astro-ph0107392/ms.tex
1: 
2: % ------------------------------------------------------------------
3: 
4: \def\asca{{\it ASCA\/}}
5: \def\chandra{{\it Chandra\/}}
6: \def\conx{{\it Constellation-X\/}}
7: \def\einstein{{\it Einstein\/}}
8: \def\hst{{\it {\it HST}\/}}
9: \def\iras{{\it IRAS\/}}
10: \def\rosat{{\it ROSAT\/}}
11: \def\xeus{{\it XEUS\/}}
12: \def\xmm{{\it XMM-Newton\/}}
13: 
14: \def\todo{{\Huge $\bullet$}}
15: 
16: \def\ltsima{$\; \buildrel < \over \sim \;$}
17: \def\simlt{\lower.5ex\hbox{\ltsima}}
18: \def\gtsima{$\; \buildrel > \over \sim \;$}
19: \def\simgt{\lower.5ex\hbox{\gtsima}}
20:         
21: % ------------------------------------------------------------------
22: 
23: % \documentclass{aastex}
24: \documentclass[preprint]{aastex}
25: 
26: \begin{document}
27: 
28: % ------------------------------------------------------------------
29: 
30: \title{The Chandra Deep Field North Survey. VII. X-ray Emission from Lyman Break Galaxies}
31: 
32: % ------------------------------------------------------------------
33: 
34: \author{W.N.~Brandt,$^1$ 
35: A.E.~Hornschemeier,$^1$ 
36: D.P.~Schneider,$^1$
37: D.M.~Alexander,$^1$ 
38: F.E.~Bauer,$^1$ 
39: G.P.~Garmire,$^1$ and  
40: C.~Vignali$^1$}
41: 
42: \footnotetext[1]{Department of Astronomy \& Astrophysics, 525 Davey Laboratory, 
43: The Pennsylvania State University, University Park, PA 16802}
44: 
45: % ------------------------------------------------------------------
46: 
47: \begin{abstract}
48: %
49: We present results from stacking analyses, using the 1~Ms \chandra\ Deep Field North
50: data, that constrain the X-ray emission of Lyman break galaxies at $z\approx$~2--4.  
51: Stacking the counts from 24 individually undetected 
52: Lyman break galaxies located within the Hubble Deep Field
53: North, we have obtained average detections of these objects in the resulting
54: 0.5--8.0~keV and 0.5--2.0~keV images; these images have effective exposure times of 
55: 22.4~Ms (260 days). Monte Carlo testing empirically shows
56: the detections to be highly significant. The average rest-frame 2--8~keV luminosity 
57: of a Lyman break galaxy is derived to be $\approx 3.2\times 10^{41}$~erg~s$^{-1}$,
58: comparable to that of the most X-ray luminous starbursts in the local Universe.
59: The observed ratio of X-ray to $B$-band luminosity is somewhat, but not
60: greatly, higher than that seen from local starbursts. The X-ray emission probably 
61: arises from a combination of high-mass X-ray binaries, ``super-Eddington''
62: X-ray sources, and low-luminosity active galactic nuclei. 
63: %
64: \end{abstract}
65: 
66: % ------------------------------------------------------------------
67: 
68: \keywords{
69: surveys~--
70: cosmology: observations~--
71: X-rays: galaxies~--
72: X-rays: general.}
73: 
74: % ------------------------------------------------------------------
75: 
76: \section{Introduction}
77: 
78: Over the last few years, the Lyman break technique has been used extensively to
79: isolate galaxies at $z\approx$~2--4 (e.g., Steidel et~al. 1996; 
80: Lowenthal et~al. 1997; Dickinson 1998; Steidel et~al. 1999) observed near 
81: the peak in the cosmic star-formation rate (e.g., Blain et~al. 1999). 
82: These objects often exhibit stellar and interstellar absorption lines
83: characteristic of local starburst galaxies. Their 
84: morphologies are varied, with multiple knots of emission
85: and diffuse wispy tails that suggest nonrelaxed systems. The Lyman break
86: technique has found particular application in the Hubble Deep Field North
87: (HDF-N; Williams et~al. 1996) due to the excellent imaging data 
88: and photometry available there, although the small angular size of 
89: the HDF-N limits somewhat its utility for statistical studies 
90: (e.g., Dickinson 1998).  
91: 
92: We have recently completed an $\approx 1$~Ms observation of the HDF-N and its
93: environs with the {\it Chandra X-ray Observatory\/} (Weisskopf et~al. 2000): 
94: the \chandra\ Deep Field North (\hbox{CDF-N}; Brandt et~al. 2001b, hereafter
95: Paper~V). In addition to probing luminous active galactic nuclei (AGN) 
96: throughout the Universe, this survey is also useful for studying
97: X-ray emission from more ``normal'' galaxies where the emission originates
98: from X-ray binaries, supernova remnants, and low-luminosity AGN (LLAGN; 
99: e.g., Hornschemeier et~al. 2001, Paper~II; Brandt et~al. 2001a, Paper~IV). 
100: Stacking techniques are particularly powerful in this regard, 
101: since they allow detection of X-ray emission from objects lying below the detection
102: limit for individual sources. In Paper~IV we used stacking techniques
103: to constrain the X-ray emission from normal spiral galaxies out to
104: $z\approx$~0.5--1, finding that their X-ray luminosities were not
105: more than a factor of $\approx 2$ larger (per unit $B$-band luminosity)
106: than those of spiral galaxies in the local Universe ($z<0.01$). 
107: Stacking analyses using the 480~ks of data available at the time 
108: allowed effective exposures of 5--8~Ms to be achieved. 
109: 
110: With the additional \chandra\ data recently obtained and improvements in 
111: our data processing techniques, we are now able to extend our
112: constraints on galaxy X-ray luminosities to higher redshift. This
113: is of cosmological interest since the X-ray luminosities of normal galaxies
114: are expected to evolve with redshift due to the changing cosmic
115: star-formation rate (e.g., van~Paradijs 1978; White \& Ghosh 1998;
116: Ghosh \& White 2001; Ptak et~al. 2001). Here we present constraints 
117: on the X-ray emission from the Lyman break galaxy population found 
118: in the HDF-N itself. 
119: 
120: The Galactic column density along this line of sight
121: is $(1.6\pm 0.4)\times 10^{20}$~cm$^{-2}$ (Stark et~al. 1992).
122: $H_0=65$~km~s$^{-1}$ Mpc$^{-1}$, 
123: $\Omega_{\rm M}=1/3$, and 
124: $\Omega_{\Lambda}=2/3$
125: are adopted throughout this paper. 
126: Coordinates throughout this paper are J2000.
127: 
128: % ------------------------------------------------------------------
129: 
130: \section{Data, Analysis, and Results}
131: 
132: We have used the Advanced CCD Imaging Spectrometer 
133: (ACIS; G.P. Garmire et~al., in preparation)
134: data sets and basic analysis methods described in 
135: Paper~V. We employ three standard X-ray bands: 0.5--8.0~keV (full band), 
136: 0.5--2.0~keV (soft band), and 2--8~keV (hard band). We use 
137: data screened with the restricted ACIS grade set defined in 
138: Table~2 of Paper~V, since this screening provides significantly
139: improved signal-to-noise for faint sources. We restrict our 
140: focus to the data obtained within the HDF-N itself; our X-ray
141: coverage is deepest in this area, and the \chandra\ background 
142: and point-spread function (PSF) are relatively uniform over the 
143: HDF-N. \chandra\ positions within the HDF-N are good to
144: within $0\farcs 6$. 
145: 
146: For the stacking analysis, we selected galaxies 
147: in the HDF-N itself using the spectroscopic redshift catalogs 
148: of Cohen et~al. (2000), Cohen (2001), 
149: Dawson et~al. (2001), and references therein. We considered 
150: the $z=$~2--4 galaxies spectroscopically identified in the 
151: HDF-N; there are 28 such objects, and most 
152: were found using the Lyman break technique (the \hst\ 
153: $U_{300}$ filter has allowed Lyman break galaxies to be 
154: found down to $z\approx 2$ in the HDF-N; e.g., Dickinson 1998).  
155: %
156: Prior to stacking, we verified that none of the objects being stacked
157: was individually detected by \chandra. This was accomplished by 
158: searching for nearby (within $2\arcsec$) X-ray sources found by 
159: {\sc wavdetect} with a false-positive probability threshold of 
160: $10^{-5}$ (Dobrzycki et~al. 1999; Freeman et~al. 2001); it
161: led to the removal of one source, the $z=3.479$ AGN
162: 123639.6+621230 (associated with 
163: CXOHDFN~J123639.5+621230). We have also manually 
164: inspected \chandra\ images in all three of the
165: standard X-ray bands and removed 
166: three Lyman break galaxies lying near unrelated X-ray sources.
167: 123645.3+621153 was removed because it lies within the region
168: covered by the faint extended source CXOHDFN~J123645.0+621142. 
169: 123645.8+621411 and 123654.7+621314 were 
170: removed because they lie relatively
171: near (6--8$\arcsec$) the bright X-ray sources 
172: CXOHDFN~J123646.3+621404 and CXOHDFN~J123655.4+621311, 
173: respectively, and could be affected by the low-level 
174: wings of the \chandra\ PSF. While we consider it
175: unlikely that these three Lyman break galaxies suffer from 
176: significant contamination by the unrelated X-ray sources, 
177: we wanted to be cautious and avoid any problems with the
178: background estimation. 
179: %
180: Figure~1 shows the redshifts and $R$ magnitudes of the 24 
181: selected objects. The median redshift of the selected objects 
182: is $z=2.92$, corresponding to a median lookback time of
183: 11.9~Gyr (84\% of the age of the Universe) and a median
184: luminosity distance of 25,800~Mpc (Hogg 1999). 
185: %
186: These galaxies have a range of morphologies but generally appear 
187: nonrelaxed. 
188: %
189: They do not show signs of containing AGN in their optical spectra,
190: although the spectral quality is often limited. They were not
191: identified as AGN candidates by Jarvis \& MacAlpine (1998), 
192: Conti et~al. (1999), or Sarajedini et~al. (2000). Two of
193: the 24 galaxies have weak (6--8~$\mu$Jy) radio emission at
194: 8.5~GHz (Richards et~al. 1998); this emission could be due
195: to either starburst activity or AGN. 
196: 
197: For each of the standard X-ray bands, we stacked ``cutout''
198: images centered on the positions of the 24 selected galaxies. 
199: For the photometry below we use an approximately circular
200: aperture comprised of the 30 pixels with 
201: centers within $1\farcs 5$ of the stacking
202: position (see \S3.4 of Paper~IV). In Figure~2 we show 
203: the number of full-band and soft-band counts 
204: obtained for each of the individual galaxies being stacked, 
205: and in Figure~3 we show images created from the stacked data. 
206: %
207: The stacked full-band image has an effective exposure time of 
208: 22.44~Ms (260~days). In this image, we find 164 counts 
209: within the 30-pixel aperture, while 134.3 are expected
210: from the background. The Poisson probability of obtaining 164 counts 
211: or more when 134.3 are expected is $5\times 10^{-3}$, indicating 
212: a detection at the 99.5\% confidence level. 
213: %
214: In the stacked soft-band image (22.46~Ms exposure), we find 43 counts
215: when 26.2 are expected from background, corresponding to a detection
216: at the 99.9\% confidence level. The higher significance of the soft-band
217: detection is most likely due to the lower background in this band. 
218: No significant detection is obtained in the hard band; unfortunately, 
219: the spectral constraint implied by the hard-band nondetection is
220: not physically interesting.  
221: 
222: Following Appendix~A of Paper~IV, we performed Monte Carlo stacking
223: simulations to assess false-detection probabilities empirically. 
224: For each band, we performed 100,000 trials where 
225: we stacked 24 random positions using the same photometry aperture as 
226: was used for the Lyman break galaxies above. The random positions 
227: were chosen to lie within $15\arcsec\times 15\arcsec$ 
228: ``local background regions'' centered on each of the Lyman break 
229: galaxies (avoiding known \chandra\ sources) to reproduce the actual 
230: background as closely as possible.
231: Figure~4 displays the number of trials giving a particular number of 
232: counts for the full and soft bands. The resulting distributions 
233: are very nearly Gaussian. 
234: %
235: In the full band 518 of the 100,000 stacking trials gave 164 counts
236: or more; the corresponding detection confidence level of 99.5\%
237: is in excellent agreement with that computed from Poisson statistics 
238: above. In the soft band 103 of the 100,000 stacking trials gave 43 counts
239: or more; again the corresponding detection confidence level of 99.9\%
240: agrees well with Poisson statistics. 
241: %
242: Again, no significant detection is obtained in the hard band. 
243: 
244: We performed several additional tests to verify the robustness
245: of our results. 
246: % 
247: If we include the three Lyman break galaxies excluded above
248: because they lie near to unrelated X-ray sources (123645.3+621153,
249: 123645.8+621411, and 123654.7+621314), the statistical significance
250: of our results is improved. 
251: %
252: We varied the size of the local background regions to examine if 
253: background gradients could be confusing the stacking analysis; our 
254: results remain highly significant for local background region sizes of  
255: $5\arcsec\times 5\arcsec$, 
256: $9\arcsec\times 9\arcsec$, and 
257: $21\arcsec\times 21\arcsec$. 
258: Because use of the local background regions limits the number of 
259: independent random positions that can be chosen in the Monte Carlo 
260: simulations, we also performed simulations with 
261: ``global background regions.'' For these 
262: simulations we drew the random positions from larger areas, 
263: not restricting the positions to lie close to the Lyman break galaxies. 
264: The statistical significance of our results remained the same or was
265: slightly improved. 
266: %
267: Finally, we varied the size of our photometry aperture, and our
268: results remain significant for reasonable aperture-size choices.
269: In fact, reducing the aperture size to include only the 14 pixels
270: with centers within $1\farcs 0$ of the stacking position improves
271: the statistical significance by a further factor of $\approx 5$
272: in both the full and soft bands. 
273: 
274: We note that our results cannot be explained 
275: by instrumental effects (e.g., subtle cosmic ray afterglows) 
276: since such effects will not be preferentially correlated with 
277: Lyman break galaxies. Moreover, unrelated galaxies at lower redshift 
278: that by chance lie within our stacking apertures cannot explain the 
279: observed signal since these galaxies will be found equally in both
280: the Lyman break galaxy stacking and the Monte Carlo simulations. 
281: 
282: % ------------------------------------------------------------------
283: 
284: \section{Discussion}
285: 
286: The stacking analyses above extend our ability to detect
287: relatively normal galaxies in the X-ray band to substantially
288: higher redshifts ($z=$~2--4) than 
289: was previously possible. At the median
290: redshift of our sample of $z=2.92$, the observed-frame full and
291: soft bands correspond to rest-frame bands of 2.0--31.4~keV and
292: 2.0--7.8~keV, respectively. Provided the basic X-ray production
293: mechanisms of Lyman break galaxies are similar to those of spiral
294: galaxies in the local Universe, the emission in these bands
295: should have an effective power-law photon index of $\Gamma\approx 2$
296: (e.g., Kim, Fabbiano, \& Trinchieri 1992; Ptak et~al. 1999); 
297: this is a typical photon index for both X-ray binaries and LLAGN. 
298: In the soft band, where we obtain the highest significance 
299: detection, the measured count rate 
300: ($7.5\times 10^{-7}$~count~s$^{-1}$) corresponds to an average
301: observed-frame flux of $4.0\times 10^{-18}$~erg~cm$^{-2}$~s$^{-1}$
302: and an average rest-frame 2.0--7.8~keV luminosity of 
303: $3.2\times 10^{41}$~erg~s$^{-1}$ (at the median 
304: redshift).\footnote{For comparison, our soft-band flux
305: limit for a single source is 
306: $\approx 3.0\times 10^{-17}$~erg~cm$^{-2}$~s$^{-1}$, 
307: corresponding to a rest-frame 2.0--7.8~keV luminosity 
308: of $\approx 2.4\times 10^{42}$~erg~s$^{-1}$.}
309: %
310: Note that these Lyman break galaxies are $\approx 30$ times more
311: luminous from 2--8~keV than the 11 normal spiral galaxies 
312: at $z\approx 0.5$ studied in \S4.2 of Paper~IV. 
313: % 
314: Since these Lyman break galaxies are thought to contain
315: significant starburst activity (see \S1), it is appropriate
316: to compare their X-ray luminosities to those of local starbursts. 
317: On average the 2--8~keV luminosities of these Lyman break galaxies 
318: are $\approx 5$ times larger than that of the starburst galaxy M82 
319: (e.g., Griffiths et~al. 2000; Kaaret et~al. 2001)
320: and are comparable to that
321: of NGC~3256, one of the most X-ray luminous starbursts in the 
322: local Universe (e.g., Moran, Lehnert, \& Helfand 1999; 
323: Lira et~al. 2001). The star-formation rates in these Lyman
324: break galaxies are uncertain but are thought to be often
325: $\simgt$~30--50~M$_\odot$~yr$^{-1}$, compared to 
326: $\approx 10$~M$_\odot$~yr$^{-1}$ for M82 and
327: $\approx 40$~M$_\odot$~yr$^{-1}$ for NGC~3256. 
328: 
329: The HDF-N Lyman break galaxies are optically 
330: luminous with $B$-band luminosities somewhat greater than 
331: present-day $L^\ast$; a present-day $L^\ast$ galaxy has a 
332: $B$-band luminosity density of 
333: $l_{\rm B}^\ast=1.0\times 10^{29}$~erg~s$^{-1}$~Hz$^{-1}$
334: (e.g., Blanton et~al. 2001). 
335: The 24 Lyman break galaxies used in the stacking 
336: analysis above have a mean rest-frame 
337: $l_{\rm B}=2.4\times 10^{29}$~erg~s$^{-1}$~Hz$^{-1}$ [the
338: median $l_{\rm B}=1.7\times 10^{29}$~erg~s$^{-1}$~Hz$^{-1}$, 
339: and the range is 
340: $l_{\rm B}=$~(0.4--6.6)$\times 10^{29}$~erg~s$^{-1}$~Hz$^{-1}$]; 
341: this is $\approx 1.3$ times larger than for the 11 normal spiral 
342: galaxies at $z\approx 0.5$ studied in \S4.2 of 
343: Paper~IV.\footnote{The rest-frame $l_{\rm B}$ values for the 
344: Lyman break galaxies have been calculated 
345: using the observed-frame $H$-band data in Table~2 of 
346: Papovich, Dickinson, \& Ferguson (2001) and the irregular
347: galaxy spectral energy distribution from 
348: Coleman, Wu, \& Weedman (1980). The good match between the
349: $H$-band data and rest-frame $B$ for these objects removes
350: nearly all of the dependence of $l_{\rm B}$ on the 
351: spectral energy distribution.} Therefore, the 
352: HDF-N Lyman break galaxies are 
353: $\approx (30/1.3)\approx 23$ 
354: times more X-ray luminous
355: per unit $B$-band luminosity than normal spiral galaxies at 
356: $z\approx 0.5$ (also see Shapley, Fabbiano, \& Eskridge 2001). 
357: An elevated ratio of X-ray to $B$-band luminosity is 
358: observed in many starburst galaxies at low redshift. 
359: The starburst galaxies in \S3 of David, Jones, \& Forman (1992) have 
360: X-ray to $B$-band luminosity ratios $\approx$~3--15 times higher 
361: than for normal galaxies (for example, this ratio is 
362: $\approx 13$ times higher for M82). Thus, the X-ray to $B$-band 
363: luminosity ratios for Lyman break galaxies are somewhat, but
364: not greatly, higher than those of local starbursts. If the
365: Lyman break galaxies suffered from more dust obscuration than local 
366: starbursts, this could explain their higher X-ray to $B$-band 
367: luminosity ratios; dust would attenuate $B$-band emission 
368: substantially more than 2--8~keV emission. 
369: 
370: Theoretical efforts to predict the X-ray evolution of galaxies
371: (e.g., van~Paradijs 1978; White \& Ghosh 1998; Ghosh \& White 2001; 
372: Ptak et~al. 2001) have typically focused 
373: on their low-mass X-ray binary (LMXB) populations which are
374: predicted to peak in luminosity at $z\approx 1.5$. Much less
375: work has been carried out, however, at the higher redshifts under study here. 
376: If the X-ray emission of Lyman break galaxies is indeed produced
377: via the same mechanisms as those operating in local starburst
378: galaxies, then high-mass X-ray binaries (HMXBs) and ``super-Eddington''
379: X-ray sources (e.g., Makishima et~al. 2000) are likely 
380: to produce much of the observed emission in the rest-frame
381: 2--8~keV band. Unlike LMXBs, which have an evolutionary 
382: timescale of $\sim 10^9$~yr, the X-ray emission from HMXBs
383: should track the cosmic star-formation rate closely (any LMXBs
384: contributing to the observed X-ray emission would have formed
385: at $z\sim 5$). Given the results from submillimeter and other surveys 
386: which are sensitive to star formation in dusty environments, cosmic 
387: star formation is likely to peak at roughly the redshifts under study 
388: here (e.g., Blain et~al. 1999). The cosmic evolution
389: of super-Eddington X-ray sources 
390: is impossible to predict reliably at present given that even
391: the basic nature of these objects is poorly understood at low
392: redshift. LLAGN with 2--8~keV luminosities
393: $\simlt 2.4\times 10^{42}$~erg~s$^{-1}$
394: could also contribute to the observed X-ray emission
395: from some of these Lyman break galaxies, although none of the
396: galaxies under study here has been found to contain an LLAGN (see \S2); 
397: again, even the most basic properties of LLAGN at these redshifts 
398: are poorly constrained by current data. 
399: 
400: Finally, these results show that the \chandra\ ACIS performs well at 
401: source detection even with effective exposure times of 260 days
402: (the current longest \chandra\ exposures are 11.6~days for the
403: CDF-N and the \chandra\ Deep Field South). Any systematic effects
404: that cause the sensitivity to deviate from that expected due
405: to photon statistics appear mild. Stacking analyses 
406: using deeper observations with \chandra\ should allow this 
407: work to be extended. Missions such as \xeus\ should ultimately 
408: allow the X-ray study of individual Lyman break galaxies. 
409: 
410: % ------------------------------------------------------------------
411: 
412: \acknowledgments
413: 
414: This work would not have been possible without the enormous efforts 
415: of the entire \chandra\ and ACIS teams. 
416: %
417: We thank 
418: O.~Almaini, 
419: A.C.~Fabian,
420: E.C.~Moran,  
421: A.~Ptak, 
422: A.E.~Shapley,
423: and an anonymous referee for helpful discussions. 
424: %
425: We gratefully acknowledge the financial support of
426: NASA grant NAS~8-38252 (GPG, PI),
427: NSF CAREER award AST-9983783 (WNB, DMA, FEB),  
428: NASA GSRP grant NGT~5-50247 and the Pennsylvania Space Grant Consortium (AEH), 
429: NSF grant AST-9900703~(DPS), and 
430: NASA LTSA grant NAG5-8107 (CV). 
431: 
432: % ------------------------------------------------------------------
433: 
434: \begin{references}
435: 
436: \reference{}
437: Blain, A.W., Smail, I., Ivison, R.J., \& Kneib, J.-P. 
438: 1999, MNRAS, 302, 632
439: 
440: \reference{}
441: Blanton, M.R., et~al.  
442: 2001, AJ, 121, 2358
443: 
444: \reference{}
445: Brandt, W.N., et~al. 
446: 2001a, AJ, 122, 1 (Paper~IV)
447: 
448: \reference{}
449: Brandt, W.N., et~al. 
450: 2001b, AJ, submitted (Paper~V)
451: 
452: \reference{}
453: Cohen, J.G., Hogg, D.W., Blandford, R., Cowie, L.L., Hu, E., Songaila, A., 
454: Shopbell, P., \& Richberg, K.  
455: 2000, ApJ, 538, 29  
456: 
457: \reference{}
458: Cohen, J.G. 
459: 2001, AJ, 121, 2895
460: 
461: \reference{}
462: Coleman, G.D., Wu, C.-C., \& Weedman, D.W. 
463: 1980, ApJS, 43, 393 
464: 
465: \reference{}
466: Conti, A., Kennefick, J.D., Martini, P., \& Osmer, P.S. 1999,
467: AJ, 117, 645
468: 
469: \reference{}
470: David, L.P., Jones, C., \& Forman, W. 
471: 1992, ApJ, 388, 82
472: 
473: \reference{}
474: Dawson, S., Stern, D., Bunker, A.J., Spinrad, H., \& Dey, A.
475: 2001, AJ, in press (astro-ph/0105043)
476: 
477: \reference{}
478: Dickinson, M. 1998, 
479: in The Hubble Deep Field, 
480: eds. Livio, M., Fall, S.M., \& Madau, P.  
481: (Cambridge University Press, Cambridge), p. 219
482: 
483: \reference{}
484: Dobrzycki, A., Ebeling, H., Glotfelty, K., Freeman, P., 
485: Damiani, F., Elvis, M., \& Calderwood, T. 
486: 1999, \chandra\ Detect 1.0 User Guide. 
487: \chandra\ X-ray Center, Cambridge
488: 
489: \reference{}
490: Ebeling, H., White, D.A., \& Rangarajan, F.V.N.
491: 2001,
492: MNRAS, submitted
493: 
494: \reference{}
495: Freeman, P.E., Kashyap, V., Rosner, R., \& Lamb, D.Q.
496: 2001, ApJ, submitted
497: 
498: \reference{}
499: Ghosh, P., \& White, N.E. 2001, ApJ, in press
500: 
501: \reference{}
502: Griffiths, R.G., et~al.
503: 2000,
504: Science, 290, 1325
505: 
506: \reference{}
507: Hogg, D.W. 1999 (astro-ph/9905116)
508: 
509: \reference{}
510: Hornschemeier, A.E., et~al. 
511: 2001, ApJ, 554, 742 (Paper~II)
512: 
513: \reference{}
514: Jarvis, R.M., \& MacAlpine, G.M. 1998, AJ, 116, 2624
515: 
516: \reference{}
517: Kaaret, P., Prestwich, A.H., Zezas, A., Murray, S.S., Kim, D.-W., 
518: Kilgard, R.E., Schlegel, E.M., \& Ward, M.J. 
519: 2001, MNRAS, 321, L29
520: 
521: \reference{}
522: Kim, D.-W., Fabbiano, G., \& Trinchieri, G. 1992, ApJ, 393, 134
523: 
524: \reference{}
525: Lira, P., Ward, M., Zezas, A., Alonso-Herrero, A., \& Ueno, S. 
526: 2001, MNRAS, submitted
527: 
528: \reference{}
529: Lowenthal, J.D., et~al. 1997, ApJ, 481, 673
530: 
531: \reference{}
532: Makishima, K., et~al. 2000, ApJ, 535, 632
533: 
534: \reference{}
535: Moran, E.C., Lehnert, M.D., \& Helfand, D.J.
536: 1999, ApJ, 526, 649
537: 
538: \reference{}
539: Papovich, C., Dickinson, M., \& Ferguson, H.C.
540: 2001, ApJ, in press (astro-ph/0105087)
541: 
542: \reference{}
543: Ptak, A., Serlemitsos, P., Yaqoob, T., \& Mushotzky, R.
544: 1999, ApJS, 120, 179
545: 
546: \reference{}
547: Ptak, A., Griffiths, R., White, N.E., \& Ghosh, P. 2001, ApJ, in press
548: 
549: \reference{}
550: Richards, E.A., Kellermann, K.I., Fomalont, E.B., Windhorst, R.A., \& Partridge, R.B.
551: 1998, AJ, 116, 1039
552: 
553: \reference{}
554: Sarajedini, V.L., Gilliland, R.L., \& Phillips, M.M.
555: 2000, AJ, 120, 2825
556: 
557: \reference{}
558: Shapley, A., Fabbiano, G., \& Eskridge, P.B. 2001, ApJ, in press (astro-ph/0106361)
559: 
560: \reference{}
561: Stark, A.A., Gammie, C.F., Wilson, R.W., Bally, J., 
562: Linke, R.A., Heiles, C., \& Hurwitz, M. 1992, ApJS, 79, 77 
563: 
564: \reference{}
565: Steidel, C.C., Giavalisco, M., Pettini, M., Dickinson, M., \& Adelberger, K.L.
566: 1996, ApJ, 462, L17
567: 
568: \reference{}
569: Steidel, C.C., Adelberger, K.L., Giavalisco, M., Dickinson, M., \& Pettini, M. 
570: 1999, ApJ, 519, 1
571: 
572: \reference{}
573: van~Paradijs, J.
574: 1978, ApJ, 226, 586
575: 
576: \reference{}
577: Weisskopf, M.C., Tananbaum, H.D., Van Speybroeck, L.P., \& O'Dell, S.L.
578: 2000, \procspie, 4012, 2 
579: 
580: \reference{}
581: White, N.E., \& Ghosh, P. 
582: 1998, ApJ, 504, L31
583: 
584: \reference{}
585: Williams, R.E., et~al. 1996, AJ, 112, 1335 
586: 
587: \end{references}
588: 
589: \clearpage
590: 
591: % ------------------------------------------------------------------
592: 
593: \begin{figure}
594: % \vspace{-0.5truein}
595: \epsscale{0.8}
596: \figurenum{1}
597: \plotone{fig1.ps}
598: \vspace{-0.1truein}
599: \caption{$R$ magnitude versus redshift for the HDF-N galaxies 
600: with spectroscopic redshifts in Cohen et~al. (2000), Cohen (2001), 
601: Dawson et~al. (2001), and references therein. The circled galaxies 
602: are those used in our stacking analysis. The four objects from 
603: $z=$~2--4 that are not circled are discussed in \S2.}
604: \end{figure}
605: 
606: % ------------------------------------------------------------------
607: 
608: \begin{figure}
609: % \vspace{-0.5truein}
610: \epsscale{0.8}
611: \figurenum{2}
612: \plotone{fig2.ps}
613: \vspace{-0.1truein}
614: \caption{Histograms showing the number of (a) full-band counts and (b) soft-band 
615: counts obtained in the 30-pixel aperture for each of the galaxies used in the
616: stacking analyses. The average number of counts expected from background is 
617: 5.6 for the full band and 1.1 for the soft band; these values are shown as the
618: vertical dotted lines.}
619: \end{figure}
620: 
621: % ------------------------------------------------------------------
622: 
623: \begin{figure}
624: % \vspace{-0.5truein}
625: \epsscale{1.0}
626: \figurenum{3}
627: \plottwo{fig3a.ps}{fig3b.ps}
628: \vspace{-0.1truein}
629: \caption{Stacked \chandra\ images of 24 Lyman break galaxies in the
630: (a) full band (22.44~Ms effective exposure) and (b) soft band
631: (22.46~Ms effective exposure). These images
632: have been made using the restricted ACIS grade set, and they have been
633: adaptively smoothed using the code of Ebeling, White, \& Rangarajan (2001). 
634: The images are $11\arcsec \times 11\arcsec$ in size, and each pixel is 
635: $0\farcs 5$ on a side. North is up, and East is to the left. The black
636: circles are centered on the stacking position and have a radius of 
637: $1\farcs 5$. The white contours are drawn at 85\%, 90\%, and 95\% of 
638: the observed maximum pixel intensity in each image.}
639: \end{figure}
640: 
641: % ------------------------------------------------------------------
642: 
643: \begin{figure}
644: % \vspace{-0.5truein}
645: \epsscale{0.7}
646: \figurenum{4}
647: \plotone{fig4.ps}
648: \vspace{-0.1truein}
649: \caption{Results from Monte Carlo testing of the Lyman break galaxy stacking
650: analyses for the (a) full band and (b) soft band. Each panel shows the results
651: from 100,000 stacking trials plotted as the number of trials yielding 
652: a given number of counts. The resulting distributions are very nearly Gaussian. 
653: The arrow in each panel indicates the number of counts actually observed when the 
654: Lyman break galaxies are stacked.}
655: \end{figure}
656: 
657: % ------------------------------------------------------------------
658: 
659: \end{document}
660: 
661: 
662: