1: \documentclass[useAMS,usenatbib,usegraphicx]{mn2e}
2:
3: \usepackage{times}
4: \usepackage{amssymb}
5: \usepackage{graphics,epsfig}
6:
7: % If your system does not have the AMS fonts version 2.0 installed, then
8: % remove the useAMS option.
9: %
10: % useAMS allows you to obtain upright Greek characters.
11: % e.g. \umu, \upi etc. See the section on "Upright Greek characters" in
12: % this guide for further information.
13: %
14: % If you are using AMS 2.0 fonts, bold math letters/symbols are available
15: % at a larger range of sizes for NFSS release 1 and 2 (using \boldmath or
16: % preferably \bmath).
17: %
18: % The usenatbib command allows the use of Patrick Daly's natbib.sty for
19: % cross-referencing.
20: %
21: % If you wish to typeset the paper in Times font (if you do not have the
22: % PostScript Type 1 Computer Modern fonts you will need to do this to get
23: % smoother fonts in a PDF file) then uncomment the next line
24:
25: %%%%% AUTHORS - PLACE YOUR OWN MACROS HERE %%%%%
26:
27: \def\arcs{\mbox{\ensuremath{^{\prime\prime}}}}
28: \def\farcs{\mbox{\ensuremath{.\!\!\arcs}}}
29: \def\lya{Ly$\alpha$}
30: \def\hi{H~{\sc i}}
31: \def\nhi{N(H\,{\sc i})}
32: %\def\nhi{\rm N_{HI}}
33: \def\sii{Si~{\sc ii}}
34: \def\mgii{Mg~{\sc ii}}
35: \def\feii{Fe~{\sc ii}}
36: \newcommand{\kms}{km~s$^{-1}$}
37: \newcommand{\cm}{cm$^{-2}$}
38: \newcommand{\Dla}{ Damped Lyman-$\alpha$ }
39: \newcommand{\dla}{ damped Lyman-$\alpha$ }
40: \newcommand{\noi}{\noindent}
41: \newcommand{\lb}{\left(}
42: \newcommand{\dV}{\Delta V}
43: \newcommand{\rb}{\right)}
44: \newcommand{\ts}{{T_s}}
45: \newcommand{\beq}{\begin{equation}}
46: \newcommand{\eeq}{\end{equation}}
47:
48:
49:
50: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
51:
52: \title[A low spin temperature DLA at $z = 2.289$]{Discovery of 21cm absorption in a $z_{\rm abs} =2.289$ DLA towards TXS~0311+430: The first low spin temperature absorber at $z>1$.}
53: \author[B. York et al.]{Brian A. York$^{1}$\thanks{E-mail: briany@uvic.ca (BY),
54: nkanekar@aoc.nrao.edu (NK), sarae@uvic.ca (SLE), pettini@ast.cam.ac.uk (MP)},
55: Nissim Kanekar$^{2}$\footnotemark[1],
56: Sara L. Ellison$^{1}$\footnotemark[1] and
57: Max Pettini$^{3}$\footnotemark[1]\\
58: $^{1}$Department of Physics and Astronomy, University of Victoria, 3800 Finnerty
59: Road, Victoria, BC V8P 1A1, Canada\\
60: $^{2}$National Radio Astronomy Observatory, 1003 Lopezville Road, Socorro, NM
61: 87801, USA\\
62: $^{3}$Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge
63: CB3 0HA}
64: \begin{document}
65:
66: \date{}
67:
68: \pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{2007}
69:
70: \maketitle
71:
72: \label{firstpage}
73:
74: \begin{abstract}
75: We report the detection of \hi~21~cm absorption from the $z=2.289$
76: damped Lyman-$\alpha$ system (DLA) towards TXS~0311+430, with the
77: Green Bank Telescope. The 21~cm absorption has a velocity spread
78: (between nulls) of $\sim 110$~km~s$^{-1}$ and an integrated optical
79: depth of $\int \tau {\rm d}V = (0.818 \pm 0.085)$~km~s$^{-1}$. We
80: also present new Giant Metrewave Radio Telescope 602~MHz imaging of
81: the radio continuum. TXS~0311+430 is unresolved at this frequency,
82: indicating that the covering factor of the DLA is likely to be high.
83: Combining the integrated optical depth with the DLA \hi~column density
84: of \nhi\ = $(2 \pm 0.5) \times 10^{20}$~\cm, yields a spin temperature
85: of $\ts = (138 \pm 36)$~K, assuming a covering factor of unity. This
86: is the first case of a low spin temperature ($< 350$~K) in a $z
87: > 1$ DLA and is among the lowest ever measured in any DLA. Indeed,
88: the $\ts$ measured for this DLA is similar to values measured in the
89: Milky Way and local disk galaxies. We also determine a lower
90: limit (Si/H)~$\gtrsim 1/3$ solar for the DLA metallicity, amongst the highest
91: abundances measured in DLAs at any redshift. Based on low redshift
92: correlations, the low $\ts$, large 21~cm absorption width
93: and high metallicity all suggest that the $z \sim 2.289$ DLA is likely
94: to arise in a massive, luminous disk galaxy.
95: \end{abstract}
96:
97: \begin{keywords}
98: galaxies: high-redshift -- galaxies: ISM -- radio lines: galaxies.
99: \end{keywords}
100:
101: \section{Introduction}
102: \label{sec:intro}
103:
104: Damped Lyman-$\alpha$ systems (DLAs) are the highest column density
105: absorbers seen along QSO lines-of-sight, with neutral hydrogen
106: column densities \nhi\ $\ge 2.0\times 10^{20}$~\cm. They have long been
107: identified as the precursors of today's galaxies and the primary gas
108: reservoir for star formation at high redshifts. Despite the
109: recognised importance of the absorbers, their typical size, structure
110: and internal physical conditions remain issues of controversy
111: (e.g. \citealt{wolfe05}). For example, DLA metallicities, now measured
112: in over 100 absorbers, show very little evolution between $z
113: \sim 3$ and $z \sim 0$, with low-metallicity ($-2<$~[Z/H]~$< -1$,
114: where Z~$\equiv$~Zn, S or Si\footnote{In the usual notation,
115: [Z/H]~$\equiv$~log(N(Z)/N(\hi))$-$log(N(Z)/N (\hi))$_{\odot}$. We use
116: solar values from \citet{lodders03}.}) absorbers the norm at all
117: redshifts (e.g. \citealt{kulkarni05}). This lack of metallicity
118: evolution runs contrary to expectations that the interstellar
119: metallicity should rise towards lower redshifts if DLAs trace the bulk
120: of gas in galaxies. Moreover, only a few tens of DLAs have their galactic
121: counterparts identified, of which only a small fraction have been
122: spectroscopically confirmed (e.g. \citealt{chen03}). Our understanding of
123: the basic physical
124: properties of the absorbing galaxies, such as their mass, size, star
125: formation rates and luminosity, remains very limited.
126:
127: \hi~21~cm absorption studies of DLAs towards radio-loud quasars
128: provide an independent probe of physical conditions in the
129: absorbers. They can be combined with optical measurements of the \hi\
130: column density of the DLA (from the Lyman-$\alpha$ line) to obtain the
131: column-density-weighted harmonic mean spin temperature $\ts$ of the
132: absorbing gas, allowing one to determine the temperature distribution
133: of the \hi\ along the line of sight. Measurements of $\ts$ in a large
134: number of DLAs may ultimately be used to infer other galactic properties
135: of high redshift systems. For example, in the local Universe, large spiral
136: disks like the Milky Way and M31 typically have low $\ts$ values ($\lesssim 300$~K;
137: \citealt{braun92}) while high $\ts$ values ($\sim 1000$~K; \citealt{young97})
138: are more common in dwarf galaxies. Tentative evidence for a similar trend has
139: been found in low~$z$ DLAs, out to $z \sim 0.7$ \citep{kanekar03}.
140: If such correlations hold for DLAs at all redshifts, measurements of $\ts$,
141: and its evolution, would provide interesting insights into galaxy evolution.
142:
143: Unfortunately, despite a number of searches over the past two and a
144: half decades (e.g. \citealt{briggs83,carilli96,chengalur00,kanekar03}), spin
145: temperature estimates are available for only $15$~DLAs (of which
146: $10$ are lower limits) at $z \gtrsim 1.7$ (Kanekar et al, in preparation).
147: All of these previous high $z$ measurements yield high spin temperatures
148: of $\ts \gg 300$ K. There are a number of reasons why the crop of $\ts$
149: estimates has increased slowly over the last 25 years, including observational
150: issues such as the frequency coverage of radio telecopes and widespread radio
151: frequency interference (RFI) at the low frequencies ($\lesssim 1$~GHz)
152: of the redshifted 21~cm line. However, an additional important reason
153: for the relatively-small present 21~cm absorption sample is simply the
154: dearth of known DLAs towards radio-loud QSOs suitable for 21~cm
155: absorption follow-up. We have hence been conducting an optical
156: survey of low-frequency-selected radio-loud quasars, specifically
157: designed to increase the number of $\ts$ estimates in the redshift
158: range $2<z<4$. While the survey is still in progress, we report here
159: its first results, the detection of 21~cm absorption from the $z \sim
160: 2.289$ DLA towards TXS~0311+430, the first case of a low spin temperature
161: in a high-$z$ DLA.
162:
163: \section{Observations and data analysis}
164:
165: \subsection{Optical observations}
166:
167: Optical observations of $\sim$ 50~QSOs selected from the Texas 365~MHz
168: survey \citep{douglas96} have been conducted at various facilities in order to
169: identify DLAs suitable for 21~cm follow-up. As part of this
170: campaign, we observed TXS~0311+430 (B=21.5, $z_{\rm em} = 2.87$) with the
171: Gemini Multi-Object Spectrograph (GMOS) on the Gemini-North
172: telescope. We obtained seven 2300-second and one 1700-second
173: long-slit spectra of TXS~0311+430, with a 1.0~arcsecond slit and the
174: B600\_G5303 disperser. The central wavelength was set to 4590\,\AA\ for four
175: of the exposures and to 4620\,\AA\ for the other four, so
176: as to achieve continuous wavelength coverage despite the
177: gap between CCD chips in the GMOS detector. The CCD
178: was binned $2 \times 2$. The final spectrum has a resolution
179: of 4.1\,\AA\ (full-width-at-half-maximum) and extends from $\sim 3550$ to
180: 6050\,\AA; this range allows the detection of DLAs in the redshift interval
181: $1.92 \leq z \leq 2.87$, with the upper limit set by the quasar
182: redshift.
183:
184: The GMOS data were reduced using standard IRAF routines including bias
185: subtraction, flatfield correction, extraction using APALL, wavelength
186: fitting [the root-mean-square (RMS) error on the wavelength fits was
187: $\leq 0.14$~\AA], and finally, vacuum and heliocentric velocity
188: wavelength corrections; the full procedure for this and the other
189: optical spectra of our survey will be described in York et al. (in
190: preparation). The final signal-to-noise ratio (S/N) ranged from $\sim
191: 8$ per pixel at 3600~\AA\ to $\sim 20$ per pixel at 6000~\AA. The GMOS
192: observations resulted in the detection of a DLA at $z \sim 2.289$ (see
193: Section~\ref{sec:lya_metals}).
194:
195:
196: \subsection{Radio observations}
197: \label{sec:radio1}
198:
199: An initial search for 21~cm absorption at the DLA redshift was carried out
200: with the PF1-450~MHz receiver of the Green Bank Telescope
201: (GBT; program AGBT-06B-042) on September~12, 2006. We used
202: the GBT Spectral Processor as the backend, with a bandwidth of
203: 1.25~MHz centred at 431.808~MHz, two circular polarizations, 1024
204: spectral channels and a spectral resolution of $\sim 0.85$~km/s
205: (before any smoothing). The data were taken in total power mode, with
206: On/Off position-switching (with a 10-minute On/Off cycle made up of
207: 2-second integrations) and online measurements of the system
208: temperature using a noise diode. The total on-source time was $\sim
209: 35$~minutes.
210:
211: Weak absorption was detected at the expected redshifted 21cm frequency
212: ($\sim 431.8$~MHz) in the September run. We hence repeated the
213: observations on October~20, 2006, and January~4, 2007, to confirm the
214: feature. The same observational setup was used in these observing
215: sessions, except for the use of linear instead of circular
216: polarizations (as laboratory calibration information was not available
217: for the circular polarizations; we will hence not further discuss the
218: September data). The total on-source time on TXS~0311+430 was $\sim 50$~minutes
219: and $\sim 65$~minutes in October and January, respectively. A calibrator,
220: PKS~B0316+162, was also observed during the October session, with the same setup,
221: to test for RFI at the observing frequency. The calibrator was observed for a
222: total of 25~on-source minutes, broken into two runs, alternating with two
223: runs on TXS~0311+430.
224:
225: The GBT data were analyzed in AIPS++ using the \texttt{dish} package
226: of single-dish routines. After the initial data-editing, to remove
227: scans with correlator problems and RFI, the data were calibrated
228: (assuming a telescope gain of 2~K/Jy) and averaged together to measure
229: the flux density of TXS~0311+430. This yielded flux densities of $\sim
230: (6.2 \pm 0.7)$~Jy in October and $\sim (6.7 \pm 0.7)$~Jy in January, where
231: the errors include those from confusing sources in the primary beam
232: (note that \citealt{ficarra85} measured $(4.94 \pm 0.10)$~Jy at
233: 408~MHz). A second-order spectral baseline was then fit to RFI- and
234: line-free channels for each 2-second spectrum (during calibration) and
235: subtracted out. The residual 2-second spectra were then averaged
236: together and Hanning-smoothed to obtain the final spectrum for each
237: epoch. A similar procedure was followed to obtain the final spectrum
238: towards PKS~B0316+162, whose flux density was measured to be $9.4 \pm
239: 0.7$~Jy in the October session. Intermittent low-level RFI was seen
240: near the absorption frequencies in all three runs and careful
241: data-editing was hence necessary, especially in the January data.
242:
243: We also obtained a 602-MHz continuum image of TXS~0311+430 with the
244: Giant Metrewave Radio Telescope (GMRT) in March~2007, to determine the
245: spatial structure of the quasar radio emission and derive an estimate of
246: the covering factor. The total on-source time was 1.5~hours, with a
247: 16~MHz bandwidth centred at a frequency of 602~MHz and sub-divided into
248: 128~channels. The standard calibrator 3C48 was used for flux density
249: and bandpass calibration. These data were analysed in classic AIPS, using
250: standard procedures (e.g. \citealt{kanekar07}).
251:
252: \section{Spectra and results}
253:
254: \subsection{Lyman-$\alpha$ and metals}
255: \label{sec:lya_metals}
256:
257: Damped Lyman-$\alpha$ absorption is clearly visible in the GMOS
258: spectrum towards TXS~0311+430, at $z = 2.289 \pm 0.002$. The
259: Lyman-$\alpha$ profile, shown in Fig.~\ref{fig:dla}[A], yields an
260: \hi\ column density of \nhi\ $= (2.0 \pm 0.5) \times 10^{20}$~\cm,
261: derived by overlaying damped profiles using the Starlink {\sc dipso}
262: software. We quote a conservative error of 25\%, to encompass the range
263: of reasonable `by-eye' profiles as well as systematic errors from the
264: continuum fit. Next, although our spectral coverage was such that only
265: a few metal lines associated with the DLA are outside the Lyman-$\alpha$
266: forest, we were able to detect the \sii~$\lambda$1526,
267: \sii~$\lambda$1808 and Al~{\sc ii}~$\lambda$1670 transitions at the
268: DLA redshift; their equivalent widths are listed in Table \ref{tab:metals}
269: and the \sii~$\lambda$1808 profile shown in Fig.~\ref{fig:dla}[B]. Despite
270: the low resolution of our spectrum ($\sim 200$~\kms), all three
271: metal lines have resolved structure, indicating a large velocity width
272: and multiple spectral components. Both the Al~{\sc ii}~$\lambda$1670 and
273: \sii~$\lambda$1526 lines are usually very strong and heavily saturated in DLAs
274: and the high rest frame equivalent widths indicate that this is indeed the case
275: for the absorber towards
276: TXS~0311+430. Conversely, \sii~$\lambda$1808 is often unsaturated in DLA
277: spectra and can therefore be used to derive an abundance for silicon. However,
278: in the present case, the high rest frame equivalent width of the
279: \sii~$\lambda$1808 line suggests that it too is likely to be
280: saturated. Nonetheless, by assuming that the \sii~$\lambda$1808 line
281: is in the linear part of the curve of growth, it is possible to
282: obtain a lower limit on the \sii\ column density and hence, on the
283: metallicity of the DLA. The Si\,{\sc ii} column density is
284: $N$(Si\,{\sc ii})\,$\geq 2.7 \times 10^{15}$\,cm$^{-2}$, giving
285: [Si/H]\,$\geq -0.48$. This is an unusually high metallicity among DLAs,
286: including those in radio-selected QSOs \citep{akerman05}. Among the 104
287: DLAs at $z_{\rm abs} > 1.6$ in the compilation by \citet{prochaska07},
288: there are only three absorbers with higher values of the silicon abundance.
289:
290: Our estimates of $N$(Si\,{\sc ii}), and hence [Si/H], may be lower limits,
291: because we have assumed no line saturation and no depletion of Si onto dust
292: grains. On the other hand, it is possible that we may have over-estimated the
293: equivalent width of the Si\,{\sc ii}~$\lambda 1808$ absorption line through
294: unrecognised blending with other features in our low resolution GMOS spectrum.
295: The work of \citet{herbertfort06} ---see, in particular, their Figure~5--- shows
296: that in low resolution spectra from the Sloan Digital Sky Survey this effect
297: can lead one to overestimate $N$(Si\,{\sc ii}) by as much as a factor of three.
298: Higher resolution spectra, and observations of other spectral lines such as
299: Zn\,{\sc ii}~$\lambda\lambda 2026,2062$, should help resolve these ambiguities.
300:
301: We also examined the GMOS spectrum for possible absorption at other redshifts
302: and detected a strong \mgii\ system at $z \sim 1.069$. The large rest frame
303: equivalent widths of the \mgii~$\lambda\lambda$2796,2803 doublet ($> 3$~\AA)
304: and the \feii~$\lambda$2600 line ($\sim 2.3$~\AA) in this absorber (see
305: Table~\ref{tab:metals}) imply that it too is likely to be a DLA \citep{rao06}.
306:
307:
308:
309: \begin{figure}
310: \centering
311: \epsfig{file=fig1.eps,height=3.5truein,width=3.5truein}
312: \caption{[A]~The damped Lyman-$\alpha$ profile at $z \sim 2.289$ towards
313: TXS~0311+430. The dashed line shows our fit to the profile, which yields
314: \nhi\ $= (2.0 \pm 0.5) \times 10^{20}$~\cm. The dotted lines show the
315: $1\sigma$ errors on the fit. The dashed vertical line
316: here (and in the lower panel) indicates $z = 2.289$. [B]~The \sii~$\lambda$1808
317: line from the $z \sim 2.289$ DLA. }
318: \label{fig:dla}
319: \end{figure}
320:
321:
322: \begin{table}
323: \begin{center}
324: \caption{Detected absorption lines in the GMOS spectrum of TXS~0311+430.
325: All equivalent widths are in the rest frame; errors are 1-$\sigma$, but
326: do not include systematics due to continuum placement ($<$5\%).
327: We do not list an equivalent width for the Mg~{\sc i}~$\lambda$2852 transition
328: from the $z \sim 1.069$ absorber, as it appears to be blended with Galactic
329: and night-sky Na~{\sc i}~$\lambda$5890 features. }
330: \begin{tabular}{ccc}
331: \hline
332: Transition & $z$ & EW (m\AA) \\
333: \hline
334: \sii\ $\lambda$1526 & $2.290 \pm 0.003$ & $761 \pm 58$ \\
335: Al~{\sc ii} $\lambda$1670 & $2.290 \pm 0.002$ & $787 \pm 45$ \\
336: \sii\ $\lambda$1808 & $2.289 \pm 0.002$ & $169 \pm 32$ \\
337: \hline
338: \feii\ $\lambda$2344 & $1.069 \pm 0.002$ & $2606 \pm 80$ \\
339: \feii\ $\lambda$2374 & $1.069 \pm 0.002$ & $1385 \pm 98$ \\
340: \feii\ $\lambda$2382 & $1.069 \pm 0.002$ & $2374 \pm 84$ \\
341: \feii\ $\lambda$2586 & $1.069 \pm 0.002$ & $1595 \pm 68$ \\
342: \feii\ $\lambda$2600 & $1.068 \pm 0.002$ & $2313 \pm 61$ \\
343: \mgii\ $\lambda$2796 & $1.068 \pm 0.001$ & $3115 \pm 70$ \\
344: \mgii\ $\lambda$2803 & $1.069 \pm 0.001$ & $3109 \pm 72$ \\
345: Mg~{\sc i}~$\lambda$2852 & $1.069 \pm 0.001$ & $-$ \\
346: \hline
347: \end{tabular}\label{tab:metals}
348: \end{center}
349: \end{table}%
350:
351:
352: \subsection{Radio spectroscopy and imaging}
353: \label{sec:radio_results}
354:
355: \begin{figure}
356: \centering
357: \epsfig{file=fig2.eps,height=3.5truein,width=3.5truein}
358: \caption{GBT \hi~21~cm spectra towards TXS~0311+430, with optical
359: depth (in units of $10^3 \times \tau$) plotted against barycentric
360: frequency, in MHz. The dashed vertical line is at $z =
361: 2.289766$. Panels~[A] and [B] show the spectra from the October and
362: January runs while panel~[C] shows the final GBT spectrum, obtained by
363: combining the spectra of panels [A] and [B]. The shaded regions in the
364: top two panels show frequency ranges affected by RFI; both these
365: ranges have been blanked out in panel~[C]. }
366: \label{fig:21cm}
367: \end{figure}
368:
369: The final \hi~21~cm spectra from the October and January observing sessions are shown
370: in panels~[A] and [B] of Figure~\ref{fig:21cm}, with the vertical shaded region
371: in each panel marking a frequency range that was affected by regular RFI in both runs.
372: The RMS noise values are $\sim 15.1$~mJy and $\sim 16.5$~mJy per 1.7~\kms\ channel in
373: October and January, respectively. A strong absorption feature is visible in both
374: spectra between $\sim 431.7$ and $\sim 431.86$~MHz. No spectral features were seen
375: in this frequency range in the October spectrum towards the calibrator PKS~B0316+162 (not
376: shown here), which has an RMS noise of $\sim 30.0$~mJy per 1.7~\kms\ channel. The
377: absorption towards TXS~0311+430 also showed the expected doppler shift ($\sim 50$~kHz)
378: between the October and January runs. The lack of any absorption towards
379: the calibrator source, combined with the expected doppler shift over a 3-month
380: period, rule out RFI as a possible cause for the absorption seen towards TXS~0311+430.
381:
382: Figure~\ref{fig:21cm}[C] shows the final GBT 21cm spectrum towards TXS~0311+430,
383: obtained by averaging the spectra from October and January with appropriate weights
384: (based on the RMS noise values), after scaling the January spectrum to a flux density
385: of $6.2$~Jy. Regions affected by RFI in either spectrum have been blanked out (note
386: that these are well-removed from the absorption feature and thus do not affect
387: our optical depth measurement). The RMS noise on this spectrum is $\sim 10.4$~mJy
388: per $\sim 1.7$~\kms\ channel. The 21~cm absorption is complex, extending over $\sim 110$~\kms\
389: (between nulls), with peak opacity at 431.765~MHz, i.e. at $z = 2.289766 (19)$.
390: The integrated optical depth is $\int \tau_{21} {\rm dV} = (0.818 \pm 0.085)$~\kms,
391: with the error dominated by the uncertainty in the source flux density.
392:
393: Figure~\ref{fig:im610} shows the GMRT 602~MHz continuum image of
394: TXS~0311+430. The image has a resolution of $6\farcs0 \times
395: 4\farcs1$ (i.e. a spatial resolution of $\sim 50 \times 34 \: h_{71}^{-2}$~kpc$^2$ at
396: $z = 2.289$\footnote{We use the standard $\Lambda$CDM cosmology, with
397: $\Omega_m = 0.27$, $\Omega_\Lambda = 0.73$ and H$_0 =
398: 71$~\kms~Mpc$^{-1}$.}) and an RMS noise of 0.6~mJy/Bm; no evidence for
399: extended structure can be seen. A single elliptical Gaussian component
400: provides a good fit, yielding an integrated flux density of $3.58$~Jy and a
401: deconvolved angular size of $\sim 1\farcs36 \times 0\farcs63$; this
402: gives a spatial extent of $\lesssim 11.3 \times 5.2 \: h_{71}^{-2}$~kpc$^2$ at $z =
403: 2.289$. Of course, this estimate of the source size should be treated
404: as an upper limit, because any residual phase errors will increase the
405: observed size. Further, while the continuum image is at a somewhat
406: different frequency from the redshifted 21~cm line, the lack of any
407: detected extended emission suggests that the quasar is also very
408: compact at the latter frequency.
409:
410:
411:
412: \begin{figure}
413: \centering
414: \epsfig{file=fig3.eps,height=3.2truein,width=3.2truein}
415: \caption{GMRT 602-MHz continuum image of TXS~0311+430, with a resolution of
416: $6\farcs0 \times 4\farcs1$ and an RMS noise of 0.6~mJy/Bm. The contours
417: are at $4.0 \times (-1, 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024)$~mJy.}
418: \label{fig:im610}
419: \end{figure}
420:
421: \subsection{The Spin Temperature}
422: \label{sec:tspin}
423:
424: For 21~cm absorption studies of DLAs towards radio-loud QSOs, the 21~cm optical depth,
425: \hi\ column density \nhi\ and spin temperature $\ts$ are related by the expression
426: \begin{equation}
427: \label{eqn:tspin}
428: N(HI) = 1.823 \times 10^{18} \left[ \frac{\ts}{f} \right] \intop \tau {\rm d}V,
429: \end{equation}
430: where \nhi\ is in cm$^{-2}$, $\ts$ in K, d$V$ in km~s$^{-1}$ and the
431: 21~cm absorption is assumed to be optically thin. In the case of
432: multiple \hi\ clouds along the line of sight, $\ts$ is the
433: column-density-weighted harmonic mean of the spin temperatures of the
434: individual clouds. Note that the use of the above equation to estimate
435: the spin temperature of the DLA implicitly assumes that the \hi\
436: column densities along the optical and the (usually more extended)
437: radio lines of sight are the same. The covering factor $f$ gives the
438: fraction of the radio flux density covered by the foreground DLA; this
439: can be estimated by very long baseline interferometric (VLBI)
440: observations at or near the redshifted 21cm line frequency, to measure
441: the fraction of flux density arising from the compact radio core, as
442: well as its spatial extent (e.g. \citealt{kanekar07}).
443:
444: The $z \sim 2.289$ DLA towards TXS~0311+430 has an \hi\ column density
445: \nhi\ $= (2 \pm 0.5) \times 10^{20}$~\cm\ and an integrated 21~cm
446: optical depth of $\int \tau {\rm d}V = (0.818 \pm 0.085)$~km/s. No
447: VLBI measurements are available in the literature at any radio
448: frequency. However, the sub-arcsecond size estimate from the
449: GMRT 602~MHz image indicates that the quasar is strongly
450: core-dominated and suggests a high covering factor, $f \sim 1$. Using
451: this, we obtain $\ts = (138 \pm 36) \times f$~K, one of the lowest
452: spin temperatures ever found in a DLA and the only low value currently
453: known at $z>1$.
454:
455: \section{Discussion}
456: \label{sec:discuss}
457:
458: Spin temperature estimates have so far been obtained in more than
459: thirty DLAs at all redshifts, including fifteen detections of 21~cm
460: absorption (Kanekar et al, in preparation). Only four absorbers, all
461: at $z < 0.6$, show low spin temperatures, $\ts < 350$~K, typical of
462: lines of sight through the Milky Way and local disks
463: \citep{kanekar03}\footnote{Note that this does not include the $z \sim
464: 0.656$ DLA towards 3C336 \citep{curran07}, where the background radio
465: source is strongly lobe-dominated, with a very small core
466: fraction. The very extended radio structure (corresponding to a
467: spatial extent of $\sim 195 \: h_{71}^{-1}$~kpc at $z \sim 0.656$)
468: implies that the radio and optical absorption almost certainly arise
469: from very different lines of sight (a possibility discussed by Curran
470: et al.).} Another eleven systems, all with
471: $\ts > 500$~K, have been detected in 21~cm absorption (of which only
472: four are at $z \gtrsim 2$;
473: \citealt{wolfe79,wolfe81,kanekar06,kanekar07}) while the remaining
474: DLAs without detectable 21~cm absorption all have $3\sigma$ lower
475: limits of $> 700$~K on the spin temperature.
476:
477: The preponderance of high spin temperature estimates in high-$z$ DLAs
478: has usually been attributed to a relatively low fraction of the cold
479: neutral phase of \hi\ (the CNM), with most of the gas in the warm
480: phase (the WNM) (e.g. \citealt{carilli96,chengalur00}). This assumes
481: that the absorbers have a similar two-phase structure to that seen in
482: the ISM of the Milky Way (e.g. \citealt{wolfe03b}). The low spin temperature of
483: the $z \sim 2.289$ DLA towards TXS~0311+430 would then imply that a
484: higher fraction of the \hi\ along the line of sight is in the CNM
485: compared with the majority of DLAs at $z \gtrsim 2$. \citet{chengalur00}
486: argued that the high derived $\ts$ values of high-$z$ DLAs could be
487: due to the low metallicities of typical high-$z$ DLAs, with the
488: paucity of metals resulting in fewer radiation pathways for gas
489: cooling (see also \citealt{young97}). If this is correct, one would
490: expect DLAs with low spin temperatures (such as the absorber towards
491: TXS~0311+430) to have significantly higher metallicities than those of
492: the general DLA population \citep{kanekar01a}. As expected, and assuming
493: that the Si\,{\sc ii}~$\lambda 1808$ equivalent width has not been
494: over-estimated, the $z \sim 2.289$ DLA has [Si/H]~$\ge -0.48$, one
495: of most metal-rich DLAs yet discovered.
496:
497: It has also been suggested that the observed low 21~cm optical depth
498: in high-$z$ DLAs arises due to covering factor effects
499: (e.g. \citealt{curran05}). In the present case, any reduction in the
500: covering factor below the assumed value of unity can only strengthen
501: the case for a low spin temperature (since $\ts = (138 \pm 36) \times
502: f$~K). The only way to alter the conclusion that a sizeable fraction
503: of \hi\ along the line of sight is in the cold phase is if there are
504: large spatial differences in the \hi\ column densities along the
505: optical and radio lines of sight (e.g. \citealt{wolfe03b}). For
506: example, if the optical QSO lies behind a ``hole'' in the \hi\ column
507: density distribution, the average \hi\ column density against the
508: radio QSO could be larger than that measured from the Lyman-$\alpha$
509: line, implying a higher spin temperature from
510: Eqn.~\ref{eqn:tspin}. Unfortunately, it is very difficult to directly
511: test this possibility. While Galactic \hi\ column densities derived
512: from Lyman-$\alpha$ absorption studies are in excellent agreement with
513: those obtained from \hi\ 21~cm emission observations in the same
514: directions, despite the very different spatial resolutions in the two
515: methods, such comparisons have only been carried out for a fairly
516: small number of high latitude lines of sight \citep{dickey90}. A
517: similar comparison between \nhi\ values derived from Lyman-$\alpha$
518: absorption and 21~cm emission has only been possible in one DLA, the
519: $z \sim 0.009$ absorber towards SBS~1543+593. Here, the \hi\ column
520: densities agree to within a factor of $\sim 2$, despite the extremely
521: poor spatial resolution ($\sim 5.3 \: h_{71}^{-1}$~kpc) of the radio observations
522: \citep{chengalur02}. While both of these studies suggest that the
523: \nhi\ values along the radio and optical lines of sight are likely to
524: be comparable, we cannot formally rule out the possibility of
525: differences in an individual absorber. However, the fact that the
526: expected high metallicity is indeed seen in the $z \sim 2.289$ DLA
527: (assuming that our Si~{\sc ii}~$\lambda$1808 measurement is accurate)
528: is consistent with the interpretation of a high CNM fraction.
529:
530: Finally, \citet{kanekar03} noted a relationship between spin
531: temperature and absorber morphology, in that, at low redshifts, low
532: $\ts$ values are only found in DLAs identified with luminous disk
533: galaxies, while low-$z$, low-luminosity DLAs, associated with dwarf or
534: LSB galaxies, are all found to have high spin temperatures ($\ts
535: \gtrsim 700$~K). It has not been possible to test this empirical
536: relationship at $z > 1$, as the host galaxies of high-$z$ DLAs are
537: rarely detectable. However, if the relationship does extend out to
538: high redshifts, we would expect the $z \sim 2.289$ DLA to be a
539: massive, luminous disk galaxy. We note that both the high metallicity
540: and the large velocity spreads seen in the optical low-ionization
541: metal lines and the 21cm absorption are consistent with the absorption
542: arising in a massive galaxy. If so, it should be possible to
543: detect the absorber host with deep imaging; this would be the first direct
544: test of the $\ts$-morphology relationship at high redshifts.
545:
546: In summary, we have detected damped Lyman-$\alpha$ and \hi~21~cm absorption at
547: $z = 2.289$ towards the quasar TXS~0311+430. We obtain a DLA spin temperature of
548: $\ts = (138 \pm 36) \times f$~K, the first case of a low spin temperature estimate
549: in a high redshift DLA. The low spin temperature, high metallicity and large velocity
550: spread of the 21~cm and metal lines all suggest that the absorber is likely to be a
551: massive disk galaxy.
552:
553: \section{Acknowledgments}
554: We thank Carl Bignell and Bob Garwood for much help with the GBT observations and
555: AIPS++ data analysis. The NRAO is operated by Associated Universities, Inc.,
556: under cooperative agreement with the NSF. We thank the staff of the GMRT who
557: made these observations possible; the GMRT is run by NCRA-TIFR. Based on
558: observations obtained at the Gemini Observatory (program ID GN-2005B-Q-60),
559: which is operated by the AURA, Inc., under a cooperative agreement with the NSF on
560: behalf of the Gemini partnership: the NSF (United States), the STFC (United Kingdom),
561: the NRC (Canada), CONICYT (Chile), the ARC (Australia), CNPq (Brazil) and CONICET
562: (Argentina). BY was supported in part by a grant from the NSERC (Canada).
563:
564: \bibliographystyle{mn2e}
565: \bibliography{ms}
566:
567: \label{lastpage}
568: \end{document}
569: