1: %%
2: %\documentclass[12pt,preprint]{aastex}
3:
4: %% manuscript produces a one-column, double-spaced document:
5:
6: %\documentclass[manuscript]{aastex}
7:
8: \documentclass[preprint2, 10pt]{aastex}
9: %%\documentclass[preprint, 10pt]{aastex}
10: %% preprint2 produces a double-column, single-spaced document:
11: %\usepackage{psfig}
12: %\usepackage{graphics}
13: %\usepackage{multicol}
14: %% Sometimes a paper's abstract is too long to fit on the
15: %% title page in preprint2 mode. When that is the case,
16: %% use the longabstract style option.
17:
18: %% \documentclass[preprint2,longabstract]{aastex}
19:
20:
21: \newcommand{\vdag}{(v)^\dagger}
22: \newcommand{\myemail}{ccarilli@nrao.edu}
23:
24: %% You can insert a short comment on the title page using the command below.
25:
26: \slugcomment{Resubmitted to the Astronomical Journal, March 9, 2007}
27:
28: \shorttitle{A search for HI 21cm absorption
29: toward the highest redshift radio loud objects}
30: \shortauthors{Carilli et al.}
31:
32: \begin{document}
33:
34: %% LaTeX will automatically break titles if they run longer than
35: %% one line. However, you may use \\ to force a line break if
36: %% you desire.
37:
38: \title{A search for HI 21cm absorption toward the highest
39: redshift ($z \sim 5.2$) radio loud objects}
40:
41: \author{C.L. Carilli}
42: \affil{National Radio Astronomy Observatory, Socorro, NM, USA, 87801}
43: \email{ccarilli@nrao.edu}
44:
45: \author{Ran Wang}
46: \affil{National Radio Astronomy Observatory, Socorro, NM, USA, 87801}
47:
48: \author{M.B. van Hoven}
49: \affil{Leiden Observatory, PO Box 9513, 2300 RA, Leiden, The Netherlands}
50:
51: \author{K. Dwarakanath}
52: \affil{Raman Research Institute, Bangalore 560080, India}
53:
54: \author{Jayaram N. Chengalur}
55: \affil{National Centre for Radio Astrophysics, Post Bag 3,
56: Ganeshkhind, Pune 411 007, India}
57:
58: \author{Stuart Wyithe}
59: \affil{School of Physics, University of Melbourne, Parkville,
60: Victoria, Australia}
61:
62: \begin{abstract}
63:
64: We have searched for HI 21cm absorption toward the two brightest radio
65: AGN at high redshift, J0924--2201 at $z = 5.20$, and J0913+5919 at
66: $z=5.11$, using the Giant Meter Wave Radio Telescope (GMRT). These
67: data set a 3$\sigma$ upper limit to absorption of $< 30\%$ at 40 km
68: s$^{-1}$ resolution for the 30 mJy source J0913+5919, and $< 3\%$ for
69: the 0.55 Jy source J0924--2201 at 20 km s$^{-1}$ resolution. For
70: J0924--2201, limits to broader lines at the few percent level are set
71: by residual spectral baseline structure. For J0924--2201 the column
72: density limit per 20 km s $^{-1}$ channel is: N(HI) $< 2.2\times
73: 10^{18} \rm T_s$ cm$^{-2}$ over a velocity range of $-700$ km s$^{-1}$
74: to $+1180$ km s$^{-1}$ centered on the galaxy redshift determined
75: through CO emission, assuming a covering factor of one. For
76: J0913+5919 the column density limit per 40 km s$^{-1}$ channel is:
77: N(HI) $< 2.2\times 10^{19} \rm T_s$ cm$^{-2}$ within $\pm 2400$ km
78: s$^{-1}$ of the optical redshift. These data rule out any cool, high
79: column density HI clouds within roughly $\pm 1000$ km s$^{-1}$ of the
80: galaxies, as are often seen in Compact Steep Spectrum radio AGN, or
81: clouds that might correspond to residual gas left over from cosmic
82: reionization.
83:
84: \end{abstract}
85:
86: \keywords{cosmology; radio - lines; galaxies - high redshift}
87:
88: \section{Introduction}
89:
90: Observations of HI 21cm absorption at cosmologically significant
91: redshifts provide important probes into a number of physically
92: interesting problems (Kanekar \& Briggs 2004). First, associated
93: systems probe the immediate environment of the AGN and the ISM in the
94: host galaxy. In particular, searches for HI 21cm absorption toward
95: Compact Symmetric Radio sources (CSOs) show that in the overwhelming
96: majority of sources ($\ge 80\%$) absorption is detected at optical
97: depth levels between 4$\%$ and 40$\%$, with line widths ranging from
98: about 50 to 500 km s$^{-1}$ (Peck et al. 2000). CSO's are thought to
99: be young radio jet sources ($< 10^7$ yrs), possibly confined by
100: a dense ISM in the host galaxy (Conway 2002). The high HI 21cm
101: detection rate is thought to be due to the fact that the radio
102: emission is confined to the inner regions of the galaxy (scales $\le
103: 10$ kpc), and hence HI 21cm absorption searches probe the dense
104: gaseous environs in the centers of the host galaxies.
105:
106: Second, studies of intervening HI 21cm absorption systems, in concert
107: with observations of damped Ly$\alpha$ absorption, have been used to
108: constrain the evolution of the excitation, or spin temperature of the
109: neutral gas. Results suggest an increasing fraction of warm neutral
110: gas with increasing redshift for damped Ly$\alpha$ systems (Kanekar \&
111: Chengalur 2003; Carilli et al. 1996). This increase might be due to a
112: change in the nature of the parent galaxy of the absorption line
113: system from large spirals at low $z$, to smaller dwarf and irregular
114: galaxies at higher $z$ (Kanekar \& Chengalur 2003), or it could be a
115: cosmological geometry effect, relating to the very slowly changing
116: value of the angular diameter distance for redshifts greater than
117: unity or so (Curran \& Webb 2006).
118:
119: And third are the recent observational constraints on the epoch of
120: reionization (EoR), suggesting cosmic reionization started around
121: $\sim 11$, with the last neutral structures being etched away around
122: $z\sim 6$ (Fan, Carilli, \& Keating 2006). It has been pointed out
123: that the existence of radio loud AGN within the near edge of cosmic
124: reionization could be used as sensitive probes of intermediate to
125: small scale structures in the neutral IGM, through HI 21cm absorption
126: observations (Carilli, Gnedin, \& Owen 2002; Furlanetto \& Loeb 2002),
127: as opposed to the very large scale that can be studied in emission.
128:
129: In this paper we present a search for HI 21cm absorption toward two $z
130: > 5$ radio sources. While not quite extending into the EoR, these
131: sources are the highest redshift bright radio AGN known, and the
132: redshifts are close enough to the end of reionization (within 200
133: Myr), that we can explore the possibility of enhanced residual cool
134: neutral gas content in the vicinity of the host galaxy due to
135: not-fully complete reionization of the densest regions of the universe
136: (Gnedin 2000; Paschos \& Norman 2005).
137:
138: \section{The sources}
139:
140: The source TN J0924--2201 is a radio galaxy with an optical redshift
141: of $z = 5.19 \pm 0.01$ (van Breugel et al. 1999). The source was
142: selected from low frequency radio surveys as an 'ultra steep spectrum'
143: source, with a flux density at 1.4 GHz of 73 mJy and a spectral index
144: between 0.37 and 1.4 GHz of --1.63. Near IR imaging has revealed a
145: faint, clumpy source, while radio imaging shows a compact double with
146: a separation $\sim 1''$ (= 7 kpc), making this source a CSO. The
147: Ly$\alpha$ emission from this galaxy is clearly weaker than is normal
148: for powerful radio galaxies. van Breugel et al. (1999) suggest that
149: the likely cause for the abnormally weak Ly$\alpha$ line is associated
150: Ly$\alpha$ absorption, thereby indicating the presence of neutral HI
151: on kpc-scales in the host galaxy. The complex near-IR structure, and
152: the associated Ly$\alpha$ absorption, make this source a prime
153: candidate for a forming giant elliptical galaxy within a cosmically
154: over-dense region (Pentericci et al. 2002). Venemans et al. (2004)
155: estimate a star formation rate in the host galaxy of $\sim 11.4
156: M_\odot$ yr$^{-1}$, based on the extended (rest-frame) UV continuum
157: flux as well as on the Ly$\alpha$ line flux. However, this star
158: formation rate has not been corrected for reddening, which may be
159: substantial. Most recently, Klamer et al. (2005) have detected CO
160: emission from this galaxy, indicating a large reservoir of molecular
161: gas ($\sim 10^{11}$ M$_\odot$),
162: and providing a very accurate host galaxy redshift of $z = 5.202 \pm
163: 0.001$.
164:
165: The source SDSS J0913+5919 is a luminous quasar at $z = 5.11 \pm 0.02$
166: (Anderson et al. 2002). Radio imaging at 1.4 GHz and 5 GHz reveals a
167: radio source with a flux density of 18 mJy at 1.4 GHz and a spectral
168: index of -0.7 (Petric et al. 2003). The radio source is compact, but
169: possibly marginally resolved at $1''$ resolution, suggesting this
170: source is a CSO. The optical spectrum of the source shows a relatively
171: narrow Ly$\alpha$ emission line for a quasar, again likely due to
172: strong associated absorption, and again indicating the presence of
173: neutral HI in the host galaxy (Anderson et al. 2001). The optical
174: luminosity of the QSO implies an Eddington limited mass to the black
175: hole $\ge 10^9$ M$_\odot$, which would imply a massive host galaxy
176: ($\ge 10^{11}$ M$_\odot$), based on the black hole mass -- bulge
177: velocity dispersion relation (Gebhardt et al. 2000).
178:
179: \section{Observations}
180:
181: Observations were made in the 230 MHz band of the GMRT. Absolute flux
182: density calibration was done using 3C48 and 3C286. The source
183: J0834+555 was observed for phase calibration for J0913+555, and
184: J0837-198 was used for phase calibration for J0924--2201. All data
185: were acquired using the full array of 30 antennas (projected baselines
186: up to 18 k$\lambda$), giving a synthesized beam $\sim 20"$ FWHM. The
187: standard line correlator set-up was employed, using 128 spectral
188: channels.
189:
190: J0913+5919 has an optical redshift of $5.11 \pm 0.02$, which puts the
191: HI 21cm line at a frequency of $232.5 \pm 0.8$ MHz. For this source we
192: observed two polarizations of 4 MHz bandwidth each, centred on the
193: expected frequency of the redshifted 21cm line. The 128 spectral
194: channels then yields a spectral resolution of 31.25 kHz, or 40.6 km
195: s$^{-1}$. 3C286 was used for bandpass calibration.
196:
197: For J0924--2201 the host galaxy redshift is $z = 5.202 \pm 0.001$,
198: placing the 21cm line at a frequency of $229.0 \pm 0.1$ MHz. An
199: initial short (2 hour) test observation of this source was made in
200: September, 2003, to investigate the radio frequency interference (RFI)
201: environment in the 229 MHz band of the GMRT. Although there was strong
202: interference over much of the 4 MHz band, by careful flagging we were
203: able to generate a spectrum of the spectral region immediately
204: defined by the CO redshift, and interestingly, a potential (3$\sigma$
205: over a few 40 km s$^{-1}$ channels) absorption line was seen.
206:
207: We reobserved this source in February 2004, with a longer integration
208: time (three observations of 7 hours each), and a narrower bandwidth
209: (2MHz) to verify this line. Unfortunately, the interference
210: environment for this second observation was much worse than during the
211: test observations, with only 1 or 2 late-night/early morning hours per
212: observation being useful. Moreover, the observations of the bandpass
213: calibrators 3C48 and 3C286 were completely wiped-out, and the phase
214: calibrator 0837--198 had to be used for bandpass and gain calibration,
215: assuming the flux density derived from the test observations. These
216: latter (admittedly marginalized) data did not verify the line seen in
217: the test data.
218:
219: Given the problems with the second data set, we observed J0924--2201
220: a third time in November 2005, separating the observations over three
221: nights, but observing only during the two hours per night
222: corresponding to the lowest RFI levels, and limiting the velocity
223: range to just the 1MHz band centered on the CO line redshift in order
224: to avoid strong RFI at the edge of the wider band.
225:
226: All data processing was performed using the wide-field imaging and
227: self-calibration capabilities in AIPS.
228:
229: \section{Data analysis}
230:
231: For the February 2004 observations of J0924--2201, all baselines
232: shorter than 2 k$\lambda$ ($\sim 50\%$ of all baselines) were
233: completely corrupted by wideband RFI, and were therefore flagged.
234: Also, all data were flagged except for roughly two hours in the middle
235: of each the night. For the November 2005 observations all baselines
236: shorter than 1 k$\lambda$ were flagged. Flagging of residual
237: channel-by-channel interference removed $\sim 30\%$ of the
238: channels of the remaining data.
239:
240: For both sources wide field images were then generated using standard
241: wide-field imaging routines in AIPS, and the data were self-calibrated
242: using the resulting models. For the first (phase only) self
243: calibration iteration a model based on the NRAO VLA Sky Survey
244: was used (Condon et al. 1998).
245:
246: A general issue in low frequency spectral line imaging is side-lobe
247: confusion from bright sources in the primary beam. In our case this is
248: especially a problem for J0913+5919, where side lobe confusion is
249: caused by two relatively bright sources ($\sim 2.4$ Jy and $\sim 1$
250: Jy) well within the primary beam. In order to overcome this
251: difficulty, proper modeling and subtraction of continuum sources is
252: required. We used a combination of bright source model subtraction
253: from the spectral line data using the AIPS task 'UVSUB', plus spectral
254: baseline fitting in the uv-plane ('UVLIN') and in the image-plane
255: ('IMLIN').
256:
257: For the February 2004 observations of J0924--2201, the bandpass
258: calibration using only the phase calibrator resulted in significant
259: residual structure across the band. This was removed using a second
260: order polynomial fit in IMLIN, leaving a $\sim 2\%$ residual bandpass
261: ripple across the spectrum. For November 2005 we found that bandpass
262: calibration using 3C286 was adequate to allow for a simple linear
263: bandpass fit in the image-plane using IMLIN.
264:
265: In the case of J0913+5919, RFI on short baselines was
266: limited to specific channels, the data could therefore be edited quite
267: efficiently using the AIPS tasks 'SPFLG', 'UVFLG' and 'FLGIT',
268: ultimately resulting in the removal of about 35$\%$ of all data.
269: A linear baseline was adequate for residual continuum subtraction
270: with IMLIN.
271:
272: \section{Results}
273:
274: Figure 1 shows the continuum image of each source. Both sources are
275: unresolved at the resolution of the GMRT ($\sim 20''$). J0913+5919 was
276: found to have a flux density of $30 \pm 3$ mJy at 230MHz, while
277: J0924--2201 had a flux density at of $0.55\pm0.05$ Jy. In both cases
278: the errors represent observation-to-observation differences, likely
279: due to residual errors in the flux density boot-strap process.
280:
281: The continuum subtracted spectrum of J0913+5919 is shown in Figure 2a.
282: The rms noise is 3 mJy per 40.7 km s$^{-1}$ channel. At this
283: resolution, we find no absorption to a 3$\sigma$ optical depth limit
284: of 0.3, over a velocity range of $\pm 2400$ km s$^{-1}$ centered on the
285: optical redshift of $z = 5.11 \pm 0.02$.
286:
287: There is a broad depression in the J0913+5919 spectrum, covering much
288: of the velocity range dictated by the optical redshift plus
289: uncertainty. We have smoothed the spectrum to 164 km s$^{-1}$, and the
290: result is shown in Figure 2b. The rms per channel is 2 mJy, and the
291: depression can be seen centered at about 232.3 MHz, or $z = 5.114$,
292: with a width of about 9 channels, or 1500 km s$^{-1}$. The mean depth
293: of the depression is about 2 mJy, or just 1 sigma per channel,
294: although this persists over 9 channels. Note that we have used all of
295: the channels shown for linear spectral baseline subtraction. If we
296: exclude the 9 'line' channels from the spectral baseline fit, the mean
297: depression depth increases to almost 3mJy. We do not consider
298: this a line detection, but a potential result that requires further
299: observation.
300:
301: Figure 3 shows the results for J0924--2201. Figure 3a shows the
302: wide-band spectrum from February 2004, for which a second order
303: spectral baseline has been removed using all of the channels shown.
304: We do not detect any absorption over the full band to an rms level of
305: 11 mJy per 20 km s$^{-1}$ channel, corresponding to a 3$\sigma$
306: optical depth limit of 6\%. The frequency range covered corresponds to
307: $-700$ km s$^{-1}$ to $+1180$ km s$^{-1}$, centered on the CO redshift
308: for the host galaxy. Figure 3b shows the spectrum from November 2005
309: at 20 km s$^{-1}$ resolution. Again, no absorption is detected to an
310: rms level of 6 mJy per 20 km s$^{-1}$ channel, corresponding to a
311: 3$\sigma$ optical depth limit of 3.3\%. The frequency range covered
312: corresponds to $\pm 460$ km s$^{-1}$. We have combined the two
313: spectra, weighting by the rms, and the result is show in Figure
314: 3c. The rms per 20 km s$^{-1}$ channel is now 5 mJy, implying a
315: 3$\sigma$ optical depth limit of 2.7\%. Residual spectral baseline
316: structure limits any broad ($\sim 1000$ km s$^{-1}$) lines to $< 2\%$.
317:
318: \section{Discussion}
319:
320: These GMRT observations set $3\sigma$ upper limits to the HI optical
321: depth toward J0924--2201 of 3\% at 20 km s$^{-1}$ resolution within
322: $\pm$460 km s$^{-1}$ of the host galaxy redshift, and 6\% over a wider
323: velocity range of $-700$ km s$^{-1}$ to $+1180$ km s$^{-1}$. The
324: column density limit per channel is then: N(HI) $< 1.1\times 10^{18}
325: \rm T_s$ cm$^{-2}$ over the narrow range, and $< 2.2\times 10^{18} \rm
326: T_s$ cm$^{-2}$ over the broader range, where $\rm T_s$ is the HI spin
327: temperature. Residual broad spectral baseline structure limits any
328: broad absorption line to $< 2\%$. The column density limit is then $<
329: 1.8\times 10^{19} \rm T_s$ cm$^{-2}$ for a 500 km s$^{-1}$ line.
330:
331: For the weaker radio source, J0913+5919, the optical depth limit is
332: 0.3 at 40 km s$^{-1}$ resolution, within $\pm 2400$ km s$^{-1}$ of the
333: optical redshift. The column density limit per channel is then: N(HI)
334: $< 2.2\times 10^{19} \rm T_s$ cm$^{-2}$. There is a broad, weak
335: depression in the center of the spectrum of about 1500 km s$^{-1}$
336: width with a depth of about 7\%, although this signal is really only
337: 1$\sigma$ in 9 channels, or roughly 3$\sigma$ averaged over all of the
338: 'line' channels. If real, the implied column density is: N(HI) $\sim
339: 2\times 10^{20} \rm T_s$ cm$^{-2}$. We consider this an upper limit
340: to a broad line in this source, although further observations of this
341: source would be very interesting.
342:
343: We note that both sources in our study show evidence for an extended
344: radio continuum source of $\sim 1"$ in size ($\sim 6$ kpc; see section
345: 2). Low HI covering factors of the continuum sources could also lead
346: to low apparent optical depths.
347:
348: For J0924--2201, assuming that the very weak Ly$\alpha$ emission line
349: from the host galaxy is due to high column density (damped) associated
350: Ly$\alpha$ absorption line system (N(HI)$ \sim 10^{21} $cm$^{-2}$),
351: and using the HI 21cm limits above, suggests that the HI spin
352: temperature is relatively high ($>$few hundred K). For comparison,
353: Galactic clouds with N(HI) $\ge 10^{21} $cm$^{-2}$ typically have spin
354: temperature values $\sim 100$ K for the absorbing gas (Dwarakanath,
355: Carilli \& Goss 2001). For damped Ly$\alpha$ systems at $z = 3$
356: typical values for $T_{spin}$ on the order of $10^{3}$K are found
357: (Carilli et al., 1996; Kanekar \& Chengalur 2003).
358:
359: These are (by far) the two brightest radio sources known at $z >
360: 5$. The redshifts of these two sources place them within 200 Myr of
361: the end of reionization ($z \sim 6$), as determined from eg. the
362: Gunn-Peterson effect toward high z QSOs (Fan et al. 2006). Further,
363: these luminous AGN are likely associated with denser regions of the
364: universe, ie. proto-clusters, at the earliest epochs (Venemans et
365: al. 2004). Given that reionization is likely a process extended in
366: space and time (Fan, Carilli, Keating 2006), two possibilities exist
367: for the large scale environments of the systems. There may be a
368: significant residual population of dense HI clouds associated with the
369: denser regions of the universe (Gnedin 2000; Paschos \& Norman
370: 2005). In this case, one might expect to see 21 cm absorption not just
371: by gas in the inner kpc's of the radio source host galaxy, but perhaps
372: by the densest residual HI clouds surrounding the host galaxy on
373: scales of tens to hundreds of kpc, left over from the EoR and still
374: being ionized by the increasing uv background radiation
375: field. Conversely, these densest regions may ionize earliest, due to
376: the biased formation of luminous structure, and the associated higher
377: ionizing radiation field (Wyithe \& Loeb 2006).
378:
379: Our observations have not detected any absorption by neutral systems
380: in the IGM surrounding the $z \sim 5.2$ quasars, implying that there
381: are no residual neutral hydrogen clouds along their lines of
382: sight. Indeed, we can certainly rule-out any large (kpc-scale), high
383: column density, cool HI clouds within 1000 to 2000 km s$^{-1}$ along
384: the lines of site to either of these these sources. This lack of
385: absorption is qualitatively consistent with expectations based both on
386: optical observations of high redshift ($z\sim 6$) quasars, and on
387: theoretical expectations from models of the reionization process. As
388: seen in the SDSS quasars at $z \sim 6$ (White et al. 2003), there
389: should be a proximity zone extending to $\sim$ 3000 km s$^{-1}$
390: blueward of the quasar redshift, within which the IGM is sufficiently
391: ionized that Ly-alpha photons are transmitted.
392:
393: On the theoretical side, we would expect the ionizing background
394: (following reionization) to be much larger than average within 1 MHz
395: of the quasar redshift. For example, the enhancement of the
396: ionizing backgound as a function of radius from a massive halo has
397: been computed by Dijkstra, Lidz, \& Wyithe (2006). When applied to a
398: quasar, they find that the typical enhancement predicted by this model
399: at a distance of 1Mpc ranges between factors of 10 and 100, depending
400: on the quasar host mass. Much of this enhancement is due to clustering
401: of sources around the quasar, which preferentially form in the
402: overdense infall region (Wyithe \& Loeb 2005). This large enhancement
403: in the ionizing background radiation would suppress any absorption
404: signature close to the quasar.
405:
406: Recent models describe the process of reionization as patchy, meaning
407: in part that overlap is completed at different times in different
408: regions of the IGM. However the time of reionization within a region
409: of IGM is not expected to be random. Rather, galaxy bias will ensure
410: that overdense regions are reionized earlier than the average IGM
411: (e.g. Wyithe \& Loeb 2006). Rare objects, such as the massive halos
412: that house quasars at $z\sim5.2$, formed preferentially in overdense
413: regions. It follows that the distribution of overdensities on 1Mpc
414: scales surrounding these quasars is significantly skewed towards
415: positive values (Wyithe \& Loeb 2006). We therefore expect
416: reionization to occur earlier around such objects. As an example,
417: using the model described in Wyithe \& Loeb (2006), we can calculate
418: the time at which the IGM within 1Mpc of a quasar is reionized
419: relative to the average IGM. Even in a case where the quasar flux is
420: ignored, such a region would only be reionized at a time later than
421: the average IGM in 5% of cases. As a result, these, and other, data,
422: and current models of reionization, suggest that the immediate infall
423: regions around bright quasars ($\sim$ a few Mpc) are not the best
424: sites to search for residual cool neutral gas clouds near the end of
425: cosmic reionization. Future searches for such primordial HI clouds
426: toward $z > 6$ radio sources are best done at redshifts a few thousand
427: km s$^{-1}$ below the AGN host galaxy redshift.
428:
429: Currently, the GMRT is the largest area telescope capable of observing
430: the HI 21cm line into cosmic reionization, ie. at frequencies close
431: to, or below, about 200 MHz. The interference environment clearly
432: makes such studies difficult, and any further progress in very high
433: redshift HI 21cm absorption searches using existing instrumentation will
434: require discovery of bright ($> 100$mJy) radio sources at redshifts
435: that happen to put the HI line in the ever decreasing RFI-free parts
436: of the low frequency spectrum. However, within a few years telescopes
437: such as the Mileura Widefield Array, the Low Frequency Array, the
438: Precision Array for Probing the Epoch of Reionization, and the 21
439: Centimeter Array, some of which will operate at sites chosen to
440: minimize terrestrial interference, should allow for the study of
441: fainter sources (few mJy), over wider frequency ranges (Carilli 2006).
442:
443: \acknowledgments We acknowledge support from the Max-Planck Society
444: and the Alexander von Humboldt Foundation through the Max-Planck
445: Forshungspreise 2005. We thank the referee for useful comments. The
446: National Radio Astronomy Observatory is a facility of the National
447: Science Foundation, operated by Associated Universities, Inc.
448:
449: \begin{thebibliography}{}
450:
451: \bibitem[]{424} Anderson, S., Fan, X., Richards, G., Schneider, D.,
452: Strauss, M. et al. 2001, AJ, 122, 503
453:
454: \bibitem[]{427} Carilli, C., Lane, W., de Bruyn, A.G., Braun, R., \&
455: Miley, G. 1996, AJ, 111, 1830
456:
457: \bibitem[]{430} Carilli, C., Gnedin, N., \& Owen F. 2002, ApJ, 577, 22
458:
459: \bibitem[]{} Carilli, C.L. 2006, New Astr. Rev. 50, 162
460:
461: \bibitem[]{432} Condon, J. J., Cotton, W. D., Greisen, E. W., Yin, Q. F.,
462: Perley, R. A., Taylor, G. B., \& Broderick, J. J. 1998, AJ, 115, 1693
463:
464: \bibitem[]{435} Conway, J. 2002, NewAR, 46, 263C
465:
466: \bibitem[]{} Curran, S.J. \& Webb, J.K. 2006, MNRAS, 371,356
467:
468: \bibitem[]{437} Dijkstra, M., Lidz, \& Wyithe, S. 2006, ApJ, in press
469:
470: \bibitem[]{439} Dwaraknath, K., Carilli, C., Goss, W.M. 2002, ApJ, 567, 940
471:
472: \bibitem[]{441} Fan, X., Strauss, M., Becker, R., White, R., Gunn, J. et al.
473: 2006, AJ, 132, 117.
474:
475: \bibitem[]{444} Fan, X., Carilli, C.L, Keating, B. 2005, ARAA, 44, 41
476:
477: \bibitem[]{446} Furlanetto, S. \& Loeb, A. 2002, ApJ, 579, 1
478:
479: \bibitem[]{448} Gebhardt, K., Bender, R., Bower, G., Dressler, A.,
480: Faber, S. et al. 2000, ApJ, 539, L13
481:
482: \bibitem[]{451} Gnedin, N.Y. 2000, ApJ, 535, 530
483:
484: \bibitem[]{453} Kanekar, N. \& Chengalur, J. 2003, A\&A, 399, 857
485:
486: \bibitem[]{455} Kanekar, N. \& Briggs, F. 2004, New AR, 48, 1259
487:
488: \bibitem[]{457} Klamer, I., Ekers, R., Sadler, E., Weiss, A.,
489: Hunstead, R., de Breuck, C. 2005, ApJ, 621, L1
490:
491: \bibitem[]{460} Lane, W. \& Briggs, F. 2002, ApJ, ApJ, 561, L27
492:
493: \bibitem[]{462} Paschos, P. \& Norman, M. 2005, ApJ, 631, 59
494:
495: \bibitem[]{464} Peck, A., Taylor, G., Fassnacht, C., Readhead, A.,
496: Vermeulen, R.. 2000, ApJ, 534, 104
497:
498: \bibitem[]{467} Pentericci, L., McCarthy, P. Rottgering, H.,
499: Miley, G., van Breugel, W. 2001, ApJS, 135, 638, L61
500:
501: \bibitem[]{470} Petric, A., Carilli, C., Bertoldi, F., Fan X.,
502: Strauss, M. et al. 2003, AJ, 126, 15
503:
504: \bibitem[]{473} van Breugel, W., De Breuck, C., Stanford, S.
505: Stern, D., Rottgering, H. et al. 1999, ApJ, 518, L61
506:
507: \bibitem[]{476} Venemans, B.P., Rottgering, H., Overzier, R.,
508: Miley, G., De Breuck, C. et al. 2004, A\&A, 424, L17
509:
510: \bibitem[]{479} White, R., Becker, R., Fan, X., Strauss, M.
511: 2003, AJ, 126, 1
512:
513: \bibitem[]{482} Wyithe, S. \& Loeb, A. 2006, ApJ, 646, 696
514:
515: \bibitem[]{484} Wyithe, S. \& Loeb, A. 2005, ApJ, 625, 1
516:
517:
518: \end{thebibliography}
519:
520: \clearpage
521: \newpage
522:
523: \begin{figure}[ht]
524: %\psfig{figure=F1a.ps,width=2.7in}
525: %\psfig{figure=F1b.ps,width=2.7in}
526: %\vskip -0.2in
527: \includegraphics[width=2.7in]{f1a.ps}\\
528: \includegraphics[width=2.7in]{f1b.ps}
529: \caption{ {\bf Upper}: The GMRT continuum image at 232 MHz
530: of J0913+5919 at $z = 5.11 \pm 0.02$. The source continuum flux density
531: at this frequency is $30\pm3$mJy. The
532: spatial resolution is $23"\times16''$ with major axis
533: position angle= 16$^o$. The contour levels are a geometric
534: progression in the square root of two starting at 4 mJy.
535: Negative contours are dashed.{\bf Lower}: The GMRT continuum image
536: at 229 MHz of J0924-2201 at $z = 5.202 \pm 0.001$. The source continuum
537: flux density at this frequency is $0.55\pm0.05$Jy. The
538: spatial resolution is $26"\times15''$ with major axis
539: position angle= 59$^o$. The contour levels are a geometric
540: progression in the square root of two starting at 30 mJy.
541: Negative contours are dashed.
542: }
543: \label{}
544: \end{figure}
545:
546: \begin{figure}[ht]
547: %\psfig{figure=F2a.ps,width=4in,angle=-90}
548: %\psfig{figure=F2b.ps,width=4in,angle=-90}
549: \includegraphics[angle=-90,width=4in]{f2a.ps}\\
550: \includegraphics[angle=-90,width=4in]{f2b.ps}
551: \caption{ {\bf Upper}: The continuum subtracted
552: GMRT spectrum of J0913+5919 at $z = 5.11 \pm 0.02$
553: at 40 km s$^{-1}$ resolution. The source continuum flux density
554: at this frequency is $30\pm3$mJy. The horizontal bar shows the
555: redshift range dictated by the uncertainty in the optical
556: redshift ($=2000$ km s$^{-1}$).
557: {\bf Lower}: The same spectrum, smoothed to 164 km s$^{-1}$ resolution.
558: }
559: \label{}
560: \end{figure}
561:
562: \clearpage
563: \newpage
564:
565: \begin{figure}[ht]
566: %\psfig{figure=F3a.ps,width=3.2in,angle=-90}
567: %\psfig{figure=F3b.ps,width=3.2in,angle=-90}
568: %\psfig{figure=F3c.ps,width=3.2in,angle=-90}
569: \includegraphics[angle=-90,width=3.2in]{f3a.ps}\\
570: \includegraphics[angle=-90,width=3.2in]{f3b.ps}\\
571: \includegraphics[angle=-90,width=3.2in]{f3c.ps}
572: %\vskip -0.2in
573: \caption{ {\bf Upper}: The continuum subtracted
574: GMRT spectrum of J0924--2201
575: at 20 km s$^{-1}$ resolution for data taken in February 2004.
576: The source continuum flux density at this frequency is $0.55\pm0.05$Jy.
577: The host galaxy redshift is $z = 5.202 \pm 0.001$, determined
578: from CO emission (Klamer et al. 2005). The horizontal bar shows the
579: redshift range dictated by the width of the CO emission line
580: (=250 km s$^{-1}$).
581: {\bf Middle}: The same as above, but for data taken in November 2005 over
582: a narrower bandwidth.
583: {\bf Lower}: The combined spectrum from November 2005 and February 2004.
584: }
585: \label{}
586: \end{figure}
587:
588: \end{document}
589:
590:
591:
592:
593: