XDEEP2: the Chandra X-ray point-source catalog

The Chandra X-ray point-source catalog in the DEEP2 Galaxy Redshift Survey fields


We present the X-ray point-source catalog produced from the Chandra Advanced CCD Imaging Spectrometer (ACIS-I) observations of the combined  deg DEEP2 (XDEEP2) survey fields, which consist of four –1.1 deg fields. The combined total exposures across all four XDEEP2 fields range from ks–1.1Ms. We detect X-ray point-sources in both the individual ACIS-I observations and the overlapping regions in the merged (stacked) images. We find a total of 2976 unique X-ray sources within the survey area with an expected false-source contamination of sources (%). We present the combined log N – log S distribution of sources detected across the XDEEP2 survey fields and find good agreement with the Extended Chandra Deep Field and Chandra-COSMOS fields to . Given the large survey area of XDEEP2, we additionally place relatively strong constraints on the log N – log S distribution at high fluxes (), and find a small systematic offset (a factor ) towards lower source numbers in this regime, when compared to smaller area surveys. The number counts observed in XDEEP2 are in close agreement with those predicted by X-ray background synthesis models. Additionally, we present a Bayesian-style method for associating the X-ray sources with optical photometric counterparts in the DEEP2 catalog (complete to ) and find that 2126 (%) of the 2976 X-ray sources presented here have a secure optical counterpart with a % contamination fraction. We provide the DEEP2 optical source properties (e.g., magnitude, redshift) as part of the X-ray–optical counterpart catalog.

Subject headings:
galaxies: active – surveys – X-rays: galaxies

1. Introduction

Understanding the role of active galactic nuclei (AGN) in galaxy evolution is a major focus in present day astrophysics. It is now becoming increasingly clear that, despite their vastly differing size-scales, the evolution of massive host galaxies and the growth of their central supermassive black holes (SMBHs) may not be independent events (e.g., Boyle & Terlevich 1998; Hopkins et al. 2006; Silverman et al. 2009; Hopkins et al. 2008; Smolčić et al. 2009). Indeed, AGN activity and galaxy properties, such as luminosity, color and morphology, are shown to evolve with time. The redshift range –2 is a crucial epoch: (1) galaxies are evolving strongly as a function of stellar mass (e.g., Zheng et al. 2009; Franceschini et al. 1999; Serjeant et al. 2010); (2) AGN activity is prevalent (e.g., Ueda et al. 2003; Hasinger et al. 2005; La Franca et al. 2005; Barger et al. 2005; Richards et al. 2006); (3) massive clusters are forming (e.g., Lidman et al. 2008; Hilton et al. 2009; Papovich et al. 2010; Fassbender et al. 2011; Bauer et al. 2011; Mehrtens et al. 2012; Nastasi et al. 2011) and (4) the red sequence is becoming established (e.g., Bell et al. 2004; Faber et al. 2007; Willmer et al. 2006; Brand et al. 2005; Domínguez Sánchez et al. 2011). To unambiguously determine the dominant physical processes that are driving the growth and evolution of galaxies and their central black holes requires sensitive, wide-field spectroscopic surveys of AGN.

Sensitive blank-field X-ray surveys arguably provide the most efficient selection of AGN that is unbiased to moderate-to-high obscuration, and in general is not readily contaminated by host-galaxy emission. Indeed, as star-formation is relatively weak at X-ray energies (; Moran et al. 1999; Lira et al. 2002), selection of AGN at these wavelengths can identify many of the most low-luminosity and/or obscured systems (e.g., Fukazawa et al. 2001; Done et al. 1996; Risaliti et al. 1999; Matt et al. 1996; Maiolino et al. 1998; Georgantopoulos et al. 2009). By harnessing the unprecedented angular resolution provided by the Chandra X-ray Observatory, both deep and wide-field X-ray surveys have been instrumental in our current understanding of AGN evolution (e.g., Kenter et al. 2005; Nandra et al. 2005; Worsley et al. 2005; Brandt & Hasinger 2005; Brand et al. 2006; Hasinger et al. 2007; Laird et al. 2009). To date, the two deepest X-ray surveys are the pencil-beam ( deg)  Ms Chandra Deep Field South (CDF-S; Giacconi et al. 2002; Luo et al. 2008; Xue et al. 2011) and the  Ms Chandra Deep Field North (CDF-N; Alexander et al. 2003b) which have successfully identified AGN across more than 95% of cosmic time (out to ). Complementary to the highest redshift sources detected in the deep fields, nearby () AGN, identified in the relatively shallow contiguous wide-field surveys, such as the 5 ks  deg XBootes field (Murray et al. 2005; Kenter et al. 2005; Brand et al. 2006), have provided the ability to measure environment, a key component in galaxy and AGN evolution (e.g., Cooper et al. 2005, 2006; Georgakakis et al. 2008; Coil et al. 2009; Hickox et al. 2009; Cappelluti et al. 2010; Gilli et al. 2009). Furthermore, these wide-field X-ray surveys serendipitously detect significant numbers of rare, extremely luminous AGN and dozens of extended groups and clusters, which allow for a more complete understanding of the most massive SMBHs and cosmic structures in the Universe.

However, the peak of AGN activity, both in total luminosity and relative abundance is believed to occur at –2 (e.g., Hopkins et al. 2007; Zheng et al. 2009; Serjeant et al. 2010). The  deg DEEP2 Galaxy Redshift Survey (Davis et al. 2003; Madgwick et al. 2003) provides one of the most detailed censuses of the Universe. DEEP2 is currently one of the widest area and most complete spectroscopic surveys of galaxies, making it the ideal survey to target large numbers of AGN at –2. Indeed, the fourth data release (DR4) of the survey contains spectra for distant galaxies (with ) within four –1.1 deg fields, which are primarily in the redshift range –1.4; these were collected using the DEIMOS spectrograph ( in the wavelength range Å) on the Keck II telescope. A complete description of the DEEP2 DR4 spectroscopic catalog is available in Newman et al. (2012).

We have used the Chandra Advanced CCD Imaging Spectrometer (ACIS-I) to provide high-angular resolution X-ray coverage across almost the entire  deg survey area covered by the four DEEP2 fields (Field 1 - PI:K.Nandra; Fields 2, 3 and 4 - PI:S.Murray). Here we present the X-ray source catalog for our Chandra ACIS-I observations of the combined  deg DEEP2 (XDEEP2) survey. The four contiguous XDEEP2 fields have combined total exposures ranging from ks – 1Ms. In section 2 we present a brief introduction to the construction of the survey fields and the data reduction and processing of the X-ray observations. In section 3 we provide an in-depth methodology for the detection of the point sources and the building of the final XDEEP2 catalog. In section 4, we compare our new catalog of Field 1 (the Extended Groth strip), which now includes three recent  ks ACIS-I observations, to the previous catalog of Laird et al. (2009), and compare the XDEEP2 catalog to the Chandra Source Catalog (Evans et al. 2010). We further present the flux band ratio and density of number count distributions for the XDEEP2 catalog. In section 5, we outline the optical–X-ray source matching technique used to compare our new X-ray catalog with the Fourth Data Release of the optical DEEP2 photometric catalog. Finally, in section 6 we present a summary of our findings. Throughout the manuscript we adopt a standard flat CDM cosmology with and .

When combined, the redshift and galaxy property information established using the DEEP2 optical spectra and the AGN identified using the new Chandra X-ray observations provide one of the most complete views of AGN activity and the growth of large scale structure at –2. In forthcoming papers we will present a statistically complete and obscuration-independent view of the evolution of AGN and their host-galaxies identified across the entire electromagnetic spectrum, in the epoch to the present-day.

Field #16 Pointings17 18 19 Total Area20 Exp21 Exp22
(deg) (deg) (deg) (ks) (ks)
1 96 214.7388 +52.7838 0.66 662.3 139.2
2 12 252.4470 +34.9300 0.74 15.8 8.1
3 17 352.4711 +0.1869 1.13 10.1 8.1
4 12 37.2497 +0.5916 0.75 15.9 8.2

XDEEP2 field number
Number of Chandra pointings within field
Center co-ordinates of field in degrees as projected onto the sky in J2000 system
Total projected area of field in square degrees
Effective exposure in kilo-seconds at 20% of total field area
Effective exposure in kilo-seconds at 80% of total field area

Table 1XDEEP2 Field Properties

2. Chandra X-ray observations

2.1. Construction of the XDEEP2 fields

Figure 1.— Median offsets between optical and X-ray source positions with associated rms uncertainties in arc-seconds are plotted for the eight merged sub-fields in XDEEP2 Field 1 and the individual ObsIDs for Fields 2, 3 and 4. These median offsets were used to calculate astrometric corrections for the sub-fields in Field 1 and the individual X-ray observations in Fields 2–4. Chandra ObsIDs containing five or fewer X-ray–optical counterparts within 5 arc-minutes of the observation aim-point are shown with dotted error bars.

The XDEEP2 survey region consists of four contiguous –1.1 deg fields covered by Chandra ACIS-I observations. The field positions, arrangements and main properties are outlined in Table 1. The total area covered by XDEEP2 is  deg. X-ray catalogs for the previous 200 ks observations in Field 1, also referred to as the Extended Groth Strip, have been presented in Nandra et al. (2005) and Laird et al. (2009). For consistency and ease of reference to the previous Field 1 catalogs, we adopt the same sub-field naming convention defined in Laird et al. (2009) (see column 2 of Table 2). Additionally, in this manuscript we include the more recent 600 ks ACIS-I observations within three sub-fields (EGS-3; EGS-4; EGS-5) of the Groth Strip centered at , ; 214.808, +52.806; 214.527, +52.622, which for distinction between this and the previous catalogs, we rename as AEGIS-1, AEGIS-2 and AEGIS-3, respectively.

The catalog presented here was derived from multi-epoch observations taken during AO3 (PI K. Nandra), AO6 and AO9, combined with Guaranteed Time Observations (PI S. Murray; AO9). All Chandra observations for XDEEP2 are publicly available through the Chandra X-ray Center Archive. XDEEP2 consists of 126 separate pointings with varying individual exposures (–85 ks). With the exception of three exposures, all XDEEP2 observations were performed in vfaint mode to allow for the best possible background rejection. ObsIDs 3305, 4537 and 4365 were taken in faint mode. In Table 2 we provide the individual pointing details for each observation and field.

Field ObsID Sub-field Obs. Start Exp Roll Mode
(UT) (ks) (deg) (deg) (deg)
1 3305 EGS-8 2002-08-11 21:43:57 29.40 214.42932 52.47367 84.74 FAINT
1 4357 EGS-8 2002-08-12 22:32:00 84.36 214.42932 52.47367 84.74 FAINT
1 4365 EGS-8 2002-08-21 10:56:53 83.75 214.42933 52.47367 84.74 FAINT
1 5841 EGS-1 2005-03-14 00:04:09 44.45 215.67386 53.43149 229.11 VFAINT
1 5842 EGS-1 2005-03-16 15:54:34 46.42 215.67348 53.43141 226.09 VFAINT
1 5843 EGS-2 2005-03-19 17:13:09 44.46 215.38301 53.22857 222.38 VFAINT
1 5844 EGS-2 2005-03-21 22:37:40 45.85 215.38274 53.22848 219.86 VFAINT
1 5845 AEGIS-1 (EGS-3) 2005-03-24 14:33:31 48.40 215.11277 53.03769 216.61 VFAINT
1 5846 AEGIS-1 (EGS-3) 2005-03-27 04:51:15 49.40 215.11244 53.03755 213.67 VFAINT
1 5847 AEGIS-2 (EGS-4) 2005-04-06 20:01:09 44.55 214.84408 52.84563 200.99 VFAINT
1 5848 AEGIS-2 (EGS-4) 2005-04-07 21:03:59 44.45 214.84409 52.84563 200.99 VFAINT
1 5849 AEGIS-3 (EGS-5) 2005-10-11 12:47:43 49.46 214.59049 52.64737 19.89 VFAINT
1 5850 AEGIS-3 (EGS-5) 2005-10-14 05:15:37 45.55 214.59075 52.64755 16.94 VFAINT
1 5851 EGS-6 2005-10-15 03:03:18 35.68 214.10808 52.33121 14.80 VFAINT
1 5852 EGS-6 2005-12-03 13:00:33 10.62 214.10950 52.33506 324.80 VFAINT
1 5853 EGS-7 2005-10-16 20:16:24 42.57 213.84975 52.13788 14.04 VFAINT
1 5854 EGS-7 2005-09-30 23:52:23 50.07 213.84816 52.13692 31.03 VFAINT
1 6210 EGS-1 2005-10-03 14:56:50 45.94 215.68094 53.42341 29.70 VFAINT
1 6211 EGS-1 2005-10-12 11:43:28 35.64 215.68196 53.42394 19.80 VFAINT
1 6212 EGS-2 2005-10-04 22:56:06 46.28 215.39109 53.22076 28.00 VFAINT
1 6213 EGS-2 2005-10-06 06:52:13 47.51 215.39126 53.22084 26.47 VFAINT
1 6214 AEGIS-1 (EGS-3) 2005-09-28 08:09:03 47.50 215.12071 53.02977 35.13 VFAINT
1 6215 AEGIS-1 (EGS-3) 2005-09-29 15:58:09 48.63 215.12088 53.02982 33.67 VFAINT
1 6216 AEGIS-2 (EGS-4) 2005-09-20 09:35:13 49.48 214.85259 52.83816 43.79 VFAINT
1 6217 AEGIS-2 (EGS-4) 2005-09-23 01:34:59 49.50 214.85259 52.83816 43.79 VFAINT
1 6218 AEGIS-3 (EGS-5) 2005-10-07 05:31:36 40.58 214.59005 52.64709 24.71 VFAINT
1 6219 AEGIS-3 (EGS-5) 2005-09-25 15:57:04 49.48 214.58864 52.64648 37.70 VFAINT
1 6220 EGS-6 2005-09-13 09:17:01 37.63 214.10431 52.32963 49.79 VFAINT
1 6222 EGS-7 2005-08-28 17:20:24 34.69 213.84478 52.13607 59.04 VFAINT
1 6223 EGS-7 2005-08-31 05:06:47 49.51 213.84477 52.13608 59.04 VFAINT
1 6366 EGS-7 2005-09-03 06:30:11 14.58 213.84476 52.13605 59.04 VFAINT
1 6391 EGS-6 2005-09-16 20:43:01 8.45 214.10440 52.32956 49.79 VFAINT
1 7169 EGS-6 2005-12-06 02:29:46 16.03 214.10951 52.33509 324.25 VFAINT
1 7180 EGS-1 2005-10-13 05:16:04 20.43 215.68191 53.42394 19.80 VFAINT
1 7181 EGS-6 2005-10-15 21:17:21 15.98 214.10803 52.33122 14.80 VFAINT
1 7187 EGS-7 2005-10-17 19:07:08 6.59 213.84962 52.13785 14.04 VFAINT
1 7188 EGS-6 2005-12-05 04:50:40 2.58 214.10909 52.33499 324.80 VFAINT
1 7236 EGS-6 2005-11-30 19:29:34 20.37 214.10952 52.33505 324.80 VFAINT
1 7237 EGS-6 2005-12-04 05:26:20 16.93 214.10947 52.33504 324.80 VFAINT
1 7238 EGS-6 2005-12-03 10:02:10 9.53 214.10956 52.33502 324.80 VFAINT
1 7239 EGS-6 2005-12-11 08:31:06 16.03 214.10932 52.33545 319.59 VFAINT
1 9450 AEGIS-1 2007-12-11 04:24:07 28.78 215.07183 53.00951 319.80 VFAINT
1 9451 AEGIS-1 2007-12-16 10:52:06 25.21 215.07180 53.00951 319.80 VFAINT
1 9452 AEGIS-1 2007-12-18 05:45:49 13.29 215.07001 53.01006 311.30 VFAINT
1 9453 AEGIS-1 2008-06-15 21:28:03 44.69 215.05924 52.99529 130.79 VFAINT
1 9454 AEGIS-2 2008-09-11 04:47:10 59.35 214.81134 52.80632 49.30 VFAINT
1 9455 AEGIS-2 2008-09-13 19:38:46 99.72 214.81134 52.80633 49.30 VFAINT
1 9456 AEGIS-2 2008-09-24 08:15:30 58.35 214.81276 52.80818 34.80 VFAINT
1 9457 AEGIS-2 2008-06-27 07:08:38 32.74 214.79607 52.80288 124.29 VFAINT
1 9458 AEGIS-3 2009-03-18 12:20:16 6.65 214.52536 52.62140 223.14 VFAINT
1 9459 AEGIS-3 2008-09-30 19:20:28 69.55 214.55046 52.61607 30.30 VFAINT
1 9460 AEGIS-3 2008-10-10 06:17:49 21.36 214.55050 52.61613 29.80 VFAINT
1 9461 AEGIS-3 2009-06-26 09:30:12 23.73 214.53241 52.61042 129.79 VFAINT
1 9720 AEGIS-1 2008-06-17 05:14:02 27.79 215.05922 52.99527 130.79 VFAINT
1 9721 AEGIS-1 2008-06-12 08:09:14 16.55 215.05741 52.99587 139.79 VFAINT
1 9722 AEGIS-1 2008-06-13 07:02:28 19.89 215.05735 52.99589 139.79 VFAINT
1 9723 AEGIS-1 2008-06-18 13:42:40 34.47 215.05923 52.99528 130.79 VFAINT
1 9724 AEGIS-1 2007-12-22 13:37:26 14.08 215.07007 53.01006 311.30 VFAINT
1 9725 AEGIS-1 2008-03-31 05:21:42 31.13 215.05145 53.00445 209.78 VFAINT
1 9726 AEGIS-1 2008-06-05 08:45:04 39.62 215.05737 52.99587 139.79 VFAINT
1 9727 AEGIS-2 2008-09-12 16:44:12 34.94 214.81132 52.80634 49.30 VFAINT
1 9729 AEGIS-2 2008-07-09 16:47:58 48.09 214.79710 52.80272 119.79 VFAINT
1 9730 AEGIS-2 2008-09-25 16:50:54 53.72 214.81277 52.80817 34.80 VFAINT
1 9731 AEGIS-2 2008-07-03 10:58:47 21.38 214.79688 52.80275 120.79 VFAINT
1 9733 AEGIS-2 2008-09-27 01:15:33 58.36 214.81275 52.80818 34.80 VFAINT
1 9734 AEGIS-3 2008-09-16 11:01:21 49.47 214.54931 52.61415 44.80 VFAINT
1 9735 AEGIS-3 2008-09-19 03:14:15 49.47 214.54930 52.61415 44.80 VFAINT
1 9736 AEGIS-3 2008-09-20 11:07:10 49.48 214.54930 52.61416 44.80 VFAINT
1 9737 AEGIS-3 2008-09-21 17:53:00 49.48 214.54931 52.61415 44.80 VFAINT
1 9738 AEGIS-3 2008-10-02 06:56:22 61.39 214.55047 52.61607 30.30 VFAINT
1 9739 AEGIS-3 2008-10-05 11:28:12 42.59 214.55049 52.61614 29.80 VFAINT
1 9740 AEGIS-3 2009-03-09 22:24:18 20.37 214.52625 52.62221 229.78 VFAINT
1 9793 AEGIS-1 2007-12-19 02:53:51 23.83 215.07005 53.01008 311.30 VFAINT
1 9794 AEGIS-1 2007-12-20 04:27:59 10.03 215.07009 53.01004 311.30 VFAINT
1 9795 AEGIS-1 2007-12-20 21:36:20 8.91 215.07008 53.01009 311.30 VFAINT
1 9796 AEGIS-1 2007-12-21 20:28:33 16.33 215.07004 53.01008 311.30 VFAINT
1 9797 AEGIS-1 2007-12-23 13:12:28 12.60 215.07007 53.01011 311.30 VFAINT
1 9842 AEGIS-1 2008-04-02 21:01:59 30.44 215.05145 53.00445 209.78 VFAINT
1 9843 AEGIS-1 2008-04-02 01:11:09 13.48 215.05143 53.00448 209.78 VFAINT
1 9844 AEGIS-1 2008-04-05 13:07:54 19.78 215.05147 53.00443 209.78 VFAINT
1 9863 AEGIS-1 2008-06-07 00:33:47 22.01 215.05733 52.99587 139.79 VFAINT
1 9866 AEGIS-1 2008-06-03 22:43:14 25.83 215.05737 52.99588 139.79 VFAINT
1 9870 AEGIS-1 2008-06-10 15:11:23 11.00 215.05736 52.99583 139.79 VFAINT
1 9873 AEGIS-1 2008-06-11 14:22:06 30.75 215.05737 52.99588 139.79 VFAINT
1 9875 AEGIS-1 2008-06-23 22:54:14 25.20 215.05968 52.99517 128.77 VFAINT
1 9878 AEGIS-2 2008-06-28 06:03:20 15.73 214.79613 52.80289 124.29 VFAINT
1 9879 AEGIS-2 2008-06-29 03:39:20 26.80 214.79612 52.80288 124.29 VFAINT
1 9880 AEGIS-2 2008-07-05 17:00:17 29.50 214.79688 52.80274 120.79 VFAINT
1 10769 AEGIS-3 2009-03-20 13:38:26 26.68 214.52497 52.62063 216.98 VFAINT
1 10847 AEGIS-3 2008-12-31 05:06:27 19.27 214.54102 52.62566 302.79 VFAINT
1 10848 AEGIS-3 2009-01-01 17:11:57 17.91 214.54109 52.62567 302.79 VFAINT
1 10849 AEGIS-3 2009-01-02 21:25:57 15.92 214.54106 52.62570 302.79 VFAINT
1 10876 AEGIS-3 2009-03-11 01:37:20 17.21 214.52626 52.62222 229.78 VFAINT
1 10877 AEGIS-3 2009-03-12 15:15:57 16.22 214.52630 52.62223 229.78 VFAINT
1 10896 AEGIS-3 2009-06-15 18:46:14 23.29 214.53123 52.61075 135.32 VFAINT
1 10923 AEGIS-3 2009-06-22 07:38:22 11.62 214.53239 52.61039 129.79 VFAINT
2 8631 - 2007-11-26 00:59:04 8.87 253.14712 35.06573 10.90 VFAINT
2 8632 - 2007-11-26 03:52:42 8.60 252.85635 35.06034 10.90 VFAINT
2 8633 - 2007-11-26 06:30:13 8.60 253.14626 34.84466 10.90 VFAINT
2 8634 - 2007-11-26 09:07:44 8.60 252.57252 35.05619 10.90 VFAINT
2 8635 - 2007-11-26 11:45:15 8.60 252.29343 35.04576 10.90 VFAINT
2 8636 - 2007-11-26 14:22:46 8.60 252.00739 35.04026 10.90 VFAINT
2 8637 - 2007-11-26 17:00:17 8.60 251.71914 35.03216 10.90 VFAINT
2 8638 - 2007-11-26 19:37:58 8.60 252.86086 34.84118 10.90 VFAINT
2 8639 - 2007-11-26 22:15:40 8.60 252.57546 34.83892 10.90 VFAINT
2 8640 - 2007-11-27 00:53:12 8.61 252.29954 34.82097 10.90 VFAINT
2 8641 - 2007-11-28 05:53:06 8.92 252.01425 34.81738 10.90 VFAINT
2 8642 - 2007-11-28 08:41:27 8.66 251.72431 34.81683 10.90 VFAINT
3 8601 - 2008-08-05 04:20:00 9.06 353.25281 0.24568 242.49 VFAINT
3 8602 - 2008-08-05 07:12:25 8.93 352.66172 0.28185 242.49 VFAINT
3 8603 - 2008-08-05 09:48:46 8.84 353.46809 0.20782 242.49 VFAINT
3 8604 - 2008-08-05 12:24:02 8.84 351.64001 0.25772 242.49 VFAINT
3 8605 - 2008-08-05 14:59:49 8.84 353.37170 0.01529 242.49 VFAINT
3 8606 - 2008-08-05 17:35:02 8.83 352.97330 0.21938 242.49 VFAINT
3 8607 - 2008-08-05 20:09:51 8.83 351.89874 0.25007 242.49 VFAINT
3 8608 - 2008-08-05 22:44:39 8.84 351.72303 0.01783 242.49 VFAINT
3 8609 - 2008-08-06 01:19:39 8.84 353.09031 -0.01102 242.49 VFAINT
3 8610 - 2008-08-06 03:54:42 8.84 352.15938 0.30473 242.49 VFAINT
3 8611 - 2008-08-06 06:29:27 8.83 352.01198 0.02339 242.49 VFAINT
3 8612 - 2008-08-06 09:04:08 8.84 351.47375 0.02736 242.49 VFAINT
3 8613 - 2008-08-06 11:38:49 8.84 352.25372 0.06485 242.49 VFAINT
3 8614 - 2008-08-06 14:13:30 8.84 352.80328 0.06085 242.49 VFAINT
3 8615 - 2008-08-06 16:48:28 8.84 351.48138 0.26156 242.49 VFAINT
3 8616 - 2008-08-06 19:23:35 8.83 352.54266 0.05529 242.49 VFAINT
3 8617 - 2008-08-06 21:58:23 8.84 352.42188 0.29140 242.49 VFAINT
4 8619 - 2007-11-28 13:18:37 9.04 36.63964 0.70242 50.76 VFAINT
4 8620 - 2007-11-28 16:09:09 8.66 36.72970 0.48135 50.76 VFAINT
4 8621 - 2007-11-29 01:03:58 9.07 36.88713 0.70250 50.76 VFAINT
4 8622 - 2007-11-29 03:53:59 8.66 37.14407 0.70447 50.76 VFAINT
4 8623 - 2007-11-29 06:32:28 8.66 37.39158 0.70452 50.76 VFAINT
4 8624 - 2007-11-29 09:10:59 8.66 37.63909 0.70268 50.76 VFAINT
4 8625 - 2007-11-29 11:49:28 8.66 37.89038 0.76129 50.76 VFAINT
4 8626 - 2007-12-01 11:55:28 8.86 36.97907 0.48142 50.76 VFAINT
4 8627 - 2007-12-01 14:50:45 8.46 37.22656 0.47393 50.76 VFAINT
4 8628 - 2007-12-01 17:25:55 8.47 37.48350 0.47209 50.76 VFAINT
4 8629 - 2007-12-01 20:01:05 8.47 37.72722 0.47402 50.76 VFAINT
4 8630 - 2007-12-01 22:36:15 8.47 37.88970 0.59744 50.76 VFAINT


XDEEP2 Field number.

Chandra observation identification number.

Due to missing gain files within the CALDB, ObsID 6221 is not included in the analyses of Field 1 which are presented here. The exposure time of 6221 is only 4.15 ks, hence, its rejection is relatively insignificant compared to the total exposure time within Field 1 and will have a negligible effect on our conclusions.

Sub-field name for observations in Field 1, adopted from Laird et al. (2009).

Observing date and start time in UT.

Exposure time in kiloseconds after appropriate screening.

Aim point position of observation in degrees in J2000 coordinates.

Spacecraft roll angle in degrees in standard north-east co-ordinate system.

Chandra observing mode.

Table 2Observation Log.

2.2. Data reduction

Basic processing was carried out using the Chandra X-ray Center (CXC) pipeline software. In addition, further processing of the X-ray data was carried out using the chav (v4.3)23 and ciao (v4.3) 24 software packages combined with custom idl scripts. Each ACIS-I observation was analyzed separately. Individual ACIS-I pointings were reduced from the Level-1 event file products of the standard Chandra data pipeline. We use the ciao tool acis_process_events to remove the standard pixel randomization, and status=0 was used to remove streak events, bad pixels and cosmic ray afterglow features.

Figure 2.— Example of a full-band (0.5–7keV) merged raw counts image of an XDEEP2 field. Region shown is Field 2. Individual Chandra pointings have been merged using the ciao tool merge_all. The image has been smoothed using a Gaussian kernel for presentation purposes only. Many sources are clearly evident throughout the image. Due to the presentation smoothing process, edge-effects (correlated streaks) can be seen along the positions of the chip gaps.
Figure 3.— Merged source count image of the sub-field AEGIS-3 located in Field 1. Aim points of the individual Chandra ObsIDs within the sub-field are shown with green crosses matched to the roll angle of the space craft. The angular separation of the aim-points is sufficiently small ( arc-seconds) that they allow for the combining of the individual ObsIDs into stacked images.

All observations were visually inspected for flaring and periods of high background. The majority of the observations were found to not be significantly contaminated. As also noted in Nandra et al. (2005) and Laird et al. (2009), observation 4365 does exhibit an interval ( ks;  % of the observation) of elevated background. However, unlike the previous analyses, here we conservatively screen-out this period of high background. Final effective exposures in good-time intervals for each observation were generally found to be  % of the “on-time” (see Table 2).

2.3. Creation of individual images & exposure maps

Figure 4.— Merged full-band (0.5-7 keV) Chandra ACIS-I exposure map for XDEEP2 Field 1. Reference spectra in monochromatic bands of , 4.0 and 2.5 keV and a spectral slope of were assumed in the creation of the exposure maps from the aspect histograms. The effective exposure (and hence sensitivity depth; see §4.1) across Field 1 is non-uniform and varies dramatically from  ks–1.1 Ms due to the large number of overlapping observations. Overlaid is the nominal survey area covered by the DEEP2 Galaxy Redshift Survey optical observations (solid black line).
Figure 5.— Same as Fig. 4, except field shown is XDEEP2 Field 2
Figure 6.— Same as Fig. 4, except field shown is XDEEP2 Field 3
Figure 7.— Same as Fig. 4, except field shown is XDEEP2 Field 4

Events files were screened using a standard grade set (grade=0,2,3,4,6) to construct images for each individual ObsID. Images were constructed in the Full (FB; 0.5–7 keV), Soft (SB; 0.5–2 keV) and Hard (HB; 2–7 keV) bands at the full ACIS-I spatial resolution, 0.492 arcsec/pixel. Here we limit the photon energy to  keV to allow a more direct comparison to sources detected in the XBootes survey. Given the small effective area of the ACIS-I detector at  keV, relatively few  keV photons are detected, and thus this choice of energy boundary is somewhat arbitrary and will have little effect on our conclusions. The chav tool aspecthist was used to create aspect histograms in all three bands. These aspect histograms were used to generate exposure maps by convolving them with the standard ACIS-I chip-map (ccd_id=0,1,2,3) and reprojecting to the previously created counts images. Reference spectra in monochromatic bands of , 4.0 and 2.5 keV (i.e., the median energies of the SB, HB and FB, respectively) were used in the creation of the exposure maps.

Figure 8.— a (left): Cumulative survey solid angle (in degrees) as a function of effective exposure in the full-band (0.5–7keV) for the four separate XDEEP2 fields. b (right): Total effective exposure across the combined XDEEP2 survey compared to the E-CDFS (Lehmer et al. 2005) and C-COSMOS (Elvis et al. 2009) survey fields. Minimum effective exposure for 20% of the total XDEEP2 area is highlighted with a dotted line.

2.4. Astrometric calibration & observation merging

Due to differing observing strategies and the sizes of exposure area overlaps between individual Chandra observations within each XDEEP2 field, X-ray observations were combined using separate methods for Field 1 and Fields 2–4. As stated previously, Field 1 contains eight sub-fields (see §2), with marginal overlap (–0.02 deg) between one another. Each of these eight sub-fields consists of several (3–28) individual Chandra ACIS-I exposures with significant overlap between the observations within a particular designated sub-field. We used the ciao Perl script, merge_all to create contiguous raw X-ray images and exposure maps within each of the eight Field 1 sub-fields. Briefly, this script searches for bright X-ray sources within two events tables which spatially overlap and compares the astrometric co-ordinates of the detected sources. By computing the average offset between the sources within the tables, and guarding against rogue outliers, the events table and associated aspect histograms are reprojected to the world co-ordinate system (WCS) of the first reference observation within the sub-field.

Given the limited area overlap (which occurs only at large off-axis radii) between the eight sub-fields, a resultant merged events table and images from a further use of merge_all to combine the sub-fields, is likely to be highly uncertain. However, one of the primary goals for this XDEEP2 X-ray catalog is to compare the X-ray detected sources with the previously astrometrically-calibrated optical DEEP2 catalog presented in Coil et al. (2004). Hence, we may consider the WCS astrometry of the DEEP2 optical catalog to be an absolute reference frame. Thus, here we use the DEEP2 optical source positions to correct the X-ray sub-fields for any systematic offsets that may be present in the combined X-ray data. Following Brand et al. (2006), we use a counterpart-matching algorithm (described in detail in § 5 of this manuscript) to match X-ray sources detected within 3 arc-minutes of the nominal observation aim-point to optical counterparts. We calculated the median offset between the X-ray and optical positions for the respective sources to identify any necessary translation for the X-ray sub-fields. We present these offsets and their associated rms uncertainties in Figure 1. Typically, –30 X-ray–optical sources were used to determine the necessary translations; the offsets were generally found to be arc-seconds (i.e., % of the ACIS-I pixel scale). While rotations were also allowed in the calculation of the relative astrometries, the magnitude of the angular rotation was always found to be negligible ( degree) and consistent with no rotation. Hence, we did not include angular rotations and used only linear transformations for the final corrections of the X-ray WCS to that of DEEP2. The required positional offsets for the merged X-ray images were applied using the ciao tool wcsupdate. The ciao tool reproject_aspect was used to reproject the events table and aspect solution files.

Figure 9.— Chandra false color image of XDEEP2 Field 1. The image is a merged composite of the exposure corrected 0.5–2 (red), 2–4 (green) and 4–7 keV (blue) images within Field 1. The color-band images have been adaptively smoothed with varying smoothing scales determined from the average background counts in the stacked images.

In Fields 2–4, the relatively shallow 9–10 ks X-ray observations include little or no overlap area between exposures. As such, and similar to the merged sub-fields in Field 1, merge_all cannot be used to accurately co-align the relative astrometries within the individual X-ray observations in these three fields. Hence, again we consider the WCS reference frame of the optical DEEP2 catalog to be absolute, and use the optical sources to align individual X-ray observations following the same methodology described above. Given the far shallower depth of the X-ray observations in Fields 2–4, we include all X-ray sources with optical counterparts to a distance of arc-minutes from the aim-point. This larger off-axis distance encompasses sufficient X-ray–optical source numbers (5–20 per observation) to accurately constrain any required systematic astrometric correction. Four of the Chandra ObsIDs (8637; 8614; 8604; 8628) included five or fewer X-ray–optical sources, and hence we consider any astrometric corrections for these four observations to be sufficiently uncertain that we subsequently include all detected X-ray sources (at all off-axis distances within the observation) to further constrain any median offset. The calculated median offsets and associated uncertainties are also included in Figure 1. Clearly, using our adopted methodology, we do not account for any possible field-to-field (or intra-field) variations in the astrometric accuracy of the optical DEEP2 catalog. However, given the low number of X-ray sources within individual Chandra observations, further investigation and/or necessary correction to the DEEP2 catalog are beyond the scope of this study. Overall, the required astrometric corrections (average correction of 0.24”) for the whole of XDEEP2 are consistent, if not slightly lower, than those found in previous wide-field X-ray surveys (e.g., XBootes: 0.41”; Brand et al. 2006) and can be considered sufficiently precise for our purposes.

2.5. Merged XDEEP2 field maps

In Figures 2 and 3, we show examples of the merged full-band (0.5–7 keV) counts images and in Figures 47, we present the merged full-band exposure maps for the four survey fields. As shown in Figure 8a, the effective exposure (and hence sensitivity depth; see § 4.1) across Field 1 is non-uniform and varies dramatically from 20 ks–1.1 Ms. The effective exposure in Field 1 is dependent on the number of repeat exposures, the large number of overlapping regions and the varying space-craft roll-angles between separate pointings. We show that at the 80th percentile, the effective exposure in Field 1 is 140 ks. By contrast, the effective exposures in Fields 2, 3 and 4 are relatively uniform ( ks at 80%) with constant spacecraft roll angle and only small overlap regions between the individual ACIS-I pointings (%). In Figure 8b, we also show the effective exposure time across the combined XDEEP2 area and compare this to the Chandra-COSMOS (Elvis et al. 2009) and Extended-Chandra Deep Field South fields (Lehmer et al. 2005). It is clear that XDEEP2 complements these previous surveys: the survey depth of XDEEP2 extends well beyond  ks (the limiting effective exposure of the E-CDF-S) to  ks at similar survey area ( deg); and XDEEP2 covers a survey area which is a factor greater than that of Chandra-COSMOS.

The raw merged count images for each of the four XDEEP2 fields were adaptively smoothed using custom idl software based on the kernel-smoothing program, asmooth (Ebeling et al., 2006). Given the wide range in exposure times across Field 1, we include a weighting algorithm based on the average number of counts within binned background images (see §4.1) to account for changes in background count rate in overlapping regions. This background-weight is applied to the calculation of the smoothing radii within our custom version of asmooth. The smoothing scales, which are calculated from analysis of the merged counts images, are then applied directly to the respective exposure maps. We use these to create false-color exposure-corrected smoothed images in each field (see Figure 9).

Figure 10.— a, left: Positional offsets between the wavelet-centers produced from wvdecomp and the events centroids. Contours encompass 50, 65, 80, 90 and 95% of XDEEP2 sources. We find a small systematic offset of ” between the median source positions produced from the events centroid and wavelet center methods. Median 90% uncertainties of ” were derived following Murray et al. (2005), see §3.3. b, right: Angular separation (wavelet – centroid) as a function of off-axis distance for XDEEP2. Median offset distances and the associated median RMS scatter are given in bins of 2’ (open squares).

3. Point source detection & spurious sources

In this section we outline the methods used to detect point-like sources throughout the XDEEP2 fields. Following earlier analogous methods for numerous wide-field and deep X-ray surveys, we used wavelet decomposition software to detect sources across XDEEP2. Indeed, previous analyses of Field 1 have used the ciao tool wavdetect to detect X-ray source candidates. Here, we chose to use wvdecomp which is publicly available in the zhtools package (see Vikhlinin et al. 1998). In §4.2 we perform a comparison of the X-ray sources detected in Laird et al. (2009) which used wavdetect and additional signal-to-noise criteria to the sources detected in this work using wvdecomp. Briefly, we find little or no difference between the number of sources detected in either analyses. We find that % of the unique X-ray sources found in the previous AEGIS-X catalog are included in our new catalog (presented here) which now includes the more recent longer exposure ACIS-I observations. We find that the majority of the sources which are not included in our new catalog are relatively low significance with few counts () and, in general, are detected in only one energy-band in the Laird et al. catalog. Sources similar to these were conservatively removed as possibly spurious detections in our new catalog based on our extensive MARX simulations (see §3.4).

3.1. Point source detection in individual Chandra ObsIDs

Point sources were detected in the individual (non-merged) counts images for the SB, HB and FB energy ranges. We used a point source detection threshold in wvdecomp of 4.5 (equivalent to a probability threshold of ). Point sources were detected over wavelet scales of {1, , 2, 4} . After detection of a source candidate, the event data at the approximate wavelet position was iterated up to five times to accurately determine the final events centroid, and hence, source position. In Figure 10, we present the offset distances between the wavelet and event centroid positions. We find that % of the X-ray sources have offset distances ” from the wavelet position. Indeed, the vast majority of the sources are consistent with zero offset. Furthermore, we find that the median offset distance between the wavelet and centroid positions are mildly correlated with the on-chip distance of the source from the observation aim-point. Those sources at ’ have ”, while those sources closer to the edge of the FOV, at ’, have ”. These increased offsets at large off-axis distances were most likely due to asymmetries in the ACIS PSF shape.

Source lists, generated from the separate energy bands in the individual observations, were cross-correlated based on their source positions. Two-dimensional Gaussian profiles were used to represent the sources detected in the separate energy bands with full-width half maxima (FWHM) determined by the physical size of the 90% encompassed energy fraction (EEF) within the Chandra energy-band images with the assumption of a spherically symmetric model for the ACIS-I PSF. The centers of the Gaussian profiles were allowed to shift within the 1 centroid error (see §3.3) of the source positions to maximise the statistical likelihood of a source match. A unique source was determined to exist when the summed 2-D Gaussian profile was well-fit at the 90% confidence level by a single (approximately symmetrical) 2-D Gaussian profile with FWHM % EEF.25 This methodology has the advantage that the ‘matching-radius’ naturally becomes a function of both the off-axis position and the energy-band of the source detection. Hence, it incorporates the size increase and rotation of the ACIS-I PSF radius which, while assumed to be symmetrical about the aim-point, still increases significantly for large off-axis distances and simultaneously changes as a function of both azimuthal angle and effective energy.

3.2. Sources in overlapping observations in Field 1

As stated previously, sources were detected in each of the individual ObsIDs. In Fields 2,3 and 4 there are small regions of significant exposure ( ks), where individual observations overlap. However, given the large systematic uncertainties brought about by significant differences in Chandra PSF radii, we did not attempt to combine these observations to search for faint sources, which would be detected in the merged deeper exposure regions. Instead, where duplicate sources in these overlap regions appear (see previous section), the source which is radially closest to the aim point in a particular Chandra observation (i.e., the source which has the smallest point spread function), is included as a unique source in the final catalog. By contrast, given the large overlap between the Chandra observations in the sub-fields of Field 1, it is highly likely that the same physical X-ray source is detected in multiple individual exposures and that many fainter sources would be detected in merged X-ray images. Hence, we have created merged events files of the sub-field regions, which were defined in Laird et al. (2009) (see Table 1 of Laird et al. 2009 and Table 2 and Figure 3 in this work).26

When combined, the Field 1 ‘EGS’ sub-fields show a significant increase in the overall exposure and depth. Each of the observations in these sub-fields have varying space-craft roll angles. However, as shown in Table 2, the pointing co-ordinates are similar ( arc-seconds; see also Figure 3). As such, these stacked sub-fields do not suffer from significant sensitivity degradation due to large changes in the Chandra point spread function (i.e., the observational setup was similar to that of the CDF-N and CDF-S; e.g., Alexander et al. 2003a; Xue et al. 2011). We used wavelet decomposition to search for additional faint sources in these merged (stacked) sub-field images which would otherwise not be detected in the individual observations. Candidate source lists for Field 1, which were compiled from each of the individual ObsIDs and those lists derived from the merged sub-field images were compared using the same unique-source detection method outlined in the previous section. The final unique source position and associated centroid errors were determined by averaging and combining in quadrature the previously calculated positions and uncertainties in the individual and merged X-ray observations.

3.3. Source extraction

Once the unique source locations were determined across each of the XDEEP2 fields, we counted the number of events (, ) within the 50% () and 90% () encircled energy fraction regions of the merged sub-field images (Field 1) and the individual ACIS-I observations (Fields 2–4) for each of the soft, hard and full bands. Within the sub-fields of Field 1, the radii for circular extraction regions were calculated from the off-axis radial distances in PSF simulations. We used the MARX simulator to model a point-source, within a specific energy-band, at varying roll angles and off-axis distances from an observation aim-point. The modeled energy-band images were combined using the method outlined in §2.4, and the spatial extent of the merged point-source was measured using a circular aperture to determine accurate extraction radii for the candidate sources identified in the Field 1 sub-fields.

Figure 11.— (a; left) Average total number of spurious sources (N) detected in MARX simulated background images as a function of net photon counts in the full-band (C) measured in the spurious source. (b; right) Required net count threshold to ensure the average total number of spurious sources (derived from Monte-Carlo simulations) are ; 0.25; 0.5; 1.0 (dot-dot-dash line; dash line; dot-dash line; solid line, respectively) in a Chandra ACIS-I observation plotted as a function of exposure time in kiloseconds.
Figure 12.— Source count threshold cut in the 0.5–7 keV band as a function of exposure time in MARX simulated Chandra ACIS-I imaging. For a fixed total number of spurious sources of within simulated observations, we show the dependence of the threshold cut on the off-axis position of the detected spurious sources.

We calculated average effective exposures for each candidate source in the and extraction regions. Background counts were determined for each source by extracting photon counts in annuli at inner and outer radii , respectively, in background images (see §4.1). Background counts were scaled by the ratio of the areas of the EEF extraction region and the background extraction region. Scaled background counts were subtracted from the respective and to give final net source counts (; , respectively). The 50% and 90% encircled energy fraction regions were chosen to match those used in the XBootes survey (Murray et al. 2005; Kenter et al. 2006; Brand et al. 2006) allowing direct comparisons to be made between the catalogs in future publications.

For a source detected in a particular energy band image, we computed the total number of source counts in the other energy band images using the analyses described above. We converted the net count rates in each band (SB; HB; FB) to total fluxes (; ; , respectively). To build a homogeneous X-ray catalog, we assumed a single simple absorbed power-law spectrum with (i.e., the typical intrinsic slope of an AGN) for all sources and for those sources in Fields 1, 2, 3 and 4, respectively. Here, we use PIMMS to calculate the count-rate–flux conversion factors assuming the simulated ARFs from AO9 of the Chandra program. The use of the AO9 ARFs compared to AO6 results in a % decrease in the calculated 0.5–7 keV flux. Total galactic HI column densities were determined using Stark et al. (1992). Uncertainties on the counts and fluxes were calculated using the formalism of Gehrels (1986).

Following Murray et al. (2005), we estimated the 90% uncertainty on the source locations as . For those sources with counts, we set a minimum centroid error of 0.8 arc-seconds (i.e., the 99% positional accuracy on the ACIS-I detector27), which takes into account the systematic uncertainties associated with the space-craft and detector astrometry. Random uncertainties also become negligible for sources with large numbers of counts.

3.4. Spurious sources

Given the widely varying exposure times, and hence varying background levels of individual observations within XDEEP2, it is important to apply further restrictions to the detected-source lists based on the number of counts for a given source. For those observations with large exposure times, the number of spurious sources with seemingly low numbers of counts increases (see Figure 11a). To limit the number of spurious sources within our final catalog, we applied a minimum photon count threshold of , where was determined through simulations of sourceless background ACIS-I images. We used the MARX software package to simulate 100,000 Chandra ACIS-I images of the unresolved Cosmic X-ray background (XRB), including instrumental effects for exposure times of 3, 6, 9, 12, 15, 20, 30, 50, 75 and 100ks. To approximate the expected emission from the unresolved CXB, we employed a simple absorbed power-law spectrum with (e.g., Hickox & Markevitch 2006) and in the 0.5–7 keV band; i.e., the XRB surface brightness measured in the ROSAT all-sky survey in a blank-sky region of XDEEP2 Field 1, which was then scaled to the projected area of ACIS-I. We note that this simplification assumes the CXB emission is homogeneous across an ACIS-I observation. We searched each of the simulated XRB ACIS-I images for spurious sources using the same wavelet detection thresholds defined above (see Figure 11a). To build source lists which were both relatively complete while limiting the number of spurious sources, we cut the source-lists where the expected total number of spurious sources for a given exposure was (see Figure 11b). By adopting a threshold of , we expect a spurious source detection rate of % in the final catalog.

As we show in Figure 12, we find that the spurious net count threshold is both a function of exposure time () and off-axis position () of the source within an ACIS-I observation. This count threshold can be approximated by the empirical formula,


and we use this to derive for a given fixed off-axis position and exposure. To verify that this parametrization of the count threshold can be extrapolated to larger exposure times (i.e., for the merged AEGIS-1, 2 and 3 sub-fields in XDEEP2 Field 1), we simulated 100 1 Ms ACIS-I exposures using MARX. On average, we detected spurious source in each 1 Ms simulation by using . Hence, within the Poisson error, we detected the same number of spurious sources expected when extrapolating the above equation to  Ms. By conservatively adopting a threshold of across the 126 XDEEP2 pointings we expect spurious sources in the final XDEEP2 catalog.

4. The XDEEP2 catalog

The XDEEP2 point source catalog contains 2976 unique sources, with 1720, 342, 528 and 386 sources in Fields 1, 2, 3 and 4, respectively. For the purposes of our point source catalog, we do not discuss those sources which are extended (e.g., the galaxy clusters) as these will be the subject of a future publication. In Table 5 we show a short extract from the main source table, which is available electronically. In Figure 13, we show the distribution of source counts across the four XDEEP2 counts in the soft, hard and full bands. It is clear that both the wide-spatial area of XDEEP2 combined with the smaller regions of sensitive long-exposures, are extremely complementary to one another. A significant cut-off is observed for sources with in Field 1 since relatively few sources () are detected with 5–10 counts within due to the long integrated exposures, even in the soft-band. However, many more sources, down to are detected when the other three XDEEP2 fields are included. Hence, within the point source catalog we detect sources down to net counts in the SB, with a completeness to 20 net counts in the HB and 15 counts in the FB. Furthermore, we detect 70 rare bright sources with , which is due to the advantage of the wide-area across the XDEEP2 survey.

Figure 13.— Distribution of X-ray counts for sources detected in each of the four XDEEP2 fields in the full-band (0.5–7keV; top panel), hard-band (2–7keV; middle panel), and soft-band (0.5–2keV; bottom panel). Median source counts for each energy band in the associated field are shown with vertical lines.
Detection band 28 # Non-detection band 29
(keV) Full Soft Hard
Full 2849 - 661 1196
Soft 2301 111 - 1006
Hard 1663 12 372 -

Energy band which a source has been detected in
Number of sources where there is a non-detection in a particular energy-band when it has been detected in a different band.

Table 3Sources detected in separate energy bands

In Table 3 we show the breakdown of the numbers of sources detected and formally undetected in individual energy bands within the main XDEEP2 source catalog. Those X-ray sources which are not formally detected in a particular energy band are denoted by “-1” in the relevant net count and flux error columns of Table 5 (e.g., NET COUNTS ERROR SB/HB/FB and FLUX ERROR SB/HB/FB). For these ‘non-detections’, we use the formalism of Gehrels (1986) to dervie upper-limits from the number of counts observed in the background images (see § 4.1) within the source region. There upper-limits are given in the appropriate NET COUNT and FLUX energy-band columns.

4.1. Background & sensitivity analysis

As is clearly evident from the merged exposure maps, many of the XDEEP2 ObsIDs spatially overlap with one another; however, a subset of these observations, specifically in Field 1, were performed up to seven years apart. Hence, care was taken to analyze changes between the overlapping images as a result of the physical changes in the detector and varying background levels. Background images were constructed separately for each ObsID in the SB, HB and FB energies. Source counts for candidates which were identified as being significant in a particular energy-band using wvdecomp were masked. Background annuli, with inner radii and outer radii centered at the source position, were used to calculate the mean local background surrounding the candidate source. The masked source region was re-populated with Poisson noise with a mean distribution equal to that of the local background. The same procedure was additionally used to create background maps of the merged sub-fields in Field 1. While this procedure will remove the count contributions from all point-sources, it will not remove extended emission from sources such as clusters (e.g., Bauer et al. 2002). Hence, the background count levels derived from this method are somewhat conservative, as they will be slightly over-estimated.

Field #30
Sub-field31 Energy band32
Mean background 33
Background 34 Total Background 35
(counts pixel) (counts pixel) ( counts)
1 AEGIS 1 Full 0.0841 0.2898 52.5
1 AEGIS 1 Soft 0.0242 0.1539 15.1
1 AEGIS 1 Hard 0.0599 0.2425 37.4
1 AEGIS 2 Full 0.0842 0.2900 51.6
1 AEGIS 2 Soft 0.0236 0.1524 14.5
1 AEGIS 2 Hard 0.0605 0.2448 37.1
1 AEGIS 3 Full 0.0991 0.3033 61.3
1 AEGIS 3 Soft 0.0284 0.1609 17.6
1 AEGIS 3 Hard 0.0706 0.2548 43.7
1 EGS 1 Full 0.0243 0.1438 13.1
1 EGS 1 Soft 0.0070 0.0773 3.8
1 EGS 1 Hard 0.0165 0.1186 8.9
1 EGS 2 Full 0.0235 0.1419 12.0
1 EGS 2 Soft 0.0068 0.0765 3.5
1 EGS 2 Hard 0.0159 0.1167 8.1
1 EGS 6 Full 0.0271 0.1531 14.8
1 EGS 6 Soft 0.0077 0.0815 4.2
1 EGS 6 Hard 0.0184 0.1257 10.0
1 EGS 7 Full 0.0241 0.1429 13.3
1 EGS 7 Soft 0.0070 0.0769 3.9
1 EGS 7 Hard 0.0163 0.1176 9.0
1 EGS 8 Full 0.0332 0.1684 14.7
1 EGS 8 Soft 0.0111 0.0980 4.9
1 EGS 8 Hard 0.0202 0.1310 8.9
2 - Full 0.0018 0.0428 11.1
2 - Soft 0.0005 0.0231 3.2
2 - Hard 0.0013 0.0361 7.9
3 - Full 0.0018 0.0432 7.4
3 - Soft 0.0005 0.0234 2.1
3 - Hard 0.0013 0.0363 5.2
4 - Full 0.0018 0.0431 7.6
4 - Soft 0.0005 0.0231 2.2
4 - Hard 0.0013 0.0364 5.4

XDEEP2 field number
XDEEP2 sub-field name
X-ray energy band of background image: Full 0.5–7keV; Soft 0.5–2keV; Hard 2–7keV
Mean number of background counts per pixel within the non-zero exposure area of the merged images.
Standard deviation of the background counts within the merged images.
Total number of background counts within the merged images.

Table 4Background analysis of XDEEP2 fields

The mean background counts, their associated standard deviation and total number of background counts for each field (and sub-field) are shown in Table 4. As expected, the average background counts are a factor of –50 greater in Field 1 than those in Fields 2–4, owing to the much longer exposure times in Field 1. We find that the average backgrounds appear to be relatively stable across the deep sub-fields AEGIS-1 and AEGIS-2, with a slightly higher (%) average background count in AEGIS-3. However, we note that the observations in AEGIS-3 occurred 6–12 months after those observations in AEGIS-1 and AEGIS-2. The background levels in XDEEP2 Fields 2, 3 and 4 are almost identical for each of the three energy-bands.

For the purposes of comparing the X-ray point sources detected within each of the XDEEP2 fields, as well as comparing with previous X-ray surveys, it is important to understand the flux sensitivity limitations of a particular X-ray field. The faintest sources detected in the XDEEP2 fields have and . While these fluxes are good indicators of the ultimate sensitivity of the survey, sources similar to these may only be detected in stacked images close to the center of several ObsID aim-points where exposure levels are sufficiently high ( ks) and the combined PSF is relatively small. Hence, given an observing strategy with varying levels of exposure across the fields, X-ray sources at these low flux levels cannot be uniformly detected across the whole of each field. To quantify the expected number of sources as a function of survey area, we have constructed flux sensitivity maps for each merged field in the 0.5–2 keV, 2–7 keV and 0.5–7 keV bands.

Figure 14.— Example of an exposure-corrected full-band flux sensitivity map for an XDEEP2 field. The sensitivity map has been created as described in section 4.1. Areas with lightest (darkest) colors correspond to those regions of the map with the greatest (poorest) sensitivity.
Figure 15.— Survey solid angle as a function of the limiting flux in the soft-band, hard-band and full-band (left, center and right panels, respectively) for each XDEEP2 field. Limiting fluxes in the full-band where at least 10% of the survey field area are sensitive are , , and .

Maps of the Chandra point spread functions for an enclosed energy fraction of 90% were simulated at , 4.0 and 2.5 keV (mean SB, HB and FB energies, respectively) for each ObsID using the ciao tool mkpsfmap. These maps were then merged for all overlapping fields to calculate the mean in each image pixel for a merged counts image in the soft, hard and full-bands. We used the formalism of Lehmer et al. (2005) and employed a Poisson model to calculate the average number of counts () required to detect a source in a given image pixel for the background counts () enclosed within the mean calculated in the merged PSF model,


where , , and (Lehmer et al. 2005). Using equation 2 we convolve the merged PSF and background images at each image pixel and normalize to the appropriate merged exposure maps to create final fluxed sensitivity images in each energy band (three per field; an example sensitivity image is shown in Figure 14).

We calculate empirical sensitivity curves in the SB, HB and FB for each of the four XDEEP2 fields using the sensitivity images derived above (see Figure 15). Due to the small overlapping regions in Fields 2–4, the sensitivity curves are found to be relatively smooth over the entire survey region with relatively sharp cut-offs at , and . Hence, the sensitivity limit is approximately uniform across the majority of the survey area in Fields 2, 3 and 4. By contrast, the wedding-cake style observational setup of Field 1 combined with changing roll angles produces small ( deg) regions of high sensitivity, which combine over the field to produce a much more shallow sensitivity curve (i.e., the sensitivity is non-uniform). However, as the average exposure across Field 1 is –100 times greater than Fields 2–4, the mean sensitivity to the detection of faint sources is vastly improved in Field 1. We find that the limiting flux in the 0.5–7 keV band for source detection, which includes at least 10% of the survey area, is a factor lower in Field 1 () than in Fields 2–4 (, and , respectively).

4.2. Comparison of X-ray sources in Field 1 to Laird et al. (2009)


Positional36 Radii37 Net Counts Soft38 Net Counts Hard39 Net Counts Full40 Flux41 Hardness Ratio42 Flux Ratio43
DEEP244 DEEP245 CSC46 Err 47
Name Field Name () () (”) (”) (”) (”) (counts) (counts) (counts) ()

F1_AEG1 CXOJ141907.7+525946 214.78246 52.99710 0.84 4.48 10.36 632.0 16.62 5.77 31.17 8.86 23.17 7.15 48.97 12.25 39.81 8.74 80.26 14.70 4.80 1.44 20.6 5.62 19.2 3.78 0.20 0.04 0.43 4.75 3.18 7.66
aeg1_002 F1_AEG1 CXOJ141907.8+530025 214.78334 53.00712 0.32 4.45 10.30 628.4 162.4 13.99 311.9 19.55 57.06 9.42 95.73 13.98 219.1 16.46 406.0 23.62 50.6 3.21 40.6 6.48 102. 6.14 -0.53 -0.59 -0.47 0.88 0.69 0.97
aeg1_003 F1_AEG1 - 214.79521 52.98033 1.18 6.91 12.42 611.5 14.31 -1 22.97 -1 35.58 9.37 54.85 13.96 46.79 10.76 71.22 16.18 3.66 -1 25.1 6.30 18.3 4.09 0.53 0.30 0.84 12.37 7.12 -1
aeg1_004 F1_AEG1 CXOJ141911.2+530320 214.79699 53.05600 1.24 4.48 10.33 623.2 12.51 5.10 20.42 7.36 12.09 -1 23.29 -1 21.19 6.82 27.06 10.65 474. 91.2 824. -1 1710 208. -0.55 -1.00 -0.40 1.17 0.18 3.78
aeg1_005 F1_AEG1 CXOJ141919.9+530254 214.83506 53.04790 0.84 3.42 7.95 536.8 10.13 4.84 43.10 9.10 15.51 6.08 46.83 11.03 25.57 7.31 89.53 13.87 5.98 1.31 17.7 4.55 19.3 3.18 0.02 -0.11 0.20 3.26 2.25 4.34
aeg1_006 F1_AEG1 CXOJ141920.6+530028 214.83600 53.00792 0.29 3.03 7.11 514.7 117.3 12.04 226.4 16.77 12.68 -1 28.79 10.21 127.6 12.91 255.2 19.24 23.7 1.76 8.15 3.07 42.3 3.22 -0.79 -0.85 -0.70 0.41 0.25 0.56
aeg1_007 F1_AEG1 CXOJ141922.8+530132 214.84506 53.02555 0.21 2.86 6.75 498.7 100.1 11.24 161.1 14.59 121.0 12.44 222.9 17.61 221.2 16.32 384.7 22.44 18.4 1.66 73.1 5.75 70.0 4.07 0.16 0.10 0.21 4.05 3.54 4.44
aeg1_008 F1_AEG1 - 214.85376 52.99871 2.53 3.14 5.65 477.5 5.61 4.29 7.76 5.82 4.46 -1 7.09 -1 5.03 -1 8.13 -1 0.76 0.59 0.62 -1 0.78 -1 -0.54 -1.00 -0.45 2865. 0.06 -1
aeg1_009 F1_AEG1 - 214.85694 53.00549 0.43 2.53 6.01 469.5 20.15 5.88 42.87 8.77 27.37 6.90 41.56 10.01 47.39 8.60 83.71 12.87 4.32 0.88 12.0 2.90 13.4 2.07 -0.03 -0.17 0.13 2.88 2.14 3.96
aeg1_010 F1_AEG1 - 214.85765 53.01971 0.96 2.53 6.06 469.6 9.02 -1 16.26 -1 12.63 5.56 34.74 9.91 13.20 -1 36.99 11.10 1.63 -1 9.82 2.86 5.75 1.78 0.80 0.74 1.00 9117. 18.87 -1
aeg1_011 F1_AEG1 - 214.86239 53.03122 0.54 2.56 5.98 464.7 22.14 6.08 44.39 8.83 11.39 -1 24.43 9.07 33.24 7.62 69.73 12.22 4.56 0.91 7.06 2.70 11.3 2.02 -0.34 -0.51 -0.10 1.92 1.11 3.12
aeg1_012 F1_AEG1 - 214.86615 53.02515 0.56 2.44 5.75 453.7 10.17 4.71 29.25 7.81 18.13 6.08 27.73 9.26 28.58 7.23 58.53 11.67 2.91 0.78 7.87 2.67 9.24 1.86 -0.06 -0.23 0.20 3.14 1.96 4.95
aeg1_013 F1_AEG1 CXOJ141928.0+525840 214.86670 52.97822 0.25 2.42 5.70 461.1 76.61 9.93 149.7 13.82 37.22 7.62 58.93 10.64 113.9 12.08 208.9 17.02 15.9 1.47 18.0 3.26 35.5 2.89 -0.43 -0.51 -0.35 1.18 0.97 1.44
aeg1_014 F1_AEG1 - 214.87337 53.03977 0.29 2.37 5.63 448.2 52.24 8.47 87.44 11.22 29.69 7.07 59.41 10.85 81.87 10.58 146.5 15.17 8.73 1.13 16.8 3.15 23.2 2.44 -0.19 -0.29 -0.09 2.12 1.62 2.50
aeg1_015 F1_AEG1 - 214.87634 53.04383 1.07 3.21 5.78 446.2 13.05 5.34 22.20 7.27 4.52 -1 12.87 8.22 15.98 6.83 33.72 10.53 2.26 0.75 0.41 -1 5.35 1.74 -0.36 -0.70 0.05 0.24 0.09 0.64
aeg1_016 F1_AEG1 - 214.87842 53.00748 1.44 3.34 6.01 422.9 12.95 5.34 21.81 7.27 4.74 -1 7.56 -1 11.05 6.47 17.48 9.83 2.11 0.71 0.64 -1 2.67 1.53 -0.81 -1.00 -0.78 0.27 0.02 1.31
aeg1_017 F1_AEG1 CXOJ141930.8+525915 214.87886 52.98781 0.42 2.11 5.00 428.3 30.90 6.81 62.29 9.67 9.90 -1 18.33 -1 36.53 7.62 79.83 11.87 6.30 0.98 5.35 -1 13.0 1.91 -0.61 -0.75 -0.43 0.86 0.53 1.58
aeg1_018 F1_AEG1 - 214.88704 53.04167 1.12 2.96 5.33 421.8 10.14 4.85 14.38 6.31 4.26 -1 6.75 -1 13.21 6.28 20.96 9.27 1.51 0.65 0.61 -1 3.62 1.53 -0.43 -1.00 -0.23 0.78 0.16 4.05
aeg1_019 F1_AEG1 - 214.88706 52.99963 0.82 1.86 4.58 405.0 7.57 -1 12.61 -1 10.46 4.84 17.61 7.37 10.76 5.10 22.31 8.48 1.23 -1 4.92 2.08 3.44 1.32 0.52 0.39 1.00 7412. 6.61 -1
aeg1_020 F1_AEG1 - 214.88917 53.09005 1.37 4.92 8.88 496.5 20.36 6.37 24.09 8.12 16.40 -1 26.88 -1 21.16 8.00 31.18 -1 3.20 1.29 12.2 -1 7.86 -1 -0.71 -1.00 -0.64 2.09 0.16 6.51

Unique source identifier
XDEEP2 ObsID/sub-field name
Unique source identifier for matched XDEEP2 sources present in the Chandra X-ray Source Catalog (CSC)
X-ray position in J2000 co-ordinates (degrees) and associated centroid positional error (arc-seconds)
Aperture radius in arc-seconds at 50% and 90% the effective area of ACIS-I given the off-axis distance of the X-ray source
Off-axis distance in arc-seconds of X-ray source from aim-point of observation
Soft-band (0.5–2 keV) net counts and associated errors in the and apertures
Hard-band (2–7 keV) net counts and associated errors in the and apertures
Full-band (0.5–7 keV) net counts and associated errors in the and apertures
Total soft-band (S), hard-band (H) and full-band (F) fluxes and associated errors in units of
Classical hardness ratios () and associated 1 upper and lower limits calculated using the BEHR method
Flux ratios () and associated 1 upper and lower limits calculated using the BEHR method

Table 5XDEEP2 source catalog

As stated previously, while the analyses presented here include the recent 600ks observations of AEGIS 1–3, the 200ks X-ray source catalog for Field 1 (AEGIS-X) has been previously presented in Laird et al. (2009). Furthermore, the new 600ks observations will also be presented in a forthcoming paper (Nandra et al. in prep.) using similar detection and Bayesian-style sensitivity analyses to that used for the previous AEGIS-X catalog. Since the source detection and extraction analyses differ significantly between AEGIS-X and the XDEEP2 catalog presented here, we now compare the detection methods and results.

Figure 16.— a (left): Comparison of X-ray source fluxes in the 0.5–2 keV energy band for the 1260 sources in common between the previous AEGIS-X source catalog and the XDEEP2 catalog presented here. The AEGIS-X sources are corrected for galactic absorption. The XDEEP2 fluxes are converted to to match the AEGIS-X catalog. The dashed-line is the best-fit linear-bisector to the logarithms of the source fluxes. The blue-shaded region (dotted-lines) represents the 3 uncertainty on the linear correlation calculated using 2-dimensional linear regression analyses. The yellow-shaded region (dash-dot-lines) represents the Poissonian error on the source fluxes due to photon-counting statistics, derived using the formalism of Gehrels (1986). We find a very strong agreement at the 99.99% level between XDEEP2 and AEGIS-X source fluxes. The % of outliers are significantly extended in the ACIS-I images and/or have counts, suggesting that these sources are strong candidates for galaxy clusters and/or moderately variable QSOs. b (right): Soft-band source flux distributions (0.2 dex bin-width) for all X-ray sources detected in XDEEP2 (gray-shaded region) and AEGIS-X (blue-dash). The flux distribution of the 59 AEGIS-X source candidates which lack secure matches in XDEEP2 are highlighted with blue-hashed shading.

Briefly, detection of sources in AEGIS-X was carried out using a custom implementation of the CIAO wavdetect tool. Laird et al. (2009) perform several runs of the detection software using different probability thresholds to build seed catalogs and to derive multiple estimates of the X-ray background in the observation. The final probability threshold for which a particular candidate is determined to be false in AEGIS-X is comparable to that used in our analyses. Laird et al. detected source candidates separately in the full, soft, hard and ultra-hard band images.48 These source candidates were then combined into an individual source catalog using Bayesian techniques to statistically associate the source candidates and calculate the fluxes in the respective energy bands. The AEGIS-X catalog contains 1325 sources. Two sources (EGS4_258; EGS7_204, nomenclature adopted from Laird et al. 2009) in the AEGIS-X catalog were only detected in the ultra-hard band, i.e., an energy-band which we do not use due to the relatively small effective area of the telescope at these higher energies. Furthermore, four AEGIS-X sources (EGS4_240; EGS6_185; EGS7_194; EGS8_127) have 90% effective-area extraction regions which significantly (%) overlap with those extraction radii of other sources in the AEGIS-X catalog; from visual inspection we find that these four AEGIS-X sources (and their neighbors) are consistent with being single point sources. Hence, we remove these six sources from further comparison between the XDEEP2 and AEGIS-X catalogs.

We compared the 1319 unique source candidates identified in AEGIS-X to the 1720 source candidates identified in Field 1 of our XDEEP2 catalog solely on the basis of source position using the same varying matching radius method described in §5. We find that 1260 (%) of the source candidates identified in AEGIS-X are included in our new catalog. We have visually inspected each of the 59 AEGIS-X sources which were not identified in the XDEEP2 catalog. We find that the majority (44/59; %) of the AEGIS-X sources, which are not included as part of the XDEEP2 catalog, were detected by wvdecomp as source candidates in one energy band. However, on the basis of our MARX simulations, these 44 non-matched AEGIS-X sources did not meet our ultimate and more conservative count detection threshold and were removed as possibly spurious based on their low net counts (–10). A further seven of the 59 non-matched sources were flagged as ‘non-standard’ and possibly spurious; we discuss these seven sources below. Finally, eight of the 59 non-matched AEGIS-X source candidates are not detected using wvdecomp after the inclusion of the more recent 600ks data.

We now briefly discuss the seven of the 59 non-matched AEGIS-X source candidates (EGS2_052; EGS5_105; EGS6_073; EGS6_093; EGS7_180; EGS8_093; EGS8_134) that were initially detected by wvdecomp and then highlighted by our routine as ‘possibly spurious’. Visual inspection shows that three of these seven source candidates, EGS6_093, EGS5_105 and EGS7_180) have their expected source PSFs partially blended with secondary brighter sources. Indeed, EGS6_093, is located between ( arc-seconds) two significantly brighter X-ray sources (EGS6_164 and EGS6_165; both these sources are included in the AEGIS-X and XDEEP2 catalogs) causing sufficient detection ambiguity and EGS7_180 has an X-ray morphology consistent with that of a jet. These three sources, while initially detected in XDEEP2, are not included in our final catalog due to our inability to accurately separate the flux contribution from the neighboring bright source. Furthermore, EGS6_073 falls on a chip-gap; EGS2_052 has ; and EGS8_134 has % of its low source counts () in one ACIS-I pixel, and is conservatively removed based on our MARX simulation analyses. Finally, EGS8_093 is possibly part of an extended source which appears extremely diffuse and only has a marginal detection () in the AEGIS-X catalog.

As above, eight of the 59 non-matched AEGIS-X source candidates are not detected using wvdecomp after the inclusion of the 600ks data; while these sources were detected in the AEGIS-X analyses with , we note here that these sources may still be real, but are no longer detected due to intrinsic variability of the source. Similarly, Nandra et al. (in prep.) find that from a re-analysis of AEGIS-X, 17 of the AEGIS-X sources are no longer detected in the three sub-fields which include the new 600ks data. Assuming a similar number of non-detected source candidates across all of Field 1, this would suggest a false source contamination rate of sources (%) in AEGIS-X.

In Figure 16 we show a comparison between the soft-band fluxes for isolated and formally-detected point-sources in the AEGIS-X (classically derived flux) and XDEEP2 catalogs. We have converted the fluxes we derived using in XDEEP2 to , as used in AEGIS-X, using a conversion of 1.031 and we have corrected the AEGIS-X fluxes for galactic absorption (a factor of 1.042; Laird et al. 2009). We find excellent agreement between the fluxes derived in the XDEEP2 and AEGIS-X catalogs with a Spearman’s rank coefficient of which is significant at % level. Additionally, we have used two-dimensional linear-regression analyses to calculate the uncertainty on the derived correlation between the XDEEP2 and AEGIS-X source fluxes (dotted-lines in Figure 16), and the error region on the photon counts used to derive the source fluxes (dash-dot-lines in Figure 16). As expected, the Poissonian error due to low source counts significantly dominate the uncertainty towards low fluxes. We find that 12 (%) of the matched XDEEP2–AEGIS-X sources lie substantially outside the error region. These outlying sources have large numbers of counts () and/or are significantly extended beyond the expected 90% EEF angular size in the merged ACIS-I images. This suggests that these outlying sources are strong candidates for galaxy clusters and/or moderately variable quasi-stellar objects (QSOs).49 Furthermore, variations in extraction radii at large offaxis distances, due to the introduction of the more recent ACIS-I observations (which were performed with substantially different roll-angles) may potentially cause significant differences in measured counts/flux for bright X-ray sources with non-point-like profiles, such as galaxy clusters. Indeed, we find that when considering only the previous 200ks observations studied in Laird et al., with matched extraction apertures, the fluxes for all of the matched XDEEP2–AEGIS-X sources are consistent to within 1.

In Figure 16 we show the source flux distributions in AEGIS-X and XDEEP2 including the 460 new XDEEP2 sources which are detected in the new deeper 600 ks observations. As expected, the majority of these 460 new XDEEP2 sources have , extending the distribution of the previous catalog to lower source fluxes. We additionally highlight the fluxes of the 59 AEGIS-X source candidates, which we conservatively do not include in XDEEP2. Each of these non-matched sources have , with the vast majority at the extreme low-flux end of the main AEGIS-X source-flux distribution (). Using a Bayesian counterpart matching algorithm, which we present in § 5, we have attempted to assign DEEP2 optical counterparts to the 59 AEGIS-X source candidates. We find that the majority (35/59; %) of these AEGIS-X source candidates lack secure optical counterparts; this is a factor larger than the fraction of X-ray sources which lack counterparts across the entire XDEEP2 sample (%). However, based on simulations of a purely random set of 59 source positions, we would expect only –7 spurious counterpart matches using our Bayesian matching algorithm. Hence, the 24 AEGIS-X sources found to coincide with an optical counterpart is a factor –8 larger than the random expectation of spurious counterparts, suggesting that some of these X-ray sources may be real.

Based on our rigorous comparison of the AEGIS-X catalog and our XDEEP2 catalog, we suggest that the two catalogs appear to be in excellent agreement, despite the use of different detection algorithms (wavdetect versus wvdecomp). In general, the small (%) discrepancy between the catalogs can be attributed to the removal of low significance sources in the XDEEP2 catalog based on our MARX simulations. Additionally, we stress that since 51 of the 59 low significance sources are initially identified by both wavdetect and wvdecomp, we cannot rule out that they are real sources, although they ultimately did not meet our more conservative detection criteria.

4.3. Comparison of X-ray sources in Fields 2–4 to the Chandra Source Catalog

Figure 17.— a (left): Soft-band (0.5–2 keV) flux distributions for all X-ray sources identified in the Chandra ACIS-I observations of XDEEP2 Fields 2 (red), 3 (blue) and 4 (green) within the XDEEP2 (solid lines) and CSC (dashed-lines) catalogs. b (right): Comparison of 0.5–2 keV fluxes for all XDEEP2 sources associated with CSC sources in Fields 2 (red dots), 3 (blue dots) and 4 (green dots). Error bar represents the median uncertainty in the flux estimates for the CSC and XDEEP2 sources.

The Chandra Source Catalog (CSC) is a compilation of all relatively bright X-ray sources detected in single ACIS and HRC imaging observations by the Chandra X-ray Observatory in the first eight years of the mission (Evans et al. 2010). In principle, the X-ray sources detected in XDEEP2 Fields 2–4 and by the CSC are likely to be equivalent. Similar to the CSC, we have not attempted to merge events in overlapping regions of Fields 2–4 as, in general, these regions occur at large off-axis distances where the Chandra PSF is poor. In this section, we compare the XDEEP2 source properties to those detected in the CSC release 1.1. The current CSC data release contains X-ray data products and information (positions; spatial; temporal multi-band count rates; fluxes) for distinct point sources and compact sources, with observed spatial extents ” observed in publicly released data to the end of 2009. Highly extended sources, and sources located in selected fields containing bright, highly extended sources are excluded in the CSC. See http://cxc.cfa.harvard.edu/csc/index.html for further information.

We have used the publicly available java-applet, CSCview to associate the XDEEP2 X-ray sources in Fields 2–4 to the CSC Master Catalog. Although we do not attempt to merge the individual ACIS-I observations in Fields 2–4, we find that only % (i.e., 150/342; 218/528; 158/386 sources in Fields 2, 3 and 4, respectively) of the XDEEP2 sources are identified in the CSC. In Figure 17a we show a comparison of the flux distributions for all XDEEP2 sources and CSC sources within the area covered by Fields 2, 3 and 4 of XDEEP2. The 90% EEF aperture fluxes produced by the CSC are derived using a simple absorbed powerlaw with and . Hence, for the purposes of comparison we convert the field-specific used to derive the XDEEP2 fluxes to match the CSC fluxes.

We find that while all CSC sources with are identified in the XDEEP2 catalog, the vast majority of the lower flux XDEEP2 sources are not included in the CSC. By design, the detected CSC X-ray sources have counts for an on-axis source (–30 counts for an off-axis source), i.e., the CSC catalog only includes sources whose flux estimates are greater than three times their estimated 1 uncertainties. However, as we have shown in Figures 11 and 12, and has been shown conclusively by many other deep and wide-field X-ray surveys (e.g., CDF-N; CDF-S; C-COSMOS; AEGIS-X; XBootes), many X-ray sources can be significantly identified with only –5 net counts, although the source flux will remain relatively unconstrained due to Poisson uncertainties. Indeed, % of the XDEEP2 sources not identified in the CSC catalog have counts. Furthermore, to within , we find excellent agreement for the X-ray fluxes of the sources in common between XDEEP2 and the CSC (see Figure 17b).

Given that all of the CSC sources within the survey area are identified in the XDEEP2 catalog and the non-matched sources have lower counts/flux which lie above the thresholds derived from our extensive simulation analyses, we find that the CSC provides a more conservative identification of X-ray sources within the XDEEP2 fields. For completeness, we have also associated the X-ray sources identified in Field 1 to the CSC catalog, and find there are 689 distinct X-ray sources in common between the catalogs. The faintest CSC sources in Field 1 have , but with the majority at (i.e., an average factor more sensitive per individual observation than Fields 2–4). For ease of comparison with future surveys, we include the CSC source identifiers as part of the XDEEP2 catalog, for all XDEEP2 sources with CSC counterparts.

4.4. Source spectral properties: hardness ratios

Using the Bayesian Estimator of Hardness Ratio (BEHR) method (Park et al. 2006), hardness count ratios (HR), defined as , where and are the counts in the soft and hard bands respectively, as well as the hardness flux ratios (FR), defined as , were calculated for all detected sources in the XDEEP2 catalog. FR and HR and their associated uncertainties calculated using BEHR are available in the main XDEEP2 source table. Briefly, BEHR treats the detected source and background X-ray photons as independent Poisson random variables, and uses a Monte Carlo based Gibbs sampler to select samples from posterior probability count distributions to correctly propagate the non-Gaussian uncertainties, which derive from the calculation of hardness ratios. BEHR is particularly powerful in the low-count Poisson regime, and computes a realistic uncertainty for the HR and FR, regardless of whether the X-ray source is detected in both energy bands. In Table 5, we include the FR and HR ratios with the associated upper and lower limits for all XDEEP2 sources. Sources with unconstrained upper or lower limits due to non-detections are denoted by “-1” in the appropriate uncertainty column.

In Figure 18 we show the FR distribution for the XDEEP2 sources. Typically, the XDEEP2 sources which are detected in both the hard and soft-bands have FR in the range –7, with distribution tails at low and high values of FR. Following previous studies (e.g., Bauer et al. 2002; Alexander et al. 2003a; Luo et al. 2008), we divide the X-ray sources with low and high-flux at (i.e., the 10% flux limit of the shallow exposure XDEEP2 fields). While the choice of cut is somewhat arbitrary, clearly we find the same general trend towards higher values of FR for X-ray sources with low-fluxes as has been observed previously (e.g.,Hasinger et al. 1993; Vikhlinin et al. 1995; Giacconi et al. 2002; Tozzi et al. 2006). We find that the distribution of FR values is moderately peaked at sources with high flux (). By contrast, lower flux sources have a more extended distribution, with a median value of and tailing to higher values of FR. Using the ciao spectral analysis package, sherpa, we have simulated X-ray spectra for AGN populations at in order to quantify the evolution of X-ray spectral slopes due to the k-correction of the observed AGN spectra towards high-z. Based on these simulations, we find that the two peaks observed in the FR distributions are co-incident with the spectral slopes expected for two separate AGN populations with –1.4 and –1.8. Further, we find that the majority of the 460 low-flux sources in Field 1, which were not previously identified in AEGIS-X due to insufficient survey depth (see § 4.2), have a similarly wide FR distribution (–10) to the AEGIS-X source candidates and the sources identified in Fields 2–4. However, the median FR for the new faint Field 1 sources is shifted slightly higher with FR3 (i.e., harder spectral indices), suggesting that these new sources have flatter X-ray spectral slopes, and are likely to be more heavily obscured. Hence, their previous non-detection in the 200ks data is due to the combined result of AGN luminosity, distance and intrinsic obscuration.

Figure 18.— Main panel: Flux band ratio defined as as a function of full-band counts () for all XDEEP2 sources detected in the soft and hard energy bands. Average spectral slopes for fixed values of FR established from X-ray spectral simulations using Sherpa are highlighted with horizontal dotted lines. Right panel: FR distributions for all detected sources within XDEEP2 with (dot-dashed) and (dotted).
Figure 19.— Main panel: Logarithm of the number of detected sources within the XDEEP2 catalog brighter than a given soft-band flux (; black solid line) versus the logarithm of soft-band flux (i.e., the log–log distribution). For comparison to previous surveys, source fluxes were converted using . We use a Monte-Carlo simulation to assess the 90% uncertainty on the XDEEP2 distribution due to flux errors and Poisson counting statistics (shaded region). We compare this distribution to previous surveys fields [CDF-N (Bauer et al. 2004); Extended-CDF-S (Lehmer et al. 2005); Chandra-COSMOS (Puccetti et al. 2009)] and to an X-ray background synthesis model (Gilli et al. 2007). Top panel: Logarithm of the residuals between XDEEP2 and the comparison curves. The logarithm of the uncertainty for the log–log is shown by the shaded region. We find good agreement (% deviation) between XDEEP2 and all previous observation surveys in the regime . However, we show a mild systematic offset towards lower for sources with , in closer agreement with XRB models.

4.5. XDEEP2 source number counts

We have calculated the cumulative number of sources in the XDEEP2 catalog () detected per square degree that are brighter than a given flux in the soft (0.5–2 keV) band, i.e., the distribution (see Figure 19). This provides a good check that the merging of the datasets and the extensive calibrations were performed correctly, as well as an excellent comparison to previous X-ray surveys. We choose to compare in the soft-band as this is the most sensitive energy and the specific energy range definition of the soft-band (0.5–2 keV) is consistent across previous surveys. As a consequence of (1) the changing slope of the distribution towards fainter fluxes, and (2) observationally fainter sources possibly being more obscured and/or lower accretion rate AGN than brighter sources, the so-called ‘Eddington bias’ introduces many statistically low-significance sources at the sensitivity limit of the X-ray survey. Hence, we have empirically restricted our analyses presented in this section to only those sources detected with , i.e., on-axis 0.5–2 keV fluxes of in Field 1 and in Fields 2–4 (equivalent to and , respectively). For the purpose of comparison, we have converted all source and field fluxes to and use the combined flux limits (see §4.1) to construct the distribution.

To quantify the uncertainties on the derived , we have used a Monte-Carlo (MC) style simulation. Using the formal error on the source flux, we built symmetrical probability flux distributions () for each source to be input to 10,000 realizations of our simulation. Within the MC simulation, we randomly assign fluxes to each source within the sample based on the individual , and recompute the distribution. The total 90% uncertainty on the is then defined as the mean absolute deviation of the 10,000 simulated distributions combined in quadrature with the 90% Poissonian error on the main distribution, defined using the formalism of Gehrels (1986). From our MC simulations, in Figure 19 we show that the XDEEP2 is very well constrained ( dex) in the flux range owing to the large sensitive area in Field 1 around the ‘knee’ of the at 6–8. However, we find that the uncertainty on the distribution increases to  dex towards the bright flux tail () of the . We determined that this is caused by the decrease in the space-density of the far rarer bright sources, combined with the relatively large uncertainties on the fluxes for those sources identified in the more shallow exposure Fields 2–4. For these particular sources, which dominate the distribution within this moderate–high flux regime, the majority are detected with relatively few counts (–15) and hence, flux errors are –50% of the overall flux. In turn, these relatively large flux uncertainties cause significant scatter of the sources within the simulated distributions.

In Figure 19, we additionally compare the derived from XDEEP2 to the distributions found in previous wide and deep Chandra surveys [CDF-N (Bauer et al. 2004); Extended-CDF-S (Lehmer et al. 2005); Chandra-COSMOS (Puccetti et al. 2009)]. In the flux range , we find excellent agreement with these previous surveys. We confirm previous results (e.g., Luo et al. 2008), that the CDF-N field may be subject to mild cosmic variance, as it appears to over-estimate (a factor –4) the number count distribution of sources with . Furthermore, using the X-ray background (XRB) synthesis models of Gilli et al. (2007), we have simulated the expected distribution of both obscured and unobscured populations of AGN with  cm, in the redshift range –8. In accordance with previous surveys, we consistently underestimate the number counts of AGN with in comparison to that expected from the XRB (see upper panel of Figure 19), suggesting that many heavily obscured sources are still being missed in even the most sensitive surveys. Indeed, multi-wavelength studies of deep and wide field X-ray surveys find a large population of seemingly obscured AGN which remain undetected using X-ray data alone (e.g., Alonso-Herrero et al. 2006; Donley et al. 2007; Daddi et al. 2007; Meléndez et al. 2008; Fiore et al. 2009; Brusa et al. 2010; Goulding et al. 2011; Georgantopoulos et al. 2011; Alexander et al. 2011). However, for XDEEP2 sources with , we find a mild (–50%) systematic offset from previous X-ray surveys (e.g., E-CDFS; C-COSMOS), resulting in number counts closer to those predicted by XRB models, although the results from each of these surveys are all consistent at the 90% significance level.

5. Optical DEEP2 & X-ray XDEEP2 source identification

By design, the XDEEP2 Chandra survey is within the same spatial region as the DEEP2 Galaxy Spectroscopic Redshift survey fields. In this section we identify optical counterparts to the sources in the XDEEP2 catalog using a custom Bayesian style analysis. For DEEP2, optical , and -band photometry was obtained with the Canada-France-Hawaii Telescope (CFHT) 12k camera. The main photometric catalog contains over ,000 sources with a typical absolute astrometric accuracy of arc-seconds and is complete to (see Coil et al. 2004). In Table 6 we show the breakdown for the approximate number of optical sources within the XDEEP2 survey fields. In DEEP2 Field 1, all galaxies which have magnitudes of were targeted for spectroscopy using the DEIMOS spectrograph on Keck (see Davis et al. 2003 for further information on the observational setup of DEEP2). However in Fields 2–4, only those galaxies which meet both a simple color-cut threshold and have magnitudes of were targeted. The 4th data release of the DEEP2 spectroscopic catalog contains 50,319 unique sources (Newman et al., 2012).

5.1. A Bayesian optical–X-ray matching routine

Figure 20.— Fraction of counterparts found between the DEEP2 optical catalog and 100 randomly simulated X-ray catalogs. We find that the fraction of spurious matches decreases rapidly as a function of the probability threshold () calculated in our Bayesian-style matching algorithm. For , we expect a spurious matching fraction of % (dashed-line).

Given the unique observational construction of the combined XDEEP2 survey, in that it is both relatively shallow in wide areas, while simultaneously being extremely deep in smaller regions across the fields, we require a method of source matching which will account for changes in both the optical and X-ray source densities and statistically associate bright X-ray sources in the shallow fields with optical counterparts which are likely to contain bright AGN (i.e., QSOs). To this end, we have extended the Bayesian source-matching algorithm of Brand et al. (2006) to now include the X-ray source density and the properties of the candidate optical counterparts. Briefly, this method uses Bayesian-style statistics to calculate the probability of a random association occurring between two counterparts given the angular and magnitude distributions of the optical sources in a specific region of the sky. Simultaneously this algorithm accounts for the distribution of matching radii appropriate for a given off-axis position of the X-ray source in a Chandra observation. Furthermore, we allow modifications to the optical source positions, assuming a Gaussian probability based on the centroid and astrometric error of the DEEP2 data. As stated previously, median offsets between DEEP2 and XDEEP2 have been removed a-priori (see § 2.4). We use a Gaussian prior based on the characteristics of the Chandra PSF for the positional uncertainty of the X-ray source to derive the probability, of an X-ray source having an optical counterpart within the catalog (i.e., the survey mean completeness). We combine these posterior assumptions with information specific to the X-ray source (total counts, background level, proximity to other X-ray sources) and the optical properties (star, normal galaxy, quasar etc.) of possible counterparts to assign likelihood association probabilities between pairs of sources. In our new implementation of the algorithm, the probability of identifying an X-ray source with optical source is then,


where is the simple Gaussian probability of associating an X-ray source with an optical counterpart at a given separation including the X-ray and optical positional uncertainties; are the Poisson-idealized number counts as a function of optical magnitude within a region encompassing the X-ray position, in effect, accounts for both the changing -band magnitude depth and source density within the optical DEEP2 catalog; is the probability that an X-ray source of a given flux and flux limit has an optical association which is then marginalized over the -band magnitude of the proposed optical counterpart; and is the probability function containing the optical classification of the source, and is essentially a weighting based on the probabilistic galaxy classification of the source ( of 0 (star) to 1 (galaxy) defined in Coil et al. 2004) derived from the optical photometry and SED fitting. We determine the priors for by randomly selecting from a cumulatively summed set of Poisson distributions in Markov-Chain simulations of the X-ray and optical catalogs. For computation speed, we limit the counterpart selection to only optical sources detected in the -band. This also conforms with the selection method used to determine targets for optical spectroscopy. We note here, that while this method increases our ability to include optical sources with R-band magnitudes fainter than the completeness limit of the DEEP2 survey ( mags), the identification of X-ray sources with optically-faint counterparts is still incomplete at (e.g., Alexander et al. 2001; Brusa et al. 2010).

Figure 21.— Positional offset between optical and X-ray positions for the 2126 XDEEP2 X-ray sources with secure DEEP2 optical counterparts found using our Bayesian-style matching algorithm. The spread in residuals is approximately Gaussian across all four DEEP2 fields with a mean positional offset of arc-seconds between the X-ray and optical source catalogs.

To compute the probability threshold required to accept the optical source as a counterpart to the X-ray source and to quantitatively assess the false association fraction, we simulated mock XDEEP2 catalogs and compared them to the optical DEEP2 catalog. Following Brand et al. (2006), we randomized the positions of the XDEEP2 sources by offsets and compared the number of false matches produced. In Figure 20 we show the behaviour of the fraction of spurious counterparts for a given matching probability threshold () produced by our association routine. We find that using produces one spurious optical counterpart for % of the X-ray sources in the randomized catalogs (see Figure 20). The spurious counterpart fraction of % is chosen specifically to be consistent with that found for the previous AEGIS-X catalog which was matched using the Maximum Likelihood technique (see Civano et al. 2012, and references there-in); in turn, this also allows for further comparison between the catalogs. In 100 Markov-Chain Monte-Carlo (MCMC) simulations, we find that the spurious fraction remains relatively constant for across all four XDEEP2 fields with an overall dispersion of % within the MCMC simulations.

Field #50 51 52 53 54 Median 55 Median 56
1 1720 1183 68.8 -0.02 0.01
2 342 254 74.3 0.05 0.02
3 528 381 72.2 0.04 0.04
4 386 308 79.8 0.04 -0.04

XDEEP2 field number
Number of X-ray sources in the XDEEP2 field
Approximate number of optical sources in the XDEEP2 field region
Number of X-ray sources with secure optical counterparts
Percentage fraction of X-ray sources with secure optical counterparts
Median positional offsets between the DEEP2 optical source co-ordinates and the X-ray source co-ordinates in arc-seconds

Table 6X-ray sources with optical counterparts

In Figure 21 we present the offset in astrometric co-ordinates between the X-ray source position and that of the optical counterpart from the XDEEP2 catalog. We find that the spread in positional offsets is approximately Gaussian across all four DEEP2 fields with a mean positional offset of arc-seconds with an approximately zero systematic offset between the two catalogs. This mean offset is consistent with that found in previous deep-wide surveys (e.g., C-COSMOS with 0.81” for 90% of the sources; Elvis et al. 2009; Civano et al. 2012) Furthermore, we find that the positional offset between the X-ray source and optical counterpart appears to be a moderately-strong function of the ACIS-I off-axis position with on-axis () and off-axis () X-ray sources having median offsets of and , respectively.

Figure 22.— R-band (AB) magnitude versus full-band (0.5–7.0 keV) flux for all XDEEP2 sources. X-ray sources are divided between those with galaxy probabilities () (i.e., optically extended sources; open circles) and (i.e., point-like sources; open stars). X-ray sources which lack optical counterparts are shown with upper-limits at , i.e., the magnitude-limit of the DEEP2 catalog. Additionally, constant X-ray–optical flux ratios () are shown for log, calculated using the relation of McHardy et al. (2003).
X-ray Optical Photometry57
XDEEP258 Field59 60 61 DEEP262 63 64 65 66 P(gal)67
Name () () Objno () () (”) (AB mag)

1 214.78246 52.99710 - - - - - - - - -
aeg1_002 1 214.78334 53.00712 13036677 214.78314 53.00728 0.72 0.5646 3 21.62 20.62 20.10
aeg1_003 1 214.79521 52.98033 13027633 214.79494 52.97998 1.38 0.7309 0.55 26.23 23.32 21.93
aeg1_004 1 214.79699 53.05600 13036612 214.79712 53.05598 0.29 - 3 20.31 18.55 17.91
aeg1_005 1 214.83506 53.04790 - - - - - - - - -
aeg1_006 1 214.83600 53.00792 13036601 214.83597 53.00809 0.63 - -2 16.67 16.36 16.23
aeg1_007 1 214.84506 53.02555 13035495 214.84502 53.02559 0.20 - 3 24.37 24.62 23.94
aeg1_008 1 214.85376 52.99871 13027346 214.85321 52.99912 1.90 - 3 24.13 23.49 23.26
aeg1_009 1 214.85694 53.00549 13100779 214.85669 53.00582 1.28 - 3 24.69 24.47 24.27
aeg1_010 1 214.85765 53.01971 13035756 214.85777 53.02011 1.45 - -2 22.79 21.16 20.41
aeg1_011 1 214.86239 53.03122 13035995 214.86253 53.03141 0.73 - 3 23.59 21.68 21.09
aeg1_012 1 214.86615 53.02515 - - - - - - - - -
aeg1_013 1 214.86670 52.97822 13027372 214.86674 52.97823 0.10 0.5608 0.81 23.00 22.87 22.61
aeg1_014 1 214.87337 53.03977 13035981 214.87335 53.03982 0.19 - 3 25.15 23.66 22.84
aeg1_015 1 214.87634 53.04383 13035650 214.87601 53.04362 1.03 0.3722 3 26.80 23.31 21.95
aeg1_016 1 214.87842 53.00748 13035444 214.87830 53.00769 0.81 - 1.00 26.06 23.14 22.27
aeg1_017 1 214.87886 52.98781 13027475 214.87888 52.98786 0.20 - 0.00 23.15 20.26 18.03
aeg1_018 1 214.88704 53.04167 - - - - - - - - -
aeg1_019 1 214.88706 52.99963 13027149 214.88704 52.99970 0.25 - 3 24.25 23.77 22.93
aeg1_020 1 214.88917 53.09005 - - - - - - - - -

XDEEP2 unique source identifier
XDEEP2 field number
X-ray source position in J2000 co-ordinates (degrees)
DEEP2 optical source identifier (Coil et al. 2004)
Optical source position in J2000 co-ordinates (degrees)
Angular separation between optical and X-ray source positions (arc-seconds)
Redshift of optical counterpart
Bayesian probability of being a galaxy based on -band image (: star/compact; : galaxy/extended; see Coil et al. 2004)
Optical photometry in the , and -bands (AB-magnitude)

Table 7DEEP2 X-ray–optical counterpart catalog

5.2. X-ray–optical source properties

Of the 2976 X-ray sources in XDEEP2, we find that 2126 (%) have at least one secure optical counterpart in the DEEP2 optical catalog. Multiple candidate counterparts are found for % of the X-ray sources in XDEEP2. When multiple optical counterparts are associated with one X-ray source, we accept the DEEP2 optical counterpart with the largest . Given the cumulative distribution of found for the XDEEP2 counterpart catalog, we expect a final spurious counterpart fraction of %. In Table 6 we show the breakdown by field of the number of X-ray sources with optical counterparts, the percentage identified and the median positional offset between the optical and X-ray source positions. We find that 943 (%) of the X-ray sources in Fields 2–4 have secure optical counterparts compared with 1183 (%) in Field 1. This higher fraction of secure counterparts in Fields 2–4 is, in all likelihood, due to the relatively shallow exposure of the Chandra observations in Fields 2–4 compared to those in Field 1, and hence, brighter X-ray sources tending towards bright optical host galaxies (i.e., X-ray-to-optical flux ratios –10) which has been found previously in very shallow wide-field X-ray surveys (e.g., Maccacaro et al. 1988; Stocke et al. 1991; Akiyama et al. 2000; Lehmann et al. 2001; Murray et al. 2005). Indeed, AGN and QSOs are typically found to have similar ratios of  log (e.g., Schmidt et al. 1998; Akiyama et al. 2000; Lehmann et al. 2001). In Figure 22 we show the full-band X-ray flux versus the DEEP2 -band magnitude for the sources with secure optical counterparts. We illustrate approximate X-ray-to-optical flux ratios for the sources assuming the relation of McHardy et al. (2003), and we divide the sample between those optical sources identified in DEEP2 to be extended/galaxy (; see Coil et al. 2004; Newman et al. 2012) and point-like sources (stellar or QSO; ). Of the 1559 optically extended X-ray-optical sources, % (1425) have log, suggesting a significant fraction are bright AGN. We also find that 77 X-ray sources are also detected with very low X-ray-to-optical flux ratios (i.e., log. These X-ray sources generally include normal galaxies, stars, and low-luminosity AGN, and as we show in Figure 22, all 77 X-ray–optical sources with are point-like suggesting a stellar origin for the X-ray emission. As is clearly evident from the distribution of galaxies in Figure 22, our X-ray–optical source matching becomes incomplete towards optically-faint () systems for due to the flux-limit of the optical DEEP2 data when compared to the depth of the X-ray observations within Field 1.

Figure 23.— Redshift histograms for all 510 XDEEP2 galaxies with optical spectroscopic counterparts (solid-line) and all optical DEEP2 galaxies (dashed-line). The optical DEEP2 distribution is divided by a factor for 60 for ease of comparison to the X-ray sources. On the top-axis we show the present-day look-back times as a function of redshift, with