1: %% AASTeX v5.x LaTeX 2e macros.
2:
3:
4: %% The command below calls the preprint style
5: %% which will produce a one-column, single-spaced document.
6: %% Examples of commands for other substyles follow. Use
7: %% whichever is most appropriate for your purposes.
8: %%
9: %%\documentclass[12pt,preprint]{aastex}
10:
11: %% manuscript produces a one-column, double-spaced document:
12:
13: %%\documentclass[manuscript]{aastex}
14:
15: %% preprint2 produces a double-column, single-spaced document:
16:
17: \documentclass[preprint2]{aastex}
18:
19: %% Sometimes a paper's abstract is too long to fit on the
20: %% title page in preprint2 mode. When that is the case,
21: %% use the longabstract style option.
22:
23: %% \documentclass[preprint2,longabstract]{aastex}
24:
25: \usepackage{rotating}
26: \usepackage{subfigure}
27:
28: %% If you want to create your own macros, you can do so
29: %% using \newcommand. Your macros should appear before
30: %% the \begin{document} command.
31:
32: \newcommand{\myemail}{d.dicken@sheffield.ac.uk}
33: \newcommand{\moo}{\rm $\mu$m}
34:
35:
36: \shorttitle{Origin of the MFIR emission in radio galaxies II}
37: \shortauthors{Dicken et al.}
38:
39: \begin{document}
40:
41: \title{The origin of the infrared emission in radio galaxies II: \\ analysis of mid- to far-infrared Spitzer observations of the 2Jy sample}
42:
43:
44:
45: \author{D. Dicken\altaffilmark{1}, C. Tadhunter\altaffilmark{1}, D. Axon\altaffilmark{2}, R.
46: Morganti\altaffilmark{3,4}, K. J. Inskip\altaffilmark{5}, J. Holt\altaffilmark{6}, R.Gonz\'alez Delgado\altaffilmark{7}, B. Groves\altaffilmark{6} }
47:
48:
49:
50: \altaffiltext{1}{Department of Physics and Astronomy, University of
51: Sheffield, Hounsfield Road, Sheffield, S3 7RH; d.dicken@sheffield.ac.uk, c.tadhunter@sheffield.ac.uk}
52: \altaffiltext{2}{Department of Physics and Astronomy, Rochester
53: Institute of Technology, 84 Lomb Memorial Drive, Rochester NY
54: 14623; djasps@rit.edu}
55: \altaffiltext{3}{ASTRON, P.O. Box 2,
56: 7990 AA Dwingeloo Netherlands; morganti@astron.nl}
57: \altaffiltext{4}{Kapteyn Astronomical Institute, University of Groningen Postbus 800, 9700 AV Groningen, Netherlands}
58: \altaffiltext{5}{Max-Planck-Institut f\"{u}r Astronomie, K\"{o}nigstuhl 17, D-69117 Heidelberg, Germany, inskip@mpia-hd.mpg.de}
59: \altaffiltext{6}{Leiden Observatory, Leiden University, P.O. Box 9513, 2300 RA Leiden, Netherlands, jholt@strw.leidenuniv.nl, brent@strw.leidenuniv.nl}
60: \altaffiltext{7}{Instituto de Astrofisica de Andalucia (CSIC), Apdto.3004, 18080 Granada, Spain; rosa@iaa.es}
61:
62: \begin{abstract}
63: We present an analysis of deep mid- to far-infrared (MFIR) Spitzer
64: photometric observations of the southern 2Jy sample of powerful radio
65: sources (0.05 $<$ z $<$ 0.7), conducting a statistical investigation
66: of the links between radio jet, AGN, starburst activity and MFIR
67: properties. This is part of an ongoing extensive study of powerful
68: radio galaxies that benefits from both complete optical emission line
69: information and a uniquely high detection rate in the far-infrared
70: (far-IR). We find tight correlations between the MFIR and
71: [OIII]$\lambda$5007 emission luminosities, which are significantly
72: better than those between MFIR and extended radio luminosities, or
73: between radio and [OIII] luminosities. Since [OIII] is a known
74: indicator of intrinsic AGN power, these correlations confirm AGN
75: illumination of the circum-nuclear dust as the primary heating
76: mechanism for the dust producing thermal MFIR emission at both 24 and
77: 70\moo. We demonstrate that AGN heating is energetically feasible, and
78: identify the narrow line region clouds as the most likely location of
79: the cool, far-IR emitting dust. Starbursts make a major
80: contribution to the heating of the cool dust in only 15-28\% of our
81: targets.
82:
83: We also investigate the orientation dependence of the continuum
84: properties, finding that the broad- and narrow-line objects in our
85: sample with strong emission lines have similar distributions of MFIR
86: luminosities and colours. Therefore our results are entirely
87: consistent with the orientation-based unified schemes for powerful
88: radio galaxies. However, the weak line radio galaxies (WLRG) form a
89: separate class of objects with intrinsically low luminosity AGN in
90: which both the optical emission lines and the MFIR continuum are weak.
91:
92: \end{abstract}
93:
94: \keywords{galaxies:active - infrared:galaxies}
95:
96: \section{Introduction}
97: \label{sec:intro}
98: In the last few decades space-based infrared observatories have
99: provided new opportunities to understand the physics of AGN. In
100: particular, mid- to far-infrared (MFIR) observations can address
101: many outstanding issues relating to the origin of the prodigious
102: emission, as well as bringing insight to questions relating to the
103: unification of active galaxies, the triggering of the activity, and
104: the evolution of AGN in general.
105:
106: Several efforts have been made to model the MFIR spectral energy
107: distributions (SEDs) of radio-loud AGN in order to understand the
108: origin of the MFIR emission. The thermal emission from warm dust
109: radiating in the mid-infrared (mid-IR) is readily explained in models
110: by AGN heating of dust close to the active core (\citealp{pier92},
111: \citealp{pier93}, \citealp{granato94}), but the origin of the thermal
112: far-infrared (far-IR) emission is less well understood. This is
113: because the far-IR continuum is emitted by cooler dust ($<50$K), and
114: models that attempt to explain the heating of the cool dust in terms
115: of AGN illumination of a compact optically thick dust torus have
116: struggled to explain how radiation from the AGN can heat dust at
117: sufficiently large radii to produce emission in the
118: far-IR. Alternative theoretical studies that focus on adjusting the
119: geometry and size of the torus, or modelling it as a collection of
120: discrete dust clouds surrounding the AGN (e.g. \citealp{nenkova02},
121: \citealp{vanbemmel03}, \citealp{fritz06}), can account for the far-IR
122: emission in terms of AGN illumination. However, the main problem with
123: testing such models is our lack of direct knowledge of the spatial
124: distribution of the emitting dust. Due to their large distances, the
125: circum-nuclear dust structures remain unresolved for the vast majority
126: of AGN. Without constraints on the radial distribution of the dust,
127: there is a limit to how much the SED modelling approach can inform us
128: about the origin of the MFIR emission.
129:
130: Statistical studies that correlate MFIR, optical and radio properties
131: provide a promising alternative to direct, spatially resolved, studies
132: of the dust. This is due to the fact that the MFIR continuum is likely
133: to be more isotropic than the shorter wavelength UV-optical-near
134: infrared continuum, and also because the thermal emission from
135: circum-nuclear dust acts as a bolometer for the AGN activity. MFIR
136: observations of radio-loud AGN are particularly important
137: because the extended radio lobe emission from such galaxies is
138: generally considered to be emitted isotropically, providing an
139: opportunity to select orientation-unbiased samples.
140:
141: Many previous studies have acknowledged these benefits
142: (\citealp{golombek88}, \citealp{impey93},
143: \citealp{heckman92,heckman94}, \citealp{hes95}, \citealp{haas04},
144: \citealp{shi05}, \citealp{cleary07}). However, definitive results were
145: hampered in the past by the relatively low sensitivity of the infrared
146: observatories, as well as the lack of complete, homogeneous samples of
147: powerful radio galaxies in the local Universe. It is notable that IRAS
148: and ISO detected less than 30\% and 50\% respectively of powerful 3C
149: radio galaxies at moderate redshifts ($z<0.3$). IRAS-based studies
150: such as those by \citet{heckman94} and \citet{hes95}, established
151: putative correlations between low frequency radio and MFIR emission
152: for radio-loud AGN, indicating a link between the AGN power and the
153: MFIR continuum emission. However, due to the limited detection rate in
154: the MFIR, the heating mechanism for the dust producing the prodigious
155: MFIR emission remained uncertain.
156:
157: It is now widely accepted that the mid-IR ($<$30\moo) continuum is
158: heated by direct AGN illumination of dust structures close to the AGN
159: (e.g. the circum-nuclear torus). However, in light of the large
160: scatter of correlations between optical and far-IR ($>30\mu \rm{m}$)
161: continuum properties, it has been suggested that illumination by a
162: starburst component may provide the principal heating mechanism for
163: the dust producing the far-IR emission (\citealp{rowan95}). More
164: direct observational evidence for starburst heating of the far-IR
165: continuum in AGN is presented in \citet{schweitzer06} and
166: \citet{netzer07}, based on an analysis of starburst-sensitive
167: polycyclic aromatic hydrocarbon (PAH) features. Finding a correlation
168: between PAH 7.7\moo\ and 60\moo\ luminosity in a sample of nearby PG
169: quasars, \citet{schweitzer06} and \citet{netzer07} attribute this to a
170: link between far-IR luminosity and star formation. However, their
171: sample is modest, and the PAH star formation signature remains
172: undetected in 60\% of the objects. Therefore the putative correlation
173: between PAH and far-IR continuum properties lacks a solid statistical
174: foundation. While there is no doubt that starburst heating of cool
175: dust can account for a substantial fraction of the far-IR flux in
176: \emph{some} AGN, the starburst contribution to the far-IR continuum in
177: the general population of AGN remains uncertain.
178:
179: As well as their importance for understanding the main dust heating
180: mechanism, MFIR observations can also be used to test the
181: orientation-based unified schemes \citep{barthel89}, under the
182: assumption that the MFIR emission is isotropic. To date, the results
183: from such tests have been ambiguous. Early IRAS-based studies
184: (\citealp{heckman94}, \citealp{hes95}) presented evidence for stronger
185: MFIR emission in broad line radio galaxies and quasars (BLRG/Q)
186: compared with narrow line radio galaxies (NLRG), suggesting that the
187: MFIR is not, in fact, isotropic. On the other hand, additional studies
188: using ISO data (\citealp{meisenheimer01}, \citealp{haas04}) found no
189: evidence for differences between the MFIR luminosities of the two
190: optical classes. Unfortunately, these studies were hampered by the
191: relatively poor sensitivity of the IRAS and ISO observatories. Further
192: studies using Spitzer found evidence for a difference between the MFIR
193: emission of BLRG/Q and NLRG. \citet{shi05} attribute this difference
194: to anisotropic emission at mid-IR wavelenghts, while \citet{cleary07}
195: attribute it to a combination of non-thermal contamination of the MFIR
196: emission as well as anisotropic emission at mid-IR
197: wavelenghts. However, both these results were based on samples that
198: were heterogeneous and/or incomplete in terms of far-IR detections.
199:
200: Many of the past studies of MFIR emission from radio galaxies have
201: selected samples based on the 3C radio catalogue. Exploitation of this
202: catalogue is at present hampered by the lack of published high quality
203: optical spectroscopic observations for many of the objects, with which
204: one can confidently classify and identify possible links
205: between MFIR emission, starburst and AGN activity. In contrast, the
206: southern 2Jy sample \citep{tadhunter93} is unique in the sense that
207: deep spectra have been published for the whole sample
208: (\citealp{tadhunter93,tadhunter98,tadhunter02}; \citealp{wills04},
209: \citealp{holt07}: See \S \ref{sec:sample}). The completeness and
210: availability of deep spectroscopic and radio data make this sample
211: well suited to investigating the nature of the MFIR emission and
212: testing the unified schemes. Therefore we have undertaken a program of
213: deep imaging with Spitzer/MIPS of the 2Jy sample, in order to address
214: the sensitivity problems of previous MFIR observatories leading
215: to incomplete sample statistics (see \citealp{dicken08}, hereafter
216: D08).
217:
218: A preliminary analysis of the Spitzer/MIPS 2Jy data set was presented
219: in \cite{tadhunter07} and the measured Spitzer MFIR fluxes are
220: presented in D08. In this paper we conduct an in-depth analysis of the
221: results: sections 2 and 3 present the sample and the data; section 4
222: is concerned with the origin of the MFIR emission; section 5 analyses
223: the far-IR emission and the contribution of starburst heating to the
224: emission at these wavelengths; and section 6 discusses these results
225: in the context of the heating mechanism, the covering factor of the
226: MFIR-emitting dust, the slopes of the correlations and the unified
227: schemes for powerful radio galaxies.
228:
229: \section{The sample}
230: \label{sec:sample}
231: The sample selected for this study comprises a complete sample of all
232: 46 powerful radio galaxies and steep-spectrum quasars ($ F_{\nu}
233: \propto \nu^{-\alpha},\alpha^{4.8}_{2.7} > 0.5 $)\footnote{In addition
234: to excluding quasars with $\alpha^{4.8}_{2.7} > 0.5 $, we also
235: excluded the quasars PKS0159-11 and PKS0842-75 on the basis that they
236: have relatively strong unresolved radio core emission that pushes them
237: above the 2Jy flux limit for the sample as a whole.} selected from the
238: 2Jy sample of \citet{wall85} with redshifts 0.05 $<$ z $<$ 0.7, flux
239: densities $S_{2.7\rm{GHz}}>$ 2Jy and declinations $\delta<10^o$. This
240: sample is a complete, redshift limited, sub-set of that presented in
241: \citet{tadhunter93}, with the addition of \object{PKS 0347$+$05},
242: which has since proved to fulfil the same selection criteria
243: \citep{diserego94}. Our selection criteria cut out all the 16
244: quasar-like objects in the full 2Jy sample that only meet the 2Jy flux
245: criterion because of the strength of their beamed flat spectrum radio
246: core/inner jet components\footnote{The only potentially ambiguous
247: cases are PKS0521-36 and 3C273, which have flat spectra based on their
248: integrated radio emission, but extended steep spectrum emission
249: components with flux $S_{2.7GHz}>2$Jy. However, given that the
250: extended steep spectrum emission in these objects is concentrated in
251: one-sided jets that are likely to be strongly beamed and hence likely
252: to be dominated by non-thermal emission, we have decided not to
253: include them in the analysis presented in this paper. We find that the
254: inclusion/exclusion of these objects in the statistical tests makes no
255: difference to the main conclusions of the paper.}; the remaining
256: objects in the sample are all dominated by their extended steep
257: spectrum lobe/hotspot emission. Therefore, given that there is no
258: strong evidence for beaming and anisotropy in extended steep spectrum
259: radio components, our sample is unlikely to be significantly biased
260: towards a particular orientation of the jets to the line of sight. The
261: lower redshift limit has been set to ensure that these galaxies are
262: genuinely powerful sources. Further discussion of the sample selection
263: can be found in D08.
264:
265: In the detailed analysis of the spectral energy distributions presented
266: in D08, we showed that a maximum of 30\% of our complete sample have
267: the possibility of contamination of their MFIR flux by non-thermal
268: synchrotron emission. This is consistent with several previous studies
269: of samples of radio sources (\citealp{polletta00},
270: \citealp{cleary07}), which have indicated that the number of
271: objects with possible non-thermal contamination of the MFIR is small, and
272: generally confined to quasars with flat spectrum radio cores.
273:
274: \clearpage
275: \begin{deluxetable}{c@{\hspace{0mm}}l@{\hspace{-2mm}}c@{\hspace{-3mm}}c@{\hspace{-1mm}}c@{\hspace{-0mm}}c@{\hspace{-3mm}}r@{\hspace{0mm}}r@{\hspace{0mm}}c@{\hspace{0mm}}c}
276: \tabletypesize{\scriptsize}
277: %%\rotate`
278: \tablecaption{The Sample \label{tbl-1}}
279: \tablewidth{0pt}
280: \tablehead{
281: \colhead{PKS}{\hspace{0mm}} & \colhead{Other}{\hspace{-2mm}} &\colhead{Optical}{\hspace{-3mm}} & \colhead{Radio}{\hspace{-1mm}} & \colhead{z}{\hspace{0mm}}& \colhead{$L_{24}$(W/Hz)}{\hspace{-2mm}} & \colhead{$L_{70}$(W/Hz)}{\hspace{-2mm}} & \colhead{$L_{[\rm{OIII}]}$(W)}{\hspace{-3mm}} & \colhead{$L_{radio}^{5GHz}$(W/Hz)} & \colhead{Starburst}
282: }
283: \startdata
284: 0023$-$26 & \phantom{aa} & \phantom{a} NLRG & CSS & \phantom{a} 0.322 &$ \phantom{a} 4.6\times10^{23 \phantom{a} }$&$ 5.6\times10^{24 \phantom{aa} }$&$ 6.5\times10^{34 }$&$ 4.7\times10^{26 }$& \phantom{aa} SB \\
285: 0034$-$01 & \phantom{aa} 3C015 & \phantom{a} WLRG & FRII & \phantom{a} 0.073 &$ \phantom{a} 4.2\times10^{22 \phantom{a} }$&$ 1.0\times10^{23 \phantom{aa} }$&$ 1.4\times10^{33 }$&$ 9.0\times10^{24 }$& \phantom{aa} No \\
286: 0035$-$02 & \phantom{aa} 3C17 & \phantom{a} BLRG & (FRII) & \phantom{a} 0.220 &$ \phantom{a} 7.1\times10^{23 \phantom{a} }$&$ 1.4\times10^{24 \phantom{aa} }$&$ 5.4\times10^{34 }$&$ 7.6\times10^{25 }$& \phantom{aa} No \\
287: 0038$+$09 & \phantom{aa} 3C18 & \phantom{a} BLRG & FRII & \phantom{a} 0.188 &$ \phantom{a} 9.9\times10^{23 \phantom{a} }$&$ 1.2\times10^{24 \phantom{aa} }$&$ 6.8\times10^{34 }$&$ 7.3\times10^{25 }$& \phantom{aa} No \\
288: 0039$-$44 & \phantom{aa} & \phantom{a} NLRG & FRII & \phantom{a} 0.346 &$ \phantom{a} 5.3\times10^{24 \phantom{a} }$&$ 1.1\times10^{25 \phantom{aa} }$&$ 5.1\times10^{35 }$&$ 2.0\times10^{26 }$& \phantom{aa} No \\
289: 0043$-$42 & \phantom{aa} & \phantom{a} WLRG & FRII & \phantom{a} 0.116 &$ \phantom{a} 1.5\times10^{23 \phantom{a} }$&$ 1.3\times10^{23 \phantom{aa} }$&$ 2.2\times10^{33 }$&$ 4.2\times10^{25 }$& \phantom{aa} No \\
290: 0105$-$16 & \phantom{aa} 3C32 & \phantom{a} NLRG & FRII & \phantom{a} 0.400 &$ \phantom{a} 1.8\times10^{24 \phantom{a} }$&$< 2.2\times10^{24 \phantom{aa} }$&$ 1.1\times10^{35 }$&$ 3.0\times10^{26 }$& \phantom{aa} No \\
291: 0117$-$15 & \phantom{aa} 3C38 & \phantom{a} NLRG & FRII & \phantom{a} 0.565 &$ \phantom{a} 3.7\times10^{24 \phantom{a} }$&$ 1.2\times10^{25 \phantom{aa} }$&$ 7.2\times10^{35 }$&$ 7.3\times10^{26 }$& \phantom{aa} No \\
292: 0213$-$13 & \phantom{aa} 3C62 & \phantom{a} NLRG & FRII & \phantom{a} 0.147 &$ \phantom{a} 8.9\times10^{23 \phantom{a} }$&$ 8.2\times10^{23 \phantom{aa} }$&$ 5.6\times10^{34 }$&$ 4.4\times10^{25 }$& \phantom{aa} No \\
293: 0235$-$19 & \phantom{aa} OD-159 & \phantom{a} BLRG & FRII & \phantom{a} 0.620 &$ \phantom{a} 5.5\times10^{24 \phantom{a} }$&$ 7.1\times10^{24 \phantom{aa} }$&$ 8.6\times10^{35 }$&$ 9.8\times10^{26 }$& \phantom{aa} No \\
294: 0252$-$71 & \phantom{aa} & \phantom{a} NLRG & CSS & \phantom{a} 0.566 &$ \phantom{a} 1.7\times10^{24 \phantom{a} }$&$< 5.5\times10^{24 \phantom{aa} }$&$ 6.4\times10^{34 }$&$ 9.7\times10^{26 }$& \phantom{aa} No \\
295: 0347$+$05 & \phantom{aa} & \phantom{a} BLRG & FRII & \phantom{a} 0.339 &$ \phantom{a} 8.0\times10^{23 \phantom{a} }$&$ 7.1\times10^{24 \phantom{aa} }$&$ 9.1\times10^{33 }$&$ 2.0\times10^{26 }$& \phantom{aa} U \\
296: 0349$-$27 & \phantom{aa} & \phantom{a} NLRG & FRII & \phantom{a} 0.066 &$ \phantom{a} 4.2\times10^{22 \phantom{a} }$&$ 2.0\times10^{23 \phantom{aa} }$&$ 5.3\times10^{33 }$&$ 9.2\times10^{24 }$& \phantom{aa} U \\
297: 0404$+$03 & \phantom{aa} 3C105 & \phantom{a} NLRG & FRII & \phantom{a} 0.089 &$ \phantom{a} 2.6\times10^{23 \phantom{a} }$&$ 6.0\times10^{23 \phantom{aa} }$&$ 1.3\times10^{34 }$&$ 2.0\times10^{25 }$& \phantom{aa} No \\
298: 0409$-$75 & \phantom{aa} & \phantom{a} NLRG & FRII & \phantom{a} 0.693 &$ \phantom{a} 2.3\times10^{24 \phantom{a} }$&$ 1.7\times10^{25 \phantom{aa} }$&$ 5.6\times10^{34 }$&$ 3.6\times10^{27 }$& \phantom{aa} SB \\
299: 0442$-$28 & \phantom{aa} & \phantom{a} NLRG & FRII & \phantom{a} 0.147 &$ \phantom{a} 5.1\times10^{23 \phantom{a} }$&$ 7.2\times10^{23 \phantom{aa} }$&$ 3.1\times10^{34 }$&$ 5.6\times10^{25 }$& \phantom{aa} No \\
300: 0620$-$52 & \phantom{aa} & \phantom{a} WLRG & FRI & \phantom{a} 0.051 &$ \phantom{a} 1.3\times10^{22 \phantom{a} }$&$ 1.4\times10^{23 \phantom{aa} }$&$< 1.1\times10^{32 }$&$ 3.3\times10^{24 }$& \phantom{aa} SB \\
301: 0625$-$35 & \phantom{aa} OH-342 & \phantom{a} WLRG & FRI & \phantom{a} 0.055 &$ \phantom{a} 7.6\times10^{22 \phantom{a} }$&$ 1.4\times10^{23 \phantom{aa} }$&$ 1.3\times10^{33 }$&$ 6.5\times10^{24 }$& \phantom{aa} No \\
302: 0625$-$53 & \phantom{aa} & \phantom{a} WLRG & FRII & \phantom{a} 0.054 &$ \phantom{a} 5.5\times10^{21 \phantom{a} }$&$< 3.4\times10^{22 \phantom{aa} }$&$< 4.9\times10^{32 }$&$ 4.7\times10^{24 }$& \phantom{aa} No \\
303: 0806$-$10 & \phantom{aa} 3C195 & \phantom{a} NLRG & FRII & \phantom{a} 0.110 &$ \phantom{a} 3.4\times10^{24 \phantom{a} }$&$ 6.4\times10^{24 \phantom{aa} }$&$ 2.6\times10^{35 }$&$ 2.2\times10^{25 }$& \phantom{aa} No \\
304: 0859$-$25 & \phantom{aa} & \phantom{a} NLRG & FRII & \phantom{a} 0.305 &$ \phantom{a} 9.2\times10^{23 \phantom{a} }$&$ 8.3\times10^{23 \phantom{aa} }$&$ 4.2\times10^{34 }$&$ 2.4\times10^{26 }$& \phantom{aa} No \\
305: 0915$-$11 & \phantom{aa} Hydra A & \phantom{a} WLRG & FRI & \phantom{a} 0.054 &$ \phantom{a} 2.9\times10^{22 \phantom{a} }$&$ 3.8\times10^{23 \phantom{aa} }$&$ 1.3\times10^{33 }$&$ 4.2\times10^{25 }$& \phantom{aa} SB \\
306: 0945$+$07 & \phantom{aa} 3C227 & \phantom{a} BLRG & FRII & \phantom{a} 0.086 &$ \phantom{a} 3.3\times10^{23 \phantom{a} }$&$ 1.3\times10^{23 \phantom{aa} }$&$ 3.5\times10^{34 }$&$ 2.1\times10^{25 }$& \phantom{aa} U \\
307: 1136$-$13 & \phantom{aa} & \phantom{a} Q & FRII & \phantom{a} 0.554 &$ \phantom{a} 6.1\times10^{24 \phantom{a} }$&$ 1.1\times10^{25 \phantom{aa} }$&$ 2.4\times10^{36 }$&$ 8.9\times10^{26 }$& \phantom{aa} U \\
308: 1151$-$34 & \phantom{aa} & \phantom{a} Q & CSS & \phantom{a} 0.258 &$ \phantom{a} 1.5\times10^{24 \phantom{a} }$&$ 4.7\times10^{24 \phantom{aa} }$&$ 1.2\times10^{35 }$&$ 2.3\times10^{26 }$& \phantom{aa} U \\
309: 1306$-$09 & \phantom{aa} & \phantom{a} NLRG & CSS & \phantom{a} 0.464 &$ \phantom{a} 2.0\times10^{24 \phantom{a} }$&$ 9.1\times10^{24 \phantom{aa} }$&$ 6.3\times10^{34 }$&$ 5.9\times10^{26 }$& \phantom{aa} U \\
310: 1355$-$41 & \phantom{aa} & \phantom{a} Q & FRII & \phantom{a} 0.313 &$ \phantom{a} 6.0\times10^{24 \phantom{a} }$&$ 7.5\times10^{24 \phantom{aa} }$&$ 3.4\times10^{35 }$&$ 2.0\times10^{26 }$& \phantom{aa} U \\
311: 1547$-$79 & \phantom{aa} & \phantom{a} BLRG & FRII & \phantom{a} 0.483 &$ \phantom{a} 2.9\times10^{24 \phantom{a} }$&$ 7.0\times10^{24 \phantom{aa} }$&$ 1.2\times10^{36 }$&$ 5.1\times10^{26 }$& \phantom{aa} U \\
312: 1559$+$02 & \phantom{aa} 3C327 & \phantom{a} NLRG & FRII & \phantom{a} 0.104 &$ \phantom{a} 2.8\times10^{24 \phantom{a} }$&$ 5.5\times10^{24 \phantom{aa} }$&$ 7.8\times10^{34 }$&$ 3.4\times10^{25 }$& \phantom{aa} No \\
313: 1602$+$01 & \phantom{aa} 3C327.1 & \phantom{a} BLRG & FRII & \phantom{a} 0.462 &$ \phantom{a} 2.2\times10^{24 \phantom{a} }$&$ 3.5\times10^{24 \phantom{aa} }$&$ 2.9\times10^{35 }$&$ 3.4\times10^{26 }$& \phantom{aa} No \\
314: 1648$+$05 & \phantom{aa} Herc A & \phantom{a} WLRG & FRI & \phantom{a} 0.154 &$ \phantom{a} 6.6\times10^{22 \phantom{a} }$&$< 6.2\times10^{23 \phantom{aa} }$&$ 2.0\times10^{33 }$&$ 3.7\times10^{26 }$& \phantom{aa} No \\
315: 1733$-$56 & \phantom{aa} & \phantom{a} BLRG & FRII & \phantom{a} 0.098 &$ \phantom{a} 3.3\times10^{23 \phantom{a} }$&$ 1.7\times10^{24 \phantom{aa} }$&$ 2.9\times10^{34 }$&$ 3.5\times10^{25 }$& \phantom{aa} U \\
316: 1814$-$63 & \phantom{aa} & \phantom{a} NLRG & CSS & \phantom{a} 0.063 &$ \phantom{a} 2.5\times10^{23 \phantom{a} }$&$ 5.9\times10^{23 \phantom{aa} }$&$ 1.9\times10^{33 }$&$ 1.4\times10^{25 }$& \phantom{aa} U \\
317: 1839$-$48 & \phantom{aa} & \phantom{a} WLRG & FRI & \phantom{a} 0.112 &$ \phantom{a} 4.5\times10^{22 \phantom{a} }$&$ 1.6\times10^{23 \phantom{aa} }$&$< 1.0\times10^{32 }$&$ 1.8\times10^{25 }$& \phantom{aa} No \\
318: 1932$-$46 & \phantom{aa} & \phantom{a} BLRG & FRII & \phantom{a} 0.231 &$ \phantom{a} 2.1\times10^{23 \phantom{a} }$&$ 1.5\times10^{24 \phantom{aa} }$&$ 1.1\times10^{35 }$&$ 2.4\times10^{26 }$& \phantom{aa} SB \\
319: 1934$-$63 & \phantom{aa} & \phantom{a} NLRG & CSS & \phantom{a} 0.183 &$ \phantom{a} 6.2\times10^{23 \phantom{a} }$&$ 7.1\times10^{23 \phantom{aa} }$&$ 5.2\times10^{34 }$&$ 2.6\times10^{26 }$& \phantom{aa} No \\
320: 1938$-$15 & \phantom{aa} & \phantom{a} BLRG & FRII & \phantom{a} 0.452 &$ \phantom{a} 2.3\times10^{24 \phantom{a} }$&$ 6.6\times10^{24 \phantom{aa} }$&$ 3.4\times10^{35 }$&$ 9.2\times10^{26 }$& \phantom{aa} No \\
321: 1949$+$02 & \phantom{aa} 3C403 & \phantom{a} NLRG & FRII & \phantom{a} 0.059 &$ \phantom{a} 6.9\times10^{23 \phantom{a} }$&$ 1.2\times10^{24 \phantom{aa} }$&$ 3.2\times10^{34 }$&$ 8.6\times10^{24 }$& \phantom{aa} No \\
322: 1954$-$55 & \phantom{aa} & \phantom{a} WLRG & FRI & \phantom{a} 0.060 &$ \phantom{a} 1.0\times10^{22 \phantom{a} }$&$ 3.3\times10^{22 \phantom{aa} }$&$< 4.5\times10^{31 }$&$ 6.1\times10^{24 }$& \phantom{aa} No \\
323: 2135$-$14 & \phantom{aa} & \phantom{a} Q & FRII & \phantom{a} 0.200 &$ \phantom{a} 4.5\times10^{24 \phantom{a} }$&$ 4.9\times10^{24 \phantom{aa} }$&$ 5.9\times10^{35 }$&$ 6.5\times10^{25 }$& \phantom{aa} U \\
324: 2135$-$20 & \phantom{aa} OX-258 & \phantom{a} BLRG & CSS & \phantom{a} 0.635 &$ \phantom{a} 6.2\times10^{24 \phantom{a} }$&$ 5.3\times10^{25 \phantom{aa} }$&$ 6.4\times10^{35 }$&$ 1.1\times10^{27 }$& \phantom{aa} SB \\
325: 2211$-$17 & \phantom{aa} 3C444 & \phantom{a} WLRG & FRII & \phantom{a} 0.153 &$ \phantom{a} 1.8\times10^{22 \phantom{a} }$&$< 3.4\times10^{23 \phantom{aa} }$&$ 1.1\times10^{33 }$&$ 6.2\times10^{25 }$& \phantom{aa} No \\
326: 2221$-$02 & \phantom{aa} 3C445 & \phantom{a} BLRG & FRII & \phantom{a} 0.057 &$ \phantom{a} 7.4\times10^{23 \phantom{a} }$&$ 5.9\times10^{23 \phantom{aa} }$&$ 7.6\times10^{34 }$&$ 7.6\times10^{24 }$& \phantom{aa} No \\
327: 2250$-$41 & \phantom{aa} & \phantom{a} NLRG & FRII & \phantom{a} 0.310 &$ \phantom{a} 1.4\times10^{24 \phantom{a} }$&$ 2.7\times10^{24 \phantom{aa} }$&$ 2.2\times10^{35 }$&$ 1.7\times10^{26 }$& \phantom{aa} No \\
328: 2314$+$03 & \phantom{aa} 3C459 & \phantom{a} NLRG & FRII & \phantom{a} 0.220 &$ \phantom{a} 4.0\times10^{24 \phantom{a} }$&$ 4.0\times10^{25 \phantom{aa} }$&$ 7.2\times10^{34 }$&$ 8.1\times10^{25 }$& \phantom{aa} SB \\
329: 2356$-$61 & \phantom{aa} & \phantom{a} NLRG & FRII & \phantom{a} 0.096 &$ \phantom{a} 4.0\times10^{23 \phantom{a} }$&$ 7.3\times10^{23 \phantom{aa} }$&$ 4.0\times10^{34 }$&$ 4.7\times10^{25 }$& \phantom{aa} No \\
330:
331: \enddata
332:
333: \tablecomments{Table\ref{tbl-1}: Column 3 definitions; Q - quasar,
334: BLRG - broad line radio galaxy, NLRG - narrow line radio galaxy,
335: WLRG - weak line radio galaxy. Column 4 definitions; FRI \&\ FRII -
336: Fanaroff-Riley class 1 and 2 respectively, CSS - compact steep
337: spectrum, C/J - core/jet. Column 8 [\rm{OIII}]$\lambda5007$
338: luminosities calculated from flux presented in \citet{tadhunter93}
339: and \citet{wills04}. Column 10 gives the 15-17Ghz monochromatic core
340: luminosities from the fluxes presented in D08. Column 11 gives
341: information about whether a spectroscopic young stellar population
342: (YSP) has been detected at optical wavelengths indicating the
343: possibility of starburst activity. SB - YSP detected, No - No YSP, U
344: - uncertain YSP component (references for YSP/starburst:
345: \citealp{tadhunter02};\citealp{wills04,wills08}; \citealp{holt07}). Positions
346: for the objects can be found in D08. }
347:
348: \end{deluxetable}
349: \clearpage
350:
351: \section{The data}
352: \label{sec:data}
353: We have made deep Spitzer/MIPS observations (24, 70 and 160\moo)
354: of our sample, as well as complementary high frequency radio
355: observations (15 to 22 \rm{GHz}) with the ATCA and the VLA. Full details of
356: the observations and reduction can be found in D08, along with the
357: MFIR, radio fluxes and spectral energy distributions for the entire
358: sample. We detect 100\% of our sample at 24\moo, 90\% at 70\moo\ and
359: 33\% in the lower sensitivity 160\moo\ band. This is by far the best
360: detection rate for MFIR observations of a sample
361: of an intermediate-redshift AGN published to
362: date. Additionally, the high frequency radio data, along with data
363: from the literature, have enabled us to detect the radio cores in
364: $\approx$70\% of our complete sample. We utilize these data to
365: calculate the orientation-sensitive R parameter (see D08 and \S \ref{sec:R}).
366:
367: In Table \ref{tbl-1} we present the 24 and 70\moo\ monochromatic
368: luminosities for the complete sample, along with 5 \rm{GHz} radio and
369: [\rm{OIII}] $\lambda5007$ emission line luminosities\footnote{In order
370: to calculate luminosities we used $H_{o}=71km s^{-1} Mpc^{-1},
371: \Omega_{m}=0.27$ and $\Omega_{\lambda}=0.73$ along with spectral
372: indices derived from the F(70)/F(24) flux ratios}. The analysis
373: presented here differs from the preliminary results presented in
374: \citet{tadhunter07} in that the emission line luminosities have now
375: been corrected for Galactic extinction, using E(B-V) reddening values
376: obtained from the NASA/IPAC Extragalactic Database (NED), along with
377: the parameterised Galactic extinction law of \citet{howarth83}. In Table
378: \ref{tbl-1} we also present the most up-to-date optical classification
379: for each of the sources, with objects classified as narrow-line radio
380: galaxies (NLRG), broad-line radio galaxies or quasars (BLRG/Q), and
381: weak-line radio galaxies (WLRG)\footnote{WLRGs are sometimes
382: known as low-excitation galaxies but we prefer to label them as WLRG
383: since the excitation (or, more accurately, ionization) of the emission
384: line gas is not necessarily related to the AGN luminosity. Indeed,
385: there exist examples of AGN with relatively high [OIII] luminosities,
386: but emission line ratios reflecting a low ionization state. WLRG are
387: defined as having EW([OIII])$<$10\AA\ \citep{tadhunter98}.}.
388:
389:
390: \begin{figure}[t]
391: \epsscale{1}
392: \plotone{f1.eps}
393: \caption{\label{fig:rad_z} Distribution of 5 \rm{GHz} vs. redshift
394: total radio luminosity (top); high frequency radio spectral index
395: ($\alpha^{4.8}_{2.7}$) vs. 5 \rm{GHz} radio luminosity (bottom) for
396: the 2Jy sample.}
397: \end{figure}
398:
399:
400: In Figure \ref{fig:rad_z} we plot 5 \rm{GHz} radio luminosity against
401: both redshift (top) and radio spectral index (bottom) for the complete
402: sample. The effect of the flux limit of the sample is clearly visible
403: in the upper plot in the form of the tight correlation between
404: monochromatic radio luminosity and redshift. It is noteworthy that our
405: complete sample covers three orders of magnitude in radio
406: luminosity. The plot of the high frequency radio spectral index
407: $\alpha^{4.8}_{2.7}$ against radio luminosity allows us to investigate
408: any bias in our steep spectrum selection method. Based on the similar
409: distributions of the different optical classes in $\alpha^{4.8}_{2.7}$
410: over the full 3 orders of magnitude covered by our sample, we are
411: confident that the steep spectrum selection has left us with no strong
412: bias towards BLRG/Q or NLRG classification.
413:
414: \begin{sidewaysfigure*}
415: \includegraphics[width =1\columnwidth]{f2.eps}
416: \caption{ Luminosity correlation plots: (a) $L_{24\mu m}$
417: vs. $L_{[\rm{OIII}]\lambda 5007}$; (b) $L_{70\mu m}$
418: vs. $L_{[\rm{OIII}]\lambda 5007}$; (c) $L_{24\mu m}$
419: vs. $L_{5\rm{GHz}}$; (d) $L_{70\mu m}$ vs. $L_{5\rm{GHz}}$. The cross in the
420: bottom right corner of plots (a) and (b) represents the maximum
421: uncertainties in [\rm{OIII}] and MFIR luminosity measurements,
422: demonstrating that the scatter in the plots is real and not a
423: consequence of observational uncertainties. Red circles indicate
424: BLRG/Q objects, green squares NLRG and blue triangles
425: WLRG.\label{fig:opradall}}
426: \end{sidewaysfigure*}
427:
428: \section{MFIR emission and the central engine}
429: \label{sec:corr}
430:
431: Many previous studies (e.g. \citealp{heckman94}, \citealp{hes95},
432: \citealp{cleary07}) have found evidence that MFIR emission power is
433: correlated with measured total radio power, albeit based on highly
434: incomplete samples. This result is in line with a model in which the
435: MFIR emission and radio jet power are both strongly linked through the
436: physics of the central engine. However, radio emission depends on
437: factors in addition to the intrinsic power of the AGN. For example,
438: the properties of the local ISM interacting with the radio plasma are
439: expected to have a major impact on the conversion of jet mechanical
440: power into radio luminosity (e.g. \citealp{barthel96}). In addition,
441: the low frequency radio emission may be emitted by extended
442: structures that are far from the regions emitting the MFIR
443: continuum. Therefore, a change in the intrinsic power of the AGN may
444: take a significant amount of time to be reflected in the emission of
445: the extended radio lobes.
446:
447: An alternative to comparisons based on low frequency radio emission is
448: to use optical emission line luminosity. The AGN-photoionized
449: narrow-line region (NLR) is emitted on a smaller scale ($\leq$5kpc)
450: than the extended radio lobe emission in most radio galaxies.
451: Therefore the $[\rm{OIII}]\lambda$5007 emission line is likely to
452: provide a good indication of the intrinsic power of the illuminating
453: AGN (e.g. \citealp{rawlings91}, \citealp{tadhunter98},
454: \citealp{simpson98}).
455:
456: \subsection{Comparisons with MFIR luminosity}
457: We now utilize our complete sample to address the question of how
458: the MFIR continuum depends on the AGN power, with the benefits of not
459: only completeness in the MFIR flux measurements, but also accurate
460: [\rm{OIII}]$\lambda$5007 emission line luminosities for the entire
461: sample.
462:
463: In Figures \ref{fig:opradall} (a) and (b) we plot 24\moo\ and 70\moo\
464: monochromatic luminosities ($L_{24\mu m }$, $L_{70\mu m }$) against
465: [\rm{OIII}] luminosities ($L_{[\rm{OIII}]}$), while in Figures
466: \ref{fig:opradall} (c) and (d) we plot the 24\moo\ and 70\moo\
467: luminosities against 5GHz monochromatic radio luminosities
468: ($L_{5\rm{GHz}}$). It is clear from Figures 2 (a) and 2(b) that the MFIR
469: luminosities are strongly correlated with $L_{[\rm{OIII}]}$. As
470: discussed in the preceding subsection, the [\rm{OIII}]$\lambda 5007$
471: luminosity is expected to provide a good indication of intrinsic AGN
472: power. Therefore, we propose that the 24\moo\ and 70\moo\ emission
473: are also intrinsically linked to the power of the active core and,
474: consequently, that the dust producing the MFIR emission is likely to be
475: heated directly by the central AGN.
476:
477: It is also notable from Figure \ref{fig:opradall}(a) that $L_{24\mu m
478: }$ shows a much tighter correlation with the $L_{[\rm{OIII}]}$ than it
479: does with the $L_{5\rm{GHz}}$. This result is expected, because we
480: believe that both the 24\moo\ luminosity and the [\rm{OIII}]
481: luminosity are strongly linked to the power of the active core via AGN
482: illumination of the emission line clouds and dust structures close to
483: the nucleus. The radio power, however, is dependent on additional
484: factors such as the nature of the ISM in the halo of the host galaxy
485: into which the jets and lobes expand.
486:
487: Considering next the differences between the 24\moo\ and 70\moo\
488: correlations, it is apparent that there is more scatter in the
489: $L_{70\mu m}$ vs $L_{[\rm{OIII}]}$ correlation. $L_{70\mu m}$ also
490: remains better correlated with $L_{[\rm{OIII}]}$ than $L_{5\rm{GHz}}$,
491: but the difference is not as evident as at 24\moo. Our interpretation
492: of the cause of this additional scatter at 70\moo\ lies in the origin
493: of the far-IR emission, as we will discuss in detail in \S
494: \ref{sec:origin}. However, it is noteworthy that this scatter cannot
495: be due to the reduced sensitivity of Spitzer at longer far-IR
496: wavelengths because our observational errors are small compared with
497: the overall scatter in the distribution (see error bars in plots (a)
498: and (b)). Therefore, there must be a physical cause for the increased
499: scatter in this plot.
500:
501:
502: \subsection{Rank correlation statistics}
503: \label{sec:rankstat}
504:
505: \begin{deluxetable}{l@{\hspace{0mm}}c}
506: \tabletypesize{\scriptsize}
507: \tablecaption{Statistical Analysis \label{tbl-2}} \tablewidth{0pt}
508: \tablehead{ \colhead{Rank Correlation}{\hspace{0mm}} &
509: \colhead{$r_s$}{\hspace{0mm}} }
510: \startdata
511: (1) $L_{24}$ vs $L_{[\rm{OIII}]}$ & 0.88 \\
512: (2) $L_{70}$ vs $L_{[\rm{OIII}]}$ & 0.76 \\
513: (3) $L_{24}$ vs $L_{5GHz}$ & 0.54 \\
514: (4) $L_{70}$ vs $L_{5GHz}$ & 0.63 \\
515: (5) $L_{5\rm{GHz}}$ vs z & 0.93 \\
516: (6) $L_{5\rm{GHz}}$ vs $L_{[\rm{OIII}]}$ & 0.57 \\
517: \cutinhead{Partial Rank Correlation with z}
518: (7) $L_{24}$ vs $L_{[\rm{OIII}]}$ & 0.77 \\
519: (8) $L_{70}$ vs $L_{[\rm{OIII}]}$ & 0.50 \\
520:
521: \enddata \tablecomments{Table \ref{tbl-2}: Result of various Spearman
522: rank correlation statistics. Values of
523: $0<r_s<1$ are given for each test, where a value close to 1 is highly
524: significant. This test was undertaken with a z limited sample
525: z$>$0.06, to remove most of the objects with upper limits in
526: [OIII]. In addition the object PKS1839-48 which has upper limits in
527: [OIII] was also removed, leaving 38 objects from our complete sample. See \S \ref{sec:rankstat} for discussion of
528: the 70\moo\ upper limits.}
529: \end{deluxetable}
530:
531:
532: Prior to interpreting our results in full, it is important to
533: investigate the significance of the correlations in Figure 2. We have
534: therefore calculated the Spearman rank correlation coefficient for the
535: four correlations. The test was undertaken using a redshift limited
536: sub-sample with z$>$0.06 in order to remove the majority of WLRGs with
537: upper limits in [OIII]. In addition the remaining object with an upper
538: limit in [OIII] (PKS1839-48) was also excluded\footnote{The exclusion
539: or inclusion of the [OIII] upper limit value for this object has been
540: investigated, and we find that it makes no significant difference to
541: the statistical results presented in Table \ref{tbl-2} }. In order to
542: quantify the effects that the four remaining upper limits in 70\moo\
543: have on this test, we chose to handle the correlations that included
544: 70\moo\ data in the following way: we created replacements for each of
545: the four 70\moo\ upper limits by randomly selecting a 70\moo/24\moo\
546: flux ratio value from the distribution of measured 70\moo/24\moo\ flux
547: ratio values for the sample, and then multiplying this by the measured
548: 24\moo\ flux to create a new 70\moo\ flux estimate. The 70\moo\
549: estimates were then converted to luminosities and included in the rank
550: correlation test. This process was repeated 1000 times, and the median
551: of the correlation coefficient for those cycles is presented in Table
552: \ref{tbl-2} for correlations involving 70\moo\ luminosities (i.e.rows 2,
553: 4 and 8).
554:
555: The results show a high level of significance for all the
556: correlations using a two tailed test, in the sense that we can reject
557: the null hypothesis that the variables are uncorrelated at a $>$99.9\%
558: level of significance. The rank correlation statistics, presented in
559: Table \ref{tbl-2}, clearly show that the $L_{[\rm{OIII}]}$
560: vs. $L_{24\mu m}$ and $L_{[\rm{OIII}]}$ vs. $L_{70\mu m}$ correlations
561: are more significant than those between the radio luminosity and MFIR
562: and [OIII] luminosities.
563:
564: For a flux-limited sample such as that considered here, it is natural
565: to consider the possibility that the correlations between MFIR, radio
566: and [OIII] luminosities might not be intrinsic, but rather arise
567: through their mutual dependence on redshift. For example, given the
568: strong correlation between radio power and redshift induced by the
569: radio flux limit (see Figure 1), and also the correlation between
570: emission line luminosity and radio power, it is possible that a
571: correlation between the $L_{[\rm{OIII}]}$ and the MFIR emission
572: luminosity might arise because the MFIR luminosities are independently
573: correlated with z (e.g. because of genuine redshift evolution rather
574: than being intrinsically correlated with $L_{[\rm{OIII}]}$). The
575: second part of Table \ref{tbl-2} (rows 7 and 8), shows the results of
576: a partial rank correlation test carried out in order to investigate
577: whether the correlations could be a result of a dependence on the
578: third variable z. In both cases we still find that the null hypothesis
579: that the variables are unrelated can be rejected at a $>$99.5\% level
580: of significance. This demonstrates that both $L_{24\mu m }$ and
581: $L_{70\mu m}$ are intrinsically correlated with $L_{[\rm{OIII}]}$.
582:
583: \subsection{LIRG and ULIRG comparison}
584: \label{sec:ulirg}
585:
586: \begin{figure}[t]
587: \epsscale{1}
588: \plotone{f3.eps}
589: \caption{\label{fig:ulirg} Plot of $L_{70\mu m}$ vs $L_{[OIII]}$
590: showing the comparison with LIRG and ULIRG luminosities. Approximate
591: ranges for LIRGs and ULIRGs are indicated by the horizontal lines,
592: where the dashed red lines, represent the luminosity limits of typical LIRGs and
593: ULIRGS for a warm SED model, and the dot dashed blue lines represent
594: the limits of LIRGs and ULIRGs for the cool SED model (see \S
595: \ref{sec:ulirg}). Stars represent objects with optical evidence
596: for recent star formation activity (see \S \ref{sec:looking}).}
597: \end{figure}
598:
599:
600: In order to place radio galaxies in the context of other sources of
601: prodigious MFIR emission in the local universe, it is interesting to
602: compare them with the luminous and ultra-luminous infrared galaxies
603: (LIRGs and ULRIGs: \citealp{sanders96}). Figure \ref{fig:ulirg} again
604: shows the $L_{70\mu m}$ vs $L_{[\rm{OIII}]}$ correlation for the
605: entire sample, but in this case we also indicate the typical ranges
606: that LIRGs and ULIRGs would occupy on the diagram. To calculate these
607: ranges we follow the definitions of \citet{sanders96}: that LIRGs have
608: integrated far-IR luminosities
609: $10^{11}L_{\odot}>L_{FIR}>10^{12}L_{\odot}$, while ULIRGs have
610: $10^{12}L_{\odot}>L_{FIR}>10^{13}L_{\odot}$. The ranges of 70\moo\
611: monochromatic luminosities corresponding to these integrated
612: luminosities have been estimated based on two assumptions about the
613: MFIR SEDs of the sources. In each case the lower line has been
614: calculated by assuming the SED of the radio galaxy 3C327 --- taken to
615: be representative of a source with relatively warm MFIR colours,
616: whereas the upper line has been calculated by assuming the SED of the
617: ULIRG Arp220 --- representative of a source with cool MFIR
618: colours. Considering these limits we find that 28-48\% of our sample
619: would be classified as LIRGs, and $\leq$7\% as ULIRGS.
620:
621: \subsection{Testing the unified schemes}
622: \label{sec:OpClass}
623: The MFIR luminosities can also be used to test the orientation-based
624: unified schemes for powerful radio galaxies (regardless of the
625: emission mechanism). In particular, if the distributions of the
626: observed MFIR luminosities are similar for different optical classes
627: of objects (NLRG, BLRG/Q, WLRG) this is consistent with, but does not
628: prove, that they are part of the same parent population.
629:
630: Evidence from investigations using IRAS (e.g. \citealp{heckman94},
631: \citealp{hes95}) suggested that the MFIR luminosities of narrow line
632: radio galaxies (NLRG) are lower by up to an order of magnitude
633: in comparison to broad line radio galaxies and quasars (BLRG/Q)\footnote{In
634: the following, unless otherwise specified, we consider BLRG and
635: quasars as a single class. This is justified on the basis that, among
636: objects with broad permitted emission lines, there is a continuous range in
637: luminosities between objects classed as BLRG and quasars.}. These
638: results have been supported by some studies (\citealt{vanbemmel00})
639: and rejected by others (\citealt{meisenheimer01}, \citealt{haas04}),
640: but all the previous studies were based on incomplete
641: samples. Therefore, it remains uncertain whether genuine differences
642: exist between the MFIR properties of broad and narrow line radio-loud
643: AGN.
644:
645: Any difference between the two types of radio galaxies at shorter
646: mid-IR wavelengths ($\lesssim$30\moo), might be explained in terms
647: of extinction by a dusty torus, provided that there is significant
648: optical depth in the torus at such wavelengths. However, it would
649: remain challenging to explain differences in longer wavelength
650: emission in terms of such obscuration. Alternatively, a difference
651: between the two classes of objects could be due to non-thermal beamed
652: emission contaminating the thermal MFIR, which is expected to be
653: stronger in the BLRG/Q.
654:
655: Using our complete sample we can investigate
656: whether there are significant differences between the MFIR properties
657: of broad- and narrow-line objects in the 2Jy sample.
658:
659: A visual inspection of Figures \ref{fig:opradall} (a) and (b) reveals
660: no evidence that the BLRG/Q have higher luminosities than the NLRG at
661: 24 or 70\moo, and we have shown that no more than 24\% of the sample have
662: a possibility of contamination from non-thermal beamed components (see
663: \citet{dicken08}). Therefore we conclude that the MFIR emission in our
664: sample is most likely to be emitted isotropically, at least down to an
665: observed wavelength of 24\moo\ (rest wavelength 14-23\moo, depending
666: on redshift).
667:
668:
669: \begin{figure*}[t]
670: \epsscale{2}
671: \plotone{f4.eps}
672: \caption{Plot of the orientation-sensitive R vs. 24\moo\ luminosity
673: (left) and 70\moo\ luminosity (right), where R is defined as
674: $S_{core}/(S_{tot}-S_{core})$. \label{fig:R}}
675: \end{figure*}
676:
677:
678: \subsection{The R parameter}
679: \label{sec:R}
680: We can further examine whether the MFIR continuum is isotropic by
681: investigating any links between MFIR luminosities and the relative
682: brightness of the radio core. If there is any anisotropy in the MFIR
683: continuum, we would expect to see enhanced emission in those objects
684: with strong cores, which have axes orientated closer to the line of
685: sight. Using the new high frequency radio observations presented in
686: D08, as well as data from the literature, we can investigate the
687: dependence of MFIR properties on the orientation-sensitive, core
688: dominance parameter R, defined as
689: $S_{core}/(S_{tot}-S_{core})$. Tables of R values estimated at 5GHz
690: for the complete sample are presented in D08.
691:
692: In Figure \ref{fig:R} we plot the 24 and 70\moo\ luminosities against
693: R. It is clear from these plots that, on average, the BLRG/Q have the
694: highest R values in our sample \citep{morganti97}, consistent with
695: unified schemes which predict that BLRG/Q have jet axes that are
696: pointing closer to the line of sight, leading to stronger beamed
697: core/jet emission. However, not all BLRG/Q are highly core
698: dominated. In fact, the population of such objects shows a wide range
699: of R parameter values.
700:
701: In Figure \ref{fig:colour_R} we plot R against the 70\moo/24\moo\
702: infrared flux ratio. Plotting similar parameters, \citet{shi05} found
703: an anti-correlation in the sense that warmer IR colour corresponds to
704: a smaller R parameter. In addition, they found some evidence that
705: different optical classes lie in different regions of the plot, with
706: the BLRG/Q tending to the highest R, and warmest colours.
707:
708: From a visual inspection of Figure \ref{fig:colour_R} there is little
709: evidence for a strong anti-correlation as presented in
710: \citet{shi05}. The range of infrared colours suggests that the MFIR
711: emission is relatively isotropic. However, it is worth noting that
712: many of the objects in the lower region of the diagram, forming the
713: group at low 70\moo/24\moo\ and high R, are BLRG/Q.
714:
715:
716: \begin{figure}[t]
717: \epsscale{1}
718: \plotone{f5.eps}
719: \caption{Plot of the orientation sensitive R parameter against MFIR colour,
720: represented by the 70/24\moo\ flux ratio. \label{fig:colour_R}}
721: \end{figure}
722:
723:
724: \subsection{Weak line radio galaxies}
725: It is not only the distributions of NLRG and BLRG/Q that are of
726: interest in the correlation plots presented in Figure
727: \ref{fig:opradall}. The figure also provides clues to the nature of
728: WLRG. Although many of the WLRG in our sample have been classified as
729: FRI radio sources, a significant subset have FRII morphologies (see
730: \citealp{tadhunter98}). Previous studies (e.g. \citealp{cao04}) have
731: suggested that obscuration could provide an explanation for
732: the differences between WLRG and radio galaxies with strong emission lines.
733:
734: In Figures \ref{fig:opradall} (a) and (b) it is striking that the WLRG
735: lie at the lowest luminosity end in \emph{both} emission line and MFIR
736: continuum luminosity. If the MFIR emission is primarily caused by AGN
737: heating of the dust, then this demonstrates that WLRG contain
738: \emph{intrinsically weak} AGN. Therefore, WLRG cannot be accommodated
739: in the simplest unified schemes for radio galaxies, which posit that
740: the differences between all classes of radio-loud AGN are solely due
741: to anisotropy and orientation. It is evident that, although
742: obscuration may contribute at some level to the weakness of the
743: emission lines in these objects, the idea that WLRG contain powerful
744: quasar-like AGN, heavily obscured by dust, is certainly not consistent
745: with our data. \citet{hardcastle07} suggest that the differences
746: between WLRG and radio galaxies with strong emission lines
747: (NLRG/BLRG/Q) may be accounted for by different accretion modes of the
748: AGN, with the WLRG accreting hot ISM at a relatively low Eddington
749: ratio, and NLRG/BLRG/Q accreting cold ISM at a higher Eddington ratio.
750:
751: Figures \ref{fig:opradall} (c) and (d) clearly show that, despite
752: being weak in both emission line and MFIR luminosity, the WLRG overlap
753: with the NLRG and BLRG/Q in terms of radio luminosity. This suggest
754: that other factors play an important role in boosting the radio powers
755: of WLRG. In particular, as noted above, the radio luminosity is not
756: necessarily a good measure of intrinsic AGN may power, and the
757: properties of the local ISM have a large impact on the radio
758: luminosity for a given intrinsic jet power. In this case we would
759: expect that, for a given MFIR luminosity, the WLRG with powerful radio
760: emission should be found in relatively dense cluster
761: environments. Some evidence for this effect is presented in
762: \citet{barthel96}, however, their result is based on a limited
763: sample. Future deep optical imaging observations of the 2Jy sample
764: will provide a direct indication of the significance of such
765: environmental effects.
766:
767: \section{Origin of the Far-Infrared Emission}
768: \label{sec:origin}
769:
770:
771: \begin{figure*}[t]
772: \epsscale{2}
773: \plotone{f6.eps}
774: \caption{Plots showing the variation of $L_{MFIR}$ with
775: $L_{\rm{[OIII]}}$ for the complete sample at 24 and 70\moo, with those
776: objects identified as having young stellar populations marked with
777: separate symbols (stars). Also displayed is a supplementary sample of
778: all known radio galaxies with spectroscopically-confirmed evidence for
779: star formation activity at optical wavelengths, the properties of
780: which are detailed in Table \ref{tbl-3}. The regression line has been fitted
781: only to the data from the original sample with z $>$ 0.06, in order to avoid
782: most of the objects with upper limits in their [\rm{OIII}], and also
783: excluding the 7 starburst objects.\label{fig:starburst}}
784: \end{figure*}
785:
786:
787: As discussed in \S \ref{sec:intro}, the nature of the heating
788: mechanism for the cool dust radiating at 70\moo\ (the far-IR) is not well
789: understood. However, it is widely accepted that the cool dust is
790: likely to be heated by starbursts and/or by AGN illumination. Based on
791: the tight correlation between $L_{\rm{[OIII]}}$ and $L_{MFIR}$ shown
792: in Figure \ref{fig:opradall}, we have concluded that the most probable
793: heating mechanism for the dust emitting at both mid- and far-IR
794: wavelengths is direct illumination by the AGN (\S
795: \ref{sec:corr}). However, it is of course naive to rule out starburst
796: heating as a contributor to far-IR emission altogether, specifically
797: because morphological evidence suggests that at least some powerful
798: radio galaxies are triggered in major, gas rich galaxy mergers
799: \citep{heckman86}. Such mergers are predicted to be associated with
800: powerful starbursts (e.g. \cite{dimatteo05}). The starburst-AGN connection is
801: also important for interpreting sub-millimetre observations in the
802: context of the star formation history of radio-loud AGN at high
803: redshift. Some studies, such as that by \citet{archibald01}, assume
804: that the cold dust responsible for the sub-mm emission is heated
805: entirely by starbursts. Therefore to comprehensively understand the
806: MFIR emission from powerful radio galaxies it is vital to investigate
807: the contribution of starburst heating to the observed far-IR fluxes,
808: and possible links between star formation and AGN activity.
809:
810: \subsection{Evidence for starburst heating of the far-IR continuum}
811: \label{sec:looking}
812:
813: Careful spectral synthesis modelling of the high quality optical
814: spectra for our sample (\citealp{tadhunter02}, \citealp{wills04,
815: wills08}, \citealp{holt07}) has allowed us to identify the objects
816: that show clear evidence for recent star formation activity at optical
817: wavelengths (see Table \ref{tbl-1}). Therefore, in Figure
818: \ref{fig:starburst} we are able to plot the $L_{24\mu m}$ and
819: $L_{70\mu m}$ data against $L_{\rm{[OIII]}}$ once again, but this time
820: highlighting those seven objects that we have established to have
821: unambiguous evidence for starburst activity at optical wavelengths
822: (PKS0023$-$26, PKS0409$-$75, PKS0620$-$52, PKS0915$-$11 (3C218),
823: PKS1932$-$46, PKS2135$-$20, PKS2314$+$03 (3C459)). As we have seen in
824: \S \ref{sec:corr}, there are good correlations between both $L_{24\mu
825: m}$, $ L_{70\mu m}$ and $L_{[\rm{OIII}]}$, but there is significantly
826: more scatter in the 70\moo\ correlation, consistent with the findings
827: of \citet{rowan95}. As first described in the preliminary results
828: presented in \cite{tadhunter07}, we believe this scatter is in part
829: due to the boosting by starburst heating of the cool far-IR emitting
830: dust that is not as significant for the warm mid-IR 24\moo\ dust
831: emission.
832:
833: To demonstrate the degree of starburst boosting above the main
834: correlations we have plotted regression lines on both plots in Figure
835: \ref{fig:starburst}. The lines shown are the bisector of a linear
836: least squares fit of $x$ on $y$ and $y$ on $x$. As for the sample
837: used in the Spearman rank statistics, in calculating this fit we
838: include only those objects with z $>$ 0.06, in order to avoid most of
839: the objects with upper limits in their [\rm{OIII}] emission. In addition,
840: the 7 starburst objects and one more object (PKS1839$-$48) with upper
841: an limit in [\rm{OIII}] were also removed, leaving a total sample of
842: 33 objects for the fit.
843:
844: From a visual inspection of Figure \ref{fig:starburst}(b) it is clear
845: that the majority of those objects we identify as having evidence for
846: optical star formation activity are displaced, showing enhanced far-IR
847: luminosities relative to the regression line in the 70\moo\ plot; this
848: enhancement is not apparent for most of the optical starburst objects
849: in the 24\moo\ plot. Therefore, it appears that there must be some
850: mechanism boosting the far-IR flux in starburst compared to
851: non-starburst objects. Analysing the displacements of starburst
852: objects we find that their far-IR fluxes are boosted by a factor of
853: four on average relative to the remainder of the sample, with a
854: maximum boosting factor of 20 in the case of 3C459.
855:
856: In addition to our complete sample, we have also plotted a
857: supplementary sample of starburst radio galaxies, taken from the
858: literature on Figure \ref{fig:starburst}. These supplementary sources
859: represent all known radio-loud AGN that show spectroscopic evidence
860: for star formation activity at optical wavelengths, apart from the
861: similar objects in the 2Jy sample. The properties of these sources are
862: presented in Table \ref{tbl-3}, and they are plotted as open stars in
863: Figure \ref{fig:starburst}. It is clear that these supplementary
864: objects also lie well above the regression line fitted to our data in
865: the 70\moo\ plot. A visual inspection of the supplementary starburst
866: sample in Figure \ref{fig:starburst} also shows a tendency for
867: increased 24 and 70\moo\ luminosities, relative to the majority of the
868: sample, at low $L_{[\rm{OIII}]}$ emission. This is not surprising
869: given that, at low [OIII] luminosities, the AGN itself is
870: intrinsically weak and a modest amount of ongoing star formation can
871: boost the MFIR fluxes well above the regression line.
872:
873: Due to the high level of completeness of our sample in terms of MFIR
874: detections, our data can also be used to test the statistical
875: significance of any differences between the starburst and
876: non-starburst populations in Figure \ref{fig:starburst}.
877:
878: First, for our complete sample we have considered a one dimensional
879: Kolmogorov-Smirnoff (K-S) two sample test, comparing the vertical
880: displacements from our fitted regression line in the $L_{70\mu m}$
881: vs. $L_{[\rm{OIII}]}$ plot (see also \citealp{tadhunter07}). In this
882: case we find that we can reject the null hypothesis that the starburst
883: and non-starburst samples are drawn from the same parent population at
884: a better than 1\% level of significance. We have made the same test,
885: this time including both the 2Jy starburst and the supplementary
886: starburst objects, and investigating whether they are significantly
887: displaced relative to the regression line for the non-starburst 2Jy
888: sample. In that case we find that we can reject the null hypothesis
889: that the starburst and non-starburst samples are drawn from the same
890: parent population at a level of significance of better than 0.1\%.
891:
892: Secondly, we consider the significance of the displacement between the
893: starburst and non-starburst samples using a two dimensional K-S test
894: (\citealp{peacock83}, \citealp{fasano87}). The method we apply here
895: is the generalisation of the 2D K-S test developed by
896: \citet{fasano87}. Again for the $L_{70\mu m}$ vs $L_{[\rm{OIII}]}$
897: plot, initially considering the starburst objects against the
898: non-starburst objects (in our complete 2Jy sample), we find that we
899: can reject the null hypothesis that the starburst and non-starburst
900: samples are drawn from the same parent population at a level of
901: significance of better than 5\%. We find exactly the same significance
902: level when including the supplementary starburst objects. In addition,
903: applying the same statistical test to the $L_{[\rm{OIII}]}$
904: vs. $L_{24\mu m}$ plot we find no differences between the starburst
905: and non-starburst samples, as expected given that the 24\moo\
906: luminosity is less likely to be effected by starburst heating.
907:
908: There are three notable outliers from the correlation, which are
909: labelled on Figure \ref{fig:starburst}(b). These sources add to the
910: increased scatter in the far-IR, and we describe them in more detail
911: below.
912:
913: \begin{itemize}
914:
915: \item{{\bf \object{PKS0347+05}}. This object lies behind the Galactic
916: plane and, despite the relatively large correction of the [OIII] flux
917: for Galactic extinction, it still lies above our fitted regression
918: line. In a recent analysis of near-IR images for the source it was
919: found that there are, in fact, two AGN within our MFIR photometric
920: aperture for this object. If both AGN radiate at MFIR wavelengths this
921: could perhaps explain some of the apparent excess relative to the
922: correlation. Additionally, since \object{PKS0347+05} is a BLRG, its
923: AGN is likely to outshine any starburst signatures at optical
924: wavelengths. Therefore, it is possible that this object could also
925: have star formation activity that has not so far been detected at
926: optical wavelengths.}
927:
928: \item{{\bf \object{PKS0945+07}}. This object lies below the
929: correlation, and is an example of an object with an inverted MFIR
930: spectrum that apparently lacks a cool dust component
931: (e.g. \citealp{miley84}, \citealp{vanbemmel01}).}
932:
933: \item{{\bf \object{PKS1306-09}}. This object was discussed in
934: \citet{tadhunter02}, and shows marginal polarization at optical
935: wavelengths, which could be due to scattered light or a non-thermal
936: optical continuum component. In addition, the extrapolation of the
937: radio component in the SED (see D08) is consistent with non-thermal
938: contamination of the thermal MFIR continuum. So far, the optical
939: spectra are inconclusive regarding the presence of a young stellar
940: component in this object, but it is difficult to entirely rule out the
941: presence of such a component.}
942:
943: \end{itemize}
944:
945: \clearpage
946: \begin{deluxetable}{c@{\hspace{2mm}}c@{\hspace{0mm}}l@{\hspace{0mm}}l@{\hspace{0mm}}l@{\hspace{0mm}}l@{\hspace{0mm}}c@{\hspace{0mm}}c@{\hspace{0mm}}}
947: \tabletypesize{\scriptsize}
948: \tablecaption{The Supplementary Starburst Sample \label{tbl-3}}
949: \tablewidth{0pt}
950: \tablehead{
951: \colhead{Object}{\hspace{2mm}} & \colhead{z}{\hspace{0mm}} &\colhead{$L_{24}$(W/Hz)}{\hspace{0mm}} & \colhead{$L_{70}$(W/Hz)}{\hspace{0mm}} & \colhead{$L_{[\rm{OIII}]}$(W)}{\hspace{0mm}}& \colhead{$L_{5GHz}$(W/Hz)}{\hspace{0mm}}& \colhead{SB ref}{\hspace{0mm}} & \colhead{[OIII] ref}{\hspace{0mm}}
952: }
953: \startdata
954: 3C48 & 0.367 &$ \phantom{aa} 3.0 \times10^{25 }$&$ \phantom{aa} 1.6\times10^{26 }$&$ \phantom{aa} 3.0 \times10^{36 }$&$ \phantom{aa} 1.6\times10^{26 }$& 6 & 4 \\
955: 3C213.1 & 0.194 &$ \phantom{aa} 5.1\times10^{24 } a$&$ \phantom{aa} 5.0\times10^{24 } a$&$ \phantom{aa} 5.6\times10^{33 }$&$ \phantom{aa} 3.1\times10^{25 }$& 18 & 11 \\
956: 3C236 & 0.101 &$ \phantom{aa} 2.0 \times10^{23 }$&$ \phantom{aa} 5.7 \times10^{23 }$&$ \phantom{aa} 1.8 \times10^{33 }$&$ \phantom{aa} 1.7\times10^{25 }$& 1,5 & 2 \\
957: 3C285 & 0.079 &$ \phantom{aa} 9.3\times10^{23 } a$&$ \phantom{aa} 1.7 \times10^{24 } a$&$ \phantom{aa} 3.1 \times10^{33 } a$&$ \phantom{aa} 4.1\times10^{24 }$& 1,2 & 2 \\
958: 3C293 & 0.045 &$ \phantom{aa} 9.2 \times10^{22 }$&$ \phantom{aa} 6.7 \times10^{23 }$&$ \phantom{aa} 1.3 \times10^{32 }$&$ \phantom{aa} 4.0\times10^{24 }$& 7 & 1 \\
959: 3C305 & 0.042 &$ \phantom{aa} 9.4\times10^{22 } a$&$ \phantom{aa} 5.1 \times10^{23 } a$&$ \phantom{aa} 1.4 \times10^{34 } a$&$ \phantom{aa} 1.7\times10^{24 }$& 7 & 3 \\
960: 3C321 & 0.096 &$ \phantom{aa} 3.3 \times10^{24 }$&$ \phantom{aa} 1.0 \times10^{25 }$&$ \phantom{aa} 8.8 \times10^{34 }$&$ \phantom{aa} 1.2 \times10^{25 }$& 1,3 & 2 \\
961: 3C433 & 0.102 &$ \phantom{aa} 2.1 \times10^{24 }$&$ \phantom{aa} 3.5 \times10^{24 }$&$ \phantom{aa} 2.2 \times10^{34 }$&$ \phantom{aa} 4.2 \times10^{25 }$& 1,4 & 3 \\
962: B2 0648$+$27 & 0.041 &$ \phantom{aa} 1.5 \times10^{24 } a$&$ \phantom{aa} 4.6 \times10^{24 } a$&$ \phantom{aa} 1.2 \times10^{34 } a$&$ \phantom{aa} 8.6\times10^{22 }$& 12 & 7 \\
963: B2 0722$+$30 & 0.019 &$ \phantom{aa} 1.6 \times10^{23 } a$&$ \phantom{aa} 1.1 \times10^{24 } a$&$ \phantom{aa} 5.6 \times10^{31 } a$&$ \phantom{aa} 2.4\times10^{22 }$& 13,14 & 8 \\
964: Cen A & 3.4Mpc &$ \phantom{aa} 8.5\times10^{21 } a$&$ \phantom{aa} 1.2\times10^{22 } a$&$ \phantom{aa} 5.0\times10^{33 }$&$ \phantom{aa} 2.4\times10^{23 }$& 15,16 & 9,10 \\
965: PKS0131$-$36 & 0.030 &$ \phantom{aa} 1.5 \times10^{23 } a$&$ \phantom{aa} 1.6 \times10^{24 } a$&$ \phantom{aa} 2.5 \times10^{32 } a$&$ \phantom{aa} 3.5\times10^{24 }$& 1 & 5 \\
966: PKS0320$-$37 & 0.006 &$ \phantom{aa} 7.8 \times10^{21 }$&$ \phantom{aa} 1.5 \times10^{23 }$&$ \phantom{aa} 3.0 \times10^{31 }$&$ \phantom{aa} 2.5\times10^{24 }$& 8,9 & 5 \\
967: PKS0453$-$20 & 0.035 &$ \phantom{aa} 1.2 \times10^{23 } a$&$ \phantom{aa} 8.8 \times10^{23 } a$&$ \phantom{aa} 6.9 \times10^{31 } a$&$ \phantom{aa} 2.1\times10^{24 }$& 11 & 5 \\
968: PKS1345$+$12 & 0.122 &$ \phantom{aa} 1.2 \times10^{25 } a$&$ \phantom{aa} 3.4 \times10^{25 } a$&$ \phantom{aa} 9.0 \times10^{34 } a$&$ \phantom{aa} 4.6\times10^{25 }$& 1,10 & 6 \\
969: PKS1549$-$79 & 0.152 &$ \phantom{aa} 1.2\times10^{25 } a$&$ \phantom{aa} 2.7\times10^{25 } a$&$ \phantom{aa} 8.9\times10^{34 } a$&$ \phantom{aa} 1.1\times10^{26 }$& 17 & 5 \\
970:
971: \enddata
972:
973:
974: \tablecomments{ Table \ref{tbl-3} supplementary sample comprising of
975: all radio-loud objects from the literature with spectroscopic evidence
976: for recent star formation activity at optical wavelengths. All
977: infrared fluxes were obtained from NED apart from the object 3C305 for
978: which Spitzer/MIPS fluxes were extracted from the archive images. 24
979: and 70\moo\ luminosities for 3C213.1 are derived from IRAS upper
980: limits. SB references: (1) \citet{holt07}, (2) \citet{aretxaga01}, (3) \citet{tadhunter96}, (4) \citet{wills02}, (5) \citet{odea01}, (6)
981: \citet{canalizo01}, (7) \citet{tadhunter05}, (8) \citet{goudfrooij01}, (9) \citet{kuntschner00}, (10) \citet{rodriguez07},
982: (11)\citet{wills04}, (12)\citet{emonts06}, (13) \citet{emonts_thesis}, (14) Emonts et al. in preperation, (15) \citet{peng02},
983: (16) \citet{peng04}, (17) \citet{holt07}, (18) \citet{wills08}. [OIII] flux references:
984: (1) \citet{emonts05}, (2) \citet{saunders89}, (3) Robinson PhD thesis,
985: 2001, Universtiy of Sheffield, (4) \citet{chatzichristou99},
986: (5) \citet{tadhunter93}, (6) \citet{gelderman94}, (7) \citet{emonts06},
987: (8) Emonts private communication, (9) \citet{simpson98}, (10) \citet{storchi97}, (11) \citet{holt_thesis}.}
988:
989: \tablenotetext{a}{derived from IRAS measurements at 24\moo\ and 60\moo.}
990: \end{deluxetable}
991: \clearpage
992:
993:
994: \subsection{Correlation slope statistics}
995: In the calculation of our fitted regression line plotted in Figure
996: \ref{fig:starburst}, we have chosen not to include those objects identified as
997: having evidence for a starburst component. Using a bootstrap technique
998: we have investigated the uncertainty in the slopes of our fit to these
999: correlations and tested the effect that the removal of the 70\moo\
1000: upper limits and starburst objects would have on the regression lines.
1001:
1002: In order to investigate the uncertainties, a sample of N data points ($x$ and
1003: $y$ points from Figure \ref{fig:starburst}) were labelled and then
1004: drawn at random to create a group of N replacements for the original
1005: sample. We then recalculated the slope and repeated this process 1000
1006: times. Provided the data points are independent, the distribution of
1007: the slopes estimated in the bootstrap trials provide an indication of
1008: the uncertainty in the estimated slopes.
1009:
1010: The data fitted for this bootstrap are identical to those used in the
1011: Spearman rank correlation statistics (\S \ref{sec:rankstat}) i.e., with
1012: redshifts limited to z $>$ 0.06 in order to avoid most of the objects
1013: with upper limits in their [\rm{OIII}] luminosity and
1014: PKS1839$-$48. Here we investigate the effects of including starburst
1015: objects and the upper limits in the sample drawn for the N data
1016: replacements, in order to quantify their effect on the fitted slopes.
1017:
1018: The starburst objects are simply included or excluded when calculating
1019: the slope however, exactly as the Spearman rank test (\S
1020: \ref{sec:rankstat}), the 70\moo\ upper limits were included by randomly
1021: choosing a ratio of 70\moo/24\moo\ to derive new 70\moo\ upper
1022: limits. This process was repeated for each of the 1000
1023: bootstrap trials. The results of our analysis are presented in Table
1024: \ref{tbl-4}.
1025:
1026: Column 2 of Table \ref{tbl-4} contains the values of the slopes we
1027: have fitted to the data in Figure \ref{fig:starburst} (the bisector of
1028: a linear least squares fit of $x$ on $y$ and $y$ on $x$) as well as
1029: the slope values calculated with starburst objects excluded. Columns
1030: 3 and 4 present the mean value for 1000 slopes of the sample, created
1031: using a bootstrap technique, column 3 includes the 70\moo\ upper
1032: limits and column 4 is for the sample bootstrapped without upper
1033: limits.
1034:
1035: Considering, in the first instance, the results for the $L_{24\mu m}$
1036: vs. $L_{\rm{[OIII]}}$ correlation, we find that the addition/removal
1037: of the starburst objects and the objects with upper limits in 70\moo\
1038: has little effect on the resulting mean slope values, where all values
1039: are consistent within the estimated 1$\sigma$ uncertainties.
1040:
1041: Secondly, looking at the results in Table \ref{tbl-4} for the
1042: $L_{70\mu m}$ vs. $L_{\rm{[OIII]}}$ slopes, it evident, that although
1043: the upper limits do not significantly affect the mean regression line
1044: slopes and associated uncertainties, the introduction of the starburst
1045: sample does. This further supports our hypothesis that the starburst
1046: objects are responsible for much of the scatter in the
1047: $L_{\rm{[OIII]}}$ vs. $L_{70\mu m}$ correlation. Excluding the optical
1048: starburst objects, the slopes and associated uncertainties of the
1049: $L_{70\mu m}$ vs $L_{\rm{[OIII]}}$ correlation are entirely consistent
1050: with those of the $L_{24\mu m}$ vs $L_{\rm{[OIII]}}$ correlation,
1051: reinforcing the idea that the dust radiating at both mid-IR and far-IR
1052: wavelengths has a common heating mechanism: AGN illumination.
1053:
1054: Given that the bootstrap sample in column 3 is identical to that used
1055: to plot the regression line in Figure \ref{fig:starburst}, these
1056: statistics map the uncertainty in our slope fit. For the sample with
1057: upper limits but not including the starburst objects, we calculate
1058: that the uncertainties in our regression line slopes are 0.05 and 0.07
1059: for the 24\moo\ and 70\moo\ correlations respectively, from the
1060: standard deviations of the bootstrap slopes.
1061:
1062: \begin{deluxetable}{c@{\hspace{0mm}}c@{\hspace{0mm}}c@{\hspace{0mm}}c}
1063: \tabletypesize{\scriptsize}
1064: \tablecaption{Bootstrap Analysis \label{tbl-4}}
1065: \tablewidth{0pt}
1066: \tablehead{
1067: \colhead{Correlation plot}{\hspace{0mm}} & \colhead{Bisector}{\hspace{0mm}} & \colhead{with 70\moo\ UL}{\hspace{0mm}} &\colhead{without 70\moo\ UL}{\hspace{0mm}}
1068: }
1069: \startdata
1070: 24\moo\ vs [\rm{OIII}] with starburst & 0.78 & 0.79 $\pm$ 0.05 & 0.76 $\pm$ 0.08 \\
1071: 24\moo\ vs [\rm{OIII}] without starburst & 0.75 & 0.75 $\pm$ 0.05 & 0.71 $\pm$ 0.07 \\
1072: \cutinhead{}
1073: 70\moo\ vs [\rm{OIII}] with starburst & 0.83 & 0.88 $\pm$ 0.11 & 0.93 $\pm$ 0.13 \\
1074: 70\moo\ vs [\rm{OIII}] without starburst & 0.71 & 0.72 $\pm$ 0.07 & 0.78 $\pm$ 0.09 \\
1075:
1076: \enddata
1077:
1078: \tablecomments{Table \ref{tbl-4}: Slopes for the complete sample excluding
1079: objects with z $>$ 0.06 and PKS1839$-$48. The uncertainties on the
1080: values presented in columns 3 and 4 are the standard deviations of the
1081: slopes derived from the 1000 bootstrap cycles. }
1082: \end{deluxetable}
1083:
1084:
1085: \section{Discussion}
1086: \subsection{Heating mechanism}
1087:
1088: On the basis of our observational results, we have concluded that the
1089: MFIR continuum emitted by powerful radio galaxies is predominantly a
1090: consequence of AGN heating of the circum-nuclear dust. However, far-IR
1091: emission from dust heated by starbursts is apparent in the minority of
1092: the objects that fall above the main correlations involving 70\moo.
1093:
1094: The link between far-IR bright radio sources and optical evidence for
1095: starburst activity has been noted in previous
1096: studies. \citet{tadhunter02} found that the two objects with the
1097: strongest evidence for optical starburst activity in their sample are
1098: also the most luminous in the far-IR. Subsequently
1099: \citet{wills02,wills04}, while investigating the UV excess in radio
1100: galaxies, also found that those objects with the best evidence for young
1101: stars have significantly larger far-IR luminosities than the rest of
1102: their sample. The Spitzer results for this 2Jy sample strongly
1103: reinforce these earlier results with a much larger sample.
1104:
1105:
1106: \begin{figure}[h]
1107: \epsscale{1}
1108: \plotone{f7.eps}
1109: \caption{Histograms of MFIR colours (70\moo/24\moo\ flux ratio) for
1110: the non-starburst sample, our starburst sample and the supplementary
1111: starburst sample. The black line shows the dividing line between what we define as cool
1112: and warm colours (70\moo/24\moo\ = 5).\label{fig:starburst_hist}}
1113: \end{figure}
1114:
1115:
1116: Contrary to our results, previous studies such as those by
1117: \citet{rowan95}, \citet{vanbemmel01, vanbemmel03}, put forward
1118: starburst heating as the primary heating mechanism of the dust
1119: producing far-IR emission in powerful AGN, albeit without direct
1120: evidence. This has encouraged some investigators to search for more
1121: direct observational evidence of a starburst/AGN link. For example,
1122: \citet{schweitzer06} and \citet{netzer07} analyzed PAH starburst
1123: tracers and MFIR emission. Although unable to detect PAH features in
1124: 60\% their sample, \citet{schweitzer06} suggest that star formation
1125: heating most likely accounts for at least 30\% of the far-IR emission
1126: of their sample of PG quasars, finding a correlation between PAH
1127: luminosity at 7.7\moo\ and 60\moo\ luminosity similar to that of
1128: starburst dominated ULIRGs. Although \citet{schweitzer06} provide some
1129: of the best empirical evidence for a starburst/AGN connection, albeit
1130: with many upper limits both in the far-IR and PAH emission, it is not
1131: clear that PAH features are exclusively starburst signatures. Indeed,
1132: studies such as that by \citet{siebenmorgen04} have suggested that it
1133: may be possible to produce PAH emission from AGN heating alone if the
1134: dust is situated at sufficiently large radii from the AGN.
1135:
1136: We have shown that the far-IR continuum luminosities ($L_{70\mu m}$)
1137: of the 2Jy radio sources are correlated with their [OIII] luminosities
1138: ($L_{[OIII]}$) with a similar slope to the correlation between
1139: $L_{24\mu m}$ and $L_{[OIII]}$. The similarity of the two correlations
1140: points towards a common mechanism for producing both the mid and
1141: far-IR emission. If we assume, as is generally accepted, that the
1142: continuum emission at 24\moo\ is heated almost exclusively by the AGN
1143: then, in order for the cool dust emitting in the far-IR to be heated
1144: predominantly by starbursts rather than the AGN, a remarkable degree
1145: of coordination between AGN and starburst activity would be
1146: implied. Although we cannot entirely rule out this possibility, we
1147: regard it as unlikely.
1148:
1149: Empirically, starburst galaxies are known to be associated with cool
1150: MFIR colours, as expected in the case of illumination of extended dust
1151: structures by spatially distributed star forming regions. In Figure
1152: \ref{fig:starburst_hist} we present histograms showing the
1153: 70\moo/24\moo\ MFIR colours of the non-starburst, starburst and
1154: supplementary samples separately. It is immediately apparent from the
1155: plot that \emph{all} the 2Jy starburst objects have cool colours
1156: ($70\mu m/24\mu m > 5$), whereas all but two of the non-starburst
1157: sample have much warmer colours ($70\mu m/24\mu m < 5$). The
1158: non-starburst object in Figure \ref{fig:starburst_hist} with the
1159: coolest colour is PKS 0347+05; as discussed above, it is possible that
1160: this object may contain as yet undetected star formation at optical
1161: wavelengths. Moreover, one of the non-starburst objects whose starburst
1162: status is uncertain -- PKS1306-09 -- also has relatively cool colours
1163: (70\moo/24\moo\ = 4.7).
1164:
1165: The data presented in Figure \ref{fig:starburst} can also be used to
1166: consider links between AGN and starburst activity. As discussed in \S
1167: \ref{sec:looking} at low emission line luminosities ($L_{\rm{[OIII]}}
1168: < 10^{33}$W) it is evident that $L_{\rm{[OIII]}}$ is only weakly
1169: correlated with the far-IR luminosity. However, it is notable that all
1170: the objects with large 70\moo\ luminosities ($L_{70\mu m} \geq
1171: 10^{25}$W/Hz) --- many of which we have identified with starburst
1172: heating of the cool dust --- also have large [OIII] luminosities
1173: ($L_{\rm{[OIII]}} \geq 10^{35}$W). This suggest a weak link between
1174: starburst and AGN activity in the sense that only the most powerful
1175: radio-loud AGN are associated with powerful, ULIRG-like
1176: starbursts. However, the reverse is not true since many of the highest
1177: emission line luminosity objects --- some of which are luminous enough
1178: at MFIR wavelengths to be classified as LIRGs --- do not show evidence
1179: for starburst activity, suggesting that powerful AGN are not always
1180: accompanied by massive star formation activity. Similar trends have
1181: been found in a study of nearby type 2 Seyfert galaxies by
1182: \citet{gonzalez01}, who show that, for a given [OIII] luminosity,
1183: there are two populations of Seyferts: one with optical evidence for
1184: prodigious recent star formation, large far-IR luminosities and cool
1185: MFIR colours; the other with little evidence for enhanced star
1186: formation activity. However, in the case of the Seyferts, a higher
1187: proportion ($>$50\%) of objects show evidence for recent star
1188: formation activity compared with our sample of radio galaxies.
1189:
1190: We can also estimate the rate of energetically significant starburst
1191: activity in our sample by considering the main optical and infrared
1192: indicators of starbursts: 7 (15\%) of the objects in the 2Jy sample
1193: show unambiguous spectroscopic evidence for recent star formation
1194: activity at optical wavelengths; 9 (20\%) have cool MFIR colours
1195: ($L_{70\mu m}/L_{24\mu m} > 5$); 12 (26\%) of the objects lie more
1196: than 0.3 dex ($\approx$factor$\times$2) about the regression line in
1197: the $L_{70\mu m}$ vs $L_{\rm{[OIII]}}$ correlation in Figure 6; and 13
1198: objects (28\%) show at least one of these indicators. Therefore we
1199: estimate that the proportion of powerful radio-loud AGN showing
1200: evidence for energetically significant recent star formation activity
1201: is in the range 15-28\%.
1202:
1203: As first described in \citet{tadhunter07}, this brings us to a key
1204: result: given the lack of a correlation between starburst and AGN
1205: activity, and the fact that only a minority of objects in our sample
1206: show any evidence for recent star formation activity, it is unlikely
1207: that all powerful radio galaxies are triggered at the peaks of major
1208: gas rich mergers. This conclusion is strengthened by the fact that not
1209: all powerful radio galaxies show tidal tails or other evidence for
1210: recent galaxy merger activity (e.g \citealp{heckman86},
1211: \citealp{tadhunter89}, \citealp{mclure99}).
1212:
1213: \subsection{Energetic feasibility}
1214:
1215: Having found preliminary evidence, based on the correlations between
1216: MFIR and [OIII] luminosities, that the MFIR emitting dust is
1217: predominantly heated by AGN, it is important to assess whether this
1218: heating mechanism is energetically feasible. We start by assuming a
1219: simple model in which the far-IR continuum, mid-IR continuum and
1220: [OIII] emission lines are produced by AGN illumination of structures
1221: with covering factors $C_{fir}$, $C_{mir}$ and $C_{nlr}$
1222: respectively. We make no assumptions about the radial distribution of
1223: dust, although the far-IR emitting dust must be situated at larger
1224: radial distances from the AGN than the mid-IR emitting dust in order
1225: to produce its cooler temperature.
1226:
1227: Based on simple recombination theory, the total $H\beta$ luminosity
1228: ($L_{H\beta}$), generated in the narrow line region (NLR) of an AGN,
1229: is related to the ionising luminosity ($L_{ion}$) by
1230:
1231: \begin{displaymath}
1232: {L_{ion}} = L_{H\beta}\frac{\langle h\nu \rangle_{ion}}{h\nu_{H\beta}} \frac {\alpha^{B}_{eff}}{\alpha_{H\beta}^{B}} C_{nlr}^{-1}
1233: \end{displaymath}
1234:
1235: \noindent where $\langle h\nu \rangle_{ion}$ is the mean ionising
1236: photon energy, $h\nu_{H\beta}$ is the energy of an $H\beta$ photon,
1237: $\alpha^{B}_{eff}$ is the total case B recombination coefficient and
1238: $\alpha_{H\beta}^{B}$ is the effective $H\beta$ recombination coefficient. From
1239: the results of \citet{elvis94} the bolometric luminosity $L_{bol}$ is
1240: related to the ionising luminosity by $ L_{bol} \approx 3.1 L_{ion}$,
1241: therefore
1242:
1243: \begin{displaymath}
1244: {L_{H\beta}} = 0.32 L_{bol}\frac{h\nu_{H\beta}}{\langle h\nu \rangle_{ion}} \frac {\alpha_{H\beta}^{B}}{\alpha^{B}_{eff}} C_{nlr}.
1245: \end{displaymath}
1246:
1247: At high luminosities and ionization parameters, $L_{[\rm{OIII}]}
1248: \approx 12 L_{H\beta}$. In addition, fits to the line ratios of nearby
1249: radio galaxies are consistent with $\langle h\nu \rangle_{ion}
1250: $=$6.2\times 10^{-11}J$, corresponding to an ionising continuum shape
1251: with $\beta = 1.5$ ($F_{\nu} \propto \nu^{-\beta}$)
1252: \citep{robinson87}. Also, given that $\alpha^B_{H\beta}/\alpha_{eff} = 0.09$
1253: \citep{osterbrock}, we find:
1254:
1255: \begin{equation}
1256: L_{[\rm{OIII}]} = 2.2 \times 10^{-2} L_{bol} C_{nlr}.
1257: \end{equation}
1258:
1259: \noindent This result can be related to the MFIR luminosities by assuming that the
1260: structures producing the MFIR absorb the short wavelength radiation
1261: and re-radiate it in the MFIR, then $L^{bol}_{mir} = C_{mir}L_{bol}$,
1262: and similarly $L^{bol}_{fir} = C_{fir}L_{bol}$. Defining the
1263: wavelength range of the mid-IR as 2 -- 30\moo, and that of the far-IR
1264: as 30 -- 100\moo, the mid-IR and far-IR bolometric luminosities are
1265: defined as:
1266:
1267: \begin{displaymath}
1268: L_{mir}^{bol} = \int_{1\times10^{13}Hz}^{1.5\times10^{14}Hz} L_{\nu} d\nu
1269: \end{displaymath}
1270:
1271: and
1272:
1273: \begin{displaymath}
1274: L_{fir}^{bol} = \int_{3\times10^{12}Hz}^{1\times10^{13}Hz} L_{\nu} d\nu.
1275: \end{displaymath}
1276:
1277: Representing $L_{\nu}$ as a power law $L_{\nu}=K\nu^{\gamma}$, we can then write
1278: \begin{displaymath}
1279: L_{mir}^{bol} = \int_{1\times10^{13}Hz}^{1.5\times10^{14}Hz} K\nu^{\gamma} d\nu
1280: \end{displaymath}
1281:
1282: \noindent and
1283:
1284: \begin{displaymath}
1285: K_{mir} = \frac{(\gamma+1)L_{bol}C_{mir}}{[\nu^{\gamma+1}]_{1\times10^{13}Hz}^{1.5\times10^{14}Hz}},
1286: \end{displaymath}
1287:
1288: \noindent and we can write a similar expression for the
1289: $K_{fir}$. Given our assumption of a simple power-law shape for the MFIR SED, the
1290: MFIR spectral index is directly calculated from the MFIR colour:
1291:
1292: \begin{displaymath}
1293: \gamma=\frac{log (L_{70}/ L_{24})}{log(24/70)}.
1294: \end{displaymath}
1295:
1296: \noindent For our complete sample we find a median MFIR colour of
1297: $L_{70\mu m}/L_{24\mu m}=2.1$, leading to $\gamma_{median}\approx
1298: 0.7$. Therefore the MFIR monochromatic luminosities can be related to
1299: the AGN bolometric luminosities by:
1300:
1301: \begin{displaymath}
1302: L_{mir}= 3\times10^{-5} L_{bol} C_{mir} \nu^{-0.7}
1303: \end{displaymath}
1304: and
1305: \begin{displaymath}
1306: L_{fir}= 1.25\times10^{-4} L_{bol} C_{fir} \nu^{-0.7}.
1307: \end{displaymath}
1308:
1309: By substituting equation (1) into these expressions we find:
1310:
1311: \begin{equation}
1312: L_{(24\mu m)}=1\times10^{-12} L_{[\rm{OIII}]} \frac{C_{mir}}{C_{nlr}}
1313: \end{equation}
1314: and
1315: \begin{equation}
1316: L_{(70\mu m)}=9\times10^{-12} L_{[\rm{OIII}]} \frac{C_{fir}}{C_{nlr}}.
1317: \end{equation}
1318:
1319:
1320: \begin{figure*}[t]
1321: \epsscale{2}
1322: \plotone{f8.eps}
1323: \caption{Plots of the $L_{[\rm{OIII}]}$ vs.$L_{24\mu}$ (left) and
1324: $L_{[\rm{OIII}]}$ vs.$L_{70\mu}$ (right) for the complete 2Jy sample
1325: with lines showing the range of possible ratios of covering factors
1326: required to explain the trends in terms of the AGN illumination
1327: model. $C_{mir}/C_{nlr}$ in plot (a) and $C_{fir}/C_{nlr}$ in plot
1328: (b). Note our calculation of the covering factors does not seek to
1329: explain the non-linear slope (discussed further in \S
1330: \ref{sec:slopes}). Also the ranges of covering factor ratios plotted
1331: are different in plots (a) and (b). The arrow in the upper left hand
1332: corners of the plots show the effect that correcting for an intrinsic
1333: extinction of E(B-V) = 0.3 would have on the [OIII] emission line
1334: luminosities.
1335: \label{fig:covering}}
1336: \end{figure*}
1337:
1338:
1339: \noindent It is likely that the major uncertainty in $C_{mir}/C_{nlr}$
1340: is associated with the assumed $ L_{bol}/L_{ion}$ ratio that may be
1341: uncertain by a factor of $\approx 2$ \citep{elvis94}.
1342:
1343: In Figure \ref{fig:covering} we have plotted the $L_{70\mu m}$ vs
1344: $L_{\rm{[OIII]}}$ and $L_{24\mu m}$ vs $L_{\rm{[OIII]}}$ correlations
1345: with predictions from our calculations for a range of
1346: $C_{mir}/C_{nlr}$ and $C_{fir}/C_{nlr}$ covering factor ratios. This
1347: allows us to investigate whether the MFIR emission can be explained
1348: solely in terms of AGN illumination of the dust. Note that our calculations
1349: do not take into account the fact that the $L_{MFIR}$ vs $
1350: L_{[\rm{OIII}]}$ relationships are non-linear (discussed further in \S
1351: \ref{sec:slopes}); the calculated ratio slopes are most relevant to
1352: the high luminosity ends of the correlations. Also, in the calculation
1353: so far we have assumed that the NLR, the mid-IR and far-IR emitting
1354: regions represent discrete structures at different locations in each
1355: galaxy, whereas, in fact, it is more likely that there exists a
1356: continuous distribution from one region to the next, or an overlap
1357: between the regions.
1358:
1359: Firstly, Figure \ref{fig:covering}(a) shows that $C_{mir}/C_{nlr}
1360: \approx$ 5-25 would account for the mid-IR luminosities of
1361: the majority of the objects in our sample. However, a value of
1362: $C_{mir}/C_{nlr} \approx 12.5$ provides the best description of the
1363: high luminosity end of the main correlation (from Figure
1364: \ref{fig:starburst}). Therefore NLR clouds
1365: cannot be responsible for the bulk of the mid-IR emission of the radio
1366: galaxies in our sample, a result consistent with detailed models for the mid-IR
1367: emission from NLR clouds \citep{groves06}. We further note that much
1368: of the scatter in the $L_{24\mu m}$ vs $L_{\rm{[OIII]}}$ correlation
1369: can be explained in terms of variations in $C_{mir}/C_{nlr}$.
1370:
1371: The absolute covering factor of the dusty torus structure emitting the
1372: mid-IR continuum can be estimated by assuming that the NLR has a
1373: covering factor in the range $C_{nlr} \approx $ 0.02-0.08
1374: (\citealp{netzer93}, \citealp{maiolino01}). For $C_{mir}/C_{nlr}
1375: \approx 12.5$ this gives $0.25<C_{mir}<1$. A large $C_{mir}$ is
1376: entirely feasible in the context of obscuration by the putative dusty
1377: torus required by the orientation-based unified schemes
1378: \citep{barthel89}, since a torus with a typical opening half angle in the
1379: range $45^{\circ}-60^{\circ}$ would have a covering factor of
1380: 0.5-0.7. This calculation also indicates that the covering factor of
1381: the NLR must be small ($C_{nlr}<0.08$) if we require
1382: $C_{mir}<$1. Therefore, it is evident that the 24\moo\ continuum
1383: emission from the warm dust component can be feasibly powered by AGN
1384: heating alone for most of the objects; the relative covering factor of
1385: the mid-IR emitting structures is consistent with that expected for a
1386: circum-nuclear torus, which is not itself expected to radiate
1387: significant forbidden line radiation due to high densities and
1388: temperatures.
1389:
1390: Secondly, it is clear from Figure \ref{fig:covering}(b) that the ratio
1391: of covering factors required for the emitting dust structures is much
1392: less in the far-IR than at mid-IR wavelengths: typically
1393: $C_{fir}/C_{nlr} \approx 1-2 $ for the main non-starburst sample,
1394: corresponding to $C_{fir} = 0.02-0.16$. Such covering factors are
1395: feasible if cool dust is associated with the outer parts of the torus,
1396: a kpc-scale dust lane such as those found in many nearby radio
1397: galaxies \citep{dekoff00}, or indeed the NLR itself. In the latter
1398: context we note that, if we correct the emission line luminosities for
1399: the reddening typical of the NLR of radio galaxies
1400: (0.3$<$E(B-V)$<$1.0: \citealp{tadhunter94}, \citealp{robinson00},
1401: \citealp{robinson_thesis})\footnote{These reddening estimates are based on
1402: Balmer line ratio measurements obtained after careful subtraction of
1403: the underlying stellar continuum.} this will shift many of the points
1404: in Figure \ref{fig:covering}(b) close to the $C_{fir}/C_{nlr}=1$
1405: line. Moreover, the NLR in many nearby radio galaxies and PG quasars
1406: are extended on sufficiently large spatial scales ($\sim$1kpc or
1407: larger: \citealp{bennert02}, \citealp{privon08}) that the dust
1408: associated with them would be heated by the AGN to the relatively cool
1409: temperatures required to produce the far-IR radiation. Therefore,
1410: rather than being separate entities, it is entirely plausible that the
1411: NLR clouds \emph{are} the dusty structures emitting the far-IR
1412: continuum.
1413:
1414: We have again marked the starburst objects from our sample in Figure
1415: \ref{fig:covering}. Many of these starburst objects require
1416: $C_{fir}/C_{nlr} \approx 10-50$. If we attempt to explain the far-IR
1417: emission in such objects in terms of AGN illumination alone, this
1418: raises the question of how the AGN could illuminate the far-IR
1419: emitting regions without producing substantial emission line radiation
1420: from the gas associated with the dusty structures (i.e. leading to
1421: $C_{fir}/C_{nlr} = 1 $). One explanation is that the $C_{nlr}$ is
1422: underestimated, and $C_{fir}/C_{nlr}$ overestimated, due to dust
1423: extinction of the NLR. However, this would require more substantial
1424: reddening (E(B-V) $>$ 1) than is supported by most observations of the
1425: NLR. Therefore the objects that lie above the $C_{fir}/C_{nlr}$= 1 line
1426: are likely to have their far-IR flux boosted by starburst heating or
1427: non-thermal emission. Incidentally, this includes \object{PKS0347+05}
1428: and \object{PKS1306-09}, which we discussed in \S \ref{sec:looking} as
1429: possible starburst candidates.
1430:
1431: \subsection{Correlation slopes}
1432: \label{sec:slopes}
1433: The fitted regression lines from Table \ref{tbl-4} (0.75 $\pm0.05$ at
1434: 24\moo\ and 0.72 $\pm0.07$ at 70\moo) are plotted on Figure
1435: \ref{fig:starburst}. Taking into account the bootstrap uncertainties,
1436: the slopes of the correlations are significantly non-linear. It is
1437: notable that \citet{maiolino07} find a similar slope (0.82$\pm0.02$)
1438: between continuum luminosity at 5100A and infrared luminosity at
1439: 6.7\moo\ for a sample of 25 high luminosity QSOs. We now consider
1440: possible explanations for the non-linear slopes.
1441:
1442: First we consider explanations related to the emission line physics of
1443: the NLR clouds. The most basic AGN photoionization models assume that
1444: the emission-line regions are photoionised by the AGN and that the
1445: properties of the emission line region, such as density and covering
1446: factor, do not change significantly with AGN power. Considering such a
1447: model for a simple optically thick slab, \citet{tadhunter98} found
1448: that at high luminosities, the relation between $L_{\rm{[OIII]}}$ and
1449: ionising luminosity can be represented as a power-law $L_{ion} \propto
1450: (L_{\rm{[OIII]}})^{\alpha}$, where $\alpha = 0.8\pm0.1$. Assuming that
1451: the properties of the torus do not change with luminosity (but see
1452: below) and that the MFIR continuum luminosity is directly proportional
1453: to $L_{bol}$ and $L_{ion}$, we therefore expect $L_{MFIR}\propto
1454: L_{\rm{[OIII]}}^{\alpha} \propto L_{\rm{[OIII]}}^{0.8\pm0.1} $. This
1455: result is consistent with the slopes of our $L_{MFIR}$ vs
1456: $L_{\rm{[OIII]}}$ correlations as well as that found by
1457: \citet{maiolino07}. However, based on several detailed studies of the
1458: emission line spectra of radio galaxies (e.g. \citealp{robinson00},
1459: \citealp{taylor03}), it is likely that the single slab
1460: photoionization models are over-simplistic, and that multi-component
1461: models for the NLR are more appropriate. In this case the relationship
1462: between $L_{\rm{[OIII]}}$ and $L_{ion}$ is less clear. Moreover,
1463: investigations of the relationship between $L_{\rm{[OIII]}}$ and the
1464: continuum luminosities in large samples of AGN provide direct
1465: observational evidence for steeper slopes in the relationship between
1466: $L_{bol}$ and $L_{\rm{[OIII]}}$, albeit with larger
1467: scatter ($\alpha \geq 1$ \citealp{netzer04, netzer06}).
1468:
1469: Alternatively, we can also consider whether the non-linear slopes are
1470: consistent with the basic receding torus model that has been suggested
1471: to explain the variation in the BLRG+Q/NLRG fraction with luminosity
1472: in the context of the unified schemes (\citealp{lawrence91}, \citealp{hill96},
1473: \citealp{simpson98, simpson05}). This solution is attractive because it is widely
1474: accepted that, if the mid-IR emitting dust lies close to the AGN, the
1475: inner radius of the torus at which the dust is sublimated due to
1476: illumination by the AGN must depend on AGN power. From simple thermal
1477: equilibrium arguments we can relate the distance of the illuminating
1478: AGN from the inner face of the torus (r) to the bolometric luminosity
1479: ($L_{bol}$) and the sublimation temperature of the dust ($T_{sub}$) as
1480: follows:
1481:
1482: \begin{equation}
1483: r \propto \left(\frac{L_{bol}}{T_{sub}^4}\right)^{1/2}.
1484: \end{equation}
1485:
1486: The area of the inner face of the obscuring torus can be written as $
1487: 4\pi rh$ where $h$ is the thickness of the torus from its mid-plane to its
1488: top surface. The covering fraction of the torus is then:
1489:
1490: \begin{equation}
1491: C_{tor} = \frac{4\pi rh}{4\pi r^2} = \frac{h}{r},
1492: \end{equation}
1493:
1494: assuming all the radiation absorbed by the face of the torus is
1495: re-emitted by dust, and $h$ is fixed, then the relationship between
1496: emission in the mid-IR at 24\moo\ emitted by dust close to the AGN and
1497: the bolometric luminosity can be written as:
1498:
1499: \begin{displaymath}
1500: P_{24\mu m} \propto \frac{h}{r}\, L_{bol} \propto L_{bol}^{1/2}.
1501: \end{displaymath}
1502:
1503: \noindent having substituted equation (5) into equation (4). Finally, assuming that $L_{bol} \propto
1504: \left(L_{[\rm{OIII}]}\right)^{\alpha}$ we obtain:
1505:
1506: \begin{displaymath}
1507: P_{24\mu m} \propto \left(L_{[\rm{OIII}]}\right)^{\alpha/2}.
1508: \end{displaymath}
1509:
1510: In the context of the receding torus model we now consider three cases with different
1511: relationships between $L_{\rm{[OIII]}}$ and $L_{bol}$:
1512:
1513: \begin{enumerate}
1514:
1515: \item {\emph{The relationship between $L_{\rm{[OIII]}}$ and $L_{bol}$ is
1516: linear with $\alpha$=1, and therefore the non-linear slopes of
1517: $L_{\rm{[OIII]}}$ vs $L_{MFIR}$ correlations are entirely due to the
1518: receding torus.} In this case $P_{24\mu m} \propto
1519: \left(L_{[\rm{OIII}]}\right)^{0.5}$ which is too shallow to account
1520: for the measured correlation slopes, even when considering the
1521: uncertainties on the measured slope we have estimated from the
1522: bootstrap analysis ($\pm$ 0.05). \citet{netzer04} found evidence for
1523: such a linear relationship between $L_{\rm{[OIII]}}$ and $L_{AGN}$,
1524: from the lack of variation of the equivalent width of [OIII] as a function of
1525: continuum luminosity. However, there are many upper limits for the
1526: high luminosity objects in their sample. }
1527:
1528: \item{\emph{The correlation between $L_{\rm{[OIII]}}$ and $L_{bol}$ is not
1529: linear and has a slope $\alpha<1$, as predicted by simple AGN
1530: photoionization models \citep{tadhunter98}}. If $\alpha$=0.8, as
1531: discussed above, then $P_{24\mu m} \propto
1532: \left(L_{[\rm{OIII}]}\right)^{0.4}$. This value is also significantly
1533: shallower than measured from our data.}
1534:
1535: \item{\emph{The correlation between $L_{\rm{[OIII]}}$ and $L_{AGN}$ is
1536: non-linear and $\alpha > 1$}. In this context it is notable that
1537: \citet{netzer06} find evidence for $\alpha \approx 1.7\pm0.07$, based
1538: on an analysis of X-ray and emission line properties of luminous
1539: AGN. This then gives $P_{24\mu m} \propto
1540: \left(L_{[\rm{OIII}]}\right)^{0.85\pm0.04} $, which is consistent with the
1541: measured slopes within the uncertainties. }
1542:
1543: \end{enumerate}
1544:
1545: Option 1, considering only a receding torus model, cannot
1546: satisfactorily explain our results. \citet{simpson05} also shows that,
1547: assuming isotropic [OIII] emission, the fraction of BLRG/Q increases
1548: with AGN luminosity, but with a shallower slope than predicted by a
1549: pure receding torus model. He proposes an increase in torus height
1550: with luminosity to account for the discrepancy. Although this may not
1551: be the only explanation for the observed trends, the \citet{simpson05}
1552: result shows that there remain questions about the application of the
1553: most basic receding torus model.
1554:
1555: In this study we have shown that the slope of the $L_{70\mu m}$ vs
1556: $L_{\rm{[OIII]}}$ correlation is similar to that of the $L_{24\mu m}$
1557: vs $L_{\rm{[OIII]}}$ correlation. If the far-IR radiation is emitted
1558: by extended cool dust structures it is clear that these must lie far
1559: from the AGN for the typical luminosities of our sources
1560: ($>0.1$kpc). In this case it is unlikely that the dust structures will
1561: recede in the same way as predicted for the torus. Therefore, while a
1562: receding torus might provide a plausible explanation for the
1563: non-linearity of the $L_{24\mu m}$ vs $L_{\rm{[OIII]}}$ correlation
1564: (for $\alpha>1$) such an explanation is unlikely to apply to the
1565: $L_{70\mu m}$ vs $L_{\rm{[OIII]}}$ correlation.
1566:
1567: \subsection{Unification}
1568: \label{sec:unify_disc}
1569:
1570: Given the completeness of our sample, the similarities between the
1571: MFIR properties of the NLRG and BLRG/Q classes provide excellent
1572: support for the orientation-based unified schemes for radio-loud
1573: AGN. In addition, our results provide an important constraint for
1574: understanding the structure and optical depth of the obscuring
1575: torus. The question then remains: how do we reconcile our results with
1576: previous studies that did find evidence for significant obscuration at
1577: MFIR wavelengths? Apart from sample selection and incompleteness (see
1578: \S \ref{sec:intro}), there are several factors that may have biased
1579: previous studies and led to the finding of a difference between the
1580: two classes of radio-loud AGN:
1581:
1582: \begin{itemize}
1583:
1584: \item{{\bf Classification of BLRG objects}: The deep spectra available
1585: for all our sample objects allow us to make secure classifications of
1586: objects into NLRG, BLRG/Q and WLRG categories. Past studies may have
1587: only included those objects with the most luminous broad lines;
1588: objects with weaker broad lines may have been missed. Therefore,
1589: excluding the weak BLRG from the broad line samples would have biased
1590: BLRG/Q towards higher luminosities compared with the NLRG.}
1591:
1592: \item{{\bf Weak line radio galaxies}: From Figure \ref{fig:opradall}
1593: (a) \& (b), it is clear that the WLRG lie at the low luminosity end of
1594: both the MFIR bands. Many previous studies have not made the
1595: distinction between NLRG and WLRG objects. In not recognising these
1596: radio-loud galaxies as a separate population, it is clear that the
1597: inclusion of WLRG with NLRG would lower the mean MFIR luminosity of
1598: samples of NLRG/WLRG, making the BLRG/Q appear relatively stronger
1599: emitters at MFIR wavelengths (see \citealp{laing94} for a a similar
1600: discussion relating to the [OIII] luminosities).}
1601:
1602: \item{{\bf Beamed non-thermal MFIR components}: When considering the
1603: MFIR emission from radio-loud galaxies, it is important to quantify
1604: the degree of non-thermal contamination of the thermal infrared
1605: emission from dust. Past investigations that did not account for such
1606: contamination (e.g. including flat spectrum objects) may have suffered
1607: a bias in their results. Contamination from a beamed non-thermal
1608: component is more likely in BLRG/Q oriented close to the line of
1609: sight, and therefore could have boosted the MFIR luminosities of some
1610: of the BLRG/Q objects in previous studies. As mentioned in Section
1611: \ref{sec:OpClass} we find no more than 24\% of our 2Jy sample have a
1612: possibility of contamination from non-thermal beamed components (see
1613: \citet{dicken08}), consistent with the results presented in
1614: \citet{cleary07}.}
1615: \end{itemize}
1616:
1617: Although the results of our study support the unified schemes, further
1618: data are required at shorter wavelengths ($\lambda < 24 \mu m$) to
1619: provide necessary evidence for greater optical depth in NLRG compared
1620: with BLRG/Q objects. Additionally, our results have been based on the
1621: assumption that the [OIII] emission line luminosity is isotropic, and
1622: it is important to consider whether this assumption is valid.
1623:
1624:
1625: \begin{figure}[t]
1626: \epsscale{1}
1627: \plotone{f9.eps}
1628: \caption{Plots of [OIII]$\lambda$5007 and [OII]$\lambda$3727 vs 5 \rm{GHz} radio total radio
1629: luminosity. Not including 4 upper limits in [OIII] and 5 upper
1630: limits in [OII] \label{fig:opOIIIRad}}
1631: \end{figure}
1632:
1633: Previous studies have presented evidence that quasars
1634: have higher [OIII]$\lambda$5007 luminosities than radio galaxies
1635: (\citealp{jackson90}, \citealp{haas05}), similar to the results in
1636: opposition to BLRG/Q and NLRG unification in the MFIR
1637: (\citealp{heckman94}, \citealp{hes95}). In fact, returning to Figure
1638: \ref{fig:opradall}, it is notable that there is a weak tendency for
1639: the BLRG/Q sources to have higher [OIII]$\lambda$5007 luminosities
1640: than the NLRG sources for a given MFIR luminosity. On the other hand, \citet{hes93, hes96}
1641: demonstrate that the [OII]$\lambda3727$ line, with its lower
1642: ionization potential and critical density, shows no evidence for a
1643: difference in luminosity between BLRG/Q and NLRG.
1644:
1645:
1646: \begin{figure}[t]
1647: \epsscale{1}
1648: \plotone{f10.eps}
1649: \caption{Plots of 24\moo\ (top) and 70\moo\ (bottom) luminosities normalized by [OIII]$\lambda$5007 luminosity vs 5 \rm{GHz} radio total radio
1650: luminosity. \label{fig:Rdr}}
1651: \end{figure}
1652:
1653:
1654: In Figure \ref{fig:opOIIIRad} we present $L_{\rm{[OIII]}\lambda5007}$
1655: and $L_{\rm{[OII]}\lambda3727}$ plotted against total radio luminosity
1656: ($L{5GHz}$). A visual inspection of this figure reveals that the
1657: distributions of the NLRG and BLRG/Q are similar across several orders
1658: of magnitudes. Despite the BLRG/Q having some of the highest
1659: luminosities in [\rm{OIII}] emission lines, there are no significant
1660: differences between the two populations in this plot. This is
1661: consistent with the findings of \citet{laing94}, \citet{mulchaey94},
1662: \citet{tadhunter98} and \citet{jackson97}.
1663:
1664: In Figure \ref{fig:Rdr} we again present $L_{24\mu}$ and $L_{70\mu}$
1665: plotted against $L_{5\rm{GHz}}$, however in this Figure we have
1666: normalized the MFIR luminosities by $L_{\rm{[OIII]}\lambda5007}$. If
1667: we make the assumption that both the [OIII] and MFIR emission are
1668: isotropic then normalising the MFIR luminosity by
1669: $L_{\rm{[OIII]}\lambda5007}$ should reveal no difference in the
1670: distribution of the different optical classifications in the
1671: plot. However, from a visual inspection of Figure \ref{fig:Rdr}, it is
1672: clear that, although there is a large scatter, the BLRG/Q objects
1673: clearly tend towards lower MFIR/[OIII] ratios than NLRG; the three
1674: lowest $L_{24\mu}$/$L_{\rm{[OIII]}\lambda5007}$ ratios, and six lowest
1675: $L_{70\mu}$/$L_{\rm{[OIII]}\lambda5007}$ ratios are associated with
1676: BLRG/Q. The BLRG/Q clearly tend to lower normalised MFIR luminosities
1677: than the NLRG. There are three explanations for such a result.
1678:
1679: \begin{itemize}
1680:
1681: \item{ The [OIII] emission is isotropic, but the MFIR emission is
1682: weaker for a given [OIII] luminosity in the BLRG/Q objects than it is
1683: in the NLRG objects. This explanation is difficult to square with the
1684: orientation-based unified schemes. In particular, the optical [OIII]
1685: emission is more likely to suffer dust obscuration effects due to the
1686: circumnuclear torus than the MFIR emission. Moreover, assuming that
1687: their jets are truly pointing closer to the line of sight, the BLRG/Q
1688: are less likely to be affected by dust obscuration than the NLRG.
1689: Note, however, that it is possible that some individual BLRG/Q may
1690: have intrinsically under-luminous MFIR emission (e.g. PKS1932-46: see
1691: \citealp{inskip07}).}
1692:
1693: \item{ The [OIII] emission is affected by torus dust obscuration and
1694: therefore appears stronger in the BLRG/Q than it is in the NLRG for a
1695: given isotropic MFIR luminosity. This explanation is certainly
1696: consistent with the orientation-based unified schemes, and the
1697: conclusions of various other studies (e.g. \citealp{jackson97},
1698: \citealp{haas05}). It is also consistent with the observational
1699: evidence for narrow line variability on a timescale of years in the
1700: BLRG 3C390.3 (\citealp{clavel87}, \citealp{zheng95}), which suggests
1701: that a proportion of the narrow [OIII] emission may be emitted on a
1702: sufficiently small scale to be affected by torus obscuration. On the
1703: basis of Figure 10, the degree of anisotropy in [OIII] is a factor
1704: $\sim$2 -- 3 for the majority of the objects, only reaching a factor
1705: of 10 in a few extreme cases.}
1706:
1707: \item{The [OIII] emission is isotropic but the
1708: $L_{MFIR}/L_{\rm{[OIII]}}$ ratio decreases with luminosity because the
1709: covering factor of the torus decreases as the torus recedes. The
1710: receding torus model predicts that the BLRG/Q objects will have higher
1711: [OIII] luminosities on average than NLRG, even in the absence of dust
1712: obscuration, provided that there exists a range of intrinsic AGN
1713: luminosities for a given isotropic radio power. This is because the
1714: objects hosting the most intrinsically powerful AGN have a greater
1715: chance of being detected as BLRG/Q. As the inner face of the torus
1716: recedes with the luminosity the covering factor of the torus will
1717: decrease. This will lead to the following dependence of the
1718: $L_{MFIR}/L_{\rm{[OIII]}}$ ratio on $L_{bol}$ at fixed radio power:
1719: $L_{MFIR}/L_{\rm{[OIII]}} \propto L^{1/2}_{bol}$ (See Section
1720: \ref{sec:slopes}) assuming a linear relationship between
1721: $L_{\rm{[OIII]}}$ and $L_{bol}$. Therefore in order to explain the
1722: BLRG/Q $L_{MFIR}/L_{\rm{[OIII]}}$ ratios that are a factor $\approx3\times$
1723: lower than the average NLRG, we would require the BLRG/Q to be
1724: intrinsically more luminous than the NLRG at a given radio power by a
1725: factor of $\approx 10$. However as discussed in Section
1726: \ref{sec:slopes} it is unlikely that this explanation can apply to the
1727: $L_{70\mu m}/L_{\rm{[OIII]}}$ ratio because the structures emitting
1728: the far-IR continuum are unlikely to recede as envisaged by the
1729: receding torus model.}
1730: \end{itemize}
1731:
1732:
1733: \section{Summary}
1734:
1735: We have presented an analysis of deep Spitzer (MIPS) observations for
1736: a complete subset of the 2Jy sample. The main results are as follows:
1737:
1738: \begin{enumerate}
1739: \item{{\bf Heating mechanism}: We find tight correlations between the
1740: MFIR luminosities and the [\rm{OIII}]$\lambda5007$ luminosities, with
1741: similar slopes for both 24 and 70\moo\ correlations (0.75$\pm0.05$ at
1742: 24\moo\ and 0.72$\pm0.07$ at 70\moo). Given that [\rm{OIII}] is an
1743: indicator of intrinsic AGN strength, we conclude that direct AGN
1744: illumination is the primary heating source for the dust producing both
1745: the mid-IR and far-IR continuum. These correlations are better than
1746: those between MFIR and radio luminosities, and between radio and
1747: [OIII] luminosities. }
1748:
1749: \item{{\bf Energetics and dust geometry}: Using simple
1750: arguments we have demonstrated that heating of MFIR-emitting dust
1751: structures by AGN illumination is energetically feasible. We identify
1752: the dust structure producing the mid-IR continuum with the
1753: circum-nuclear torus, and the dust structure producing far-IR
1754: continuum with the NLR clouds. }
1755:
1756: \item{{\bf Slopes of the correlations}: Our finding of a non-linear slope in
1757: the $L_{MFIR}$ vs. $L_{[OIII]}$ correlation is quantitatively in support of the
1758: simple receding torus model; however, this result can also be explained
1759: in terms of a non-linear correlation between the strength of [OIII]
1760: optical emission line luminosity and the AGN ionising
1761: luminosity.}
1762:
1763: \item{{\bf Unified schemes}: Contrary to some previous studies we find
1764: no strong evidence for increased obscuration of MFIR emission of NLRG
1765: compared to BLRG/Q; the MFIR luminosities of NLRG and BLRG/Q also cover
1766: a similar range. These results fully support the orientation-based
1767: unified schemes, and we hypothesise that previous studies may have been affected
1768: by incomplete, heterogeneous, meager and/or low S/N data.}
1769:
1770: \item{{\bf Weak Line radio galaxies}: The WLRG in
1771: our sample are found to have weak MFIR continuum emission as well as
1772: weak [OIII] emission line emission. This implies that the AGN in WLRGs
1773: are intrinsically low luminosity objects, and
1774: that their weak optical emission lines are not simply a consequence of
1775: enhanced dust obscuration by circum-nuclear dust.}
1776:
1777: \item{{\bf Starburst heating and triggering}: We have analysed the
1778: increased scatter in the distribution of $L_{70\mu m}$
1779: vs. $L_{[\rm{OIII}]}$ compared to the $L_{24\mu m}$
1780: vs. $L_{[\rm{OIII}]}$ correlation. We find that the increased scatter
1781: is due, in part, to enhanced heating by starbursts in the objects in
1782: our sample that show evidence for recent star formation activity at
1783: optical wavelengths. The relatively low incidence of energetically
1784: significant starburst activity in our sample (15-28\%) has
1785: implications for our understanding of the triggering and evolution of
1786: radio-loud AGN. In particular, only a minority are likely to be
1787: triggered close to the peak of major gas rich mergers with associated
1788: starburst activity. Clearly, a variety of triggering mechanisms must
1789: be present.}
1790:
1791: \item{{\bf Anisotropy of [OIII] emission}: Finding lower ratios of
1792: MFIR/[OIII] of BLRG/Q compared to those of NLRG, we have concluded
1793: that the most likely explanation for this result is that there is mild
1794: anisotropy (typically factor $\approx$2-3) in the [OIII] emission due
1795: to obscuration by the circumnuclear dust in the NLRG that is not seen
1796: in the BLRG/Q.}
1797:
1798: \end{enumerate}
1799:
1800: \acknowledgments
1801: This work is based [in part] on observations made
1802: with the Spitzer Space Telescope, which is operated by the Jet
1803: Propulsion Laboratory, California Institute of Technology under a
1804: contract with NASA. This research has made use of the NASA/IPAC
1805: Extragalactic Database (NED) which is operated by the Jet Propulsion
1806: Laboratory, California Institute of Technology, under contract with
1807: the National Aeronautics and Space Administration. Based on
1808: observations made with ESO Telescopes at the Paranal Observatory.
1809:
1810: DD, JH and KJI acknowledge support from the STFC. KJI acknowledges
1811: support from the German DFG under grant JA1114/3-1 within the Emmy
1812: Noether programme.
1813:
1814: {\it Facilities:} \facility{Spitzer (MIPS)}, \facility{ATCA}, \facility{VLA}.
1815:
1816: \bibliographystyle{apj.bst}
1817: \bibliography{bib_list}
1818:
1819: \end{document}
1820:
1821: