGG_n1347_3 and 2PIGG_n1671_4) are clearly extended while 2PIGG_s1752_4 has a spatial count distribution similar to that expected due to an unresolved point source. Therefore, we are able to classify the latter as an AGN.

As an independent test to further solidify the presence of an AGN, we make use of the optical emission line properties of ZENS galaxies using the 2dFGRS spectra. We specifically measure the emission line ratios [N ii]λ6584/Hα, and [O iii]λ5007/Hβ if all lines are detected. In Figure 6, we show the distribution of their ratios as compared to emission-line galaxies from SDSS. For many galaxies in our sample, the Hα and [N ii] emission lines are clearly present while the Hβ and [O iii] lines are very faint, primarily for the quiescent galaxy population. In Figure 7, we provide examples of the 2dFGRS spectra of galaxies, classified in ZENS as either star-forming or quiescent, that have X-ray emission placing them within the locus of normal galaxies (Figure 4), and optical line ratios typical of an AGN. A limitation is that we cannot separate Seyferts and LINERS, where the latter have been heavily debated in the literature as to whether their line ratios are indicative of AGN photoionization or that due to post-AGB stars (Sarzi et al. 2010; Yan & Blanton 2012). A reasonable working hypothesis, mainly pertaining to quenched galaxies, is that a high [N ii]/Hα ratio is an indication of an AGN, irrespective of the Hβ and [O iii] line strengths. In Figure 6, we find that 10 of the 13 galaxies for which we could measure a [N ii]/Hα emission-line ratio are unlikely to be typical star-forming galaxies, given their location relative to the demarcation line of Kauffmann et al. (2003). The optical spectra of the remaining nine galaxies with X-ray detections have either very weak or non-existent Hα or [N ii]λ6584 emission.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Summarizing, our final decision on whether X-ray emission is attributed to an AGN is based on three lines of evidence. First and foremost, if there is X-ray emission substantially higher than that expected from normal star-forming or quiescent galaxies. Second, if the X-ray emission is not diffuse as determined with Chandra resolution (otherwise, if diffuse, it is attributed to thermal emission). Third, we consider the optical emission line ratios to aid in the discrimination between photoionization indicative of an AGN and UV emission from young stars. Taking these together, we find that the 16 of 22 X-ray sources are associated with emission from an AGN. Those that are not AGN are mostly quiescent galaxies with either noticeable thermal emission or unresolved stellar remnants. Our detection rate for AGNs is consistent with the steeply rising faint end of the local AGN X-ray luminosity function (XLF; Sazonov & Revnivtsev 2004; Ueda et al. 2011), for which our sample spans L_X ∼ 0.4–6.6 × 10⁴¹ erg s⁻¹, and with the area coverage of the X-ZENS observations.

It is important to recognize that the X-ray emission may be of a composite nature with AGN, stellar or thermal processes all making some contribution to the total X-ray emission. This may provide a boost in the number of X-ray detections as expected from normal galaxies, complicate the selection of AGN, and hamper the determination of a pure AGN luminosity used to determine a bolometric quantity (Section 5.4). With the limited sample in hand, there is little that can be done to quantify sufficiently such effects. A larger sample will allow us to determine at what luminosities does such dilution of the X-ray emission become problematic. There may even be X-ray data in the archive of local galaxies where this can be addressed more effectively than with the ZENS sample.

Finally, note that we detect X-rays from two (satellite) galaxies with low stellar masses M_galaxy ∼ 10⁹ M_☉ (see Figure 3). If the X-rays are attributed to black hole accretion, the black holes are likely to be of low mass M_BH ∼ 10⁶ M_☉ with Eddington rates below 10⁻² (see Figure 13, obtained assuming a typical bulge-to-disk ratio of ∼0.5 and local scaling relations; Häring & Rix 2004). Recently, Schramm et al. (2013) have reported on the discovery of three low-mass galaxies with M_galaxy < 3 × 10⁹ M_☉ in the Chandra Deep Field South survey that emit X-rays which are likely attributed to AGN activity (with M_BH ∼ a few × 10⁵ M_☉). Such low black holes mass and accretion rate regimes are still largely unexplored (see Reines et al. 2013, for a recent effort using optical selection); this shows the potential for opening them to detailed studies of X-ZENS-like surveys extended however to larger group samples.

We first determine whether galaxies in groups with X-ray luminosities above our detection threshold are preferentially hosted by any particular type of galaxy host. The left panel of Figure 8 shows, as a function of galaxy stellar mass, the bulge-to-total ratio B/T, when available, as derived from double-component fits to the bulge and disk light distributions. These were described with, respectively, a general Sérsic profile (Sérsic 1968) and an exponential profile (see Cibinel et al. 2013b). The whole ZENS galaxy sample is reported with black points, and in red are highlighted galaxies with an X-ray detected AGN.⁶ Galaxies with a purely elliptical morphology are placed at a bulge-to-total value of 1. For reference, we show the median ratio B/T for galaxies with double component fits in bins of stellar mass (10⁹ M_☉ < M < 10^11.5 M_☉; Δlog M = 0.5 M_☉). For galaxies which have both a bulge and a disk component, the right panel of the same figure shows the disk scale-length h plotted versus the half-light radius of the bulge, r_{1/2, Bulge}. Interestingly, the AGN hosts appear to be on average slightly "under-bulged" relative to the galaxy population at similar stellar masses (i.e., the AGN hosts systematically lie in the bottom half of the B/T versus galaxy stellar mass relation); on the other hand, the bulges of AGN hosts appear to be similarly "embedded" within their surrounding disks than the global galaxy population at similar mass scales (i.e., they are distributed evenly around the best fit to the h versus r_{1/2, Bulge} relation. Possible explanations are that, at a given mass, the hosts of AGNs have either relatively brighter/denser disks or relatively fainter/diffuse bulges than their non-active relatives.

Zoom In Zoom Out Reset image size

We measure the halo occupation distribution (HOD) of AGNs in the X-ZENS groups. With 16 AGNs distributed in 18 groups, the mean number of AGNs with L_X > 10⁴⁰ erg s⁻¹ per group is 0.89 ± 0.22 (with the error based on Poisson number counts). We compare the number of AGNs detected in our groups to those expected based on the XLF of the global AGN population (Ueda et al. 2011), to establish whether an over-density of AGNs is seen within group-sized potentials. We restrict the analysis to AGNs found in the Chandra observations of the 12 groups, since we implemented an identical X-ray count threshold as in the observations of the Bootes survey (Kenter et al. 2005), and we can thus make use of the well-established sky area curve as a function of X-ray flux sensitivity. We find a surface density of 9.2 AGNs per square degree considering X-ray detections within 85 of the aim point. We then integrate the XLF over the narrow redshift range 0.05 < z < 0.06 and over X-ray luminosities between 10^40.3–10⁴⁶ erg s⁻¹. Based on the XLF, we would expect to detect 3.1 AGNs per square degree with redshifts and luminosities within these ranges. The enhancement of AGNs of a factor ≈3 relative to the XLF-based estimates is in good agreement with the typical galaxy over-densities of the X-ZENS groups, as shown in Figure 1, indicating no extra AGN activity in groups relative to the global population.

Alternatively, we can compare the fraction of galaxies hosting an AGN to published studies at similar redshifts and over a range of environmental conditions. For massive clusters, Martini et al. (2006) find that the AGN fraction of galaxies with M_R < −20 is ∼5% ± 1.5%. Using a similar magnitude selection and scaling our results to match their coverage of the AGN XLF, we measure an AGN fraction of 8.2% ± 2.1%. While there might be slight evidence for an enhancement of the AGN fraction in groups relative to the more massive clusters (fully compatible with a similar analysis presented in Arnold et al. 2009), a larger X-ray selected AGN sample in ZENS groups is clearly needed to statistically substantiate such a claim. For a comparison to less dense environments, we find that the AGN fraction of galaxies in ZENS groups (∼2%; 1 out of 54 galaxies with −21 < M_i < −20) is compatible with estimates of the AGN fraction of galaxies in the field (Haggard et al. 2010); such a narrow selection of absolute magnitude is required to compare to their results (i.e., sample 2 in their study).

We then determine how the AGN HOD of X-ZENS groups compares to higher redshift measurements. Allevato et al. (2012) provide the AGN distribution in X-ray selected groups in COSMOS up to z ∼ 1. Despite the different selection criteria for the COSMOS and X-ZENS samples, the halo mass ranges are very similar thus allowing us to make such a comparison and look for evolutionary trends. For this exercise, we assume that the shape of the 0.5–8 keV XLF does not change with redshift, and that the faint-end slope is a strict power law below L_X ∼ 10⁴² erg s⁻¹. The first assumption is supported by observational evidence up to z ∼ 1 (e.g., Hasinger et al. 2005; Silverman et al. 2008; Aird et al. 2010), where the evolution has been shown to be essentially a global shift in luminosity (i.e., pure luminosity evolution). The second assumption is based on an extrapolation of the known XLF. We use the XLF of Ueda et al. (2011) to derive a scale factor which accounts for the difference in flux (or luminosity) sensitivity between the COSMOS and ZENS X-ray data sets. A scaling of the AGN HOD of 12.8 is required to match the two samples down to the luminosity limit of our X-ZENS data.

In Figure 9, we plot the mean number of AGNs per halo as a function of redshift and fit the data with a function having terms for the normalization and redshift evolution. We find the following relation with best-fit parameters and 1σ errors:

$$\begin{equation} \langle N_{{\rm AGN}}\rangle =0.80_{-0.43}^{+0.89}\times (1+z)^{3.99\pm 2.69}. \end{equation} \tag{ 1 }$$

Even considering the uncertainties, we find that there is very good agreement between the two samples as indicated by the redshift evolution of the AGN HOD, shown by the solid curve (Figure 9). Furthermore, the rate of evolution is practically identical to the global XLF of Ueda et al. (2003) which has evolution parameterized as (1 + z)^4.2, which is also well reproduced by other determinations for the AGN population (Silverman et al. 2008; Ebrero et al. 2009; Aird et al. 2010). We conclude that the rate of decline with redshift of AGN activity in galaxy groups is similar to that of the global AGN population. The origin of this similarity may rest on either a predominance of AGN activity in halos at the M_group ∼ 10¹³–10¹⁴ M_☉ mass scale, as supported by clustering analyses of AGNs, or on the independence of AGN evolution on halo mass.

Zoom In Zoom Out Reset image size

Next, we investigate whether AGN activity is typical of central or satellite galaxies within groups, and whether it is enhanced in the cores or outskirts of the typical ∼10¹³ M_☉ galaxy groups that we probe in X-ZENS. There is evidence in a number of other studies (e.g., Ruderman & Ebeling 2005; Martini et al. 2007; Martel et al. 2007; Fassbender et al. 2012) that AGN show a preference for the inner regions of massive clusters, at halo masses larger than the X-ZENS values, and that AGNs may possibly be associated with the brightest cluster members. There are however other studies which report opposite results, i.e., even possible AGN excesses in the outskirts of galaxy clusters (e.g., Gilmour et al. 2009; Pentericci et al. 2013). At the lower halo mass scales of our sample, cosmological simulations predict that AGN activity should be more closely tied to central galaxies as opposed to satellites (Richardson et al. 2013).

We quantify the fraction of galaxies, split as central and satellites, that host an AGN over the total mass range 10.4 ⩾ M_galaxy ⩽ 11.6. We find that 5 of 14 central galaxies host an AGN ($0.36_{-0.11}^{+0.14}$), and only 4 out of 30 satellite galaxies host an AGN ($0.13_{-0.04}^{+0.09}$). Thus, although there is a hint for an enhancement by a factor of ∼3 for AGNs in centrals relative to satellites, as predicted by the simulations, this is detected in our data at a low statistical level of significance. A clearer effect is however seen when we investigate the relative frequency of AGNs in centrals and satellites in narrower stellar mass bins, as listed in Table 7. In particular, at galaxy mass scales <10¹¹ M_☉, there is a clear effect for centrals hosting AGNs about four times more frequently than satellites of similar mass, as shown in Figure 10. This result is in agreement with clustering analysis of optically selected narrow-line AGNs at low redshift in SDSS (Li et al. 2006).

Zoom In Zoom Out Reset image size

Table 7. AGN Statistics in ZENS Galaxies

Type	Mass Range	No. of Galaxies	No. with AGNs	AGN Fraction
Central	10.4–11.6	14	5	$0.36_{-0.11}^{+0.14}$
	10.4–11.0	7	3	$0.42_{-0.14}^{+0.19}$
	11.0–11.6	7	2	$0.28_{-0.11}^{+0.20}$
Satellite	10.4–11.6	30	4	$0.13_{-0.04}^{+0.09}$
	10.4–11.0	28	3	$0.11_{-0.04}^{+0.08}$
	11.0–11.6	2	1	$0.50_{-0.25}^{+0.25}$

Download table as:  ASCIITypeset image

In Figure 11, we furthermore explore whether there is any differential effect with halo-centric distance in the frequency of AGNs in satellites. The figure shows halo-centric distance R in units of the group characteristic radius ${\hat{R}}_{200}$ (see Carollo et al. 2013, for an explicit definition of this parameter) as a function of stellar mass. The majority of the detected AGNs (11/15), as shown by the star symbols, are within $R/{\hat{R}}_{200}<0.5$. This may however be possibly explained by the number of galaxy and galaxy mass variations with halo-centric distance (including differences between centrals and satellites). Indeed, a 2-D K-S test returns no significant difference between the radial distribution of AGNs and galaxies at stellar masses above 10^10.4 M_☉. Also, a 2-D KS test on the stellar mass distribution of galaxies targeted in X-rays and those hosting AGNs shows that these distributions are dissimilar at the 3.3σ level, as shown in Figure 12: AGNs are indeed more prevalent in massive galaxies (log M_galaxy ≳ 10.3), as independently seen in other studies.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Finally, we assess in what part of parameter space our AGNs fall in terms of their likely black hole masses and Eddington rates. To do so, we use the local scaling relation between black hole mass and the stellar mass of its host galaxy (Häring & Rix 2004). Bolometric luminosities are derived by applying correction factor of 20 to the broadband X-ray luminosity as applicable to X-ray selected AGNs in COSMOS (Lusso et al. 2012), with a caveat that it is an area of debate whether this factor is appropriate for low luminosity AGNs. In Figure 13, we find that the majority of our sample falls well below a Eddington rate with L/L_Edd ∼ 10⁻⁴. This is two to three orders of magnitude below more luminous quasars such as those in SDSS (e.g., Shen et al. 2011; Kelly & Shen 2013) and AGNs in COSMOS (Trump et al. 2009; Lusso et al. 2012), i.e., much lower than these latter species, even assuming a large correction factor to the normalization value used above. Disentangling in Figure 13 quenched galaxies (with sSFR < 10⁻¹¹ yr⁻¹) from star forming galaxies, and centrals from satellites, it is clear that the X-ZENS AGNs with the lowest Eddington ratios (<10⁻⁴) are hosted at the high mass end of the galaxy population. Already, our small samples show that such "starved" AGNs can occur both in central and satellite galaxies (albeit with a possible preference for centrals), and in quenched as well as star forming hosts (albeit with a preference for quenched systems).

Zoom In Zoom Out Reset image size