1: \documentclass[usenatbib]{mn2e}
2:
3: \usepackage{dcolumn}% Align table columns on decimal point
4: \usepackage{graphics,epsfig}
5: \usepackage{txfonts}
6: \usepackage{times}
7: \usepackage{amssymb}
8:
9: \def\arcs{\mbox{\ensuremath{^{\prime\prime}}}}
10: \def\farcs{\mbox{\ensuremath{.\!\!\arcs}}}
11: \def\lya{Ly$\alpha$}
12: \def\hi{H{\sc i}}
13: \def\nhi{\rm N_{HI}}
14: \newcommand{\al}{{\alpha}}
15: \newcommand{\dal}{{[\Delta \alpha/ \alpha]}}
16: \newcommand{\gp}{{{g}_{p}}}
17: \newcommand{\lqcd}{{\Lambda_{\rm QCD}}}
18: \newcommand{\dmu}{{[\Delta \mu/\mu]}}
19: \newcommand{\lsb}{\left[}
20: \newcommand{\rsb}{\right]}
21:
22:
23: \newcommand{\kms}{km~s$^{-1}$}
24: \newcommand{\cm}{cm$^{-2}$}
25: \newcommand{\Dla}{ Damped Lyman-$\alpha$ } \newcommand{\dla}{ damped Lyman-$\alpha$ }
26: \newcommand{\NHI}{N_{\rm HI}}
27: \newcommand{\noi}{\noindent}
28: \newcommand{\lb}{\left(}
29: \newcommand{\dV}{\Delta V}
30: \newcommand{\rb}{\right)}
31: \newcommand{\pks}{1127$-$145}
32: \newcommand{\ao}{0235+164}
33: \newcommand{\ts}{T_s}
34: \newcommand{\tx}{T_x}
35: \newcommand{\beq}{\begin{equation}}
36: \newcommand{\eeq}{\end{equation}}
37:
38: %\thesaurus{11(11.05.2; 11.06.1; 11.09.4), 12(12.03.3), 13(13.19.3)}
39: \title[Outflowing atomic and molecular gas at $z \sim 0.67$]{Outflowing atomic and molecular gas at $z \sim 0.67$ towards 1504+377}
40: \author[Kanekar \& Chengalur]{Nissim~Kanekar$^1$\thanks{E-mail: nkanekar@aoc.nrao.edu (NK); chengalu@ncra.tifr.res.in (JNC)}
41: Jayaram~N.~Chengalur$^{2}$\\
42: $^1${}National Radio Astronomy Observatory, 1003 Lopezville Rd, Socorro, NM 87801, USA; \\
43: $^2${}National Centre for Radio Astrophysics, Ganeshkhind, Pune--411007, India}
44:
45: \begin{document}
46: \date{Received mmddyy/ accepted mmddyy}
47: \maketitle
48: \label{firstpage}
49:
50: \begin{abstract}
51:
52: We report the detection of OH~1667~MHz and wide \hi~21cm absorption at $z \sim 0.67$ towards the
53: red quasar 1504+377, with the Green Bank Telescope and the Giant Metrewave Radio Telescope.
54: The \hi~21cm absorption extends over a velocity range of $\sim 600$~km/s blueward of
55: the quasar redshift ($z=0.674$), with the new OH~1667~MHz absorption component at
56: $\sim -430$~\kms, nearly coincident with earlier detections of mm-wave absorption at
57: $z \sim 0.6715$. The atomic and molecular absorption appear to arise from a fast gas
58: outflow from the quasar, with a mass outflow rate ${\dot M} \sim 12 M_\odot$~yr$^{-1}$
59: and a molecular hydrogen fraction $f_{\rm H_2} \equiv (N_{\rm H_2}/N_{\rm HI}) \sim 0.2$.
60: The radio structure of 1504+377 is consistent with the outflow arising due to a jet-cloud
61: interaction, followed by rapid cooling of the cloud material. The observed ratio of
62: HCO$^+$ to OH column densities is $\sim 20$~times higher than typical values in
63: Galactic and high-$z$ absorbers. This could arise due to small-scale structure in
64: the outflowing gas on sub-parsec scales, which would also explain the observed
65: variability in the \hi~21cm line.
66:
67:
68: \end{abstract}
69:
70: \begin{keywords}
71: quasars: individual : quasars: absorption lines -- galaxies: ISM
72: \end{keywords}
73:
74: \section{Introduction}
75: \label{intro}
76:
77: The quasar 1504+377 is a rare case of a radio-loud active galactic nucleus (AGN) hosted by a disk
78: galaxy (e.g. \citealt{perlman96,carilli97}). The flat-spectrum radio emission arises from
79: a compact core and a one-sided jet to the southwest, with the jet axis aligned (within
80: $\sim 15^\circ$) with the major axis of the host galaxy \citep{polatidis95,fomalont00}.
81: The AGN is heavily reddened ($r$-K~=~5.1) and was not detected in a deep R-band image,
82: suggesting a high level of dust obscuration \citep{stickel96}. Consistent with
83: this, strong redshifted mm-wave molecular absorption has been detected towards the
84: radio source \citep{wiklind96}, with two absorption complexes at $z \sim 0.6734$ (system~A)
85: and $z \sim 0.6715$ (system~B), close to the redshift of the host galaxy
86: ($z = 0.674 \pm 0.001$; \citealt{stickel94}).
87:
88: Besides the mm-wave transitions, \hi~21cm, OH~1665~MHz and OH~1667~MHz absorption
89: have all been detected from system~A, with strong, wide profiles extending over a
90: velocity range of $\gtrsim 100$~\kms\ \citep{wiklind96,carilli97,kanekar02}. In contrast,
91: the mm-wave absorption in system~B is quite narrow [full-width-at-half-maximum (FWHM)$ \sim 15$~\kms]
92: and neither \hi~21cm nor OH absorption have been detected at this redshift
93: \citep{carilli97,carilli98}.
94: %These non-detections have been a puzzle, as a cloud with a high molecular column
95: %density would be expected to also have a detectable \hi\ column density.
96: %Further, a tight correlation has been found between
97: %OH and HCO$^+$ column densities in both Galactic \citep{lucas96} and redshifted
98: %molecular clouds \citep{kanekar02}, with $N_{\rm HCO^+} \approx 0.03 N_{\rm OH}$.
99: %In fact, the HCO$^+$ column density at $z \sim 0.6715$ is higher than that at
100: %$z \sim 0.6734$, suggesting that the OH column density too should be higher at the
101: %lower redshift.
102: This is the only $z > 0.1$ mm-wave absorber that has not hitherto been detected
103: in OH or \hi~21cm absorption
104: \citep{wiklind94,wiklind95,wiklind96,wiklind96b,chengalur99,kanekar02,kanekar03c}
105: and is thus an excellent candidate for a deep search in these transitions. Beside studying
106: physical conditions in the interstellar medium (ISM) of the QSO host, the detection
107: of these lines would, in principle, also allow one to test for changes in the fundamental
108: constants from $z \sim 0.67$ to the present epoch \citep{darling03,chengalur03,kanekar04b}.
109: Unfortunately, the OH~1665~MHz line from $z \sim 0.6715$ lies at the same frequency as the
110: known 1667~MHz absorption from $z \sim 0.6734$ \citep{kanekar02}, implying that it
111: (and the latter 1667~MHz line) cannot be used to probe fundamental constant evolution.
112: We report here a search for the other three redshifted OH
113: ground-state lines (at rest frequencies of 1667.3590, 1612.2310 and 1720.5299~MHz) and the
114: \hi~21cm line towards 1504+377 with the Giant Metrewave Radio Telescope (GMRT)
115: and the Green Bank Telescope (GBT), resulting in the detection of OH~1667~MHz and \hi~21cm
116: absorption at $z \sim 0.6715$.
117: %The atomic and molecular absorption apparently arise
118: %in a powerful outflow from the $z \sim 0.674$ quasar.
119:
120: \section{Observations and data analysis}
121: \label{sec:obs}
122:
123: \begin{figure*}
124: \centering
125: \epsfig{file=fig1a.eps,height=3.3truein,width=3.3truein}
126: \epsfig{file=fig1b.eps,height=3.3truein,width=3.3truein}
127: \caption{Final GBT \hi~21cm absorption spectrum towards 1504+377 (left panel; resolution $\sim
128: 0.54$~\kms), with optical depth plotted against heliocentric velocity (relative to $z = 0.674$,
129: the quasar redshift). The right panel shows the spectrum smoothed to a resolution of
130: $\sim 4.8$~\kms\ and zoomed-in. The dashed vertical line indicates $z = 0.6715$ while
131: the dashed curves show the 3-Gaussian fit of \citet{carilli98} to their WSRT spectrum.}
132: \label{fig:21cm}
133: \vskip -0.1in
134: \end{figure*}
135:
136: A search for the 1667~MHz and 1720~MHz OH lines from $z \sim 0.6715$ was
137: initially carried out with the GMRT on 26 and 27~March~2006, using the 256-channel
138: mode of the correlator. Bandwidths of 1~and 4~MHz, centred at 997.37 and 1028.77~MHz,
139: were used for the 1667 and 1720~MHz observations, respectively (also allowing a
140: search for the 1720~MHz line from system~A), yielding velocity resolutions
141: of $\sim 2.3$~\kms\ and $\sim 9.1$~\kms\ after Hanning smoothing.
142: 3C286 was used to calibrate the flux density scale
143: and the bandpass shape; no secondary calibrator was observed as 1504+377 is a phase
144: calibrator for the GMRT. The on-source times in the 1667 and 1720~MHz transitions
145: were $\sim 4.6$~hours and $\sim 1.3$~hours, respectively.
146:
147: The GMRT data were analysed in ``classic'' AIPS, using standard procedures. After initial
148: editing to remove corrupted data, continuum images were made of the field at the
149: two frequencies; both images yielded a flux density of $\sim 1.04 \pm 0.01$~Jy for 1504+377.
150: The radio continuum at each frequency was then subtracted out using the task UVLIN and
151: the residual visibilities shifted to the heliocentric frame and imaged in all channels.
152: The final spectra were then extracted by a cut through the spectral cubes at
153: the location of 1504+377.
154:
155: The GMRT observations resulted in the detection of a weak absorption feature at the expected
156: frequency of the redshifted 1667~MHz line. To confirm this and to obtain a better \hi~21cm
157: spectrum, we retrieved archival GBT datasets covering the redshifted \hi~21cm line (from December~2003)
158: and all four ground-state OH lines (from September~2004). The \hi~21cm line was later
159: re-observed with the GBT in November~2006, to confirm the wide, weak absorption seen in
160: the archival data.
161:
162: The GBT observations were carried out in total-power, position-switched mode.
163: %, with online measurements of the system temperature.
164: The OH~runs used four 12.5~MHz Auto-Correlation Spectrometer (ACS) bands, with
165: 8192~channels, centred on the redshifted OH~18cm frequencies. This allowed simultaneous
166: coverage of all four OH~18cm lines from both redshifts, with velocity resolutions of
167: $\sim 0.9$~\kms\ after Hanning smoothing. The \hi~21cm observations of 2003 and 2006
168: used a single 12.5~MHz ACS band with 16384 and 32768 channels,
169: respectively, giving resolutions of $\sim 0.54$~\kms\ (in 2003) and $\sim 0.27$~\kms\
170: (in 2006). The on-source times were 2.5~hours for the OH lines and 0.3~and 1.5~hours for
171: the \hi~21cm line in 2003 and 2006, respectively.
172:
173: All GBT data were analysed in {\sc DISH}, the {\sc AIPS++} single-dish package, using
174: standard procedures. After initial data editing and calibration, the continuum flux density
175: was measured using RFI- and line-free channels. A second-order baseline was then
176: fit to each (typically 10-second) record and subtracted out during the process of calibration;
177: the residual data were then averaged together to obtain the final spectra for each transition.
178: In the case of multiple observing epochs (e.g. the \hi~21cm line), the data from different
179: runs were averaged together, after smoothing and interpolating to the same spectral
180: resolution and frequency scale. The GBT 1720~MHz dataset was affected by strong
181: terrestrial interference (RFI) and will hence not be discussed further.
182:
183: \section{Results}
184: \label{sec:results}
185:
186: \subsection{Spectra}
187: \label{sec:spectra}
188:
189: The left panel of Fig.~\ref{fig:21cm} shows the final GBT \hi~21cm spectrum towards 1504+377,
190: with
191: optical depth (computed assuming a flux density of 1.04~Jy) plotted as a function of
192: heliocentric velocity, in \kms, relative to
193: $z = 0.674$. This has a root-mean-square (RMS) noise of 0.0032, in optical depth units,
194: per $\sim 0.54$~\kms\ channel. The strong 21cm absorption at $\sim -100$~\kms\ (system~A)
195: was detected by \citet{carilli97}; the dashed curve shows the 3-Gaussian fit of
196: \citet{carilli98} to their WSRT spectrum (which we note, in passing, is $\sim 15$~times
197: less sensitive than the GBT spectrum of Fig.~\ref{fig:21cm}). The right panel of the
198: figure shows a zoomed-in version of the spectrum, smoothed to a resolution
199: of $\sim 4.8$~\kms\ to clearly display the wide absorption tail. The \hi~21cm absorption
200: extends well beyond the absorption detected by \citet{carilli97,carilli98}, with a
201: full-width-between-nulls (FWBN) of $\sim 600$~\kms. The new extended absorption can be
202: separated into three distinct parts, a narrow component at $\sim -320$~\kms\ (i.e.
203: $z = 0.6722$), a broad feature at $\sim -430$~\kms\ [nearly the same redshift as the
204: $z \sim 0.6715$ mm-wave absorption of \citet{wiklind96}] and a smooth weak tail, extending out
205: to $-600$~\kms. The integrated \hi~21cm optical depth is
206: $\int \tau_{\rm 21} \rm{d}V = (27.20 \pm 0.04)$~\kms,
207: with around 15\% of the integrated optical depth in the new components detected here.
208:
209: \begin{figure}
210: \centering
211: \epsfig{file=fig2.eps,height=3.3truein,width=3.3truein}
212: \caption{The final redshifted OH~1667~MHz absorption spectrum from system~B towards
213: 1504+377, with optical depth (in units of $10^3 \times \tau$) plotted against
214: heliocentric velocity in \kms, relative to $z = 0.674$. The redshift $z = 0.6715$
215: is indicated by the dashed vertical line.}
216: \label{fig:oh1667}
217: \vskip -0.1in
218: \end{figure}
219:
220:
221: %Fig.~\ref{fig:oh1667}[A] shows an overlay of the final GMRT and GBT OH-1667~MHz spectra
222: %towards 1504+377, again with optical depth plotted against heliocentric velocity, in \kms,
223: %relative to $z \sim 0.6734$ and with the redshift $z \sim 0.6715$ indicated by the dashed vertical
224: %line. The RMS noise values are $\sim 0.0008$~per 2.7~\kms\ channel (GMRT) and $\sim 0.0013$
225: %per 2.7~\kms\ channel (GBT; \footnote{The GBT spectrum was smoothed to the GMRT
226: %resolution for the comparison.}), in optical depth units. Weak narrow 1667~MHz absorption
227: %was detected with $\gtrsim 5\sigma$ significance at $z \sim 0.6715$ in both spectra. It is
228: %very unlikely that the absorption is due to local RFI, given that the spectra were taken at
229: %independent telescopes, separated by a period of nearly three years. Fig.~\ref{fig:oh1667}[B]
230: %shows our final OH~1667~MHz spectrum, obtained by averaging the GMRT and GBT spectra with
231: %appropriate weights; the optical depth RMS noise is $\sim 0.00064$ per $2.7$~\kms\ channel.
232: %and the integrated 1667~MHz optical depth is $\int \tau_{rm 1667} \rm{d}V \sim 0.067 \pm 0.003$~\kms.
233: %The spectrum is well-fit by a single Gaussian model (the solid curve in the figure), with
234: %a full-width-at-half-maximum (FWHM) of $16.5 \pm 2.2$~\kms\ and a peak optical depth
235: %of $0.0032 \pm 0.0004$, at $z = 0.6714637 (51)$
236:
237: Weak narrow absorption was visible close to the expected frequency of the redshifted
238: OH~1667~MHz line in both the GMRT and the GBT spectra (at $\sim 5\sigma$ significance
239: in each spectrum, after averaging all channels). It is very unlikely that the absorption
240: is due to local RFI, given that the spectra were taken at independent telescopes, separated
241: by a period of 2.5 years and with very different doppler shifts. Fig.~\ref{fig:oh1667}
242: shows our final OH~1667~MHz spectrum, obtained by averaging the GMRT and GBT spectra with
243: weights determined by the RMS noise values [$\sim 8 \times 10^{-4}$~(GMRT) and $\sim
244: 1.3 \times 10^{-3}$~(GBT), per 2.75~\kms\ channel, in optical depth units].
245: The optical depth RMS noise on this spectrum is
246: $\sim 6.4 \times 10^{-4}$ per $2.75$~\kms\ channel. The spectrum is well-fit by a
247: single Gaussian model (the dashed curve in the figure), with FWHM$ =(16.5 \pm 2.2)$~\kms\
248: and a peak optical depth of $(3.2 \pm 0.4) \times 10^{-3}$, at $z = 0.6714637 (51)$.
249: The integrated 1667~MHz optical depth is $\int \tau_{\rm 1667}
250: \rm{d}V \sim (0.067 \pm 0.003)$~\kms.
251:
252:
253: Finally, absorption was not detected in the redshifted OH~1612 and 1720~MHz lines (not shown
254: here) at any velocity. The RMS noise on the GMRT 1720~MHz spectrum is $\sim 0.0012$ per
255: $\sim 9.1$~\kms\ channel, while that on the GBT 1612~MHz spectrum is $\sim 0.0024$
256: per $\sim 0.92$~\kms\ channel, in optical depth units. The $3 \sigma$ upper limits on the
257: velocity-integrated optical depth in the 1720 and 1612~MHz lines are $\sim 0.06$~\kms\
258: and $\sim 0.04$~\kms, after smoothing the GBT spectrum by 11~channels to a resolution
259: of $\sim 10.1$~\kms. These limits assume a Gaussian profile with FWHM$ =16.5$~\kms,
260: that of the fit to the OH~1667~MHz line. The ratio of the integrated optical depths
261: in the 1665~MHz and 1612~MHz lines of system~A is $R > 6.4$, higher than that expected
262: ($R \sim 5$) for gas in thermal equilibrium. Note that our sensitivity is insufficient
263: to detect the satellite lines from system~B, if the gas is in thermal equilibrium.
264:
265: \subsection{HI and OH column densities}
266: \label{sec:column}
267:
268: For optically thin gas, the \hi\ and OH column densities can be derived from the
269: \hi~21cm and OH~1667~MHz absorption profiles using the expressions
270:
271: \begin{equation}
272: N_{\rm HI} = 1.823 \times 10^{18} {\lb {\frac {\ts}{f_{\rm 21}}} \rb} \: \int \tau_{\rm 21}
273: \: \mathrm{d} V \:\:\:\: {\rm and}
274: \label{eqn:21cm}
275: \end{equation}
276:
277: \begin{equation}
278: N_{\rm OH} = 2.24 \times 10^{14} {\lb {\frac {\tx }{f_{\rm OH}}} \rb} \: \int \tau_{\rm 1667} \:
279: \mathrm{d} V \; ,
280: \label{eqn:noh}
281: \end{equation}
282: where $\ts$ (in K) is the \hi\ spin temperature, $\tx$ (in K), the OH excitation temperature,
283: and $f_{\rm 21}$ and $f_{\rm OH}$ are the \hi\ and OH covering factors at the respective
284: redshifted line frequencies. $N_{\rm OH}$ and $N_{\rm HI}$ have units of \cm, while the
285: integrals are over velocity, in \kms. \citet{carilli97} used 1.6 and 5~GHz VLBI
286: observations to estimate $f_{\rm 21} = 0.46$, if only the radio core (of angular size
287: $\lesssim 1.4$~mas) is covered, and $f_{\rm 21} = 0.74$, if the inner jet is covered out to
288: $\sim 10$~mas; this would require the absorbing material to have a spatial extent of
289: $\gtrsim 10 \: h_{71}^{-1}$~pc and $\gtrsim 70 \: h_{71}^{-1}$~pc, respectively\footnote{We
290: use the standard LCDM cosmology, with H$_0 = 71$~\kms~Mpc$^{-1}$, $\Omega_m = 0.27$ and
291: $\Omega_\Lambda = 0.73$.}. Typical sizes of Galactic molecular clouds range from
292: $\sim 10 - 50$~pc
293: \citep{blitz90}, somewhat smaller than the latter value; we will hence assume that at least
294: the radio core is covered in both \hi~21cm and OH lines, i.e. $f_{\rm 21} \ge 0.46$ and
295: $f_{\rm OH} \ge 0.46$.
296:
297: Next, it is not possible to determine either $\ts$ or $\tx$ using only the \hi~21cm or
298: OH~1667~MHz absorption profiles. Spin temperature estimates range from $\sim 100$~K
299: in the Galaxy and local and intermediate redshift spiral disks (e.g. \citealt{braun92})
300: to $\gtrsim 1000$~K in high redshift damped Lyman-$\alpha$ systems \citep{kanekar03}.
301: Assuming $\ts = 100$~K gives a lower limit to the \hi\ column density. Further,
302: following \citet{kanekar02}, we will assume $\tx \sim 10$~K, a typical temperature
303: in dark clouds. We then obtain $N_{\rm HI} \ge (1.08 \pm 0.15) \times (\ts/100) (0.46/f_{21})
304: \times 10^{22}$~\cm\ and $N_{\rm OH} \sim (3.26 \pm 0.15) \times (\tx/10) (0.46/f_{\rm OH})
305: \times 10^{14}$~\cm. It should be emphasized that the above \hi\ column density is for
306: the entire profile, i.e. is not restricted to the absorption from system~B, while
307: the OH column density is merely for this system. Finally, we use the empirical
308: relation $N_{\rm H_2} \sim 10^7 \times N_{\rm OH}$ \citep{liszt99} to estimate the
309: molecular hydrogen column density to be $N_{\rm H_2} \sim 3.3 \times (\tx/10) (0.46/f_{\rm OH})
310: \times 10^{21}$~\cm\ for system~B. System~A has
311: $N_{\rm H_2} \sim 2.3 \times (\tx/10) (0.46/f_{\rm OH}) \times 10^{22}$~\cm\ \citep{kanekar02},
312: giving a total ${\rm H_2}$ column density of $N_{\rm H_2} \sim 2.6 \times (\tx/10)
313: (0.46/f_{\rm OH}) \times 10^{22}$~\cm\ at $z \sim 0.67$.
314:
315: \section{Discussion}
316: \label{sec:discuss}
317:
318: \begin{figure}
319: \centering
320: \epsfig{file=fig3.eps,height=3.3truein,width=3.3truein}
321: \caption{Final difference spectrum between the \hi~21cm optical depth spectra towards
322: 1504+377 in 2003 and 2006, with optical depth difference (in units of $10 \times \tau_{diff}$)
323: plotted against heliocentric velocity in \kms\, relative to $z = 0.674$. The original
324: difference spectrum had a resolution of $\sim 0.54$~\kms; this was boxcar-smoothed to,
325: and resampled at, a resolution of $\sim 4.8$~\kms\ to produce this spectrum.}
326: \label{fig:variability}
327: \vskip -0.1in
328: \end{figure}
329:
330: \subsection{Variability in the \hi~21cm profile}
331: \label{sec:21cm-change}
332:
333: Fig.~\ref{fig:variability} shows a plot of the difference between the \hi~21cm optical depths
334: in 2003 and 2006 versus heliocentric velocity, in \kms, relative to $z = 0.674$. The
335: strong features in the difference spectrum between $\sim -150$~\kms\ and $\sim -70$~\kms\
336: indicate significant changes ($\sim 10\%$ of the line depth) in the \hi~21cm profile
337: between 2003 and 2006. Note that the difference cannot be due to a simple scaling of one or
338: both of the spectra, as different spectral components show changes of opposite sign. While
339: the possibility that the observed change might be due to RFI cannot be ruled out, no
340: evidence was seen for RFI at these frequencies, in either these or our other 850~MHz
341: GBT datasets. The profile ``variability'' is coincident with the strongest spectral
342: components, with the rest of the profile showing no evidence for changes within the noise.
343:
344: Variability in redshifted \hi~21cm profiles has been seen earlier in two damped Lyman-$\alpha$
345: systems, at $z \sim 0.524$ towards 0235+164 \citep{wolfe82} and $z \sim 0.3127$ towards 1127$-$145
346: \citep{kanekar01c}. While changes in the latter two profiles have been detected on far
347: shorter timescales (a few days) than in 1504+377, it is interesting that all three sources
348: contain highly compact ($\sim $~mas-scale) components. Possible explanations for the
349: observed changes towards 1504+377 include refractive scintillation in the Galactic
350: interstellar medium (for which the background source need not be compact; \citealt{macquart05}),
351: or transverse motion of a source component on VLBI scales \citep{briggs83c}. Both models
352: require small-scale structure in the 21cm optical depth of the absorbing gas.
353:
354:
355: \subsection{Physical conditions in the absorbing gas}
356: \label{sec:1504}
357:
358: The radio core of 1504+377 and the nucleus of the host galaxy are coincident within the errors
359: ($\sim 1''$) in the R-band image of \citet{stickel94}. At mm-wave frequencies, the core
360: dominates the quasar flux density, with very little emission coming from the steep-spectrum
361: jet \citep{wiklind96}. The core is also likely to be extremely compact at these frequencies,
362: implying that both mm-wave absorbers arise along a single line of sight, which must also pass
363: extremely close to the centre of the host galaxy. \citet{wiklind96} noted that it is
364: impossible to produce two absorption components at very different velocities in such
365: circumstances if the absorbing gas is in pure rotational motion. The large separation
366: ($\sim 330$~\kms) between the two observed
367: absorption velocities is thus suggestive of the presence of strong non-circular orbits;
368: these authors argued in favour of a scenario in which the broad absorption from system~A
369: originates close to the nucleus (in a nuclear ring or a bar), while the narrow absorption
370: of system~A arises in a more-distant cloud in the disk of the host galaxy. In this picture,
371: the systemic redshift is $z \sim 0.6715$. On the other hand, \citet{carilli97} used the
372: fact that the optical emission redshift of the host galaxy ($z = 0.674 \pm 0.001$) is in
373: excellent agreement with that of the higher-redshift complex to argue that the latter is the
374: systemic redshift. They also pointed out that the outflow velocity of system~A in this model
375: ($\sim 330$~\kms) is too large for a cloud in the outer disk of the parent galaxy and
376: suggested the possibility that it might arise in a high-velocity cloud, due to
377: tidal interactions between the host galaxy and a nearby object seen in the R-band image
378: of \citet{stickel94}. Our new GBT \hi~21cm spectrum of Fig.~\ref{fig:21cm} clearly
379: shows that the two absorption systems are, in fact, part of a continuous absorption
380: complex, spanning $\sim 600$~\kms\ and extending from the optical redshift of $z \sim 0.674$
381: out to $z \sim 0.6706$. The 21cm absorption lies entirely blueward of the optical redshift,
382: implying that it must arise in gas that is outflowing from the quasar.
383:
384: The large velocity spread of the \hi\ outflow in 1504+377 is similar to that seen in a number of
385: low-redshift AGNs \citep{morganti05}. These authors note that all known fast \hi\ outflows have
386: been detected in radio galaxies in early or re-started phases of their radio activity.
387: There is also evidence that the most likely mechanism to explain such fast \hi\ outflows
388: is interaction between the radio jets and the surrounding interstellar medium (e.g.
389: \citealt{morganti05b}), with rapid cooling taking place in the gas after a jet-cloud interaction,
390: as expected from numerical simulations (e.g. \citealt{fragile04}). The fact that 1504+377
391: shows no extended radio structure (the outer jet extends to only $\sim 55$~mas, i.e. $\sim 387$~pc,
392: from the nucleus; \citealt{polatidis95}) suggests that it too is in a early phase of its
393: radio activity. Recent 5~GHz VLBI observations \citep{bolton06} have found a new north-eastern
394: extension, which was not seen in earlier (deeper) images (e.g. \citealt{fomalont00}),
395: demonstrating that the source is currently in an active phase. Finally, the fact that the
396: radio structure in 1504+377 is strongly one-sided (e.g. \citealt{fomalont00}) indicates
397: that the jet lies close to the line of sight towards the core. The above suggestion that
398: jet-cloud interactions are responsible for local gas cooling is consistent with the fact
399: that mm-wave absorption (which
400: takes place in cold gas and, as noted earlier, must arise towards the core) is seen
401: at multiple velocities along the line of sight.
402:
403: It thus appears that the wide \hi~21cm and molecular absorption towards 1504+377
404: arise in outflowing gas from the AGN that is cooling rapidly after an interaction with
405: the south-western radio jet. This is the highest redshift at which such a high-velocity
406: outflow has been observed (e.g. \citealt{morganti05}) and, perhaps more interesting, the
407: first case where molecular gas has been detected in the outflow. The ${\rm H_2}$ fraction
408: is $f_{\rm H_2} = \left[N_{\rm H_2}/N_{\rm HI}\right] \le 2 \times (\ts/100) (\tx/10)
409: (f_{\rm OH}/f_{\rm 21})$. \citet{morganti05} assume $\ts \sim 1000$~K to estimate
410: \hi\ column densities for sources in their sample due to the proximity of the gas
411: to the AGN and the likely presence of shocks. Using this value for consistency gives
412: a molecular fraction of $f_{\rm H_2} \sim 0.2$ in the outflowing gas.
413:
414: %\citet{fragile04} find that a large fraction of the gas
415: %interacting with the jet cools below 1000~K, in both equilibrium and non-equilibrium
416: %models, while significant cooling below $\sim 100$~K occurs only in equilibrium models.
417: %While the non-equilibrium models do not include self-gravity, the results of
418: %\citet{fragile04} suggest that $\ts \sim 1000$~K is a more likely spin temperature for
419: %gas interacting with an active jet.
420:
421: We estimate the mass outflow rate ${\dot M}$ using the model of \citet{heckman00},
422: in which a constant-velocity, mass-conserving wind flows into a solid angle
423: $\Omega$ from a minimum radius $r_*$, viz.
424:
425: \begin{equation}
426: {\dot M} = 30 \left[ \frac{\Omega}{4\Pi}\right] \left[ \frac{r_*}{1 \: {\rm kpc}} \right]
427: \left[ \frac{N_{\rm H}}{10^{21} \: {\rm cm^{-2}}} \right]
428: \left[ \frac{v}{300 \:{\rm km s^{-1}}} \right] \: M_\odot \: {\rm yr}^{-1},
429: \end{equation}
430: \noi where $v$ is the outflow velocity and $N_{\rm H}$, the total hydrogen column density
431: of the outflowing gas. We will assume that the minimum radius $r_*$ is $\sim 10$~pc,
432: the size of the radio core, and, following \citet{morganti05}, that $\Omega = \Pi$~steradians
433: and $v = \:{\rm FWBN}/2 \sim 300$~\kms. The total hydrogen column density at
434: $z \sim 0.67$ is $N_{\rm H} = \left[ N_{\rm HI} + 2 \times N_{\rm H_2} \right]
435: \sim 1.6 \times 10^{23}$~\cm, again assuming $\ts \sim 1000$~K. This leads to an
436: estimated mass outflow rate of ${\dot M} \sim 12 M_\odot$~yr$^{-1}$,
437: comparable to estimates in nearby fast \hi\ outflows \citep{morganti05}.
438:
439: %at the lower end of the distribution of outflow rates for ionized superwinds from
440: %starburst galaxies \citep{heckman00}, but
441:
442: %Further, a tight correlation has been found between
443: %OH and HCO$^+$ column densities in both Galactic \citep{lucas96} and redshifted
444: %molecular clouds \citep{kanekar02}, with $N_{\rm HCO^+} \approx 0.03 N_{\rm OH}$.
445:
446: \citet{wiklind96} noted that HCO$^+$ is highly over-abundant in system~B, enhanced by at
447: least an order of magnitude relative to expected abundances in chemical models. The
448: ratios of HCO$^+$ to CO and HCO$^+$ to HCN column densities here are $3 - 5$ times
449: larger than in system~A. While such large differences in relative abundances between
450: HCO$^+$ and species such as CO, HCN, etc, have been
451: observed in Galactic clouds \citep{lucas98}, the ratio of OH to HCO$^+$ column densities
452: in both the Galaxy and a sample of four redshifted HCO$^+$ and OH absorbers has been
453: found to be fairly constant, with $N_{\rm HCO^+}/N_{\rm OH} \sim 0.03$ \citep{liszt96,kanekar02}
454: over more than two orders of magnitude in HCO$^+$ column density. \citet{liszt04}
455: found this ratio to show a large spread (by a factor of $\sim 4$) in the clouds
456: towards Cen.A and NGC1052, with the HCO$^+$ and OH lines also showing very different
457: kinematics, but argued that this could be explained by differing source structure
458: and foreground free-free opacity at the OH and HCO$^+$ frequencies, source variability
459: between observing epochs, and excitation effects at high OH column densities ($\gtrsim
460: 10^{15}$~\cm; \citealt{langevelde95}). Conversely, system~B has
461: $N_{\rm HCO^+}/N_{\rm OH} \sim 0.5 \times (\tx/10) (0.46/f_{\rm OH})$, discrepant by more than
462: an order of magnitude from the expected value. However, 1504+377 is highly compact at
463: both mm-wave and cm-wave frequencies (with a cm-wave core-fraction of $\sim 46$\%;
464: \citealt{carilli97}) and the HCO$^+$ and OH lines have very similar FWHMs
465: [$\sim 16.5 \pm 2.2$~\kms\ (OH) and $\sim 15.2 \pm 0.9$~\kms\ (HCO$^+$)], making it likely
466: that they arise from similar gas. Increasing $\tx$ by an order of magnitude could
467: resolve this problem but such high $\tx$ values have never been seen in the Galaxy
468: (e.g. \citealt{liszt96}). Similarly, the ratio of peak optical depths in the
469: HCO$^+$ and \hi~21cm lines in system~A is $R \equiv \tau_{HCO^+}/\tau_{\rm 21}
470: \sim 30$, far larger than that seen in Galactic clouds ($0.1 \le R \le 6$;
471: \citealt{lucas96,liszt96}). \citet{carilli97} point out that high values of $R$ could
472: result from either far warmer \hi\ or low molecular dissociation, but this would not
473: explain the discrepancy in the ratio of OH and HCO$^+$ column densities. If the latter
474: is not due to real chemical differences between OH and HCO$^+$ (which seems unlikely;
475: \citealt{liszt00}), a plausible explanation is extreme small-scale structure in the
476: opacity of the absorbing gas on sub-parsec scales, smaller than the size of the radio
477: core at cm wavelengths. This could arise due to internal shocks or turbulence in the
478: rapidly outflowing gas. As noted earlier, the observed variability in the \hi~21cm absorption
479: at $z \sim 0.674$ suggests similar small-scale structure at a different location in the
480: outflow, which could also account for the large velocity difference ($\sim 15$~\kms) in peak
481: OH and HCO$^+$ absorption in system~A \citep{kanekar02}.
482: Monitoring the mm-wave lines for variability would be one way of testing this hypothesis.
483:
484: Finally, comparisons between the OH, \hi~21cm and HCO$^+$ redshifts from an absorber
485: can be used to test the evolution of fundamental constants \citep{darling03,chengalur03}.
486: However, the above possibility of small-scale structure in the absorbing gas makes it
487: likely that any such comparisons in the absorbing gas towards 1504+377 will be dominated
488: by local systematic velocity offsets. We conclude that this absorber is unlikely to be
489: useful for the purpose of probing fundamental constant evolution.
490:
491: %However, the OH~1667~MHz and HCO$^+$
492: %absorption at $z \sim 0.6715$ both have simple profiles, well-fit by a single Gaussian
493: %component, and with similar FWHMs,
494: %$\sim 16.5 \pm 2.2$~\kms\ (OH) and $\sim 15.2 \pm 0.9$~\kms\
495: %(HCO$^+$), suggesting that they arise in the same gas. The peak redshifts for the
496: %two transitions are $z_{\rm OH} = 0.6714637 (51)$ and $z_{\rm HCO^+} = 0.6715076 (21)$,
497: %giving $\Delta z/(1 + {\bar z}) = (-2.63 \pm 0.33) \times 10^{-5}$, where
498: %$\Delta z = z_{\rm OH} - z_{\rm HCO^+}$ and ${\bar z} = \left[z_{\rm OH} + z_{\rm HCO^+} \right]/2$.
499: %We note that this difference should be taken with caution, since concerns remain that the
500: %frequency scale of the mm-wave spectra could contain unknown errors [WHY? WE NEED TO
501: %SPECIFY!!!]. The comparison between HCO$^+$ and OH redshifts towards
502: %1504+377 must hence await new measurements of the redshift of the HCO$^+$ line.
503: %
504:
505: \section{Acknowledgments}
506: We thank Francoise Combes for providing us with an ASCII file of the HCO$^+$ spectrum
507: towards 1504+377. We also thank the staff of the GMRT who made these observations possible.
508: The GMRT is run by the National Centre for Radio Astrophysics of the Tata Institute of
509: Fundamental Research. The National Radio Astronomy Observatory is operated by Associated
510: Universities, Inc. under cooperative agreement with the National Science Foundation.
511:
512:
513: \bibliographystyle{mn2e}
514: \bibliography{ms}
515:
516: \end{document}
517:
518: