1: \documentclass[12pt,preprint]{aastex}
2: \newcommand{\object}{CXO CDFS J033225.3-274219}
3: \newcommand{\objecta}{CXO CDFS J033236.8-274407}
4: \newcommand{\objectb}{CXO CDFS J033213.3-274241}
5: \newcommand{\chandra}{$Chandra$ }
6: \newcommand{\pkso}{PKS~0637$-$752 }
7: \newcommand{\pksiron}{PKS~2149$-$306 }
8: \newcommand{\etal}{{\it et al.} }
9: \newcommand{\myemail}{jxw@pha.jhu.edu}
10:
11:
12: \shortauthors{Wang et al.}
13: \begin{document}
14:
15: \title{A PUZZLING X-RAY SOURCE FOUND IN THE \chandra DEEP FIELD SOUTH}
16: \author{J. Wang\altaffilmark{1},
17: T. Yaqoob\altaffilmark{1,2},
18: G. Szokoly\altaffilmark{3},
19: R. Gilli\altaffilmark{1,4},
20: L. Kewley\altaffilmark{5},
21: V. Mainieri\altaffilmark{6},
22: M. Nonino\altaffilmark{7},
23: P. Rosati\altaffilmark{6},
24: P. Tozzi\altaffilmark{7},
25: W. Zheng\altaffilmark{1},
26: A. Zirm\altaffilmark{1},
27: and C. Norman\altaffilmark{1,8}
28: }
29: \altaffiltext{1}{Dept. of Physics and Astronomy, The Johns Hopkins University,
30: Baltimore, MD 21218}
31: \altaffiltext{2}{Laboratory for High Energy Astrophysics, NASA/GSFC, code 662,
32: Greenbelt, MD 20771}
33: \altaffiltext{3}{Max-Planck-Institut f\"{u}r extraterrestrische Physik, Postfach 1312, D-85741 Garching, Germany}
34: \altaffiltext{4}{Istituto Nazionale di Astrofisica (INAF) - Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, 50125 Firenze, Italy}
35: \altaffiltext{5}{Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138}
36: \altaffiltext{6}{European Southern Observatory, Karl-Schwarzschild-Strasse 2, Garchin
37: g, D-85748, Germany}
38: \altaffiltext{7}{Istituto Nazionale di Astrofisica (INAF) - Osservatorio Astronomico, Via G. Tiepolo 11, 34131 Trieste, Italy}
39: \altaffiltext{8}{Space Telescope Science Institute, 3700 San Martin Drive,
40: Baltimore, MD 21218}
41: \vspace{0.1cm}
42:
43: \begin{abstract}
44:
45: In this letter we report the detection of an extremely strong X-ray
46: emission line in the 940ks \chandra ACIS-I spectrum of \object.
47: The source was identified as a Type~1 AGN at redshift of $z$ = 1.617,
48: with 2.0 -- 10.0 keV rest frame X-ray luminosity of
49: $\sim$ 10$^{44}$ ergs s$^{-1}$. The emission line was detected at
50: 6.2$^{+0.2}_{-0.1}$ keV, with an equivalent width (EW) of 4.4$^{+3.2}_{-1.4}$
51: keV, both quantities referring to the {\it observed frame}. In the rest
52: frame, the line is at 16.2$^{+0.4}_{-0.3}$ keV with an
53: EW of 11.5$^{+8.3}_{-3.7}$ keV. An X-ray emission line at similar
54: energy ($\sim$ 17 keV,
55: rest frame) in \pksiron was discovered before using ASCA data.
56: We reject the possibility that the line is due
57: to a statistical or instrumental artifact.
58: The line is most likely due to blueshifted Fe-K
59: emission from an relativistic outflow, probably an inner X-ray jet,
60: with velocities of the order of $\sim 0.6-0.7c$.
61: Other possible explanations are also discussed.
62: \end{abstract}
63: \keywords{galaxies: active --- galaxies: emission lines ---
64: galaxies: individual (\object) --- X-rays: galaxies}
65:
66: \section{Introduction}
67:
68: Along with the 1 Ms $Chandra$ exposure of
69: \chandra Deep Field North (CDFN, Brandt et al. 2001, 2 Ms exposure
70: available now, see Alexander et al. 2002), 1 Ms exposure of
71: \chandra Deep Field South (CDFS, Rosati et al. 2002) constitute the
72: deepest X-ray exposure of a part of the sky ever taken.
73: With these two deep surveys,
74: the original X-ray glow discovered
75: by Giacconi et al. in 1962
76: is now almost completely resolved into individual sources.
77: In CDFS, 346 such X-ray sources have been detected
78: in $\sim$ 0.1 deg$^2$ (Giacconi et al. 2002), most of which are extragalatic
79: sources harboring supermassive black holes.
80: Great interest is now focusing on studying the properties of these X-ray
81: sources and understanding their physical nature.
82: In this letter, we report the discovery of an extremely strong X-ray
83: emission-line feature in one of the
84: CDFS sources (a type~1 active galactic nucleus, or AGN, at $z=1.617$).
85: The line is at 16.2$^{+0.4}_{-0.3}$ keV in the rest frame, with a huge
86: equivalent width of 11.5$^{+8.3}_{-3.7}$ keV.
87: Both of these parameters are highly unusual.
88:
89: \section{The Data}
90: The 1~Ms exposure of CDFS
91: was composed of eleven individual \chandra ACIS observations from
92: Oct. 1999 to Dec. 2000.
93: The detailed X-ray data reduction
94: and the final catalog of the X-ray detected sources were given in
95: Giacconi et al. (2002).
96: \object, 6$\arcmin$ from the center of the field, was covered by all
97: eleven exposures.
98: Its X-ray radial intensity profile
99: is consistent with that of a point-source
100: (Giacconi et al. 2002), and no other X-ray source was detected within
101: 38$\arcsec$.
102:
103: The source was covered by the deep $R$
104: band [$R$(Vega)$<$26] primary optical imaging survey of CDFS,
105: which was obtained using the FORS1 camera on the ANTU (UT-1 at VLT) telescope.
106: The $R$ band image cut-out of the source overplotted with X-ray flux
107: contours is given in Figure~1.
108: From the cut-out we can clearly see that the optical counterpart of
109: the X-ray source is located at the center of
110: the X-ray flux contours. The nearest other optical source is 5.6$\arcsec$
111: away, far outside of the 3$\sigma$ positional error circle (which has
112: a radius of 1.55$\arcsec$, see Giacconi et al. 2002 for details).
113: The CDFS was observed with the VLA at 1345 MHz and 4860 MHz in Oct,
114: 1999 and Feb, 1999 respectively (Kellermann et al. in preparation).
115: The measured flux densities at the position of CXO CDFS
116: J033225.3-274219 at the two frequencies are 71 $\pm$ 13 and 43 $\pm$ 12
117: microJy corresponding to a spectral index of -0.41.
118: The source is radio loud with radio to optical ratio R = 160.
119:
120: Low-resolution optical spectroscopy of \object\ was obtained with the
121: robotic multi-slit mode of FORS-1 on the ESO Very Large Telescope 8.2m
122: unit facility (ESO/VLT-ANTU) in November 23, 2000. A 1.2\arcsec\ slit
123: was placed on the object (seeing was 0.6\arcsec\ during most of the
124: observation). We used the lowest resolution, I150 grism and {\em no}
125: order-separation filter. The total integration time was 2.5 hours under
126: photometric conditions. The data were reduced following `standard industry
127: practices' and is shown in Figure~1. The object was classified
128: with relatively high confidence as a $z=1.617\pm0.01$ type-I AGN,
129: based on two certain and one marginal broad emission line (C-III, Mg-II, C-IV,
130: with FWHM $>$ 1800 km/s).
131: The source has
132: an optical-to-X-ray ratio $\alpha_{ox}$ of 1.2, which is right of the
133: average value of soft X-ray selected AGNs (Puchnarewicz, et al. 1996).
134:
135:
136: In Figure~2, we plotted the X-ray count rates of the source extracted from
137: eleven $Chandra$ exposures. Since each exposure has similar effective area
138: (good to within $\pm$ 3\%) at the source position, the count rates
139: are representative of relative X-ray flux levels.
140: We can see that the continuum count
141: rates of all but one of the exposures are consistent with the average value
142: (within 3$\sigma$ errorbars).
143: We also looked into the individual exposures, but due to the limited counts,
144: no evidence of variance on a smaller time scale could be
145: established.
146:
147: \section {X-ray spectral fitting}
148:
149: The \chandra ACIS X-ray spectrum of the source was extracted from
150: a circle with radius of 6.2$\arcsec$,
151: which is the 95\% encircled-energy radius of the
152: ACIS point-spread function at the source position,
153: and the background was extracted from an annulus with outer radius of
154: 18$\arcsec$ and inner radius of 8.2$\arcsec$.
155: The X-ray telescope response (ARF) and CCD ACIS-I instrument response
156: was generated for each single \chandra observation,
157: and the final time-weighted files were used for spectral analysis.
158: The summed spectrum (source plus background) and the expected
159: background were shown in Figure~3.
160: After subtracting the background, we obtained around 440 X-ray counts in
161: the soft band (0.5 -- 2.0 keV), and 130 in the hard band (2.0 -- 9.0 keV).
162: Since we don't have enough counts, especially in the hard band,
163: we use C statistics (Cash 1979, Nousek \& Shue 1989) but compare with
164: results using $\chi^2$ for reference,
165: and XSPEC v11.2, which enables
166: the utilization of C statistics in the case when a background spectrum is
167: also in use \footnote{see http://heasarc.gsfc.nasa.gov/docs/xanadu/xspec/manual/node57.html}, for the spectral-fitting analysis.
168: All the spectral fitting was done in the energy band 0.5 -- 9.0 keV.
169: All statistical errors
170: quoted in this paper are for 90\% confidence, one interesting parameter.
171:
172: The unbinned spectrum was first fitted with a simple power-law plus a neutral
173: absorber in the quasar frame. A Galactic
174: neutral hydrogen absorption column of $8 \times 10^{19} \ \rm cm^{-2}$
175: (Dickey \& Lockman 1990) was also included.
176: The X-ray spectrum is fairly steep ($\Gamma$ = 2.1$^{+0.2}_{-0.3}$)
177: with moderate absorption
178: (N$_H$ = 1.3$^{+0.7}_{-0.8}$ $\times$ 10$^{22}$ cm$^{-2}$).
179: The absorption-corrected 2.0 -- 10.0 keV luminosity in the rest frame
180: is 0.8 $\times$ 10$^{44}$ ergs.s$^{-1}$ ($H_0$ = 60.0 km s$^{-1}$Mpc$^{-1}$,
181: $\Omega_m$ = 0.3, $\Omega_\Lambda$ = 0.7).
182: The ratio of data to the model for the fit is given in Figure~4 (upper panel).
183: From both figure~3 and 4, we can see a huge bump at $\sim$ 6.2 keV (observed
184: frame), indicating a possible strong emission line.
185: A single Gaussian was then added to our continuum model.
186: With three more free parameters (the line centroid energy $E_c$, the line
187: width $\sigma$, and
188: the line intensity $I$), the fit was significantly improved
189: ($\Delta$ C = - 22.6), indicating a confidence level of 99.99\%.
190: The results are given in Table~1.
191: In the quasar frame we obtain $E_c$ = 16.2$^{+0.4}_{-0.3}$ keV, $\sigma$ =
192: 0.5$^{+0.6}_{-0.2}$ keV, and the
193: equivalent width EW = 11.5$^{+8.3}_{-3.7}$ keV.
194: The large rest-frame energy and
195: large EW are both highly unusual and present a challenge
196: for the interpretation of the origin of the line emission.
197: We attempted to perform time-resolved spectroscopy
198: on the emission line but found that
199: the data are not of sufficiently high quality
200: to investigate the line variability.
201:
202: We also tried C and $\chi^2$ statistics on the rebinned spectrum (at least 10
203: counts per bin), and present the results in table~1 for comparison.
204: We can see that three methods gave almost the same best-fit values for
205: the parameters, but different confidence levels.
206: Using the rebinned spectrum, C statistics gives higher significance
207: level of the line and slightly narrower errors than $\chi^2$ statistics does,
208: which is consistent with the prediction of Cash (1979).
209: Since rebinning the spectrum would smooth the line feature,
210: it's also reasonable that the line is more significant in the
211: unbinned spectrum than in the rebinned one.
212: \section{Discussion}
213: \label{lineorigin}
214:
215: It is clear that the emission line is not an artifact
216: of calibration uncertainty in the ACIS instrumental
217: response function.
218: No such artifact (i.e. an emission-like feature at
219: $\sim 6.2$ keV) has been reported, as far as we are aware.
220: In addition, we examined the ratios of data to a simple power-law
221: plus neutral absorber model for \objecta, and \objectb, two nearby sources
222: with similar X-ray counts. These are also shown in Figure~4 for comparison,
223: and it is clear that the feature is not observed in the other two sources.
224: Also, since the total exposure
225: was composed of eleven individual observations
226: with different roll angles, photons from each source fell into
227: different positions on the instrument during different observations.
228: It is unlikely that an unknown instrumental artifact
229: (if there is any), which occurs
230: only at certain positions on the
231: detector, would affect photons
232: from all eleven exposures. Figure~2 shows the 5.7 -- 6.7 keV count rates
233: of \object\ from the eleven exposures, which are consistent with
234: the average value within 3$\sigma$ errorbars.
235: We note that the 5.7 -- 6.7 keV count rates from exposure No.1 and 5 are about
236: 3$\sim$4 times of the average value. However, they only contribute 6\% of the
237: total exposure time and 19\% of the 5.7 -- 6.7 keV counts, and fittings with
238: the two exposures excluded yield similar results for the line
239: feature ($E_c$ = 6.1$^{+0.5}_{-0.2}$
240: keV, EW = 3.1$^{+3.8}_{-1.9}$ keV, both in the observed frame).
241: This rules out the possibility that the line is due to an unknown artifact
242: which only affects one or several of the exposures.
243: For the same reason, the line cannot either be due to a transient event
244: not associated with the source.
245: We can also eliminate inadequate background subtraction as the
246: origin of the emission line.
247: We tried using background spectra obtained from different
248: regions of the detector and obtained similar results.
249: An even more compelling reason
250: is that the spectrum {\it without} background subtraction
251: still exhibits the emission line, but the background spectrum itself does not.
252: And again the fact that we do not see the emission line in the
253: other two sources shown in Figure~4
254: indicates that it is not due to the instrument
255: background or X-ray background.
256:
257: Yaqoob et al. (1999) reported the discovery of a highly Doppler
258: blueshifted Fe-K emission line in QSO PKS 2149-306 from an ASCA
259: observation. In the quasar frame, the line has a similar center
260: energy (17.0$\pm{0.5}$ keV) with that of our source, but lower
261: equivalent width (300$\pm{200}$ eV). Furthermore, if we consider
262: both the lines as blueshifted Fe-K fluorescent lines,
263: the blueshift factors are also similar with that of
264: another emission line detected in PKS 0637-75 at 1.6 keV
265: (with EW = 60 eV, rest frame) from ASCA data, which was
266: interpreted as highly blueshifted O$_{VII}$ (Yaqoob et al. 1998).
267: The good agreements of the blueshift factors
268: from different sources and different instruments
269: make them more convincing and interesting.
270:
271: So, could the line we detected also be explained as a highly
272: Doppler blueshifted Fe-K emission line?
273: The Doppler factor of 2.3--2.5 then implies bulk velocities
274: of $\sim 0.62-0.72c$ (head-on) must be responsible for the blueshift
275: since the rest-frame energy of the line must be in the range
276: 6.4--6.97 keV (neutral to H-like Fe).
277: Inflow (observed from the
278: far side) can be ruled out quite
279: simply because the only way to produce Keplerian velocities
280: high enough is to approach a compact object to within
281: a few gravitational radii, but by that time the gravitational
282: redshift is so large that it prevents the net
283: blueshift attaining a high enough value. Exact calculations
284: show this to be the case even for a maximally rotating Kerr
285: hole in which the rotation can help increase the blueshift.
286: Therefore, outflow is required.
287: Yaqoob et al. (1999) pointed out that Fe-K line photons
288: originating in an accretion disk and Compton scattering off a leptonic
289: jet aligned along the disk axis can account the emission line
290: in PKS 2149-306. However, this model can not produce the large
291: equivalent width we observed. Thus Fe-K line originating from
292: the outflow itself is required, and, to produce the line, the
293: outflow can not be fully ionized. Even then, if the emission
294: line is produced by fluorescence, the continuum photons
295: cannot illuminate the outflowing material from behind
296: since the Doppler redshift of the continuum would make it even harder
297: than it already is to
298: attain the large equivalent width. For the same reason,
299: there must be preferential blueshifting of the line photons
300: over the continuum photons. The continuum photons must
301: illuminate the outflow from the sides.
302: So the outflow must locate very close to the X-ray continuum
303: emission region, and have comparable size to ensure the outflow
304: has enough open angle to the X-ray continuum emission to produce
305: the large equivalent width.
306: Under these conditions
307: it is possible to boost the rest-frame equivalent width
308: by the Doppler factor raised to the power of $3+\Gamma$
309: ($\Gamma \sim 2.2$ is the X-ray photon index), or $\sim 2.5^{5.2}
310: \sim 117$.
311: Thus, an emission line which would ordinarily have an
312: equivalent width of 0.1 keV could be boosted to have
313: a rest-frame equivalent width of $\sim 12$ keV.
314: Of course, the geometry is highly constrained.
315: We note that Fe-K lines have now been found in several
316: gamma-ray burst afterglows (e.g. Piro \etal 1999, 2000; Yoshida \etal 1999;
317: Antonelli \etal 2000) and some models
318: to account for these lines invoke super-solar Fe
319: abundances of as much as a factor $\sim 60$. If there is
320: any association at all with past gamma-ray bursts, or
321: if such super-solar abundances occur under other
322: circumstances, then the large equivalent width of the
323: emission line
324: would obviously also be much
325: easier to explain (although an outflow is still required).
326:
327: Similar blueshifted and strong iron emission line has not
328: been detected before in other AGNs. However, it's interesting to
329: note that Migliari et al. (2002) reported the discovery of
330: blueshifted and very strong iron emission lines (EW = 13 keV,
331: Migliari, private communication) from extended X-ray emission
332: in x-ray binary system SS 433. The X-ray jets are required to
333: reheat itself to produce the X-ray continuum and line emission.
334: We argue that if the possible outflow of \object\ discussed above was also
335: heated by itself or other unknown mechanisms and ionized, it could
336: produce iron line itself without being illuminated by the X-ray continuum.
337: Then the special geometry requirement can be released.
338: It's interesting to note that VLA observation of the source resolved
339: an extended radio emission 6\arcsec\ south to the core
340: (Kellermann, private communication), presenting a possible radio jet.
341: Based on above discussions,
342: we conclude the most likely outflow is an inner X-ray jet. It's highly
343: valuable to study the source in more details in near future, including obtaining optical
344: spectroscopy with high quality to check if there are blueshifted optical
345: emission lines, obtaining X-ray image with better spatial resolution to
346: check if there is any extended X-ray emission, etc.
347:
348: An alternative explanation is that some fraction of the X-ray
349: continuum of \object\ is heavily absorbed by a cold outflow, and
350: the observed big bump is due to the Fe absorption edge.
351: The spectral fitting results indicate we need a cold outflow with
352: bulk velocities of 0.74c and column density of
353: N$_H$ = 1.6 $\times$ 10$^{24}$ cm$^{-2}$, covering 98\%
354: of the X-ray continuum ($\Delta$ C = - 17.9).
355: Such a cold outflow is unusual either.
356:
357: Is it possible that there happens to be a low-redshift ($z \sim 0.034$)
358: type~2 AGN so close to \object\ that we cannot detect it individually?
359: Fe K lines at 6.4 keV with large equivalent widths ($>$ 1 keV) are not
360: difficult to produce theoretically by Compton-thick type~2
361: AGNs (e.g. Ghisellini \etal 1994), and already confirmed by X-ray
362: observations (e.g. Levenson \etal 2002).
363: NGC 6240, a standard Compton-thick type~2 AGN, has an Fe K line intensity
364: of $\sim$ 3.2
365: $\times$ 10$^{-5}$ photons cm$^{-2}$ s$^{-1}$ (Ikebe et al. 2000),
366: and a R band nuclei magnitude of 16.1 (Fried \& Schulz 1983).
367: Assuming a spectral energy distribution of NGC 6240, the iron line
368: flux we observed leads to a type~2 AGN with R = 22.5, which is even
369: brighter than \object\ (R=22.9). However, we didn't detect any evidence of
370: its existence from either the deep $R$ image or the optical spectroscopy.
371: Even if the source was much more heavily obscured in optical than NGC 6240,
372: it is also puzzling that we would have missed its host galaxy (which has to
373: have a low redshift, z $\sim$ 0.034) from our deep $R$ image.
374: Thus it is very unlikely that the emission line feature at 6.2 keV is
375: produced by another low redshift type~2 AGN. However, it
376: does not hurt to make a wild guess: is it possible that there is
377: a class of source consisting of an obscured Compton-thick
378: active nucleus whose host galaxy is also obscured so that the entire object
379: is optically invisible?
380: If this kind of source really exists, there might be more among the X-ray
381: sources with no optical counterparts in the \chandra deep fields.
382:
383:
384: In principle there is a simple way to distinguish the blueshifted Fe
385: K line model from the outflowing absorption model or the type~1 plus
386: type~2 model.
387: That is, the latter two models predict that the X-ray emission above
388: 10 keV should be dominated by the outflow obscured component or the type~2
389: AGN since the effect of the obscuring matter is much less than it is
390: below 10 keV.
391: In fact using the latter two model we predict that the 10--20 keV flux
392: should be $\sim 13$ times larger compared to the blueshifted Fe K line
393: model.
394: The planned hard X-ray imaging mission,
395: InFoc$\mu$s would be able to perform the simple test outlined here.
396:
397: \acknowledgments
398: We would like to thank T. Heckman, J. Krolik, C. Reynolds, Y. Lu for discussions. We also thank the referee for a helpful report, especially on the use of C statistics which significantly improved the presentation of this letter.
399: \clearpage
400: \begin{deluxetable}{lccc}
401: \tablecaption{Spectral fits to \object}
402: \tablecolumns{4}
403: \tablewidth{0pt}
404: \tablehead
405: {
406: \colhead{Parameter} & \colhead{C} & \colhead{C} &\colhead{$\chi^2$}\\
407: & \colhead{unbinned} & \colhead{rebinned} & \colhead{rebinned}
408: }
409:
410: \startdata
411:
412: $\Gamma$ & 2.2$^{+0.2}_{-0.1}$ & 2.3$^{+0.3}_{-0.3}$ & 2.3$^{+0.3}_{-0.3}$ \\
413: $N_H$(10$^{22}$cm$^{-2}$) & 1.4$^{+0.4}_{-0.3}$ & 1.6$^{+1.0}_{-0.8}$ & 1.7$^{+1.2}_{-1.0}$ \\
414: $E_c$ (keV) & 6.2$^{+0.2}_{-0.1}$ & 6.2$^{+0.3}_{-0.2}$ & 6.2$^{+0.3}_{-0.2}$ \\
415: $\sigma$ (keV) & 0.2$^{+0.2}_{-0.1}$ & 0.2$^{+0.3}_{-0.2}$ & 0.2$^{+0.4}_{-0.2}$ \\
416: EW (keV) & 4.4$^{+3.2}_{-1.4}$ & 4.9$^{+3.5}_{-2.7}$ & 5.3$^{+3.5}_{-3.0}$ \\
417: $I$ & $1.3^{+1.0}_{-0.4}$ & $1.2^{+1.0}_{-0.6}$ & $1.3^{+0.8}_{-0.7}$ \\
418: C($\chi^2$)/dof & 548.8/575 & 48.8/49 & 46.7/49 \\
419: $\Delta$C ($\Delta \chi^2$) & -22.6 & -16.1 & -11.4
420: \tablecomments{
421: Spectral fits with an absorbed power law plus Gaussian line model.
422: The line intensity $I$ is in units of $10^{-7} \ \rm photons \ cm^{-2} \ s^{-1}$,
423: and the line parameters are given in the observed frame.
424: $\Delta$C ($\Delta \chi^2$) gives the improvements of the fit
425: by adding the Gaussian line to the continuum model.
426: }
427: \enddata
428: \end{deluxetable}
429:
430: \begin{references}
431: \reference{}Alexander, D. M., et al. 2002, Astron. Nachrichten, in press, astro-ph0210308
432: \reference{}Antonelli, L. A., et al. 2000, ApJ, 545, L39
433: \reference{}Brandt, W.N., et al. 2001, AJ, 122, 2810
434: \reference{}Cash, W. 1979, ApJ, 228, 939
435: \reference{}Dickey, L. M., \& Lockman, F. J. 1990, ARAA. 28, 215.
436: \reference{}Fried, J. W., \& Schulz, H. 1983, A\&A, 118, 166
437: \reference{}Ghisellini, G., Haardt, F., \& Matt, G., 1994, MNRAS, 267, 743
438: \reference{}Giacconi, R., Gurksy, H., Paolini, F.R., \& Rossi, B.B., 1962,
439: Phys. Rev. Letters, 9, 439
440: \reference{}Giacconi, R., et al. 2002, ApJS, 139, 369
441: \reference{}Ikebe, Y., et al. 2000, MNRAS, 316, 433
442: \reference{}Levenson, N. A., Krolik, J. H., \.{Zycki}, P. T.,
443: Heckman, T, M., Weaver, K. A., \& Awaki, H. 2002, ApJ, 573, L81
444: \reference{}Migliari, S., Fender, R., \& M\'{e}ndez, M. 2002, Science, 297, 1673
445: \reference{}Nousek, J. A., \& Shue, D. R. 1989, ApJ, 342, 1207
446: \reference{} Piro, L., et al. 1999, ApJ, 514, L73
447: \reference{} Piro, L., et al. 2000, Science, 290, 955
448: \reference{}Puchnarewicz, E. M. et al. 1996, MNRAS, 281, 1243
449: \reference{}Rosati, P., et al. 2002, ApJ, 566, 667
450: \reference{}Yaqoob, T., George, I.M., Turner, T.J., Nandra, K., Ptak, A., \&
451: Serlemitsos, P.J. 1998, 505, L87
452: \reference{}Yaqoob, T., George, I.M., Nandra, K., Turner, T.J., Zobair, S., \&
453: Serlemitsos, P.J. 1999, ApJ, 525, L9
454: \reference{}Yoshida, A., et al. 1999, A\&AS, 138, 433
455: \end{references}
456:
457: \newpage
458: \begin{figure}
459: \plotone{f1.eps}
460: \caption{
461: Low resolution optical spectroscopy of \object.
462: R band cut-out is
463: also shown in the figure. The iso-intensity X-ray contours are at
464: 2, 5, 10 and 20 sigma levels above the background. The size of the cutout
465: is 20$\arcsec$. The small circle in the cutout shows the 3$\sigma$
466: positional error of the X-ray source, and the optical counterpart was
467: marked by a box.
468: }
469: \end{figure}
470:
471:
472: \begin{figure}
473: \plotone{f2.eps}
474: \caption{
475: 0.5 -- 7.0 keV and 5.7 -- 6.7 keV band count rates (source plus background)
476: from 11 $Chandra$ exposures
477: in time sequence. Note that in order to show them in one figure, the X-axis
478: does not have a linear scale.
479: 1$\sigma$ errorbars are shown in solid lines, and 3$\sigma$ errorbars shown in dotted lines.
480: Dashed lines are at the average value of the count rates.
481: }
482: \end{figure}
483:
484: \begin{figure}
485: \plotone{f3.eps}
486: \caption{
487: The summed (source plus background) X-ray spectrum of \object\
488: and the expected background (points with dotted error bars).
489: The spectra were rebinned only for display purpose.
490: }
491: \end{figure}
492:
493: \begin{figure}
494: \plotone{f4.eps}
495: \caption{
496: Ratios of \chandra ACIS spectral data to best-fitting power-law
497: plus neutral absorber models for upper panel: \object, and two nearby sources
498: with similar X-ray counts, middle panel: \objecta, and lower panel: \objectb.
499: The spectra were rebinned for display purpose.
500: }
501: \end{figure}
502:
503: \end{document}
504: