1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass[apj]{emulateapj}
3: \topmargin +.3in
4: \usepackage{lscape}
5: \shortauthors{Cardamone et al.}
6: \shorttitle{Mid-IR colors of X-ray detected AGN}
7: \begin{document}
8: \title{Mid-infrared Properties and Color Selection for X-ray Detected AGN in the MUSYC ECDF-S field}
9: \author{Carolin N. Cardamone\altaffilmark{1,2}, C. Megan Urry\altaffilmark{1,2}, Maaike Damen\altaffilmark{3}, Pieter van Dokkum \altaffilmark{1,2}, Ezequiel Treister \altaffilmark{4} , Ivo Labb$\acute{e}$\altaffilmark{5}, Shanil N. Virani \altaffilmark{1,2}, Paulina Lira\altaffilmark{6}, and Eric Gawiser\altaffilmark{7},}
10: \email{ccardamone@astro.yale.edu}
11: \altaffiltext{1}{Department of Astronomy, Yale University, New Haven, CT 06511, USA}
12:
13: \altaffiltext{2}{Yale Center for Astronomy and Astrophysics, Yale University, P.O.~Box 208121, New Haven, CT 06520}
14:
15: \altaffiltext{3}{Leiden Observatory, Universiteit Leiden, 2300 RA Leiden, Netherlands}
16:
17: \altaffiltext{4}{European Southern Observatory, Santiago, Chile}
18: \altaffiltext{5}{Carnegie Observatories, Pasadena, CA 91101}
19: \altaffiltext{6}{Departamento de Astronomķa, Universidad de Chile, Casilla 36-D, Santiago, Chile}
20: \altaffiltext{7}{Department of Physics \& Astronomy,Rutgers University, Piscataway, NJ 08854-8019}
21:
22:
23: \begin{abstract}
24: We present the mid-infrared colors of X-ray-detected AGN and explore mid-infrared selection criteria.
25: Using a statistical matching technique, the likelihood ratio, over 900 IRAC counterparts were identified with a new MUSYC X-ray source catalog that includes $\sim$ 1000 published X-ray sources in the {\it Chandra} Deep Field-South and Extended {\it Chandra} Deep Field-South.
26: Most X-ray-selected AGN have IRAC spectral shapes consistent with power-law slopes, $f_{\nu} \propto \nu^{\alpha}$, and display a wide range of colors, $-2 \le \alpha \le 2$.
27: Although X-ray sources typically fit to redder (more negative $\alpha$) power-laws than non-X-ray detected galaxies, more than 50\% do have flat or blue (galaxy-like) spectral shapes in the observed 3-8 $\mu$m band.
28: Only a quarter of the X-ray selected AGN detected at 24 $\mu$m are well fit by featureless red power laws in the observed 3.6-24 $\mu$m, likely
29: the subset of our sample whose infrared spectra are dominated by emission from the central AGN region.
30: Most IRAC color-selection criteria fail to identify the majority of X-ray-selected AGN, finding only the more luminous AGN, the majority of which have broad emission lines.
31: In deep surveys, these color-selection criteria select 10-20\% of the entire galaxy population and miss many moderate luminosity AGN.
32:
33: \end{abstract}
34:
35: \keywords{(galaxies:) quasars: general --- infrared: galaxies --- X-rays: galaxies}
36:
37: \section{INTRODUCTION}
38: Theoretical arguments and observational evidence indicate that Active Galactic Nuclei (AGN) may play an important role in galaxy evolution \citep{fabian2003,fabianetal2006,crotonetal2006,krieketal2006,sijackietal2007,schawinskietal2007}.
39: The study of this mutual feedback requires a complete sample of AGN, their host galaxies, and comparable inactive galaxies at the same epochs.
40: To discover these populations, surveys with deep multi-wavelength coverage are necessary.
41:
42: Today, AGN are identified at many wavelengths, each of which selects a slightly different sample.
43: For example, radio surveys preferentially find radio-loud AGN \citep{brinkmannetal2000}, while optical and ultraviolet-excess surveys find luminous unobscured AGN \citep{richardsetal2004,Mazzarellabalzano1986}.
44: Because they detect all but the most heavily obscured AGN, hard X-ray surveys
45: efficiently select relatively unbiased and complete samples of AGN \citep{branthasinger2005}, even if
46: X-ray background models predict a still-hidden population of Compton-thick AGN \citep{treisteretal2004,worsleyetal2005}.
47: For obscured AGN (roughly three quarters of the total population), the rest frame optical light is dominated by the host galaxy (e.g., \citealt{treisteretal2004,treisterurry2006}).
48: This obscuring gas and dust eventually re-emits the absorbed X-rays at infrared wavelengths.
49: In general the optical/infrared spectral shapes of AGN contain two broad components of thermal emission: the ``Big Blue Bump'' seen at optical and ultraviolet wavelengths, often attributed to emission from a hot accretion disk, visible only in unobscured AGN and an infrared excess attributed to re-radiation from circumnuclear dust on larger scales visible in all AGN in the far-infrared \citep{sandersetal1989,sanders1999,pierkrolik1992}.
50:
51: Mid-infrared diagnostic studies have found strong continuum emission, originating in dust heated by the intense nuclear radiation field, an important indicator of AGN emission \citep{laurentetal2000}.
52: Indeed, surveys with the {\it Spitzer Space Telescope} Infrared Array Camera (IRAC; \citealt{fazioetal2004}) are finding AGN via color-selection techniques by targeting strong red power-law continua \citep{Alonso-Herrero2006,hickoxetal2007,donleyetal2007}.
53: This emission can be used to find even Compton-thick AGN \citep{pollettaetal2006}.
54: However, many color-selection techniques developed with {\it Spitzer} data have been based on the infrared colors of optically-selected AGN samples in shallow surveys (e.g., \citealt{lacyetal2004,sternetal2005,hatziminaolgou2005}).
55: The ideal AGN selection technique is both highly reliable, meaning most objects selected are indeed AGN, and highly complete, meaning most AGN are identified.
56: To evaluate and improve infrared selection of AGN, it is important to investigate how the mid-infrared colors of X-ray-selected AGN samples compare to those of optically-selected samples.
57:
58: Finding high redshift AGN in multi-wavelength surveys requires both a large area and sensitive detection limits.
59: The MUltiwavelength Survey by Yale-Chile (MUSYC\footnote[1]{\url{http://www.astro.yale.edu/MUSYC}}; \citealt{gawiseretal2006}) is a square-degree survey of four fields to
60: limiting depths of $R_{AB}=26$ and $K_{AB}=22$ ($K_{AB}=23$ in the central areas; \citealt{quadrietal2007}), with extensive
61: follow-up spectroscopy.
62: One MUSYC field, the Extended {\it Chandra} Deep Field-South (ECDF-S),
63: has been surveyed by the {\it Chandra} X-ray observatory \citep{giacconietal2002,alexanderetal2003,lehmeretal2005,viranietal2006}.
64: The sensitivity of the {\it Chandra} data allows for
65: detection of luminous AGN out to large redshifts\footnote[2]{Luminous AGN ($L_X \sim 10^{44} \: {\rm ergs \: s^{-1}}$) can in principle be detected to very high redshifts in the ECDF-S ($z \sim 10$), but the volume sampled is too small to make such a detection likely.}
66: and more modest luminosity AGN ($\sim 10^{42-44} \: {\rm ergs \: s^{-1}}$) back to the quasar era ($z\sim2$, \citealt{branthasinger2005}).
67: Recently, the existing MUSYC data in the ECDF-S were augmented with very deep IRAC {\it Spitzer} data in the SIMPLE\footnote[3]{\url{http://www.astro.yale.edu/dokkum/SIMPLE/}} survey (Damen et al. 2008, in prep.) and by the GOODS and SWIRE surveys at 24 $\mu$m \citep{dickinson2003,lonsdale2003}.
68: These two surveys are among the deepest {\it Spitzer} observations.
69: This results in a large sample of X-ray-selected AGN with mid-infrared coverage, unique in the combination of area and X-ray and infrared depth.
70:
71:
72: In this paper we evaluate mid-infrared selection techniques using the X-ray selected sample of AGN in the ECDF-S.
73: In Section 2, we describe the X-ray and infrared data, detail the likelihood ratio-matching technique used to find {\it Spitzer} counterparts for each X-ray source, and present the infrared identifications.
74: We investigate the mid-infrared colors of these X-ray sources in Section 3.
75: In Section 4 we
76: compare our sources to published IRAC-selection criteria and assess the impact of adding the 24 $\mu$m data to the IRAC bands.
77: In Section 5, we discuss why the majority of our X-ray-selected AGN would be missed by current infrared selection criteria.
78: Conclusions are given in Section 6.
79: Throughout this paper we assume $H_0 = 71 \: {\rm km \: s^{-1} \: Mpc^{-1}}$, $\Omega_{\rm m} = 0.3$ and $\Omega_{\rm \lambda} = 0.7$.
80:
81: \section{\uppercase{OBSERVATIONS \& ANALYSIS}}
82: \subsection{{\it Chandra} Data}
83: Three X-ray surveys have overlapping coverage of the same region of the sky:
84: the very deep central {\it Chandra} Deep Field-South (CDF-S; \citealp{giacconietal2002}, hereafter G02; \citealt{alexanderetal2003}, hereafter A03), the deep Extended {\it Chandra} Deep Field-South (\citealp{lehmeretal2005}, hereafter L05; \citealp{viranietal2006}, hereafter V06), and the shallower XMM pointing that covers this region \citep{streblyanska2004}.
85: For this study, we use only the two {\it Chandra} surveys because the higher resolution of {\it Chandra} allows for fainter point source detections.
86: In contrast, the XMM data are better suited for X-ray spectroscopy of the brighter sources because of the larger effective area \citep{streblyanska2006,dwellypage2006}.
87: The CDF-S consists of 1 Ms of data reaching a limiting on-axis flux of $5.2\times10^{-17} \: {\rm ergs \: s^{-1} \: cm^{-2}}$ ($2.8\times 10^{-16} \: {\rm \: ergs \: s^{-1} \: cm^{-2}}$) in the 0.5-2.0 keV (2.0-8.0 keV) band (A03).
88: The 11 individual pointings of the CDF-S cover a central area of $\sim 300$ arcmin$^2$ (G02).
89: The four 250-ks {\it Chandra} pointings of the ECDF-S overlap the deep central region, covering a total area of $\sim 1100$ arcmin$^2$.
90: On axis the ECDF-S reaches a limiting flux of $1.7\times 10^{-16} \: {\rm \:ergs \:cm^{-2}\: s^{-1}}$ ($3.9 \times 10 ^{-16} \: {\rm \:ergs \: s^{-1} \:cm^{-2}}$) in the 0.5-2.0 (2.0-8.0) keV band (V06).
91: We combine published point source detections from these fields into a new unified MUSYC X-ray point source catalog.
92:
93: We begin with the most conservative detection catalogs for the CDF-S and ECDF-S, A03 and V06.
94: Since the CDF-S goes significantly deeper in flux than the ECDF-S and covers a much smaller area between the four ECDF-s pointings, we expect that only a fraction of sources will be common to both surveys.
95: Both catalogs used similar detection methodologies: first filtering the data to eliminate periods of high X-ray background, then using {\it wavdetect} \citep{freemanetal2002} for source selection, which is robust in detecting individual sources in crowded fields.
96: The MUSYC X-ray catalog first adopts the 326 sources presented in A03 Table 3.
97: We next search for new sources in V06 Tables 4 and 5 as described in what follows.
98: Table 4, the primary catalog in V06 uses the same conservative false-positive probability threshold for detection ($p_{\rm thresh}=10^{-7} $) as \citet{alexanderetal2003}.
99: Given this detection threshold, we expect no more than 2 or 3 false X-ray sources in either A03 or Table 4 of V06.
100: Table 5 of V06, with $p_{\rm thresh}=10^{-6} $, allows for a larger percentage of false positives but includes a greater number of bona-fide X-ray sources in the catalog that would otherwise be missed.
101: In order to improve the absolute astrometric solutions in the C06 catalogs, small shifts were applied to the X-ray sources RA and DEC, $\le 1$ arcsecond.
102: The shifts were calculated for each of the four pointings in the ECDF-S using optical counterparts in the MUSYC BVR image.
103: No shift was necessary to align A03 catalog positions.
104: For each source in V06, we calculated the distance to the nearest X-ray sources in A03 and defined combined X-ray positional uncertainties for each pair by $\sigma=\sqrt{\sigma_{A03}^2+\sigma_{V06}^2}$, where $\sigma_{XR}$ is the 1 $\sigma$ error of a Gaussian based on the positional errors reported in each catalog.
105: We included in our MUSYC X-ray catalog the 503 sources from V06 that were more than $4\sigma$ ($\sim 3$ arcseconds) from any source in A03.
106: The choice of $4\sigma$ is conservative and was determined by detailed inspections of both the X-ray and MUSYC BVR images.
107: The median separation between the reported source positions for the 148 X-ray sources found in both A03 and V06 was 0.58 arcseconds.
108: A total of 829 unique X-ray sources resulted from the union of A03 and V06, the first 771 of which were detected with the conservative false-positive probability threshold.
109:
110: We next investigated the other published X-ray catalogs in this field for the same two data sets.
111: \citet{lehmeretal2005}, created a larger source catalog for the ECDF-S region by retaining data obtained during small X-ray background flares and using the same lower significance threshold parameter as Table 5 in V06.
112: A very small $\le 1$ arcsecond positional shift was applied to the L05 catalog to align the astrometry to the MUSYC BVR image.
113: Using the X-ray positional uncertainties, we found 141 additional X-ray sources in L05.
114: Finally, \citet{giacconietal2002} used a set of unique detection methods including a modified version of Sextractor \citep{bertinarnouts1996} and running {\it wavdetect} with different selection criteria.
115: Three catalogs are presented in G02: Table 2 containing sources common to both detection methods, Table 3 containing Sextractor sources only and Table 4 containing {\it wavdetect} sources only.
116: A slightly larger fixed positional shift ($\sim 1$ arcsecond) was necessary to align the X-ray sources presented in G02 with our MUSYC BVR image.
117: Using X-ray positional uncertainties, we found a total of 47 additional X-ray sources not previously presented in A03, V06 or L05: 17 from Table 2, 19 from Table 3 and 10 from Table 4.
118: Although there is a good possibility that some of these additional X-ray sources may be spurious, those will fail to match to an infrared counterpart and thus will not affect the conclusions of this paper (see \S~2.3).
119:
120:
121: The complete MUSYC X-ray catalog contains 1017 individual sources whose X-ray properties are given in columns 1 through 5 in Table 1.
122: Column 1 presents a new unique MUSYC X-ray ID number for each source.
123: Column 2 presents the original X-ray catalog and catalog ID for each X-ray source (e.g., V06:323 is source number 323 from Virani et al. 2006). All X-ray source properties, such as flux or count rate, are taken from the referenced detection catalog.
124: Columns 3 and 4 present the RA and DEC, J2000, for each X-ray source, based on the MUSYC BVR image.
125: They are on a common system incorporating the small $\Delta$RA and $\Delta$DEC applied to each catalog to shift the coordinates to agree with the MUSYC BVR image.
126: Column 5 gives the hardness ratio, defined by $\frac{H-S}{H+S}$, where H and S are the X-ray counts in the hard (2-8 keV) and soft (0.5-2 keV) bands, respectively. For sources detected only in the hard band (soft band), the hardness ratio is set equal to 1 (-1). For sources detected only in the full band (0.5-8 keV), the hardness ratio is set equal to 10. G02 defined the hard band as 2-10 keV. Therefore, we adjusted the published hard band counts from G02 to equal those that would have been detected at 2-8 keV, the hard band defined by the other three authors, using PIMMS (version 3.9c; \citealt{mukai1993}) with the average X-ray spectrum measured by G02 ($\Gamma=1.4$; $H_{2-8}=0.9896 \times H_{2-10}$).
127:
128:
129:
130: \subsection{{\it Spitzer} Data}
131: Considerable {\it Spitzer} time has been invested in the ECDF-S with both IRAC and MIPS (the Multi-Band Imaging Photometer for {\it Spitzer}; \citealt{riekeetal2004}).
132: The GOODS Survey along with additional Guaranteed Time Observer (GTO) time gathered deep IRAC and MIPS data on the central region of this field \citep{dickinson2003}.
133: Additionally, the {\it Spitzer} Wide-area InfraRed Extragalactic survey (SWIRE; \citealt{lonsdale2003}) covers a nearly 8 square degree region around the original {\it Chandra} Deep Field-South in a shallow survey mode with both the MIPS and IRAC cameras.
134: Finally, a public deep IRAC survey was conducted in the region covering the ECDF-S (SIMPLE; PI. P. van Dokkum).
135:
136: The IRAC data cover $3-8$ $\mu$m to $5 \: \sigma$ limiting depths of 24.4 (3.6 $\mu$m), 24.0 (4.5 $\mu$m), 22.2 (5.8 $\mu$m) and 22.2 (8.0 $\mu$m) in AB magnitude.
137: In units of flux density, these values are 0.63 $\mu$Jy, 0.91 $\mu$Jy, 4.7 $\mu$Jy, and 4.7 $\mu$Jy, respectively.
138: A catalog of more than 60,000 sources was created with SExtractor using a weighted sum of the 3.6 and 4.5 $\mu$m data (Damen et al. 2008, in prep.).
139: Of these, we have reliable photometry in all four IRAC bands for $38,755$ sources (3$\sigma$), which will be used in this study.
140: The astrometry was calibrated using the MUSYC BVR detection image ($R_{AB}=27.1$);
141: the resulting positional accuracy for individual sources is $\lesssim 0.3$ arcseconds ($1 \sigma$; Damen et al. 2008, in prep.).
142:
143: We also use the SWIRE 24 $\mu$m catalog provided in data release 3 \citep{lonsdale2003,surace2005}, which cover the entire area of the X-ray data to AB magnitude 18 (0.23 mJy), and the GOODS data in the central region, which goes to $\sim5$ magnitudes deeper \citep{dickinson2003}.
144: We adopt the aperature fluxes from the SWIRE catalog which agree with the published GOODS fluxes to better than 5\%.
145: The positional accuracy of the 24 micron data is considerably less than that of the IRAC data;
146: however, all 24 micron source positions were matched to IRAC sources, decreasing the positional uncertainties.
147:
148: \subsection{Catalog Matching}
149: Following the work of \citet{brusa2005,brusaetal2007}, we use the Likelihood Ratio (LR), a statistical matching technique,
150: to find {\it Spitzer} counterparts to the ECDF-S X-ray sources.
151: Although the median positional accuracy of {\it Chandra} is $\sim 0.\arcsec6$, off axis the PSF broadens and becomes circularly asymmetric,
152: leading to larger positional uncertainties in faint off-axis sources.
153: The positional accuracy for each detected {\it Chandra} source depends on the number of counts detected and the distance of the source from the aim point.
154: For several X-ray sources, this positional uncertainty encompasses multiple potential infrared counterparts, while for others there may be no counterpart to the survey detection limits.
155: Due to the high source density of the faintest IRAC sources, there is a non-negligible random chance of
156: finding a faint IRAC object near an unassociated X-ray source.
157: Simulations showed that $\sim$ 60\% of random positions in the field fall within 4$\arcsec$ of an IRAC source, the maximum distance for which we find an IRAC counterpart for an X-ray source.
158: The Likelihood Ratio takes into account the background object density and the positional accuracy of both the {\it Spitzer} and the {\it Chandra} data, and returns the probability that a given match is the correct one.
159: A nearest neighbor approach, restricted to a typical search radius of 1.5 arcseconds, would miss 109 IRAC counterparts that we recover with our technique.
160: Although larger search radii can return more possible counterparts, a nearest neighbor approach does not indicate which counterparts fall inside the search area by chance and it can only distinguish among multiple possible candidates by their distance from the reported X-ray source position.
161: Our method weights the distance using the positional errors and adds a prior on the brightness of the prospective counterpart.
162:
163: The Likelihood Ratio is the ratio between the probability of finding the true infrared
164: counterpart at a given distance from a particular X-ray source divided by the probability of finding a
165: background object at that distance (see \citealt{sutherlandsaunders1992,ciliegietal2003}).
166: When matching the {\it i}th X-ray source, to a possible {\it j}th infrared counterpart, we calculate:
167: \vspace{-0.2cm}
168: {\small{
169: \begin{equation}
170: LR_{ij}=\frac{q(m_j) f(r_{ij})}{n(m_j)},
171: \end{equation}}}
172:
173: \noindent where {\it q(m)} is our empirically determined probability
174: distribution for infrared counterparts as a function of magnitude;
175: {\it f(r)} is the probability distribution function
176: of the positional errors in source {\it i} and counterpart {\it j} (modeled as a two dimensional Gaussian defined with $\sigma_{gaussian}=\sqrt{\sigma_{X,i}^2+\sigma_{IR,j}^2}$);
177: and {\it n(m)} is the surface density of background objects as a function of magnitude.
178: When calculating the probability distribution function
179: of the positional errors, $\sigma_{X,i}$ is unique to each X-ray source depending on its off axis angle and brightness, while no such field-dependent distinction is made for the infrared counterparts.
180: We adopt $\sigma_{IRAC} = 0.3 \arcsec$ and $\sigma_{24\mu m} = 0.5 \arcsec$ (\S~2.2).
181: Both {\it q(m)} and {\it n(m)} depend on the depth of the infrared data.
182: This technique is inherently Bayesian, and the resulting likelihood ratio
183: assumes the priors {\it q(m)}, {\it n(m)} and {\it f(r)} are known.
184:
185: When calculating the background counts of infrared sources, {\it n(m)},
186: we correct for the fact that there are actually fewer faint sources near X-ray objects than near random positions in the field \citep{brusaetal2007}.
187: This follows from the high probability of finding a bright infrared counterpart near any given X-ray position
188: and the fact that the fainter sources very close to these bright counterparts go undetected.
189: To do this we first estimate the background around a sample of IRAC objects with a magnitude
190: distribution similar to those IRAC objects located within $1 \arcsec$ of the X-ray source positions;
191: the positional uncertainties make it highly probable that IRAC objects within this distance are true counterparts.
192: Subtracting this estimated background magnitude distribution from the magnitude distribution of the actual IRAC sources near the X-ray source positions
193: provides a better estimate of the magnitude distribution of IRAC sources associated with X-ray sources.
194: Experimentation showed that using this improved estimate was stable to multiple iterations.
195: We then estimate the background around a sample of IRAC sources with this final magnitude distribution for objects associated with X-ray source positions; this is the {\it n(m)} used in our likelihood ratio calculations.
196:
197: Next, we count the total number of infrared objects lying within a distance $r$ of all X--ray sources, $N_{total}(m)$.
198: Then we multiply the background distribution by the area searched to find these sources; the latter is the number of X--ray sources, $N_{X}$, times the area of each circle searched, $\pi r^2$.
199: The difference between $N_{total}(m)$ and the expected background, $n(m) N_{X} \pi r^2$, is our expected distribution of real counterparts, $N_{real}(m)$.
200: It is important to understand this is a statistical quantity; rather than predicting the expected infrared counterpart for any individual X--ray source, it calculates the expected magnitude distribution for all $N_{X}$ sources.
201: With our expected distribution of real counterparts we derive the normalization, $Q$, as the ratio of the total expected number of real counterparts, $\sum{N_{real}(m)}$, divided by the total number of X--ray sources, $N_{X}$.
202: Then we compute our expected distribution of true infrared counterparts as a function of magnitude:
203: \begin{equation}
204: q(m)=\frac{N_{real}(m)}{{\displaystyle{\sum_{m}}}{N_{real}(m)}} Q
205: \end{equation}
206: Finally, we compute the likelihood ratio for each infrared object located near an X--ray source using Equation 1.
207:
208: When we search for infrared counterparts over a larger radius we see more possible counterparts, but the large distance between these prospective counterparts and the X--ray source drives down the likelihood ratio.
209: We must then carefully select the best likelihood ratio which can distinguish between true matches and false matches.
210: Because the presence of multiple prospective counterparts provides additional knowledge to the likelihood ratio, we use the reliability ($R$) to select counterparts \citep{sutherlandsaunders1992}.
211: The reliability is the probability that a given infrared counterpart is the correct infrared counterpart for each X-ray source.
212: For each possible infrared counterpart {\it j}, we calculate the reliability by:
213: \begin{equation}
214: R=\frac{LR_j}{\sum{LR_i} + (1-Q)}
215: \end{equation}
216: Here, ${\sum{LR_i}}$ is the sum over all possible infrared counterparts for a given X-ray source.
217: The $(1-Q)$ term is the probability of there being no infrared counterpart.
218: The calculated reliabilities form a bimodal distribution, clustering at values above R=0.9 and below R=0.1, and dividing cleanly at R=0.6.
219: Therefore, we accept all matches with $R \ge 0.6$.
220: The likelihood ratio approach assumes exact knowledge of the priors, {\it q(m)} and {\it n(m)}.
221: Because we estimate these values from the data, the likelihood ratio does not reflect uncertainties in our estimation of the background.
222: However, with a threshold value of R = 0.6, we find from Monte Carlo simulations of the background estimation that fewer than 1\% of our matches are affected by this uncertainty.
223:
224: \subsection{{\it Spitzer} Counterparts}
225: We find that more X-ray sources in the ECDF-S are detected with {\it Spitzer} than at any other wavelength.
226: For the full MUSYC X-ray catalog of 1017 sources, we find 921 {\it Spitzer} counterparts in the IRAC data: a 90\% recovery rate.
227: Of the 921 sources, 908 have $3\sigma$ detections in all four of the IRAC channels.
228: For many of the X-ray sources, blending and confusion of IRAC sources, sometimes due to proximity to a very bright source, are responsible for the lack of IRAC counterparts.
229: Excluding regions where confusion or bright stars interfere, we recover infrared counterparts for 98\% of the high-confidence X-ray source detections, i.e. the subset of the X-ray sources detected in the ECDF-S with the conservative false-probability threshold.
230: For the remaining 2\% of X-ray sources, either an IRAC source is seen on the image near the X-ray position, but this IRAC source is not above the source detection limit of the IRAC catalog (2 cases), or between 2 and 5 IRAC sources are located at larger distances from the X-ray position (2-4 arcseconds) and no single source has a large enough probability of being the correct counterpart (14 cases).
231: We find no convincing case of an undetected high significance X-ray source in the IRAC image.
232: That is, essentially all X-ray sources in our catalog are bright enough in the infrared to be detected in the deep IRAC data.
233:
234: There are 453
235: unique MIPS counterparts combing the results from the 24 micron surveys in this field.
236: We find 181 X-ray counterparts in the SWIRE catalog and 318 counterparts in the GOODS catalog from the central region.
237: Combining the counterparts found in the MIPS and IRAC data, we have 428 sources with complete SEDs from 3.0 - 24 microns.
238:
239: All matched counterparts and their mid-infrared fluxes are presented in Table 1.
240: Columns 7, 8, 9 \& 10 give the observed flux in the 3.6, 4.5, 5.8 \& 8.0 $\mu$m IRAC bands, respectively.
241: The flux units are in $\mu $Jy, and are followed by the 1$\sigma$ errors in parentheses.
242: Column 11 gives the MIPS 24 $\mu$m flux in $\mu$Jy followed by the 1$\sigma$ error in parentheses. The number is preceded by a G if it comes from the deeper GOODS data and an S if it is taken from the shallower SWIRE survey.
243:
244:
245: \subsection{Redshifts}
246: Although the majority of X-ray sources turn out to be AGN, it is also possible for deep X-ray surveys to detect Galactic stars as well as galaxies that are vigorously forming stars.
247: Luminosities above $10^{42} \: {\rm \:ergs \: s^{-1}}$ separate out sources that are primarily powered by black hole accretion rather than star formation \citep{persic2004,liraetal2002}.
248: In order to study intrinsic X-ray luminosities
249: and to apply a luminosity cut to exclude Galactic stars and starburst galaxies from our AGN sample,
250: we require some knowledge of a source's distance or redshift.
251: To this end, we are obtaining optical spectra of our X-ray detected sources using Magellan IMACS and VLT/VIMOS (Treister et al. 2008, in prep).
252: Thus far, we have 281 spectra for our X-ray sample, 110 with secure redshifts and 113 with spectral type identifications:
253: 42 broad-line sources (BLAGN), 47 non-broad line luminous ($\log L_{X} \ge 42 \: {\rm \:ergs \: s^{-1}}$) X-ray sources (non-BLAGN),
254: 15 galaxies ($\log L_{X} \le 42 \: {\rm \:ergs \: s^{-1}}$) showing no AGN feature in their spectra,
255: and nine stars (Treister et al. 2008, in prep.).
256: The BLAGN are typically referred to as Type 1 AGN, while sources showing narrow-lines are referred to as Type 2 AGN.
257: However, these spectral signatures are often hidden in obscured AGN and moderate luminosity AGN with large host galaxies \citep{moranetal2002,severgninietal2003}.
258: Therefore, we define non-BLAGN to be all sources showing narrow emission lines suggestive of Type 2 AGN emission plus those sources with $\log L_{X} \ge 42 \: {\rm \:ergs \: s^{-1}}$ that show galaxy-like optical spectra.
259: To the 110 redshifts we obtained with IMACS, we add an additional 173 sources with spectroscopic redshifts published in the literature \citep{szokolyetal2004,croometal2001,lefevreetal2004,vanzella2005}.
260: We further include spectroscopic identifications from \citet{szokolyetal2004}, which adds 33 BLAGN and 74 non-BLAGN.
261: Another 274 sources have high quality photometric redshifts from the COMBO-17 Survey \citep{wolfetal2004}.
262: We note that our sample of AGN spans a range of $L_X(0.5-8 \: {\rm keV}) \sim 10^{42-45}\: {\rm \:ergs \: s^{-1}}$, typical for deep X-ray surveys, different from the shallower survey sources which are on average more luminous.
263: This makes a total of 557 X-ray sources (60\% of the counterpart sample) for which we can calculate the X-ray luminosity in the observed 0.5-8 keV band:
264: of these 416, 76\%, have $L_X(0.5-8 \: {\rm keV}) \ge 10^{42} \: {\rm \:ergs \: s^{-1}}$ and thus are bona-fide AGN.
265:
266:
267:
268: In Table 1, we present the X-ray luminosities and spectral identifications.
269: Column 12 gives the observed 0.5-8 keV luminosity calculated from the redshift. For sources from G02, no 0.5-8 keV flux was reported. Therefore, we adjusted the published 2-10 keV counts from G02 to equal those that would have been detected at 0.5-8 keV, using PIMMS (version 3.9c; \citealt{mukai1993}) with the average X-ray spectrum measured by G02 ($\Gamma=1.4$; $F_{0.5-8}=1.13 \times F_{2-10}$).
270: Column 13 gives the source of the redshift estimate used to calculate the X-ray luminosity as follows: 0 for VLT/FORS2 Spectroscopy Version 1.0, 1 for VIMOS VLT Deep Survey (VVDS) Version 1.0, 2 for \citet{szokolyetal2004}, 3 for \citet{croometal2001}, 4 for redshifts determined from the MUSYC spectroscopy (Treister et al. 2008, in prep.), 5 for COMBO-17 \citep{wolfetal2004}. Those counterparts falling into the categories of BLAGN and non-BLAGN are also indicated.
271:
272: \section{\uppercase{The Mid-Infrared Colors of X-ray Sources}}
273: %%%%FIGURE 1%%%%%
274: \begin{figure}
275: \plotone{f1.ps}
276: \caption{Color-color plot for the MUSYC X-ray sources.
277: Also shown are two SED tracks:
278: (1) the mean type 1 quasar SED from Richards et al. (2006) red-shifted from z=0 to z=6 ({\it purple dot-dashed line, CCW});
279: (2) pure power-law SEDs, $f_{\nu} \propto \nu^{\alpha}$ ({\it thick purple line}), with stars marking power-law exponents of $\alpha = +1.5$ to $\alpha = -1.5$ ({\it bottom left to top right, $\Delta \alpha=0.5$}).
280: X-ray-selected AGN ({\it light blue crosses}) have colors consistent with power-law SEDs from the bluest to reddest colors shown here.
281: The black lines indicate the region suggested by Lacy et al. (2004) for selecting AGN, which excludes 42\% of X-ray-selected AGN ($L_X \ge 10^{42} \: {\rm \:ergs \: s^{-1}}$).}
282: \end{figure}
283:
284:
285: From 3.6 to 8.0 $\mu$m X-ray detected sources show a wide range of colors.
286: In Figure 1, a color-color plot of the infrared counterparts to the MUSYC X-ray catalog, sources with blue colors are in the lower left region and those with red colors are located in the upper right.
287: A linear relation extends from blue to red along a track occupied by power-law SED shapes ($f_{\nu}\sim \nu^{\alpha}$).
288: Along this track, stars indicate $\alpha$ values of +1.5 (lower left) to -1.5 (upper right) in steps of 0.5.
289: For each source, X-ray luminosities are indicated where known:
290: $L_X \ge 10^{42} \:{\rm \: ergs \: s^{-1}}$ plotted with light blue $\times$ symbols,
291: $L_X \le 10^{42} \:{\rm \: ergs \: s^{-1}}$ plotted with light green dashes,
292: and those with unknown redshifts plotted with black + signs.
293: It is clear that the X-ray-selected AGN, $L_X \ge 10^{42} \:{\rm \: ergs \: s^{-1}}$, extend along the full range of power-law shapes.
294: We also explore the infrared colors of the X-ray counterparts identified as BLAGN (dark blue boxes) and non-BLAGN (red circles).
295: For comparison, we plot the mean Type 1 quasar SED compiled by Richards et al. (2006), from redshift 0 to 6 (dashed purple line, CCW).
296: Unsurprisingly the BLAGN lie near the Type 1 quasar track and have the reddest infrared colors.
297: Still, many AGN have significantly bluer colors than the Type 1 track, falling down and to the left on Figure 1.
298: In contrast, the sources classified as non-BLAGN are roughly evenly split between red ($\alpha \le 0$) and blue ($\alpha \ge 0$) colors.
299: Overall, while the optically identified BLAGN tend to have the reddest infrared colors, the X-ray-selected AGN population is {\it not all} red in color.
300:
301:
302: Next, we compare the IRAC colors of X-ray sources (908) to the $\sim$40,000 IRAC sources showing no X-ray emission to the limits of the published X-ray surveys, most of which are inactive galaxies.
303: In Figure 2 we display two popular IRAC color-color plots; the boxes in the right and left panels correspond to two different proposed selection techniques (\S~4) and the contours enclose 75\%, 50\%, 25\% and 15\% of the IRAC sources.
304: The left panel shows $\log{({S_{5.8}})/({S_{3.6}})}$ versus $\log{({S_{8}})/({S_{4.5})}}$, where S is the observed flux in $\mu$Jy, and on the right we plot Vega magnitudes: $[5.8]-[8.0]$ vs $[3.6]-[4.5]$.
305: For clarity we plot only $L_X \ge 10^{42} \:{\rm \: ergs \: s^{-1}}$ sources and indicate only the BLAGN.
306: On both plots, the power-law SED track and Type 1 QSO SED track are repeated from Figure 1.
307: Although a larger percentage of the AGN have red colors than the galaxies, in neither plot are the X-ray selected AGN clearly separated from the rest of the galaxies.
308:
309: %%%%FIGURE 2%%%%%
310: \begin{figure}
311: \begin{center}
312: \plotone{f2.ps}
313: \end{center}
314: \caption{
315: Color-color plots for all sources in the MUSYC catalog (contours indicate 75\%, 50\%, 25\% and 15\% of the IRAC sources).
316: The non-X-ray detected galaxies and X-ray-selected AGN are not well separated in this color space.
317: {\it left}
318: The region defined by \citet{lacyetal2004} includes nearly 20\% of all IRAC-detected galaxies.
319: Therefore, this selection region is not effective at identifying a sample of AGN.
320: {\it right}
321: The selection box of \citet{sternetal2005} uses a different color-color plot of the IRAC bands.
322: Only $\sim 40$\% of X-ray-selected AGN ($L_X \ge 10^{42}\: {\rm \:ergs \: s^{-1}}$) fall in this region.
323: Because it eliminates 90\% of the general population of MUSYC sources, this selection box is more effective at indicating potential AGN, but misses a larger fraction of the X-ray selected AGN.
324: X-ray selected AGN occupy a far larger region in mid-infrared color-space than optically-selected BLAGN.}
325: \end{figure}
326:
327: Because the galaxies and AGN appear to lie along the locus of power-laws, we fit the four observed IRAC fluxes, 3.6-8.0 $\mu$m, to power-law spectral energy distributions.
328: Within the short baseline of the IRAC camera, nearly all of our sources are consistent with power laws, especially given the photometric uncertainties.
329: Each source with detections in all 4 IRAC channels was fit by $f_{\nu} \propto \nu^{\alpha}$; 94\%
330: of all sources and 97 \% of the X-ray-selected AGN had acceptable fits by power laws ($\chi^2$, $p\ge0.05$).
331: In Figure 3, we compare the histogram of the power-law indices, $\alpha$, for the X-ray-selected AGN and other sources.
332: Both types of sources show a very broad range in $\alpha$, but the AGN tend to be slightly redder than normal galaxies (more negative $\alpha$).
333: A KS test rejects the hypothesis that the two distributions come from the same population (P=0.0014).
334: While only 17\% of the galaxy sample have $\alpha \le 0$, 37\%
335: of $L_X \ge 10^{42} \: {\rm \:ergs \: s^{-1}}$ sources fit a negative power-law.
336: Optically-selected AGN spectra show red continuum emission at infrared wavelengths, with $\alpha \sim -1$ (e.g., \citealt{elvisetal1994}).
337: This is consistent with our finding that 100\% of the BLAGN have acceptable power-law fits, 72\% with $\alpha \le 0$ (Dark blue cross-hatched histogram).
338: As a whole, our power-law fits to the BLAGN are consistent with featureless negative power-law slopes in the mid-infrared.
339: However, we find a broad range of $\alpha$ values even for the BLAGN.
340: Similarly, the non-X-ray detected galaxies are bluer as a whole than the AGN population, even considering that the sources with the reddest $\alpha$ in the X-ray-selected AGN histogram may be broad line objects.
341: Indeed, although the X-ray-selected AGN have on average slightly more negative $\alpha$ values than galaxies, over half (59\%) of X-ray-selected AGN are fit by power laws with positive exponents similar to those of the bulk of the galaxy population.
342: Therefore, we find that most X-ray-selected AGN do not show red continuum emission at observed {\it Spitzer} IRAC wavelengths.
343:
344: %%%%FIGURE 3%%%%%
345: \begin{figure}
346: \begin{center}
347: \plotone{f3.ps}
348: \end{center}
349: \caption{
350: Histograms of power-law index, $\alpha$ ($f_{\nu} \propto \nu^{\alpha} $), for fits to the {\it Spitzer} IRAC photometry.
351: The X-ray sources with $L_{X} \ge 10^{42} \: {\rm \:ergs \: s^{-1}}$ ({\it solid light blue line}) and the subset of X-ray sources spectroscopically identified as BLAGN ({\it dark blue cross-hatched histogram}) have been normalized to the total number of X-ray sources with good estimates of redshift and good power-law fits (396).
352: Also plotted is the normalized histogram for all MUSYC sources well fit by power laws ({\it dotted black line}, 38988).
353: In total, 94\% of all galaxies and 97\% of the $L_X \ge 10^{42} \: {\rm \:ergs \: s^{-1}}$ sources are consistent with a power-law SED.
354: A KS-test rejects the hypothesis that these two distributions come from the same population (P=0.0014).
355: Although the AGN are slightly redder than the galaxies on average, both the AGN and the galaxies are fit by a broad range of power-law slopes.}
356: \end{figure}
357:
358:
359: \section{\uppercase{Selecting AGN by Infrared Colors}}
360: Although selecting AGN through X-ray emission is relatively unbiased,
361: current X-ray surveys may still be missing AGN that are heavily obscured by large columns of dust and gas.
362: All AGN, whether obscured or unobscured at optical wavelengths, reprocess and re-emit the absorbed radiation in the mid- and far-infrared \citep{granatodanese1994,nenkovaetal2002}.
363: Because AGN are expected to be luminous at infrared wavelengths, several authors have developed infrared selection techniques for AGN (e.g., \citealt{lacyetal2004,sternetal2005}).
364: Here, we examine what fraction of X-ray-selected AGN would be selected using these techniques.
365:
366: \citet{lacyetal2004} found that a sample of Sloan Digital Sky Survey quasars occupied a distinct region of mid-infrared color-color space (Figures 1, 2).
367: Because the selection box was defined using luminous Type 1 quasars from the Sloan survey, we look at the subset of the X-ray sources that have been identified as broad-line AGN.
368: Indeed, 96\% of these sources fall inside the selected Lacy region.
369: Furthermore, all 61 X-ray sources identified photometrically as quasars in the Combo-17 survey also fall inside this box.
370: However, only 58\% of all X-ray-selected AGN fall inside this region.
371: Thus, the percentage of X-ray-selected AGN that lie inside this color selection box depends on the number of luminous Type 1 objects in the sample.
372: That is, this infrared color selection finds the luminous unobscured quasars, but misses a high fraction of the lower luminosity AGN, many of which look like optically normal galaxies.
373: The brightest AGN in our sample, $L_X \ge 10^{44.5} \: {\rm \:ergs \: s^{-1}}$, are also all located inside the Lacy selection box.
374: Additionally, 18\% of all sources ($\sim 11,000$ galaxies) lie within the Lacy selection box (Figure 2).
375: This is 20 times the total number of X-ray sources in this region (529).
376: A reliable AGN selection criteria should yield a sample of sources containing mostly active galaxies. Because so many IRAC-detected sources fall inside this selection box, it is not a reliable method for selecting AGN in deep surveys.
377:
378: \citet{sternetal2005} define a similar selection region using a deeper sample of spectroscopically confirmed AGN from the Bootes field.
379: In the right panel in Figure 2, we plot X-ray-selected sources on the \citet{sternetal2005} color-color plot.
380: Again, we find that the majority (76\%) of our spectroscopically identified broad-line AGN fall inside this selection box, but only 40\% of X-ray-selected AGN do.
381: Over 90\% of the brightest AGN, $L_X \ge 10^{44.5} \: {\rm \:ergs \: s^{-1}}$, are also all located inside the Stern selection box.
382: In this case fewer galaxies lie inside this box (5163, $\sim$ 8\%), but they still out number the 340 X-ray sources by a factor of $\sim 15$.
383: The \citet{sternetal2005} selection criteria is a slightly more reliable way to select for candidate AGN, but at the same time misses a larger percentage of the total AGN population.
384:
385: Although color selection techniques may work well for shallow surveys, where a larger percentage of the AGN are bright and show broad lines, they are not as reliable in deep surveys.
386: These selection boxes correctly highlight the region containing luminous broad-line quasars but miss most of the AGN population,
387: in particular non-BLAGN and low-luminosity AGN.
388: It is possible that some of the galaxies in these selection regions hide luminous AGN.
389: However, looking at Figure 2, there does not appear to be a clear region in color-color space that selects a complete sample of X-ray detected AGN and avoids returning a large portion of the entire galaxy population.
390: In an effort to find a more reliable selection technique, we now add the 24 $\mu$m band data to our analysis of infrared colors.
391:
392: \subsection{24 $\mu$m Data}
393: Adding the 24 $\mu$m data to a given source's SED more than doubles the baseline over which we can determine the AGN spectral shape.
394: For $\sim 25$\% of our X-ray sources, consistent power-law fits are determined from the 3.6-8 $\mu$m and 3.6-24 $\mu$m data (examples shown in Figure 4, bottom panels).
395: However, for many sources the best-fit power law to the IRAC data alone is much bluer than one that includes the 24 $\mu$m data (examples shown in Figure 4, top panels).
396: We found a variety of spectroscopically identified BLAGN and non-BLAGN with 24 $\mu$m intensities that confirm the power-law slope from the IRAC data and also those whose 24 $\mu$m fluxes imply a distinct slope.
397: In Figure 4, the sources on the left are identified as non-BLAGN from optical spectra while the sources on the right are BLAGN.
398: When adding the 24 $\mu$m flux to the fit, the median non-BLAGN slope changes from a value of $\alpha_{3.6-8\mu m}=0.2$ to $\alpha_{3.6-24\mu m}=-1.0$, while the median BLAGN slope changes from a value of $\alpha_{3.6-8\mu m}=-0.3$ to $\alpha_{3.6-24\mu m}=-1.2$.
399: The non-BLAGN appear bluer on average in their 3.6-8.0 $\mu$m colors, while the BLAGN are more likely to show a similar $\alpha_{3.6-8\mu m}$ and $\alpha_{3.6-24\mu m}$.
400: While nearly all X-ray sources show a negative slope in their $3.6-24 \mu m$ spectrum, only for a minority ($\sim 25\%$) of X-ray sources is it a strictly monotonic rise.
401:
402: %%%%FIGURE 4%%%%%
403: \begin{figure}
404: \begin{center}
405: \plotone{f4.ps}
406: \end{center}
407: \caption{
408: X-ray-selected AGN show a variety of spectral shapes below 8 $\mu$m, but nearly all 24 $\mu$m-detected AGN are very red between 3.6 and 24 $\mu$m.
409: In the top plots, a power-law fit to 3.6-8 $\mu$m photometry is significantly bluer than a power-law fit including the 24 $\mu$m flux.
410: In the lower plots, the same power-law fits the 3.6-8 $\mu$m and the 3.6-24 $\mu$m data.
411: Only $\sim 25\%$ of the MIPS sample falls into the latter category.
412: The plots on the left are for sources identified spectroscopically as non-BLAGN while the plots on the right are BLAGN.
413: Blue lines show power-law fits to $3.6-8 \mu m$ and red lines show fits including the 24 $\mu$m data point.}
414: \end{figure}
415:
416:
417: In Figure 5, we directly compare the fitted index $\alpha_{3.6-8\mu m}$ with that of $\alpha_{3.6-24 \mu m}$ for all 24 $\mu$m detected sources.
418: In this figure we include all sources whether or not their spectral shape is consistent with a power law.
419: The X-ray detected sources span a large range of $\alpha_{3.6-8\mu m}$ but in their longer wavelength color, $\alpha_{3.6-24 \mu m}$, they are nearly uniformly red.
420: In contrast, all five galactic stars detected in X-rays and at 24 $\mu$m are in the upper left corner ($\alpha_{3.6-24 \mu m} \ge 1$, dark green points).
421: The histogram of $\alpha_{3.6-24 \mu m}$ peaks below -1, with $\sim$90\% of X-ray-selected AGN detected at 24 $\mu$m having $\alpha_{3.6-24 \mu m} \le 0$ (Figure 5, bottom panel).
422: This histogram is quite different than that for the best-fit $\alpha_{3.6-8\mu m}$ (Figure 5, right panel).
423: The vast majority of X-ray sources with counterparts at 24 $\mu$m have $\alpha_{3.6-24 \mu m} \le 0$: we question if this is because all AGN have negative infrared slopes, or if our shallow 24 $\mu$m coverage simply misses those AGN that would show blue colors.
424: We indicate sources without a 24 $\mu$m detection in the cross-hatched portion of the $\alpha_{3.6-8\mu m}$-histogram (Figure 5, right panel).
425: The deep GOODs data in the central portion returns 24 $\mu$m magnitudes even for sources with very blue spectral shapes in the IRAC bands; we find no significant correlation between $\alpha_{3.6-8\mu m}$ and sources lacking 24 $\mu$m detections.
426: Compared to 54\% of all X-ray counterparts, 48\% of the X-ray counterparts detected at 24 $\mu$m have blue spectral shapes below 8 $\mu$m ($\alpha_{3.6-8\mu m} \ge 0$).
427: Only 20\% of the X-ray selected AGN are consistent with a power-law SED from 3.6-24 $\mu$m ($\chi^2$, $p \ge 0.05$), but of those that fit a power law, nearly all are negative power laws (34 out of 38 sources).
428: This contrasts with only 43\% of the X-ray-selected AGN with IRAC power-law SEDs that have $\alpha_{3.6-8\mu m} \le 0$.
429: Although they have a large range in color below 8 microns, X-ray-selected AGN detected at 24 $\mu$m are uniformly red beyond 8 $\mu$m.
430: For comparison, we add all 24 $\mu$m detected galaxies in the GOODS survey to Figure 5 (gray points) and indicate those that are consistent with a power-law SED between 3.6 and 24 $\mu$m (purple points).
431: Many of these sources may be AGN not detected at X-ray wavelengths, but the rest contain significant amounts of dust which is luminous at the observed 24 $\mu$m \citep{egamietal2004}.
432: The 24 $\mu$m data are not sufficient to differentiate the X-ray selected AGN from the non-X-ray detected galaxies.
433:
434: %%%%FIGURE 5%%%%%
435: \begin{figure}
436: \begin{center}
437: \epsscale{.9}
438: \plotone{f5.ps}
439: \epsscale{1}
440: \end{center}
441: \caption{
442: We compare the power-law slope ($f_{\nu} \propto \nu^{\alpha} $) from the IRAC bands, $\alpha_{3.6-8\mu m}$, to the power-law slope to all 5 bands between 3.6 and 24 $\mu$m ({\it main panel}).
443: We include all sources whether or not $f_{\nu} \propto \nu^{\alpha} $ is a good fit to the data in order to study the approximate shape of the SED.
444: For comparison we plot the GOODS sample of 24 $\mu$m-detected sources ({\it gray points}) and indicate those that are consistent with a good power-law fit between 3.6 and 8 $\mu$m ({\it purple points}).
445: The distribution of $\alpha_{3.6-24\mu m}=$ peaks below -1.0 (bottom panel), significantly redder than either of the two peaks of the distribution of $\alpha_{3.6-8\mu m}$ for all sources (right panel).
446: The cross hatched histogram indicates sources lacking 24 $\mu$m detections.
447: The deep GOODS data detects the 24 $\mu$m flux for sources with a large variety of IRAC spectral shapes.
448: The median uncertainty in $\alpha$ for the X-ray sources is indicated in the top left corner.
449: In the IRAC bands X-ray selected AGN show a wide range of colors, but beyond the observed 8 $\mu$m band, nearly all X-ray selected AGN fit very red spectral shapes, $\alpha \le -1.0$
450: However, the 24 $\mu$m flux, which highlights cold dust, does not help separate AGN from non-active galaxies. }
451: \end{figure}
452:
453: \section{\uppercase{Discussion}}
454: If the AGN contribution to an X-ray source SED is a red power law at infrared wavelengths, the infrared emission from most AGN in the MUSYC sample must also include a
455: significant contribution from the stars and dust in the host galaxy.
456: Stellar spectra are declining in the infrared, creating a blue continuum for rest wavelengths $\lambda \ge 1.6$ $\mu$m.
457: Because the peak of the redshift distribution for X-ray-selected AGN in the ECDF-S is at $z \le 1$, the AGN host galaxies will typically appear blue in the observed $3.6-8.0 \mu m$ band.
458: For the lowest redshift objects, contributions from cool dust and PAH features can also contribute to the observed mid-infrared light \citep{laurentetal2000}.
459: The 24 $\mu$m band is sensitive to dust emission in galaxies, making both AGN and luminous star forming galaxies easily detectable \citep{treisteretal06,egamietal2004,leflochetal2004}.
460:
461:
462: Figure 6 shows X-ray luminosity versus the fitted infrared power-law slope ($\alpha_{3.6-8 \mu m}$, bottom; $\alpha_{8-24 \mu m}$, top) for all MUSYC X-ray sources.
463: Although there is a trend for brighter X-ray sources to have redder power laws below 8 $\mu$m, even some of the most luminous X-ray sources in the ECDF-S ($L_X = 10^{44} - 10^{45} \: {\rm \:ergs \: s^{-1}}$) have $\alpha_{3.6-8\mu m} \ge 0$.
464: Because sources with similar $L_X$ fit such a large range of $\alpha_{3.6-8\mu m}$, it is likely that varying contributions from the host galaxy (stars and cool dust) determine the different IRAC slopes.
465: This is in agreement with the large number of non-BLAGN that lie outside the infrared color selection box defined by optically-selected quasars, and with what is seen in modestly obscured AGN of Seyfert luminosities, where the host galaxy starlight can dominate the optical colors \citep{moranetal2002,cardamoneetal2007}.
466: In contrast, for infrared power-law selected AGN ($\alpha_{3.6-8\mu m} \le 0$), the energetics are dominated by the central AGN and thus the signatures are often visible in the optical.
467: Longward of 8 $\mu$m, there is no observed trend of spectral shape ($\alpha_{8-24\mu m}$) with X-ray luminosity.
468: Our sample agrees with other studies that find that AGN selected through IRAC power-law emission are the most luminous of those detected in X-rays \citep{donleyetal2007} and with studies that look at the most luminous hard X-ray-detected AGN and find them to have red mid-IR colors \citep{georgantopoulosetal2007}.
469:
470: %%%% FIGURE 6 %%%%%
471: \begin{figure}
472: \begin{center}
473: \plotone{f6.ps}
474: \end{center}
475: \caption{
476: {\it Top:} At wavelengths longer than the observed 8 $\mu$m, there is no observed trend of $\alpha_{8-24\mu m}$ with X-ray luminosity. The slope of the best fit line to the sources with $L_X \ge 10^{42}\: {\rm \:ergs \: s^{-1}}$ is $0.04 \pm 0.06$ ({\it dashed line}).
477: {\it Bottom:} More luminous X-ray sources have redder infrared power laws in the IRAC bands, $\alpha_{3.6-8\mu m}$, although in all cases there is a broad range.
478: The slope of the best-fit line to the sources with $L_X \ge 10^{42}\: {\rm \:ergs \: s^{-1}}$ is $-0.6 \pm 0.05$ ({\it dashed line}).
479: It is likely that the galaxy influence, which is proportionally larger for less luminous AGN, is the origin of this trend.}
480: \end{figure}
481:
482:
483: Supporting the idea that the colors below 8 $\mu$m are influenced by the host galaxy light is the fact that the X-ray spectral shape is entirely uncorrelated with the infrared spectral slope but correlated with optical AGN signatures.
484: We plot $\alpha_{3.6-8\mu m}$ and $\alpha_{3.6-24 \mu m}$ versus the X-ray hardness ratio, $(H-S)/(H+S)$, where H is the count rate in the hard band ($2-8$ keV) and S is the count rate in the soft band ($0.5-2$ keV) in Figure 7.
485: The hardness ratio is set to -1 or +1 if the source is detected in only one of the two X-ray bands, and is not shown here for detections only in the full band.
486: The subset of sources with optical spectroscopy is indicated by dark blue squares (BLAGN) and red circles (non-BLAGN) and sources consistent with power-law SEDs are indicated (purple stars).
487: %There appears to be no correlation between the shape of the observed spectrum in the X-ray and the shape of the observed spectrum in the IRAC bands.
488: There are fewer sources consistent with power-law spectral shapes between 3.6 and 24 $\mu$m, but there is no trend with X-ray hardness ratio.
489: This result may indicate different that the X-ray and infrared radiation is produced in vastly different environments.
490: The X-rays produced near the central black hole are sensitive to obscuration near the central region, while the infrared reprocessed emission at wavelengths shorter than the observed 24 $\mu$m depends on the dust distribution not only around the nucleus but throughout the galaxy.
491: In contrast, the optical identifications based on emission lines are somewhat correlated with hardness ratio: the left side of Figure 7, i.e., negative X-ray hardness ratios (relatively soft spectra), is preferentially occupied by BLAGN.
492: This reinforces the idea that the observed infrared emission is insensitive to the geometry,
493: or more likely, includes a substantial contribution from the host galaxy itself.
494:
495: %%%% FIGURE 7 %%%%%
496: \begin{figure}
497: \vspace*{-10mm}
498: \begin{center}
499: \plotone{f7.ps}
500: \end{center}
501: \caption{Power-law indices $\alpha_{3.6-24 \mu m}$ and $\alpha_{3.6-8\mu m}$ (top and bottom panels, respectively) versus X-ray hardness ratio, $(H-S)/(H+S)$, where H represents the counts in the hard {\it Chandra} band (2-8 keV) and S represents the counts in the soft band (0.5-2 keV).
502: Sources that are well fit by a power law over the given wavelength range ({\it purple $*$}) are indicated, other sources ({\it black $+$}) are plotted with a best-fit value of spectral slopes.
503: The sequence of points at -1.0 (+1.0) are from sources detected only in the soft (hard) band.
504: There is no correlation between observed slopes at X-ray and infrared wavelengths.
505: The X-ray emission is most likely produced close to the black hole, while the mid-infrared emission likely comes from dust around the nucleus and in the host galaxy.}
506: \end{figure}
507:
508:
509: Our results are consistent with the findings for other X-ray-selected samples, where a significant fraction of the X-ray detected AGN have bluer colors indistinguishable from those of galaxies (e.g., \citealt{franceschinietal2005,barmbyetal2006}).
510: \citet{treisteretal06} concluded that a fair fraction of the mid-infrared light in low-luminosity AGN arises from the host galaxies, based on the large overlap in the infrared luminosity distributions of normal galaxies and lower luminosity AGN in the GOODS sample.
511: \citet{hickoxetal2007} showed that sources were moved into mid-infrared selection boxes by combining an AGN template with increasing strength to a galaxy template.
512: Mid-infrared selection techniques can miss a significant fraction of AGN if the high column densities significantly dim the central power-law emission and/or star light significantly dilutes the AGN continuum.
513:
514: Mid-infrared selection techniques can uncover sources whose SED is dominated by the central AGN emission.
515: AGN undetected in {\it Chandra} or XMM surveys because of heavy obscuration are very likely to be bright infrared sources.
516: \citet{Alonso-Herrero2006} used the 24 $\mu$m-detected sample of sources in GOODS-South to select IRAC power-law galaxies.
517: Only half of their sample were individually detected at X-ray wavelengths.
518: \citet{pollettaetal2006} effectively used infrared-color selection techniques to uncover a sample of heavily obscured quasars.
519: Their selection was based on featureless spectral energy distribution, and they deliberately removed any source that could be fit acceptably by a galaxy spectral template.
520: A complementary study by \citet{daddietal2007} examined non-X-ray-detected galaxies, which were not dominated by power-law continuum emission in the IRAC bands, and uncovered a detectable contribution from hot dust near an AGN in 20-30\% of the sample.
521: This emission was discovered in excess mid-infrared flux above and beyond the amount predicted from star formation indicators at other wavelengths.
522: These galaxies lack any detectable signatures of AGN emission at optical wavelengths, but stacking results at X-ray wavelengths pointed towards obscured AGN emission \citep{daddietal2007}.
523:
524:
525: A well-defined subset of the X-ray counterparts detected at all infrared wavelengths are well-fit by red (negative) power laws from $3.6$ to 24 $\mu$m.
526: These sources have similar red power laws ($\alpha \le 0$), lie inside the standard infrared selection boxes for optically selected AGN, and show preferentially broad lines in their optical spectra.
527: About 25\% of the whole sample fits into this category.
528: This is an upper bound since the shallow 24 $\mu$m coverage in the SWIRE survey may bias us towards redder X-ray sources.
529: To check this, we selected a sample of sources in the deeper GOODS area that are bright enough at 3.6 $\mu$m to be detected at 24 $\mu$m, even if they have a blue slope ($\alpha=+1$).
530: Of these 26 X-ray sources with bright 3.6 $\mu$m magnitudes ($m_{3.6} \le 21.4$), 4 have a monotomic red slope between 3.6 and 24 $\mu$m. This is a slightly lower fraction (15\%) but consistent within the statistics with the 25\% seen in the large sample.
531: These mid-infared power-law sources, tend to be the most luminous X-ray sources.
532: Together with their Compton-thick counterparts undetected at X-ray wavelengths, these are the best candidates for infrared selection techniques.
533:
534:
535: \section{\uppercase{Conclusions \& Summary}}
536: AGN are luminous infrared sources and deep infrared data from {\it Spitzer} potentially provide powerful new methods for selecting AGN samples.
537: In this paper, we examined the infrared properties of a large sample of AGN at moderately high redshifts selected through their X-ray emission.
538: In the observed 3.6 - 8.0 $\mu$m band over half of these X-ray-selected AGN show infrared colors consistent with those of galaxies.
539: Across the narrow wavelength range of the IRAC filters, most galaxies are also consistent with power-law spectral shapes.
540: As a result, IRAC data alone are not sufficient for a complete or reliable selection of (moderate luminosity) AGN.
541: At 24 $\mu$m, dust from both star forming galaxies and AGN is detected.
542: Unfortunately, observed IRAC to 24 $\mu$m colors do not effectively separate galaxies powered by AGN from those experiencing star formation.
543: Previous studies with IRAS have shown the far-infrared to be valuable for discriminating active from inactive galaxies \citep{sandersmirabel1996};
544: to apply similar techniques to present deep surveys requires longer wavelength data, not just at 24 $\mu$m but at 70 $\mu$m and if possible 160 $\mu$m, as well.
545:
546: To develop complete samples of AGN, we need to understand what kinds of AGN each selection criterion returns.
547: Infrared selection returns a slightly different population than X-ray selection.
548: Future studies using complete SEDs from the ultraviolet through infrared will help disentangle the contribution from the central AGN from that of the host galaxy.
549: We have imaged the entire ECDF-S with medium-band filters, which will result in high quality photometric redshifts for most of the AGN and will allow us to disentangle the nuclear and host galaxy light.
550: These data will also reveal the origin of the variety of infrared colors shown below 8 $\mu$m.
551: The large range in color shown by AGN below 8 $\mu$m potentially holds valuable information about the star formation processes going on in the AGN host galaxies and may help us understand the AGN-galaxy connection.
552:
553: \acknowledgements We thank the anonymous referee for constructive comments that improved this paper. Support from NSF grant \#AST0407295 and Spitzer JPL Grant \#RSA1288440 is gratefully acknowledged
554:
555: \begin{thebibliography}{64}
556: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
557:
558: \bibitem[{{Alexander} {et~al.}(2003)}]{alexanderetal2003}
559: {Alexander}, D.~M. {et~al.} 2003, \aj, 126, 539
560:
561: \bibitem[{{Alonso-Herrero} {et~al.}(2006)}]{Alonso-Herrero2006}
562: {Alonso-Herrero}, A. {et~al.} 2006, \apj, 640, 167
563:
564: \bibitem[{{Barmby} {et~al.}(2006)}]{barmbyetal2006}
565: {Barmby}, P. {et~al.} 2006, \apj, 642, 126
566:
567: \bibitem[{{Bertin} \& {Arnouts}(1996)}]{bertinarnouts1996}
568: {Bertin}, E. \& {Arnouts}, S. 1996, \aaps, 117, 393
569:
570: \bibitem[{{Brandt} \& {Hasinger}(2005)}]{branthasinger2005}
571: {Brandt}, W.~N. \& {Hasinger}, G. 2005, \araa, 43, 827
572:
573: \bibitem[{{Brinkmann} {et~al.}(2000){Brinkmann}, {Laurent-Muehleisen}, {Voges},
574: {Siebert}, {Becker}, {Brotherton}, {White}, \& {Gregg}}]{brinkmannetal2000}
575: {Brinkmann}, W., {Laurent-Muehleisen}, S.~A., {Voges}, W., {Siebert}, J.,
576: {Becker}, R.~H., {Brotherton}, M.~S., {White}, R.~L., \& {Gregg}, M.~D. 2000,
577: \aap, 356, 445
578:
579: \bibitem[{{Brusa} {et~al.}(2005)}]{brusa2005}
580: {Brusa}, M. {et~al.} 2005, \aap, 432, 69
581:
582: \bibitem[{{Brusa} {et~al.}(2007)}]{brusaetal2007}
583: ---. 2007, \apjs, 172, 353
584:
585: \bibitem[{{Cardamone} {et~al.}(2007){Cardamone}, {Moran}, \&
586: {Kay}}]{cardamoneetal2007}
587: {Cardamone}, C.~N., {Moran}, E.~C., \& {Kay}, L.~E. 2007, \aj, 134, 1263
588:
589: \bibitem[{{Ciliegi} {et~al.}(2003){Ciliegi}, {Vignali}, {Comastri}, {Fiore},
590: {La Franca}, \& {Perola}}]{ciliegietal2003}
591: {Ciliegi}, P., {Vignali}, C., {Comastri}, A., {Fiore}, F., {La Franca}, F., \&
592: {Perola}, G.~C. 2003, \mnras, 342, 575
593:
594: \bibitem[{{Croom} {et~al.}(2001){Croom}, {Warren}, \&
595: {Glazebrook}}]{croometal2001}
596: {Croom}, S.~M., {Warren}, S.~J., \& {Glazebrook}, K. 2001, \mnras, 328, 150
597:
598: \bibitem[{{Croton} {et~al.}(2006)}]{crotonetal2006}
599: {Croton}, D.~J. {et~al.} 2006, \mnras, 365, 11
600:
601: \bibitem[{{Daddi} {et~al.}(2007)}]{daddietal2007}
602: {Daddi}, E. {et~al.} 2007, ArXiv e-prints, 705
603:
604: \bibitem[{Damen} {et~al.}(2007)]{damenetal2007}
605: {Damen}, M. {et~al.} 2008, in preparation
606:
607: \bibitem[{{Dickinson} {et~al.}(2003){Dickinson}, {Giavalisco}, \& {The Goods
608: Team}}]{dickinson2003}
609: {Dickinson}, M., {Giavalisco}, M., \& {The Goods Team}. 2003, in The Mass of
610: Galaxies at Low and High Redshift, ed. R.~{Bender} \& A.~{Renzini}, 324--+
611:
612: \bibitem[{{Donley} {et~al.}(2007){Donley}, {Rieke}, {P{\'e}rez-Gonz{\'a}lez},
613: {Rigby}, \& {Alonso-Herrero}}]{donleyetal2007}
614: {Donley}, J.~L., {Rieke}, G.~H., {P{\'e}rez-Gonz{\'a}lez}, P.~G., {Rigby},
615: J.~R., \& {Alonso-Herrero}, A. 2007, \apj, 660, 167
616:
617: \bibitem[{{Dwelly} \& {Page}(2006)}]{dwellypage2006}
618: {Dwelly}, T. \& {Page}, M.~J. 2006, \mnras, 372, 1755
619:
620: \bibitem[{{Egami} {et~al.}(2004)}]{egamietal2004}
621: {Egami}, E. {et~al.} 2004, \apjs, 154, 130
622:
623: \bibitem[{{Elvis} {et~al.}(1994)}]{elvisetal1994}
624: {Elvis}, M. {et~al.} 1994, \apjs, 95, 1
625:
626: \bibitem[{{Fabian} {et~al.}(2003){Fabian}, {Sanders}, {Allen}, {Crawford},
627: {Iwasawa}, {Johnstone}, {Schmidt}, \& {Taylor}}]{fabian2003}
628: {Fabian}, A.~C., {Sanders}, J.~S., {Allen}, S.~W., {Crawford}, C.~S.,
629: {Iwasawa}, K., {Johnstone}, R.~M., {Schmidt}, R.~W., \& {Taylor}, G.~B. 2003,
630: \mnras, 344, L43
631:
632: \bibitem[{{Fabian} {et~al.}(2006){Fabian}, {Sanders}, {Taylor}, {Allen},
633: {Crawford}, {Johnstone}, \& {Iwasawa}}]{fabianetal2006}
634: {Fabian}, A.~C., {Sanders}, J.~S., {Taylor}, G.~B., {Allen}, S.~W., {Crawford},
635: C.~S., {Johnstone}, R.~M., \& {Iwasawa}, K. 2006, \mnras, 366, 417
636:
637: \bibitem[{{Fazio} {et~al.} (2004)}]{fazioetal2004}
638: {Fazio}, G., et al. 2004, \apjs, 154, 10
639:
640: \bibitem[{{Franceschini} {et~al.}(2005)}]{franceschinietal2005}
641: {Franceschini}, A. {et~al.} 2005, \aj, 129, 2074
642:
643: \bibitem[{{Freeman} {et~al.}(2002){Freeman}, {Kashyap}, {Rosner}, \&
644: {Lamb}}]{freemanetal2002}
645: {Freeman}, P.~E., {Kashyap}, V., {Rosner}, R., \& {Lamb}, D.~Q. 2002, \apjs,
646: 138, 185
647:
648: \bibitem[{{Gawiser} {et~al.}(2006)}]{gawiseretal2006}
649: {Gawiser}, E. {et~al.} 2006, \apjs, 162, 1
650:
651: \bibitem[{{Georgantopoulos} {et~al.}(2007){Georgantopoulos}, {Georgakakis}, \&
652: {Akylas}}]{georgantopoulosetal2007}
653: {Georgantopoulos}, I., {Georgakakis}, A., \& {Akylas}, A. 2007, \aap, 466, 823
654:
655: \bibitem[{{Giacconi} {et~al.}(2002)}]{giacconietal2002}
656: {Giacconi}, R. {et~al.} 2002, \apjs, 139, 369
657:
658: \bibitem[{{Granato} \& {Danese}(1994)}]{granatodanese1994}
659: {Granato}, G.~L. \& {Danese}, L. 1994, \mnras, 268, 235
660:
661: \bibitem[{{Hatziminaoglou} {et~al.}(2005)}]{hatziminaolgou2005}
662: {Hatziminaoglou}, E. {et~al.} 2005, \aj, 129, 1198
663:
664: \bibitem[{{Hickox} {et~al.}(2007)}]{hickoxetal2007}
665: {Hickox}, R.~C. {et~al.} 2007, ArXiv e-prints, 708
666:
667: \bibitem[{{Kriek} {et~al.}(2006)}]{krieketal2006}
668: {Kriek}, M. {et~al.} 2006, ArXiv Astrophysics e-prints
669:
670: \bibitem[{{Lacy} {et~al.}(2004)}]{lacyetal2004}
671: {Lacy}, M. {et~al.} 2004, \apjs, 154, 166
672:
673: \bibitem[{{Laurent} {et~al.}(2000){Laurent}, {Mirabel}, {Charmandaris},
674: {Gallais}, {Madden}, {Sauvage}, {Vigroux}, \& {Cesarsky}}]{laurentetal2000}
675: {Laurent}, O., {Mirabel}, I.~F., {Charmandaris}, V., {Gallais}, P., {Madden},
676: S.~C., {Sauvage}, M., {Vigroux}, L., \& {Cesarsky}, C. 2000, \aap, 359, 887
677:
678: \bibitem[{{Le F{\`e}vre} {et~al.}(2004)}]{lefevreetal2004}
679: {Le F{\`e}vre}, O. {et~al.} 2004, \aap, 428, 1043
680:
681: \bibitem[{{Le Floc'h} {et~al.}(2004)}]{leflochetal2004}
682: {Le Floc'h}, E. {et~al.} 2004, \apjs, 154, 170
683:
684: \bibitem[{{Lehmer} {et~al.}(2005)}]{lehmeretal2005}
685: {Lehmer}, B.~D. {et~al.} 2005, \apjs, 161, 21
686:
687: \bibitem[{{Lira} {et~al.}(2002){Lira}, {Ward}, {Zezas}, {Alonso-Herrero}, \&
688: {Ueno}}]{liraetal2002}
689: {Lira}, P., {Ward}, M., {Zezas}, A., {Alonso-Herrero}, A., \& {Ueno}, S. 2002,
690: \mnras, 330, 259
691:
692: \bibitem[{{Lonsdale} {et~al.}(2003)}]{lonsdale2003}
693: {Lonsdale}, C.~J. {et~al.} 2003, \pasp, 115, 897
694:
695: \bibitem[{{Mazzarella} \& {Balzano}(1986)}]{Mazzarellabalzano1986}
696: {Mazzarella}, J.~M. \& {Balzano}, V.~A. 1986, \apjs, 62, 751
697:
698: \bibitem[{{Moran} {et~al.}(2002){Moran}, {Filippenko}, \&
699: {Chornock}}]{moranetal2002}
700: {Moran}, E.~C., {Filippenko}, A.~V., \& {Chornock}, R. 2002, \apjl, 579, L71
701:
702: \bibitem[{{Mukai}(1993)}]{mukai1993}
703: {Mukai}, K. 1993, Legacy, vol.~3, p.21-31, 3, 21
704:
705: \bibitem[{{Nenkova} {et~al.}(2002){Nenkova}, {Ivezi{\'c}}, \&
706: {Elitzur}}]{nenkovaetal2002}
707: {Nenkova}, M., {Ivezi{\'c}}, {\v Z}., \& {Elitzur}, M. 2002, \apjl, 570, L9
708:
709: \bibitem[{{Persic} {et~al.}(2004)}]{persic2004}
710: {Persic}, M. {et~al.} 2004, \aap, 427, 35
711:
712: \bibitem[{{Pier} \& {Krolik}(1992)}]{pierkrolik1992}
713: {Pier}, E.~A. \& {Krolik}, J.~H. 1992, \apj, 401, 99
714:
715: \bibitem[{{Polletta} {et~al.}(2006)}]{pollettaetal2006}
716: {Polletta}, M.~d.~C. {et~al.} 2006, \apj, 642, 673
717:
718: \bibitem[{{Quadri} {et~al.}(2007)}]{quadrietal2007}
719: {Quadri}, R. {et~al.} 2007, \aj, 134, 1103
720:
721: \bibitem[{{Richards} {et~al.}(2004)}]{richardsetal2004}
722: {Richards}, G.~T. {et~al.} 2004, \apjs, 155, 257
723:
724: \bibitem[{{Rieke} {et~al}(2004)}]{riekeetal2004}
725: {Rieke}, G., {et al.} 2004, \apjs, 154, 25
726:
727: \bibitem[{{Sanders}(1999)}]{sanders1999}
728: {Sanders}, D.~B. 1999, in IAU Symposium, Vol. 194, Activity in Galaxies and
729: Related Phenomena, ed. Y.~{Terzian}, E.~{Khachikian}, \& D.~{Weedman}, 25--+
730:
731: \bibitem[{{Sanders} \& {Mirabel}(1996)}]{sandersmirabel1996}
732: {Sanders}, D.~B. \& {Mirabel}, I.~F. 1996, \araa, 34, 749
733:
734: \bibitem[{{Sanders} {et~al.}(1989){Sanders}, {Phinney}, {Neugebauer}, {Soifer},
735: \& {Matthews}}]{sandersetal1989}
736: {Sanders}, D.~B., {Phinney}, E.~S., {Neugebauer}, G., {Soifer}, B.~T., \&
737: {Matthews}, K. 1989, \apj, 347, 29
738:
739: \bibitem[{{Schawinski} {et~al.}(2007){Schawinski}, {Thomas}, {Sarzi},
740: {Maraston}, {Kaviraj}, {Joo}, {Yi}, \& {Silk}}]{schawinskietal2007}
741: {Schawinski}, K., {Thomas}, D., {Sarzi}, M., {Maraston}, C., {Kaviraj}, S.,
742: {Joo}, S.-J., {Yi}, S.~K., \& {Silk}, J. 2007, ArXiv e-prints, 709
743:
744: \bibitem[{{Severgnini} {et~al.}(2003)}]{severgninietal2003}
745: {Severgnini}, P. {et~al.} 2003, \aap, 406, 483
746:
747: \bibitem[{{Sijacki} {et~al.}(2007){Sijacki}, {Springel}, {di Matteo}, \&
748: {Hernquist}}]{sijackietal2007}
749: {Sijacki}, D., {Springel}, V., {di Matteo}, T., \& {Hernquist}, L. 2007,
750: \mnras, 380, 877
751:
752: \bibitem[{{Stern} {et~al.}(2005)}]{sternetal2005}
753: {Stern}, D. {et~al.} 2005, \apj, 631, 163
754:
755: \bibitem[{{Streblyanska} {et~al.}(2004){Streblyanska}, {Bergeron}, {Brunner},
756: {Finoguenov}, {Hasinger}, \& {Mainieri}}]{streblyanska2004}
757: {Streblyanska}, A., {Bergeron}, J., {Brunner}, H., {Finoguenov}, A.,
758: {Hasinger}, G., \& {Mainieri}, V. 2004, Nuclear Physics B Proceedings
759: Supplements, 132, 232
760:
761: \bibitem[{{Streblyanska} {et~al.}(2006){Streblyanska}, {Bergeron}, {Brunner},
762: {Finoguenov}, {Hasinger}, \& {Mainieri}}]{streblyanska2006}
763: {Streblyanska}, A., {Bergeron}, J., {Brunner}, H., {Finoguenov}, A.,
764: {Hasinger}, G., \& {Mainieri}, V. 2006, in IAU Symposium, Vol. 230,
765: Populations of High Energy Sources in Galaxies, ed. E.~J.~A. {Meurs} \&
766: G.~{Fabbiano}, 450--454
767:
768: \bibitem[{{Surace} {et~al.}(2005){Surace}, {Shupe}, {Fang}, {Evans}, {Alexov},
769: {Frayer}, {Lonsdale}, \& {SWIRE Team}}]{surace2005}
770: {Surace}, J.~A., {Shupe}, D.~L., {Fang}, F., {Evans}, T., {Alexov}, A.,
771: {Frayer}, D., {Lonsdale}, C.~J., \& {SWIRE Team}. 2005, in Bulletin of the
772: American Astronomical Society, Vol.~37, Bulletin of the American Astronomical
773: Society, 1246--+
774:
775: \bibitem[{{Sutherland} \& {Saunders}(1992)}]{sutherlandsaunders1992}
776: {Sutherland}, W. \& {Saunders}, W. 1992, \mnras, 259, 413
777:
778: \bibitem[{{Szokoly} {et~al.}(2004)}]{szokolyetal2004}
779: {Szokoly}, G.~P. {et~al.} 2004, \apjs, 155, 271
780:
781: \bibitem[{Treister} {et~al.}(2008)]{treisteretal2008}
782: {Treister}, E. {et~al.} 2008, in preparation
783:
784: \bibitem[{{Treister} \& {Urry}(2006)}]{treisterurry2006}
785: {Treister}, E. \& {Urry}, C.~M. 2006, \apjl, 652, L79
786:
787: \bibitem[{{Treister} {et~al.}(2004)}]{treisteretal2004}
788: {Treister}, E. {et~al.} 2004, \apj, 616, 123
789:
790: \bibitem[{{Treister} {et~al.}(2006)}]{treisteretal06}
791: {Treister}, E. {et~al.} 2006, \apj, 640, 603
792:
793: \bibitem[{{Vanzella} {et~al.}(2005)}]{vanzella2005}
794: {Vanzella}, E. {et~al.} 2005, \aap, 434, 53
795:
796: \bibitem[{{Virani} {et~al.}(2006){Virani}, {Treister}, {Urry}, \&
797: {Gawiser}}]{viranietal2006}
798: {Virani}, S.~N., {Treister}, E., {Urry}, C.~M., \& {Gawiser}, E. 2006, \aj,
799: 131, 2373
800:
801: \bibitem[{{Wolf} {et~al.}(2004)}]{wolfetal2004}
802: {Wolf}, C. {et~al.} 2004, \aap, 421, 913
803:
804: \bibitem[{{Worsley} {et~al.}(2005)}]{worsleyetal2005}
805: {Worsley}, M.~A. {et~al.} 2005, \mnras, 357, 1281
806:
807: \end{thebibliography}
808:
809:
810: \clearpage
811:
812: \begin{landscape}
813: \tabletypesize{\scriptsize}
814: \begin{deluxetable}{rrrrrrrrrrrrr}
815: \tablecolumns{13}
816: \tablewidth{0pc}
817: \tablecaption{X-ray and Infrared Properties of MUSYC ECDF-S X-ray Sources \label{cat}}
818: \tabletypesize{\scriptsize}
819: %\rotate
820: \tablehead{
821: \colhead{ID} & \colhead{Cat ID$^{\rm a}$} & \colhead{RA} & \colhead{DEC} & \colhead{HR$^{\rm b}$} & \colhead{$3.6 \mu m^{\rm c}$} & \colhead{$4.5 \mu m^{\rm c}$} & \colhead{$5.8 \mu m^{\rm c}$} & \colhead{$8.0 \mu m^{\rm c}$} & \colhead{$24.0 \mu m^{\rm d}$} & \colhead{${L_X}^{\rm e}$} & \colhead{Notes$^{\rm f}$}\\
822: \colhead{} & \colhead{} & \multicolumn{2}{c}{(J2000)} & \colhead{} & \multicolumn{4}{c}{($\mu$Jy)} & \colhead{($\mu$Jy)} & \colhead{ (${\rm ergs \: s^{-1}}$)} & \colhead{} \\
823: \cline{3-4} \cline{6-9} \\
824: }
825: \small
826: \startdata
827: 1 & A03:1 &52.934208 &-27.823917 & -0.09 & 1.52 (0.14) & 1.89 (0.15) & 1.62 (0.62) & 2.31 (0.57) & \nodata (\nodata) & \nodata & \\
828: 2 & A03:2 &52.936333 &-27.865639 & -0.43 & 6.69 (0.12) & 7.94 (0.14) & 5.33 (0.56) & 5.98 (0.57) & \nodata (\nodata) & \nodata & \\
829: 3 & A03:3 &52.936875 &-27.860889 & -0.24 & 11.95 (0.13) & 14.31 (0.15) & 17.28 (0.56) & 14.77 (0.58) & \nodata (\nodata) & \nodata & \\
830: 4 & A03:4 &52.947083 &-27.887028 & -0.34 & 15.15 (0.13) & 22.73 (0.14) & 32.26 (0.58) & 46.54 (0.54) & \nodata (\nodata) & \nodata & \\
831: 5 & A03:5 &52.949917 &-27.845972 & -0.54 & 5.48 (0.14) & 4.25 (0.16) & 3.93 (0.61) & 2.71 (0.62) & \nodata (\nodata) & 43.53 & 2 BL \\
832: 6 & A03:6 &52.950083 &-27.800528 & -0.47 & 23.38 (0.13) & 28.93 (0.15) & 30.18 (0.62) & 22.41 (0.60) & S: 560.34 (28.32) & 41.37 & 5 \\
833: 7 & A03:7 &52.955958 &-27.776222 & -0.37 & 13.32 (0.13) & 16.04 (0.16) & 20.63 (0.59) & 25.57 (0.63) & \nodata (\nodata) & 44.17 & 5 \\
834: 8 & A03:8 &52.956208 &-27.842778 & -0.28 & 5.33 (0.14) & 7.03 (0.16) & 7.80 (0.61) & 7.44 (0.62) & \nodata (\nodata) & \nodata & \\
835: 9 & A03:9 &52.959958 &-27.845028 & 0.10 & 15.81 (0.14) & 19.24 (0.15) & 15.07 (0.61) & 10.65 (0.62) & \nodata (\nodata) & \nodata & \\
836: 10 & A03:10 &52.960125 &-27.864389 & -0.42 &12.34 (0.12) & 11.83 (0.15) & 8.86 (0.57) & 7.60 (0.58) & \nodata (\nodata) & \nodata & \\
837: 11 & A03:11 &52.961000 &-27.883583 & -0.33 & 2.42 (0.13) & 2.85 (0.15) & 2.54 (0.59) & 6.24 (0.59) & \nodata (\nodata) & \nodata & \\
838: 12 & A03:12 &52.963167 &-27.847667 & -0.29 &25.68 (0.14) & 16.94 (0.16) & 13.30 (0.60) & 8.44 (0.62) & \nodata (\nodata) & 42.57 & 2 non-BL \\
839:
840:
841: \enddata
842:
843: \tablecomments{This table is published in its entirety in the electronic edition of the Journal. A portion is shown here for guidance regarding its form and content.}
844: \tablenotetext{a}{Original detection catalog for X-ray source: and X-ray source ID there-in. Counts and Flux are taken from this catalog.}
845: \tablenotetext{b}{Hardness ratio, defined by $\frac{H-S}{H+S}$, where H and S are the X-ray counts in the hard (2-8 keV) and soft (0.5-2 keV) bands, respectively. For sources detected only in the hard band (soft band), the hardness ratio is set equal to 1 (-1). For sources detected only in the full band (0.5-8 keV), the hardness ratio is set equal to 10. G02 defined the hard band as 2-10 keV. Therefore, we adjusted the published hard band counts from G02 to equal those that would have been detected at 2-8 keV, the hard band defined by the other three authors, using PIMMS (version 3.9c; \citealt{mukai1993}) with the average X-ray spectrum measured by G02.}
846: \tablenotetext{c}{Infrared flux in $\mu$Jy followed by 1$\sigma$ error in parentheses, from SIMPLE Survey (Damen et al., in prep)}
847: \tablenotetext{d}{24 micron flux in $\mu$Jy followed by 1$\sigma$ error in parentheses. Source of 24 micron flux is indicated by S for SWIRE \citep{lonsdale2003} or G for GOODS \citep{dickinson2003}}
848: \tablenotetext{e}{Observed luminosity [ergs/s] in the 0.5-8 keV band, using references given in final column.}
849: \tablenotetext{f}{Source of redshift used to calculate the luminosity: 0 for VLT/FORS2 Spectroscopy Version 1.0, 1 for VIMOS VLT Deep Survey (VVDS) Version 1.0, 2 for spectroscopic redshifts from \citet{szokolyetal2004}, 3 for spectroscopic redshifts from \citet{croometal2001}, 4 for spectroscopic redshifts determined from MUSYC spectroscopy (Treister et al. 2008, in prep.), 5 for photometric redshifts from COMBO-17 \citep{wolfetal2004}. Where known, spectra showing broad emission lines (BL) are distinguished from those without broad lines (non-BL).}
850: \end{deluxetable}
851:
852:
853: \clearpage
854: \end{landscape}
855:
856:
857:
858: \end{document}
859:
860: