1: %% Last modified May 14, 2007
2: %%
3:
4: \documentclass[apjl]{emulateapj}
5: \usepackage{natbib,graphicx,apjfonts}
6: \citestyle{aj}
7: %
8: \newcommand{\vdag}{(v)^\dagger}
9: \newcommand{\myemail}{steffen@astro.psu.edu}
10: %
11: \def\arcdeg{\hbox{$^\circ$}}
12: \def\arcmin{\hbox{$^\prime$}}
13: \def\arcsec{\hbox{$^{\prime\prime}$}}
14: \def\flux{\mbox{erg cm$^{-2}$ s$^{-1}$}~}
15: \def\kms{\mbox{km s$^{-1}$}}
16: \def\alphaox{$\alpha_{\mbox{\tiny OX}}$}
17: \def\luv{$l_{\mbox{\scriptsize 2500 \AA}}$}
18: \def\lx{$l_{\mbox{\scriptsize 2 keV}}$}
19:
20: %
21: \def\phn{\phantom{0}} % Phantom numeral for aligning columns in tables
22: \def\phd{\phantom{.}} % Phantom decimal for aligning columns in tables
23: \def\phs{\phantom{$-$}} % Phantom minus sign for aligning
24: \def\simgt{\lower 2pt \hbox{$\, \buildrel {\scriptstyle >}\over {\scriptstyle \sim}\,$}}
25: \def\simlt{\lower 2pt \hbox{$\, \buildrel {\scriptstyle <}\over {\scriptstyle \sim}\,$}}
26: %
27: \def\asca{{\slshape ASCA\/}}
28: \def\chandra{{\slshape Chandra\/}}
29: \def\conx{{\slshape Constellation-X\/}}
30: \def\einstein{{\slshape Einstein\/}}
31: \def\genx{{\slshape Generation-X\/}}
32: \def\heao1{{\slshape HEAO1\/}}
33: \def\hst{{\slshape HST\/}}
34: \def\iso{{\slshape ISO\/}}
35: \def\rosat{{\slshape ROSAT\/}}
36: \def\sax{{\slshape BeppoSAX\/}}
37: \def\spitzer{{\slshape Spitzer\/}}
38: \def\swift{{\slshape Swift\/}}
39: \def\xeus{{\slshape XEUS\/}}
40: \def\xmm{{\slshape XMM-Newton\/}}
41: \def\xray{\mbox{X-ray}}
42: %\shorttitle{Revealing the \xray\ background using {\slshape Spitzer}}
43: \shorttitle{Constraints on the contribution of 24$\mu$m sources to the CXB}
44: \shortauthors{Steffen et al.}
45:
46: \submitted{Submitted to \apj\ Letters, 2007 May 14}
47: %
48: %
49: %%% This is the end of the preamble. Indicate the beginning of the
50: %%% paper itself with \begin{document}.
51: %
52: \begin{document}
53:
54: %% LaTeX will automatically break titles if they run longer than
55: %% one line. However, you may use \\ to force a line break if
56: %% you desire.exit
57:
58: \title{{\slshape Chandra} stacking constraints on the contribution of 24 micron {\slshape Spitzer} sources to the unresolved cosmic X-ray background.}
59:
60: \author{A.~T.~Steffen\altaffilmark{1},
61: W.~N.~Brandt\altaffilmark{1},
62: D.~M.~Alexander\altaffilmark{2},
63: S.~C.~Gallagher\altaffilmark{3},
64: B.~D.~Lehmer\altaffilmark{1}}
65: %
66: \altaffiltext{1}{Department of Astronomy and Astrophysics, 525 Davey Laboratory, Pennsylvania State University, University Park, PA 16802.}
67: \altaffiltext{2}{Department of Physics, University of Durham, South Road, Durham, DH1 3LE, UK.}
68: \altaffiltext{3}{Department of Physics and Astronomy, University of California, 430 Portola Plaza, Box 951547 Los Angeles CA, 90095-1547.}
69:
70: \begin{abstract}
71: We employ \xray\ stacking techniques to examine the contribution
72: from \xray\ undetected, mid-infrared--selected sources to the
73: unresolved, hard ($6-8$~keV) cosmic \xray\ background (CXB). We use
74: the publicly available, $24\mu$m {\slshape Spitzer Space Telescope}
75: MIPS catalogs from the Great Observatories Origins Deep Survey
76: (GOODS) - North and South fields, which are centered on the 2~Ms
77: \chandra\ Deep Field-North and the 1~Ms \chandra\ Deep Field-South,
78: to identify bright ($S_{24\mu{\rm m}} > 80\mu$Jy) mid-infrared
79: sources that may be powered by heavily obscured AGNs. We measure a
80: significant stacked \xray\ signal in all of the \xray\ bands
81: examined, including, for the first time, a significant ($3.2\sigma$)
82: $6-8$~keV stacked \xray\ signal from an X-ray undetected source
83: population. We find that the \xray -undetected MIPS sources make up
84: about $2\%$ (or less) of the total CXB below 6~keV, but about $6\%$
85: in the $6-8$~keV band. The $0.5-8$~keV stacked \xray\ spectrum is
86: consistent with a hard power-law ($\Gamma = 1.44 \pm 0.07$), with
87: the spectrum hardening at higher \xray\ energies. Our findings
88: show that these bright MIPS sources do contain obscured AGNs, but
89: are not the primary source of the unresolved $50\%$ of $6-8$~keV
90: CXB. Our study rules out obscured, luminous QSOs as a significant
91: source of the remaining unresolved CXB and suggests that it most
92: likely arises from a large population of obscured, high-redshift ($z
93: \simgt 1$), Seyfert-luminosity AGNs.
94:
95: \end{abstract}
96: %% Keywords should appear after the \end{abstract} command. The uncommented
97: %% example has been keyed in ApJ style. See the instructions to authors
98: %% for the journal to which you are submitting your paper to determine
99: %% what keyword punctuation is appropriate.
100:
101: \keywords{Galaxies: Active: Nuclei --- Galaxies: Active:
102: Infrared/\mbox{X-ray} --- \mbox{X-rays}: Diffuse Background}
103:
104: \section{Introduction}
105: Deep ``blank-sky'' surveys with the {\em Chandra \xray\ Observatory}
106: (hereafter \chandra ) have resolved $ \sim50 - 90\% $ of the hard ($
107: 2-8 $ keV) cosmic \xray\ background (CXB) into discrete sources
108: \citep[e.g.,][]{bauer04,de_luca04,hickox06}, with the resolved
109: fraction decreasing with increasing \xray\ energy
110: \citep{worsley04a,worsley05}. Subsequent spectroscopic observations
111: have revealed that the majority of the sources are Active Galactic Nuclei \citep[AGNs;
112: e.g.,][]{barger03b,steffen04,szokoly04}. This is consistent with CXB
113: synthesis models, which rely on the assumptions of the ``unified'' AGN
114: model \citep[see][for a review]{antonucci93}, that predict the
115: observed power-law ($\Gamma \simeq 1.4$) shape of the CXB is created
116: by the integrated \xray\ emission from both soft ($\Gamma \sim 1.8$),
117: unobscured and harder ($\Gamma < 1.4$), obscured\footnote{The term
118: `obscured' refers to \xray\ obscuration, unless otherwise noted.}
119: AGNs \citep[e.g.,][]{comastri95,gilli07}.
120:
121: Although deep \xray\ surveys find the highest sky density of AGNs to
122: date \citep[$\sim 7000$ deg$^{-2}$;][]{bauer04}, the decreasing
123: resolved fraction of the CXB with energy suggests that there exists an
124: additional, highly-obscured AGN population that is missed in even the
125: deepest \xray\ surveys \citep{worsley05}. While the majority of the
126: \xray\ emission from these sources is attenuated, a small fraction of
127: hard \xray\ photons can penetrate the obscuring torus. In addition,
128: hard (rest-frame $>10$~keV) \xray\ emission can be scattered into the observer's
129: line-of-sight via Compton reflection, which comprises a relatively
130: larger fraction of the total observed \xray\ emission from a heavily
131: obscured AGN. The fraction of hard \xray\ photons emitted from a
132: heavily obscured source would likely be too small to identify the
133: source individually, but the hard \xray\ emission from many of these
134: undetected, Compton-thick sources could comprise the unresolved CXB.
135:
136: Since the absorbed energy in obscured AGNs is re-emitted in the mid-
137: and far-infrared, obscured AGNs should be bright \spitzer\ sources.
138: The dust reprocessing of AGN accretion energy heats the dust to higher
139: temperatures than can be achieved by heating via stellar processes,
140: and this emission appears to be relatively isotropic \citep{lutz04}.
141: The strong rest-frame $3-8 \mu$m thermal continuum produced by the
142: AGN-heated dust can be used to separate luminous AGNs from starbursts
143: using infrared color-color selection \citep{laurent00,lacy04,stern05}
144: and infrared power-law selection \citep{alonso-herrero06,donley07a}.
145: Once candidate AGNs have been identified, \xray\ stacking analyses can
146: be used to measure their average X-ray properties and determine their
147: contribution to the CXB below the detection limit of individual X-ray
148: sources.
149:
150: In this Letter, we use \xray\ stacking techniques to examine the
151: contribution of MIPS-detected GOODS sources to the unresolved
152: component of the CXB. In \S2 we describe the \xray\ and infrared data
153: used in this study. Our analyses are presented in \S3. Our
154: discussion and conclusions are in \S4. We use the cosmological
155: parameters $H_{0} = 70$ km s$^{-1}$ Mpc$^{-1}$, $\Omega_{M} = 0.3$,
156: and $\Omega_{\Lambda} = 0.7$.
157:
158:
159: \section{Sample}
160:
161: The 2 Ms \chandra\ Deep Field-North and 1 Ms \chandra\ Deep
162: Field-South are currently the deepest $0.5-8$~keV surveys
163: \citep{giacconi02,alexander03}. The high spatial resolution of
164: \chandra\ ($\sim 0 \farcs 5$ at the aimpoint) and \chandra 's low
165: background make these images ideal for examining the resolved fraction
166: of the CXB. We use the \xray\ catalogs of \citet{alexander03} for
167: both the \mbox{CDF-N} and \mbox{CDF-S}, which contain 503 and 326 \xray\ sources,
168: respectively, detected using {\ttfamily wavdetect} at a significance
169: threshold of $10^{-7}$. We also consider the supplementary \xray\
170: source catalogs for both fields, which add an additional 79 (42)
171: sources to the CDF-N (CDF-S); these sources are detected with a looser
172: threshold ($10^{-5}$), but because they are coincident with optically
173: bright ($R<23$) sources, the vast majority ($>90\%$) are likely real.
174:
175:
176: %
177: % FIGURE 1
178: %
179: \begin{figure}[t]
180: \includegraphics[angle=90,scale=0.42]{f1.eps}
181: \caption{\label{mips_xray_hist}Histogram of 24$\mu$m fluxes for the
182: 2147 MIPS-selected sources in the GOODS-North \& South fields
183: ({\slshape open histogram}). The 24$\mu$m flux distribution of MIPS
184: sources with \xray\ counterparts is shown for comparison ({\slshape
185: hatched histogram}).}
186: \end{figure}
187:
188: To identify obscured AGN candidates, we use \spitzer\ data from the
189: Great Observatories Origins Deep Survey \citep[GOODS;][]{dickinson05}.
190: The GOODS infrared catalogs consist of observations in the four IRAC
191: bands \citep[3.6, 4.5, 5.8, and 8.0$\mu$m;][]{fazio04} and one MIPS
192: band \citep[24$\mu$m;][]{rieke04}. The contribution to the CXB from
193: the IRAC-selected sources will be presented in a subsequent paper
194: (Steffen et al. 2007, in prep). In this Letter we measure the
195: contribution of the MIPS sources to the CXB.
196:
197: The GOODS MIPS observations are publicly available\footnote{Available
198: at \url{http://ssc.spitzer.caltech.edu/legacy/goodshistory.html}.}
199: as part of GOODS data releases DR1 (GOODS-North) and DR3
200: (GOODS-South). The GOODS-North and GOODS-South MIPS-selected catalogs
201: contain 1199 and 948 sources, respectively, extending down to a
202: $24\mu$m flux limit of \mbox{$\sim80\mu$Jy.} We cross-correlated the
203: MIPS source positions with the aforementioned \xray\ catalogs using a
204: $3\arcsec$ matching radius. We found 253 (156) CDF-N (CDF-S) MIPS
205: sources matched to an \xray\ counterpart within $3\arcsec$ (with an
206: expectation of $\sim20$ spurious matches). In
207: Figure~\ref{mips_xray_hist}, we present histograms showing the
208: distribution of 24$\mu$m fluxes for all MIPS-selected GOODS
209: sources ({\slshape open histogram}) and for the MIPS sources with
210: \xray\ counterparts ({\slshape hatched histogram}). It is apparent
211: that while the number of MIPS sources increases at fainter
212: fluxes, the number of sources with \xray\ counterparts increases
213: slowly, so the fraction of MIPS sources with a detected
214: \xray\ counterpart decreases as one examines fainter 24$\mu$m MIPS
215: sources. This is consistent with the idea that luminous AGNs power
216: the brightest 24$\mu$m sources, while the fainter sources can be
217: powered by the more common dusty starbursts.
218:
219:
220:
221: \section{Analysis}
222: %
223: % FIGURE 2
224: %
225: \begin{figure}[b]
226: \epsscale{1.0}
227: \plotone{f2.eps}
228: \caption{\label{stacked_thumb}({\slshape a}) Stacked, adaptively
229: smoothed, $4-6$ keV \xray\ thumbnail for the $638$ MIPS-selected AGN
230: candidates. Adaptive smoothing was performed using the
231: \texttt{csmooth} algorithm in CIAO. The $3\arcsec$ diameter aperture
232: used for source photometry is shown. ({\slshape b}) Stacked,
233: adaptively smoothed $6-8$ keV \xray\ thumbnail.}
234: \end{figure}
235:
236: %
237: % TABLE 2
238: %
239: \begin{deluxetable*}{ccccccccc}[t]
240: \tablewidth{0pt}
241: \tablecolumns{9}
242: \tablecaption{\label{mips_stacking_results}Results of \xray\ Stacking Analysis of MIPS Sources}
243: \tablehead{
244: \colhead{Energy Band} &
245: \colhead{Total Counts} &
246: \colhead{Mean Background\tablenotemark{a}} &
247: \colhead{Exposure Map} &
248: \colhead{Encircled Energy} &
249: \colhead{Area\tablenotemark{c}} &
250: \colhead{Total Intensity} &
251: \colhead{\% of CXB\tablenotemark{d}} &
252: \colhead{S/N} \\
253: \colhead{[keV]} &
254: \colhead{[counts]} &
255: \colhead{[counts]} &
256: \colhead{(mean) [cm$^{2}$ s]} &
257: \colhead{Fraction\tablenotemark{b}} &
258: \colhead{[deg$^{2}$]} &
259: \colhead{[ergs cm$^{-2}$ s$^{-1}$ deg$^{-2}$]} &
260: \colhead{} &
261: \colhead{[$\sigma$]} \\
262: \colhead{[1]} &
263: \colhead{[2]} &
264: \colhead{[3]} &
265: \colhead{[4]} &
266: \colhead{[5]} &
267: \colhead{[6]} &
268: \colhead{[7]} &
269: \colhead{[8]} &
270: \colhead{[9]}
271: }
272: \startdata
273: \cutinhead{Standard \chandra\ \xray\ bands}
274: $0.5-8.0$ & $6857.2 \pm 82.8$ & $4698.9 \pm 6.2$ & $3.86 \times 10^{8}$ & 0.556 & $5.676 \times 10^{-2}$ & $(5.50\pm0.21) \times 10^{-13}$ & $2.35\pm0.09$ & 26.0 \\
275: $0.5-2.0$ & $2904.6 \pm 53.9$ & $1429.8 \pm 3.4$ & $3.85 \times 10^{8}$ & 0.614 & $5.930 \times 10^{-2}$ & $(1.64\pm0.06) \times 10^{-13}$ & $2.16\pm0.08$ & 27.3 \\
276: $2.0-8.0$ & $3952.6 \pm 62.9$ & $3269.1 \pm 5.2$ & $3.95 \times 10^{8}$ & 0.542 & $5.559 \times 10^{-2}$ & $(4.22\pm0.34) \times 10^{-13}$ & $2.43\pm0.20$ & 10.8 \\[-5pt]
277:
278: \cutinhead{Narrow \xray\ bands}
279: $0.5-1.0$ & $1202.3 \pm 34.7$ & $\phn679.2 \pm 2.4$ & $2.50 \times 10^{8}$ & 0.624 & $5.950 \times 10^{-2}$ & $(0.60\pm0.04) \times 10^{-13}$ & $1.99\pm0.13$ & 15.0 \\
280: $1.0-2.0$ & $1702.4 \pm 41.3$ & $\phn750.5 \pm 2.5$ & $6.94 \times 10^{8}$ & 0.610 & $5.920 \times 10^{-2}$ & $(0.85\pm0.04) \times 10^{-13}$ & $1.86\pm0.09$ & 23.0 \\
281: $2.0-4.0$ & $1491.7 \pm 38.6$ & $1088.5 \pm 3.0$ & $4.10 \times 10^{8}$ & 0.578 & $5.826 \times 10^{-2}$ & $(1.36\pm0.13) \times 10^{-13}$ & $1.96\pm0.19$ & 10.4 \\
282: $4.0-6.0$ & $1089.9 \pm 33.0$ & $\phn927.8 \pm 2.8$ & $4.32 \times 10^{8}$ & 0.544 & $5.588 \times 10^{-2}$ & $(0.97\pm0.20) \times 10^{-13}$ & $1.73\pm0.36$ & \phn4.9 \\
283: $6.0-8.0$ & $1371.0 \pm 37.0$ & $1252.3 \pm 3.2$ & $1.69 \times 10^{8}$ & 0.504 & $5.463 \times 10^{-2}$ & $(2.83\pm0.89) \times 10^{-13}$ & $5.79\pm1.82$ & \phn3.2 \\[-8pt]
284: \enddata
285: \tablenotetext{a}{The small error bars for the mean background result from the much larger area used to calculate this value, relative to the size of the extraction aperture.}
286: \tablenotetext{b}{This is the mean encircled-energy fraction weighted by the exposure map values of the sources.}
287: \tablenotetext{c}{This is the total area within $6\arcmin$ of the CDF-N and -S aimpoints, corrected for the masked areas within $2\times 90\%$ encircled energy of the known X-ray sources.}
288: \tablenotetext{d}{Assuming the CXB normalization of \citet{hickox06}. The CXB fraction is $11\%$ higher assuming the normalization value measured by \citet{revnivtsev05} and $6\%$ lower using the value from \citet{de_luca04}.}
289: \end{deluxetable*}
290:
291: We used \xray\ stacking techniques to examine the contribution of
292: MIPS-selected AGN candidates to the unresolved CXB. We calculated the
293: total number of photons within a $3\arcsec$ diamater circular
294: aperture, correcting for the fractional contribution from pixels only
295: partially covered by the aperture. The local background is measured
296: by extracting the total number of counts within a $30 {\rm ~pixel}
297: \times 30 {\rm ~pixel}$ ($\sim 15\arcsec \times15\arcsec$) box,
298: centered on a source and masking all of the known X-ray and MIPS
299: source positions.
300:
301: To calculate the expected background in the desired aperture, the
302: total number of background counts is scaled by the ratio of the summed
303: exposure map values within the source aperture and the summed exposure
304: map values in the local background region. This is similar to scaling
305: by the relative sizes of the source aperture and background region,
306: except this method accounts for variations in sensitivity due to chip
307: gaps and edges, and HRMA mirror effects. Our simulations have shown
308: that this method is both much faster and more accurate than measuring
309: the local background using thousands of randomly placed apertures, as
310: would be adopted in a Monte-Carlo approach.
311:
312: For each source, the encircled-energy fraction was calculated using a
313: tabular parameterization of the \chandra\ PSF, provided by the CXC,
314: that gives the radius of a circular aperture for a given off-axis
315: angle, photon energy, encircled-energy fraction, and azimuthal
316: angle.\footnote{Available at
317: \url{http://cxc.harvard.edu/cal/Hrma/psf/ECF/hrmaD1996-12-20hrci\_ecf\_N0002.fits}.}
318: Using our desired aperture size, mean photon energy, and the sources'
319: off-axis angles from the exposure-weighted mean aimpoints given by
320: \citet{alexander03}, we interpolated the HRMA FITS table to obtain the
321: encircled-energy fraction for each source. Since the individual
322: images used in the CDF-N and CDF-S mosaics were taken at a variety of
323: roll angles, we removed the small azimuthal dependence by averaging
324: the PSF parameterization over all azimuthal angles.
325:
326: To calculate the total \xray\ flux for all of the stacked sources, the
327: total background-subtracted source counts (in photons) was divided by
328: the sum of the mean exposure-map value within each extraction aperture
329: (in cm$^{2}$ s) to obtain the total photon flux for the stacked
330: sources. This value was divided by the mean encircled-energy
331: fraction, weighted by the mean exposure-map value for each source, and
332: converted to energy flux by assuming a power-law spectrum with $\Gamma
333: = 1.4$ and correcting for Galactic absorption using the \xray\ opacity
334: table of \citet{morrison83} and the exposure-weighted mean Galactic
335: column density (N$_{\rm HI}=1.16 \times 10^{20}$~cm$^{-2}$).
336:
337: In Table~\ref{mips_stacking_results}, we show the stacking results for
338: the 638 MIPS sources that lay within $6\arcmin$ of the
339: exposure-weighted mean aimpoints of the CDF-N or the CDF-S, and were
340: outside $2\times$ the radius of the 90\% encircled-energy fraction
341: of the \xray\ sources in the \citet{alexander03} main and
342: supplementary catalogs to avoid contamination from known \xray\
343: sources. The \xray\ stacking was performed in the three standard
344: \chandra\ bands, the full ($0.5-8.0$ keV), soft ($0.5-2$~keV), and
345: hard ($2-8$~keV) bands, as well as five narrower, non-overlapping
346: \xray\ bands.
347:
348: From Table~\ref{mips_stacking_results}, the \xray\
349: undetected MIPS source population makes up about 2\% of the total CXB
350: intensity below 6~keV. The resolved fraction of the CXB increases to
351: $\sim6\%$ for the hardest \xray\ energy band analyzed here.
352: This is the first time that a statistically significant, stacked
353: $6-8$~keV signal has been found for an \xray\ undetected population of
354: sources. \xray\ stacking analyses were used to measure the
355: contribution of the optical GOODS sources to the CXB, but no
356: significant $6-8$~keV signal was found \citep{worsley06}. To verify
357: the authenticity of the $3.2\sigma$ detection in the $6-8$~keV energy
358: band, we extracted thumbnail \xray\ images centered on each \xray\
359: undetected MIPS source and coadded them. In
360: Figure~\ref{stacked_thumb}, we show the resulting $15\arcsec \times
361: 15\arcsec$ stacked \xray\ thumbnail for both the ({\slshape a})
362: $4-6$~keV and ({\slshape b}) $6-8$~keV bands with our $3\arcsec$
363: diameter aperture overlaid ({\slshape black circle}). It is clear
364: from these smoothed images that there is a significant stacked \xray\
365: signal within our extraction aperture for both \xray\ bands
366: ($p=6.9\times10^{-4}$ in the $6-8$~keV band, assuming a one-tailed
367: Gaussian distribution). Our $3\arcsec$ aperture was chosen to
368: maximize the stacked signal within the aperture while minimizing the
369: background signal. While the peak of the stacked X-ray image is not
370: precisely centered within our aperture, using a larger aperture does
371: not increase the measured S/N for the stacked $6-8$~keV signal.
372:
373: %
374: % FIGURE 3
375: %
376: \begin{figure}[b]
377: \epsscale{1.0}
378: \plotone{f3.eps}
379: \caption{\label{stack_vs_mag}The total resolved $6-8$~keV CXB
380: fraction, the signal-to-noise of the stacked \xray\ signal, and the
381: total number of stacked sources as a function of 24 $\mu$m flux
382: assuming the CXB normalization of \citet{hickox06}. Error bars
383: ($1\sigma$) are shown for stacked sources that have S/N > 3, and
384: $3\sigma$ upper limits are given for the stacked sources that are
385: not significantly detected.}
386: \end{figure}
387:
388:
389: While we detect a significant $6-8$~keV \xray\ signal for the stacked,
390: \xray --undetected MIPS population we do not know how this low-level
391: \xray\ flux is distributed among the sources. Is the stacked
392: $6-8$~keV signal dominated by a small number of bright $24\mu$m
393: sources, as one might suspect from Figure~\ref{mips_xray_hist}? To
394: examine the importance of infrared flux on our \xray\ stacking
395: analysis, we break down our earlier stacking results by performing
396: stacking as a function of limiting $24\mu$m flux. In
397: Figure~\ref{stack_vs_mag}, we show the resolved fraction of the CXB,
398: the signal-to-noise (S/N), and the number of sources stacked as a
399: function of limiting $24\mu$m flux. Overall, it appears that the S/N
400: of our stacked \xray\ signal does gradually increase as we include the
401: \xray\ flux from fainter MIPS sources, which suggests that the stacked
402: \xray\ signal is not dominated by a small number of bright $24\mu$m
403: sources.
404:
405: \section{Discussion and Conclusions}
406:
407: We present in this Letter results from an \xray\ stacking analysis of
408: the \xray--undetected $24\mu$m {\slshape Spitzer} MIPS sources using
409: the GOODS catalogs and the {\chandra} Deep Fields. We find, for the
410: first time, a significant stacked \xray\ signal ($3.2\sigma$) in the
411: $6-8$~keV band for an \xray -undetected, AGN-candidate sample,
412: suggesting that at least some of these sources harbor heavily obscured
413: AGNs. Approximately $12\%$ of the unresolved fraction \citep[$\sim
414: 50\%$;][]{worsley05} of the $6-8$~keV CXB can be attributed to these
415: sources. The stacked $0.5-8$~keV spectrum has a power-law photon
416: index of $\Gamma=1.44 \pm 0.07$, consistent with the slope of the CXB
417: \citep{hickox06}. The slope of the stacked \xray\ spectrum hardens
418: with increasing \xray\ energy, suggesting that the flux from obscured
419: AGNs dominates the stacked spectrum at high \xray\ energies.
420:
421: We found evidence that there exist mid-IR-bright,
422: heavily obscured AGNs that are not individually detected in the \xray\
423: band, but we know little about the general properties of these
424: sources. By examining how the \xray\ to $24\mu$m flux relation for
425: AGNs evolves with redshift, we can infer the AGN types that are
426: detected in this mid-IR catalog. In Figure~\ref{sey2_evol}, we
427: examine the evolution of the \xray\ to $24\mu$m flux relation for
428: obscured AGNs using the observed SEDs of three obscured AGNs: the
429: luminous, Type 2 quasar CXO52 \citep[distant Compton-thin, obscured
430: QSO; $L_{2-10 \rm{~keV}} = 3.3 \times 10^{44}$ ergs s$^{-1}$,
431: intrinsic;][]{stern02}, NGC~1068 \citep[extreme Compton-thick,
432: luminous AGN; $L_{2-10 \rm{~keV}} \sim 10^{43} - 10^{44}$ ergs
433: s$^{-1}$, intrinsic;][]{cappi06}; and the Circinus Galaxy
434: \citep[Compton-thick, moderate-luminosity AGN; $L_{2-10 \rm{~keV}}
435: \sim 10^{42}$ ergs s$^{-1}$, intrinsic;][]{matt99}. This is similar
436: to Figure~1 of \citet{martinez-sansigre06}, except here we utilize the
437: observed SEDs of known obscured AGNs instead of using theoretical
438: models. The rest-frame \xray\ and IR fluxes were calculated by
439: interpolating the SEDs provided by \citet{stern02} for CXO52, and the
440: NASA/IPAC Extragalactic Database (NED) for NGC~1068 and the Circinus
441: Galaxy.
442:
443: %
444: % FIGURE 4
445: %
446: \begin{figure}[t]
447: \epsscale{1.0}
448: \plotone{f4.eps}
449: \caption{\label{sey2_evol}The redshift evolution of the observed-frame
450: $2-10$~keV flux and $24\mu$m flux density for three obscured AGNs,
451: CXO52 \citep[{\slshape dashed curve}]{stern02}, NGC 1068 ({\slshape
452: solid curve}), and the Circinus galaxy ({\slshape dotted curve}).
453: The two dotted horizontal lines and one dotted vertical line denote the average flux limits for
454: the 2 Ms CDF-N, 1 Ms CDF-S, and the GOODS MIPS catalogues, respectively.}
455: \end{figure}
456:
457: From Figure~\ref{sey2_evol} it is apparent that a luminous,
458: significantly obscured AGN such as CXO52 would be detected in both the
459: GOODS MIPS catalog and the deep \xray\ surveys, even at $z>5$, and
460: would thus would not contribute to the unresolved portion of the CXB.
461: Even if this source were Compton thick, it would easily be detected at
462: $24\mu$m and would be included in our \xray\ stacking analysis. Given
463: the small fraction of the CXB that we resolve at $6-8$~keV, the
464: unresolved CXB at these energies {\slshape cannot} be attributed to a
465: population of luminous, heavily obscured, ``Type 2'' Quasars at any
466: redshift (unless Compton-thick QSOs are significantly fainter at
467: $24\mu$m than their less obscured counterparts). In addition, it is
468: apparent that heavily obscured, low-luminosity Seyfert 2s without
469: significant star-formation can be detected in the MIPS band at
470: $z<0.8$, but fall below the GOODS detection threshold at higher
471: redshifts. This suggests that the unresolved $6-8$~keV CXB is not
472: emanating primarily from a population of low-luminosity, low-redshift
473: AGNs but could be from a population of low-luminosity, $z>0.8$ AGNs.
474: At these low $24\mu$m fluxes it is difficult to separate AGNs from
475: dusty starburst galaxies, which makes stacking analyses problematic
476: due to the increased background signal in the $6-8$~keV band from
477: obscured starbursts.
478:
479: The average $0.5-2$~keV flux for our MIPS-selected AGNs is $S_{0.5-2
480: {\rm ~keV}} \simeq 5.4\times10^{-18}$ ergs cm$^{-2}$ s$^{-1}$, a
481: factor of $\sim4.6$ lower than the flux limit of the 2~Ms CDF-N
482: \citep{alexander03}. Deeper \xray\ observations will help to improve
483: significantly the S/N of the stacked \xray\ signal and will greatly improve
484: our measurements of the $6-8$~keV CXB from these sources. In
485: addition, if it is possible to seperate better the obscured starbursts
486: from the infrared-selected AGN candidates, the significance of the
487: stacked \xray\ detections should improve. We plan to address
488: additional infrared AGN selection techniques in a subsequent paper
489: \citep{steffen07b}.
490:
491: \acknowledgements
492: %We thank the anonymous referee for helpful comments that improved the manuscript.
493: We gratefully acknowledge support from CXC grant GO4-5157A (A.~T.~S.,
494: W.~N.~B., B.~D.~L.), SAO grant AR6-7012X (A.~T.~S.), JPL grant 1278940
495: (A.~T.~S.), NASA LTSA grant NAG5-13035 (W.~N.~B.), the Royal
496: Society (D.~M.~A), and JPL grant 1268000 (S.~C.~G.).
497:
498: {\it Facilities:} \facility{{\slshape CXO} (ACIS)}, \facility{{\slshape Spitzer} (MIPS)}
499:
500: \begin{thebibliography}{30}\setlength{\itemsep}{0mm}
501: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
502:
503: \bibitem[\protect\astroncite{{Alexander} et~al.}{2003}]{alexander03}
504: {Alexander}, D.~M., et~al. 2003, \aj, 126, 539
505:
506: \bibitem[\protect\astroncite{{Alonso-Herrero} et~al.}{2006}]{alonso-herrero06}
507: {Alonso-Herrero}, A., et~al. 2006, \apj, 640, 167
508:
509: \bibitem[\protect\astroncite{{Antonucci}}{1993}]{antonucci93}
510: {Antonucci}, R. 1993, \araa, 31, 473
511:
512: \bibitem[\protect\astroncite{{Barger} et~al.}{2003}]{barger03b}
513: {Barger}, A.~J., et~al. 2003, \aj, 126, 632
514:
515: \bibitem[\protect\astroncite{{Bauer} et~al.}{2004}]{bauer04}
516: {Bauer}, F.~E., et~al. 2004, \aj, 128, 2048
517:
518: \bibitem[\protect\astroncite{{Cappi} et~al.}{2006}]{cappi06}
519: {Cappi}, M., et~al. 2006, \aap, 446, 459
520:
521: \bibitem[\protect\astroncite{{Comastri} et~al.}{1995}]{comastri95}
522: {Comastri}, A., {Setti}, G., {Zamorani}, G., \& {Hasinger}, G. 1995, \aap, 296,
523: 1
524:
525: \bibitem[\protect\astroncite{{De Luca} \& {Molendi}}{2004}]{de_luca04}
526: {De Luca}, A. \& {Molendi}, S. 2004, \aap, 419, 837
527:
528: \bibitem[\protect\astroncite{{Dickinson} \& {GOODS Team}}{2005}]{dickinson05}
529: {Dickinson}, M. \& {GOODS Team} 2005, American Astronomical Society Meeting
530: Abstracts, 207
531:
532: \bibitem[\protect\astroncite{{Donley} et~al.}{2007}]{donley07a}
533: {Donley}, J.~L., et~al. 2007, \apj, 660, 167
534:
535: \bibitem[\protect\astroncite{{Fazio} et~al.}{2004}]{fazio04}
536: {Fazio}, G.~G., et~al. 2004, \apjs, 154, 10
537:
538: \bibitem[\protect\astroncite{{Giacconi} et~al.}{2002}]{giacconi02}
539: {Giacconi}, R., et~al. 2002, \apjs, 139, 369
540:
541: \bibitem[\protect\astroncite{{Gilli} et~al.}{2007}]{gilli07}
542: {Gilli}, R., {Comastri}, A., \& {Hasinger}, G. 2007, \aap, 463, 79
543:
544: \bibitem[\protect\astroncite{{Hickox} \& {Markevitch}}{2006}]{hickox06}
545: {Hickox}, R.~C. \& {Markevitch}, M. 2006, \apj, 645, 95
546:
547: \bibitem[\protect\astroncite{{Lacy} et~al.}{2004}]{lacy04}
548: {Lacy}, M., et~al. 2004, \apjs, 154, 166
549:
550: \bibitem[\protect\astroncite{{Laurent} et~al.}{2000}]{laurent00}
551: {Laurent}, O., et~al. 2000, \aap, 359, 887
552:
553: \bibitem[\protect\astroncite{{Lutz} et~al.}{2004}]{lutz04}
554: {Lutz}, D., {Maiolino}, R., {Spoon}, H.~W.~W., \& {Moorwood}, A.~F.~M. 2004,
555: \aap, 418, 465
556:
557: \bibitem[\protect\astroncite{{Mart{\'{\i}}nez-Sansigre}
558: et~al.}{2006}]{martinez-sansigre06}
559: {Mart{\'{\i}}nez-Sansigre}, A., et~al. 2006, \mnras, 370, 1479
560:
561: \bibitem[\protect\astroncite{{Matt} et~al.}{1999}]{matt99}
562: {Matt}, G., et~al. 1999, \aap, 341, L39
563:
564: \bibitem[\protect\astroncite{{Morrison} \& {McCammon}}{1983}]{morrison83}
565: {Morrison}, R. \& {McCammon}, D. 1983, \apj, 270, 119
566:
567: \bibitem[\protect\astroncite{{Revnivtsev} et~al.}{2005}]{revnivtsev05}
568: {Revnivtsev}, M., {Gilfanov}, M., {Jahoda}, K., \& {Sunyaev}, R. 2005, \aap,
569: 444, 381
570:
571: \bibitem[\protect\astroncite{{Rieke} et~al.}{2004}]{rieke04}
572: {Rieke}, G.~H., et~al. 2004, \apjs, 154, 25
573:
574: \bibitem[\protect\astroncite{{Steffen} et~al.}{2004}]{steffen04}
575: {Steffen}, A.~T., et~al. 2004, \aj, 128, 1483
576:
577: \bibitem[\protect\astroncite{{Steffen} et~al.}{2007}]{steffen07b}
578: {Steffen}, A.~T., et~al. 2007, {in prep}
579:
580: \bibitem[\protect\astroncite{{Stern} et~al.}{2002}]{stern02}
581: {Stern}, D., et~al. 2002, \apj, 568, 71
582:
583: \bibitem[\protect\astroncite{{Stern} et~al.}{2005}]{stern05}
584: {Stern}, D., et~al. 2005, \apj, 631, 163
585:
586: \bibitem[\protect\astroncite{{Szokoly} et~al.}{2004}]{szokoly04}
587: {Szokoly}, G.~P., et~al. 2004, \apjs, 155, 271
588:
589: \bibitem[\protect\astroncite{{Worsley} et~al.}{2004}]{worsley04a}
590: {Worsley}, M.~A., et~al. 2004, \mnras, 352, L28
591:
592: \bibitem[\protect\astroncite{{Worsley} et~al.}{2005}]{worsley05}
593: {Worsley}, M.~A., et~al. 2005, \mnras, 357, 1281
594:
595: \bibitem[\protect\astroncite{{Worsley} et~al.}{2006}]{worsley06}
596: {Worsley}, M.~A., et~al. 2006, \mnras, 368, 1735
597:
598: \end{thebibliography}
599:
600: %
601: %%% The following command ends your manuscript. LaTeX will ignore any text
602: %%% that appears after it.
603: \end{document}
604:
605: