1: \documentstyle[11pt,aaspp4]{article}
2:
3: \newcommand{\noun}[1]{\textsc{#1}}
4:
5: \begin{document}
6:
7: \newcommand{\La}{\mathrm{Ly}\alpha }
8:
9: \newcommand{\Lalpha}{\mathrm{Ly}\alpha \, \lambda 1215}
10:
11: \newcommand{\Ka}{\mathrm{K}\alpha }
12:
13: \newcommand{\Lb}{\mathrm{L}\beta }
14:
15: \newcommand{\Ha}{\mathrm{H}\alpha }
16:
17: \newcommand{\Halpha}{\mathrm{H}\alpha \, \lambda 6563}
18:
19: \newcommand{\Hb}{\mathrm{H}\beta }
20:
21: \newcommand{\Hbeta}{\mathrm{H}\beta \, \lambda 4861}
22:
23: \newcommand{\Pa}{\mathrm{P}\alpha }
24:
25: \newcommand{\HeI}{\mathrm{He}\, {\textsc {i}}\, \lambda 5876}
26:
27: \newcommand{\HeII}{\mathrm{He}\, {\textsc {ii}}\, \lambda 1640}
28:
29: \newcommand{\CIII}{\mathrm{C}\, {\textsc {iii}}\, \lambda 977}
30:
31: \newcommand{\CIIIb}{\mathrm{C}\, {\textsc {iii]}}\, \lambda 1909}
32:
33: \newcommand{\CIV}{\mathrm{C}\, {\textsc {iv}}\, \lambda 1549}
34:
35: \newcommand{\bOIIIbA}{[\mathrm{O}\, {\textsc {iii]}}\, \lambda 4363}
36:
37: \newcommand{\bOIIIbB}{[\mathrm{O}\, \textsc {iii]}\, \lambda 5007}
38:
39: \newcommand{\OIIIb}{\mathrm{O}\, {\textsc {iii]}}\, \lambda 1663}
40:
41: \newcommand{\OVb}{\mathrm{O}\, {\textsc {v]}}\, \lambda 1218}
42:
43: \newcommand{\OVI}{\mathrm{O}\, {\textsc {vi}}\, \lambda 1035}
44:
45: \newcommand{\OIVb}{\mathrm{O}\, {\textsc {iv]}}\, \lambda 1402}
46:
47: \newcommand{\bOIIb}{[\mathrm{O}\, {\textsc {ii]}}\, \lambda 3727}
48:
49: \newcommand{\bOIb}{[\mathrm{O}\, {\textsc {i]}}\, \lambda 6300}
50:
51: \newcommand{\NV}{\mathrm{N}\, {\textsc {v}}\, \lambda 1240}
52:
53: \newcommand{\NIVb}{\mathrm{N}\, {\textsc {iv]}}\, \lambda 1486}
54:
55: \newcommand{\NIIIb}{\mathrm{N}\, {\textsc {iii]}}\, \lambda 1750}
56:
57: \newcommand{\MgII}{\mathrm{Mg}\, {\textsc {ii}}\, \lambda 2798}
58:
59: \newcommand{\bNeVb}{[\mathrm{Ne}\, {\textsc {v]}}\, \lambda 3426}
60:
61: \newcommand{\NeVIII}{\mathrm{Ne}\, {\textsc {viii}}\, \lambda 774}
62:
63: \newcommand{\SiIV}{\mathrm{Si}\, {\textsc {iv}}\, \lambda 1397}
64:
65: \newcommand{\bFeXb}{[\mathrm{Fe}\, {\textsc {x]}}\, \lambda 6734}
66:
67: \newcommand{\bFeXIb}{[\mathrm{Fe}\, {\textsc {xi]}}\, \lambda 7892}
68:
69: \newcommand{\FeII}{\mathrm{Fe}\, {\textsc {ii}}\, }
70:
71: \newcommand{\bSiVIIb}{[\mathrm{Si}\, {\textsc {vii]}\, \lambda 2.5\mu \mathrm{m}}}
72:
73: \newcommand{\bSiIXbA}{[\mathrm{Si}\, {\textsc {ix]}}\, \lambda 2.6\mu \mathrm{m}}
74:
75: \newcommand{\bSiIXbB}{[\mathrm{Si}\, {\textsc {ix]}}\, \lambda 3.9\mu \mathrm{m}}
76:
77: \newcommand{\bMgVIIIb}{[\mathrm{Mg}\, {\textsc {viii]}}\, \lambda 3.0\mu \mathrm{m}}
78:
79: \newcommand{\bMgIVb}{[\mathrm{Mg}\, {\textsc {iv]}}\, \lambda 4.5\mu \mathrm{m}}
80:
81: \newcommand{\bMgVIIb}{[\mathrm{Mg}\, {\textsc {vii]}\, \lambda 5.5\mu \mathrm{m}}}
82:
83: \newcommand{\bNeVIb}{[\mathrm{Ne}\, {\textsc {vi]}}\, \lambda 7.6\mu \mathrm{m}}
84:
85: \newcommand{\bArIIIb}{}
86:
87: \newcommand{\bArVIb}{[\mathrm{Ar}\, {\textsc {vi]}}\, \lambda 4.5\mu \mathrm{m}}
88:
89: \newcommand{\bSIVb}{[\mathrm{S}\, {\textsc {iv]}}\, \lambda 10.5\mu \mathrm{m}}
90:
91: \newcommand{\bNeIIb}{[\mathrm{Ne}\, {\textsc {ii]}}\, \lambda 12.8\mu \mathrm{m}}
92:
93: \newcommand{\bNeVbA}{[\mathrm{Ne}\, {\textsc {vi]}}\, \lambda 14.3\mu \mathrm{m}}
94:
95: \newcommand{\bNeIIIbA}{[\mathrm{Ne}\, {\textsc {iii]}}\, \lambda 15.6\mu \mathrm{m}}
96:
97: \newcommand{\bSIIIbA}{[\mathrm{S}\, {\textsc {iii]}}\, \lambda 18.7\mu \mathrm{m}}
98:
99: \newcommand{\bNeVbB}{[\mathrm{Ne}\, {\textsc {vi]}}\, \lambda 24.3\mu \mathrm{m}}
100:
101: \newcommand{\bOIVb}{[\mathrm{O}\, {\textsc {iv]}}\, \lambda 25.9\mu \mathrm{m}}
102:
103: \newcommand{\bSIIIbB}{[\mathrm{S}\, {\textsc {iii]}}\, \lambda 33.5\mu \mathrm{m}}
104:
105: \newcommand{\bSiIIb}{[\mathrm{Si}\, {\textsc {ii]}}\, \lambda 34.8\mu \mathrm{m}}
106:
107: \newcommand{\bNeIIIbB}{[\mathrm{Ne}\, {\textsc {iii]}}\, \lambda 36.0\mu \mathrm{m}}
108:
109: \newcommand{\EBV}{E_{\textsc {b-v}}}
110:
111: \newcommand{\Fnlr}{F_{\textsc {nlr}}}
112:
113: \newcommand{\lgs}{\log ^{2}S}
114:
115: \newcommand{\qion}{q_{\mathrm{ion}}}
116:
117: \newcommand{\tion}{\Delta \theta _{\mathrm{ion}}}
118:
119: \newcommand{\ml}{m_{\ell }}
120:
121: \newcommand{\fl}{f_{\ell }}
122:
123: \newcommand{\kl}{k_{\ell }}
124:
125: \newcommand{\Eion}{E_{\mathrm{ion}}}
126:
127: \newcommand{\maxs}{\max S_{\ell }}
128:
129: \newcommand{\Mo}{M_{\odot }}
130:
131: \newcommand{\Ro}{R_{\odot }}
132:
133: \newcommand{\Lo}{L_{\odot }}
134:
135: \newcommand{\NGC}{\mathrm{NGC}\, 1068}
136:
137: %\maketitle
138:
139: \title{Infrared spectroscopy of NGC~1068:\\
140: Probing the obscured ionizing AGN continuum}
141:
142: \author{Tal Alexander\altaffilmark{1}\\
143: Dieter Lutz\altaffilmark{2}, Eckhard Sturm\altaffilmark{2}, Reinhard Genzel\altaffilmark{2}\\
144: Amiel Sternberg\altaffilmark{3}, Hagai Netzer\altaffilmark{3}}
145:
146: \altaffiltext{1}{Institute for Advanced Study, Olden Lane, Princeton, NJ 08540,USA}
147:
148: \altaffiltext{2}{Max-Planck-Institut f\"{u}r Extraterrestrische Physik, Postfach 1603, D-85740 Garching, Germany}
149:
150: \altaffiltext{3}{School of Physics and Astronomy and Wise Observatory, Raymond and Beverly Sackler Faculty of Exact Sciences, Tel Aviv University, Ramat Aviv, Tel Aviv 69978, Israel}
151:
152: \begin{abstract}
153: The \emph{ISO-SWS}\footnote{%
154: Based on observations made with ISO, an ESA project with instruments funded
155: by ESA member states (especially the PI countries: France, Germany, The Netherlands
156: and the United Kingdom) and with the participation of ISAS and NASA. The SWS
157: is a joint project of SRON and MPE.
158: } \( 2.5 \)--\( 45\, \mu \mathrm{m} \) infrared spectroscopic observations
159: of the nucleus of the Seyfert 2 galaxy \( \NGC \) (see companion paper) are
160: combined with a compilation of UV to IR narrow emission line data to determine
161: the spectral energy distribution (SED) of the obscured extreme-UV continuum
162: that photoionizes the narrow line emitting gas in the active galactic nucleus.
163: We search a large grid of gas cloud models and SEDs for the combination that
164: best reproduces the observed line fluxes and NLR geometry. Our best fit model
165: reproduces the observed line fluxes to better than a factor of 2 on average
166: and is in general agreement with the observed NLR geometry. It has two gas components
167: that are consistent with a clumpy distribution of dense outflowing gas in the
168: center and a more extended distribution of less dense and more clumpy gas farther
169: out that has no net outflow. The best fit SED has a deep trough at \( \sim \! 4 \)
170: Ryd, which is consistent with an intrinsic Big Blue Bump that is partially absorbed
171: by \( \sim \! 6\times 10^{19}\, \mathrm{cm}^{-2} \) of neutral hydrogen interior
172: to the NLR.
173: \end{abstract}
174:
175: \section{Introduction}
176:
177: %\label{fig:temp}\label{fig:mix}\label{fig:ratios}\label{fig:abs_sed}\label{tbl:flux}\label{tbl:fit}
178:
179: The intrinsic spectral energy distribution (SED) of active galactic nuclei (AGN),
180: which extends from the radio up to \( \gamma \)-rays, cannot be directly observed
181: from the Lyman limit and up to several hundred eV due to Galactic and intrinsic
182: absorption. However, the extreme-UV (EUV) and soft X-ray continuum can be investigated
183: indirectly by the infrared coronal line emission. These lines are emitted by
184: collisionally excited forbidden fine-structure transitions of highly ionized
185: atoms, whose ionization potentials extend well beyond the Lyman limit up to
186: hundreds of eV. Unlike the strong permitted lines of these ions, which are also
187: emitted in the obscured EUV, the reddening-insensitive forbidden IR coronal
188: lines and semi-forbidden optical coronal lines can be observed. Therefore, when
189: photoionization is the main ionization mechanism, the coronal lines can provide
190: information on the intrinsic obscured SED and the accretion process that powers
191: the AGN. This information can be extracted by photoionization models of the
192: NLR.
193:
194: The coronal lines are collisionally suppressed in the dense broad line region
195: (BLR) close to the continuum source and are efficiently emitted only from the
196: more rarefied gas in the narrow line region (NLR), hundreds of pc away from
197: the center. It is well established that large quantities of gas attenuate the
198: continuum emission in many AGN. These gas clouds are detected by narrow UV absorption
199: lines (e.g. Kriss et al. \cite{Kriss92}) or by X-ray absorption features and
200: emission lines (e.g. George et al. \cite{George98}). Although their exact location
201: along the line of sight is unknown, there are reasons to believe that in some
202: cases they may be inside the NLR. In particular, the warm absorbers that block
203: the X-ray continuum appear to cover a large fraction of the continuum source
204: (George et al. \cite{George98}). This raises the possibility that in some AGN
205: the ionizing SED, which is traced by the coronal lines, is not the intrinsic
206: one produced by the accretion process, but rather one that is filtered by intervening
207: absorbers inside the NLR. It has been proposed that such absorbers are common
208: in Seyfert galaxies, and are responsible for the observed correlations between
209: the soft X-ray slope and the narrow emission line spectra of Seyfert 1.5 galaxies
210: (Kraemer, Ruiz \& Crenshaw \cite{KTCG99}).
211:
212: This study of the Seyfert 2 galaxy \( \NGC \) is part of the \emph{ISO-SWS}
213: program on bright galactic nuclei. Previous studies in this program include
214: the reconstruction of the SED of the Seyfert 2 Circinus galaxy (Moorwood et
215: al. \cite{Moorwood96}; Alexander et al. \cite{Alexander99}) and of the Seyfert
216: 1 Galaxy NGC~4151 (Alexander et al. \cite{Alexander99}). In both cases we found
217: evidence of a ``Big Blue Bump'' signature of a thin accretion disk (Shakura
218: \& Sunyaev \cite{SS73}). However, in the case of NGC~4151 this structure is
219: masked by a deep absorption trough of an absorber situated between the BLR and
220: the NLR, which filters the light that photoionizes the NLR.
221:
222: In this paper we apply our SED reconstruction method to \( \NGC \), one of
223: the closest, brightest and most extensively studied Seyfert 2 galaxies, which
224: is considered a prototype of this AGN class. The first detection of broad permitted
225: emission lines in the polarized light of \( \NGC \) (Antonucci \& Miller \cite{AM85})
226: provided a major argument for the Seyfert 1 and 2 unification scheme (Antonucci
227: \cite{Antonucci93}). This scheme postulates that the two Seyfert types have
228: both broad and narrow line regions and an obscuring torus that lies between
229: the two. When the torus is face on and the BLR is directly observed, the AGN
230: is classified as a Seyfert 1. When the BLR is obscured by the torus, the AGN
231: is classified as a Seyfert 2, and the BLR can be observed only indirectly in
232: scattered polarized light. The factors that determine the accretion properties
233: and the ionizing SED of AGN are currently unknown. However, the Seyfert unification
234: picture implies that all possibly relevant factors being equal, such as luminosity,
235: host galaxy type or redshift, the intrinsic SED of both AGN types should be
236: similar. It is therefore of interest to complement our previous study of the
237: nearby luminous Seyfert 1 galaxy NGC~4151 with a corresponding study of a nearby
238: luminous Seyfert 2 galaxy with a similar host galaxy type, such as \( \NGC \).
239: The \emph{ISO-SWS} observations of \emph{\( \NGC \)} are presented in a companion
240: paper (Lutz et al. \cite{Lutz00}) and are used there to derive the gas density
241: and to place constraints on the structure and dynamics of the NLR.
242:
243: This paper is organized as follows. In \S\ref{sec:physprop} we summarize the
244: physical properties of the nucleus of \( \NGC \) that are needed for constructing
245: the photoionization models and constraining their results. In \S\ref{s:lines} we
246: present the emission line flux compilation that we use in our modeling. In \S\ref{s:models}
247: we briefly discuss the construction and fitting of the NLR photoionization models.
248: We present the results in \S\ref{s:results} and discuss them in \S\ref{s:discuss}.
249:
250:
251: \section{The physical properties of NGC~1068}
252:
253: \label{sec:physprop}
254:
255: \label{s:n1068}\( \NGC \) is a barred spiral galaxy at \( z=0.0036 \) (distance
256: \( D=16.6\, \)Mpc for \( H_{0}=65\, \mathrm{km}\, \mathrm{s}^{-1}\, \mathrm{Mpc}^{-1} \))
257: with magnitude \( m_{B}=9.17 \) (e.g. Lipovetsky, Neizvestny \& Neizvestnaya
258: \cite{LNN88}). Observations of the nucleus of \( \NGC \) and models of the
259: nuclear line emission indicate that the gas in the nucleus forms a complex system,
260: which is composed of various spatial and dynamical components. These are excited
261: by several physical mechanisms, including photoionization by the nuclear continuum,
262: photoionization by hot stars, and possibly also by shocks and energetic particles
263: from a radio jet. In order to isolate the effects of the nuclear continuum and
264: to construct photoionization models of the NLR it is necessary to understand
265: the morphology and content of the galactic nucleus. We present here a brief
266: overview of the properties of the nucleus that are relevant to this work.
267:
268:
269: \subsection{The galactic nucleus}
270:
271: \label{s:n1068-nucleus}
272:
273: The most prominent morphological feature in the nucleus of \( \NGC \) is the
274: asymmetric bi-polar pattern of both the radio and the optical line emission.
275: The radio emission extends over \( \sim \! 15\arcsec \) and has a sharply
276: defined northern lobe and a weaker diffuse southern lobe (Wilson \& Ulvestad
277: \cite{WU83}; Muxlow et al. \cite{Muxlow96}). The southern lobe is both smaller
278: and redder, which is consistent with the picture that the large northern lobe
279: is observed above the galactic disk, generally facing the observer, and the
280: southern lobe is seen through the galactic disk (Unger et al. \cite{ULPA92};
281: Macchetto et al. \cite{Macchetto94}). Images of the NLR in low excitation lines
282: show mainly the northern cone (Cecil, Bland \& Tully \cite{CBT90}; Unger et
283: al. \cite{ULPA92}; Evans et al. \cite{Evans91}). The precise value of the
284: opening angle associated with the emission maps depends on the way the edge
285: of the cone is defined and on the assumed position of the nucleus. The location
286: of the nucleus can be determined to within \( \sim \! 0.05\arcsec \) by the
287: center of symmetry of the UV polarization pattern (Capetti et al. \cite{Capetti95},
288: Kishimoto \cite{Kishimoto99}). The position of the nucleus does not appear
289: to coincide with the maximum of the continuum emission, which implies that the
290: nucleus is heavily obscured even in the infrared. This is consistent with an
291: obscuring torus of column density in excess of \( 10^{24} \) cm\( ^{-2} \),
292: as is inferred from X-ray (Marshall et al. \cite{Marshal93}) and CO observations
293: (Tacconi et al. \cite{Tacconi94}). The positioning of the nucleus makes it
294: possible to estimate the opening angle of the radiation cone at \( \gtrsim 70\arcdeg \)
295: up to \( \sim \! 100\arcdeg \). The ionization cone appears to be only partially
296: filled.
297:
298: Marconi et al. (\cite{MWMO96}) find that the coronal line emission peaks \( 0.5\arcsec \)
299: NE of the nucleus and extends up to \( \sim \! 4\arcsec \). They note that
300: the blueshift of the emission line profile centroid increases, and the FWHM
301: of the profile decreases with \( \Eion \) (the ionization energy required
302: to produce the emitting ion from the preceding ionization stage). They interpret
303: this as evidence that the emission lines are emitted from outflowing material.
304: The high ionization lines are emitted in the inner light cone, where the velocity
305: field is relatively coherent, while the lower ionization lines are emitted from
306: slower, more extended areas with different velocities. The NLR \( \bOIIIbB \)
307: emission extends over the few inner arcseconds (Evans et al. \cite{Evans91};
308: Unger et al. \cite{ULPA92}; Dietrich \& Wagner \cite{DW98}). The extended
309: emission line region (EELR) \( \bOIIIbB \) emission extends over more than
310: \( 10\arcsec \) (Unger et al. \cite{ULPA92}).
311:
312: The unresolved BLR is seen only in scattered polarized light and has FWHM of
313: \( \sim \! 3000 \) km~s\( ^{-1} \) (Miller et al \cite{MGM91}). As is seen
314: in other Seyfert 2 galaxies (Wilson \cite{Wilson88}; Capetti et al. \cite{Capetti96}),
315: the morphology of the NLR emission maps is correlated with that of the radio
316: structure (Wilson \& Ulvestad \cite{WU83}; Capetti et al. \cite{CAM97}). In
317: particular, a 3-dimensional reconstruction of the positions of individual clouds
318: based on polarimetric measurements also indicates that the northern NLR cone
319: is directed towards the observer, and the southern part away from the observer
320: (Kishimoto \cite{Kishimoto99}). The close correspondence between the NLR and
321: the jet suggests that the jet outflow sweeps and compresses the ambient gas
322: and thereby increases the line emissivity. The NLR has a very complex structure
323: (Macchetto et al. \cite{Macchetto94}) and displays large scale clumpiness.
324: It is composed of many line emitting cloud complexes (Alloin et al. \cite{Alloin83};
325: Meaburn \& Pedlar \cite{MP1986}; Evans et al. \cite{Evans91}; Dietrich \&
326: Wagner \cite{DW98}). The individual components have FWHM ranging from \( \sim \! 200 \)
327: km~s\( ^{-1} \) to \( \sim \! 700 \) km~s\( ^{-1} \), and extend over \( \sim \! 2500 \)
328: km~s\( ^{-1} \) in velocity space, resulting in an integrated \( \bOIIIbB \)
329: profile with FWHM of \( 1150 \) km~s\( ^{-1} \) (Dietrich \& Wagner \cite{DW98}).
330: The velocity field of the NLR clouds with the lowest FWHM is consistent with
331: rotation around the nucleus. The bulk velocities of intermediate FWHM clouds
332: are clearly split relative to the symmetry axis of the radio jet, which suggests
333: that they are associated with the interaction between the radio jet and the
334: NLR gas. The highest FWHM clouds are associated with highly polarized emitting
335: structures (Capetti et al \cite{Capetti95}), and could therefore be a reflected
336: image of an inner obscured region. The EELR appears to follow the galactic rotation
337: (Unger et al. \cite{ULPA92}).
338:
339: The large \emph{ISO} apertures (\( 14\arcsec \times 20\arcsec \) to \( 20\arcsec \times 33\arcsec \))
340: includes both the outflowing inner NLR and the rotating EELR.
341:
342:
343: \subsection{The line emitting gas}
344:
345: \label{s:n1068-gas}
346:
347: There is evidence that at least three different ionization mechanisms are at
348: work in the NLR. The very high ionization states are probably due to the the
349: central continuum source. Marconi et al. (\cite{MWMO96}) find that the infrared
350: coronal line ratios point to photoionization as the main excitation mechanism
351: of the coronal gas, and that both collisional excitation or photoionization
352: by very hot stars can be ruled out. At lower ionization states, the jet / ISM
353: interaction can provide internal sources of excitation in addition to the external
354: central continuum, for example by fast shocks or cosmic rays. The morphological
355: connection between the radio and line emission suggests that the jet outflow
356: shapes the NLR. Estimates of high gas temperatures (Kriss et al. \cite{KDBFL92})
357: and the existence of dense but highly ionized clouds near the nucleus on both
358: sides of the jet axis suggest that the jet may also play a role in photoionizing
359: the clouds (Capetti et al \cite{CAM97}; Axon et al. \cite{Axon98}; Dietrich
360: \& Wagner \cite{DW98}). Hot stars are a third ionization mechanism. Unresolved
361: UV continuum point sources in the inner \( 7\arcsec \times 7\arcsec \), which
362: are not observed in \( \bOIIIbB \), could be OB associations (Macchetto et
363: al. \cite{Macchetto94}). Hot stars are certainly a component in the ring-like
364: structure that surrounds the nucleus. An ellipse of H\,\textsc{ii} emission
365: delineates the NLR at an average radius of \( 13\arcsec \) (Cecil, Bland \&
366: Tully \cite{CBT90}; Bruhweiler et al. \cite{BTA91}) and starburst knots and
367: CO emission encircle the nucleus and the NLR at an average radius of \( 18\arcsec \)
368: (Planesas, Scoville \& Myers \cite{Planesas91}). However, the overall similarity
369: in the profiles of the mid-infrared high-excitation lines, which cannot be excited
370: by hot stars, and the profiles of the intermediate excitation lines (\( \Eion >30\, \mathrm{eV} \)),
371: which could be excited by hot stars, strongly suggests that gas excited by hot
372: stars within the large \emph{ISO} aperture does not contribute more than \( \sim \! 20\% \)
373: to these NLR line fluxes (Lutz et al. \cite{Lutz00}). An additional complication
374: due to stars is contamination of the observed emission line spectrum with stellar
375: absorption features. This appears in the difference spectrum of the \( 30\arcsec \oslash \)
376: and \( 18\arcsec \oslash \) \emph{HUT} apertures, which shows both a reddened
377: early-type stellar continuum and stellar absorption features (Kriss et al. \cite{KDBFL92}).
378:
379: As is discussed by Lutz et al. (\cite{Lutz00}), the \emph{ISO-SWS} line ratios
380: indicate that the mid-IR lines are emitted from gas with a hydrogen density
381: of \( \sim \! 2000\, \mathrm{cm}^{-3} \). The NLR appears to contain also higher
382: density gas. The density of individual knots in the high ionization core is
383: estimated at \( 10^{4} \) to \( 3\times 10^{4} \) cm\( ^{-3} \) from \emph{HST}
384: measurements of the \( [\textrm{S}\, \textsc {ii}] \) doublet, while the overall
385: density \( \sim \! 1\arcsec \) from the nucleus varies between \( 10^{3} \)
386: to \( 4\times 10^{3} \) cm\( ^{-3} \) (Capetti et al. \cite{CAM97}).
387:
388: The ionization parameter\footnote{%
389: \( U\equiv Q_{\mathrm{ion}}/4\pi r^{2}nc, \) where \( Q_{\mathrm{ion}} \)
390: is the ionizing photon luminosity, \( n \) is the hydrogen density at the illuminated
391: face of the cloud, \( r \) is the distance of the face of the cloud from the
392: continuum source and \( c \) is the speed of light. The local ionization parameter
393: in the cloud falls with increasing depth due to absorption and geometrical dilution
394: of the radiation field.
395: } also appears to vary across the NLR. The \( \bOIIIbB /(\Ha +[\mathrm{N}\, {\textsc {ii]}}\, \lambda 6584) \)
396: ratio (Capetti et al. \cite{CAM97}) traces a high excitation core (\( \log U\sim -2.5 \))
397: in the inner \( 1\arcsec \times 2\arcsec \) north of the nucleus, followed
398: by a lowered ionization halo (\( \log U\sim -3.3 \)) out to \( \sim 4\arcsec \)
399: from the nucleus, and then intermediate ionization filaments (\( \log U\sim -2.8 \))
400: which extend out to the EELR.
401:
402: Estimates of dust reddening in the NLR of \( \NGC \) range from \( \EBV =0.07 \)
403: for the continuum (Kriss et al \cite{KDBFL92}), to 0.20 (Marconi et al \cite{MWMO96}),
404: 0.40 (Shields \& Oke \cite{SO75}; Neugebauer et al \cite{Neugebauer80}; Ward
405: \cite{Ward87}), and \( \EBV =0.52 \) (Koski \cite{Koski78}) for the NLR.
406: The analysis of Kraemer et al. (\cite{KRC98}) indicates that there may be some
407: dust mixed with gas in varying amounts. The observed line ratios in the NLR
408: suggest that the O/N abundance is less than solar. Netzer (\cite{Netzer97})
409: and Netzer \& Turner (\cite{NT97}) interpret this as an indication of under-abundant
410: oxygen (see also Sternberg, Genzel \& Tacconi \cite{SGT94}), and propose that
411: the abundances of He:C:N:O:Ne:Mg:Si:S:Ar:Fe relative to hydrogen are \( (100:3.7:1.1:2.7:1.1:0.37:0.35:0.16:0.037:0.4)\times 10^{-4} \),
412: respectively. Kraemer et al. (\cite{KRC98}) interpret the line ratios as showing
413: an overabundance of nitrogen, as well as hinting at higher than solar iron and
414: neon, and propose abundance ratios of \( (100:3.4:3.6:6.8:2.2:0.33:0.31:0.15:0.037:0.8)\times 10^{-4} \).
415:
416: Finally, Kraemer et al. (\cite{KRC98}) model the NLR emission with a multi-component
417: gas model, and suggest that the low ionization emission lines are emitted by
418: gas that is partially screened by a dense, optically thin component. They use
419: their models to estimate that the filling factor is \( F\sim 10^{-4} \).
420:
421:
422: \subsection{The ionizing continuum}
423:
424: \label{s:n1068-SED}
425:
426: Unlike the situation in Seyfert 1 galaxies, where it is possible to observe
427: the intrinsic SED outside the obscured range, the intrinsic SED of \( \NGC \)
428: cannot be directly observed even in the optical or X-ray bands. Only a small
429: fraction of the AGN light is scattered into the line of sight, and can be observed
430: against the host galaxy in polarized light. Pier et al (\cite{Pier94}) compiled
431: various continuum measurements in the optical, UV and X-ray, and carefully took
432: account of aperture differences, star-light contamination, reflection by dust
433: and bremsstrahlung emission from the scattering plasma. They conclude that the
434: resulting reflected SED is broadly similar to that observed in Seyfert 1 galaxies.
435: Pier et al (\cite{Pier94}) also list various estimates of the fraction of light
436: reflected by the scatterer. These values range from \( f_{\textrm{refl}}=10^{-3} \)
437: (Bland-Hawthorn \& Voit \cite{BH93}) to \( \sim \! 0.05 \) (Bland-Hawthorn,
438: Sokoloski \& Cecil \cite{BSC91}). Pier et al (\cite{Pier94}) argue that the
439: most reliable estimate is \( f_{\textrm{refl}}\sim 0.01 \) to within a factor
440: of a few. Because the reflectors are more than a hundred light years away from
441: the continuum source (Miller et al. \cite{MGM91}), short-term continuum variability
442: is unlikely to affect the reconstruction of the reflected SED.
443:
444: We adopt the Pier et al. nuclear continuum SED in the UV and X-ray as the template
445: SED (Fig.~\ref{fig:temp}), and further extend it from 10~keV to 100~keV with
446: a slope of \( F_{\nu }\propto \nu ^{-1} \). We enumerate on the unobserved
447: UV to soft X-ray range to find the best fitting SED.
448:
449:
450: \section{The observed line flux compilation}
451:
452: \label{s:lines}
453:
454: In addition to the ISO-SWS mid-IR lines fluxes presented in Lutz et al. (\cite{Lutz00}),
455: we compiled a list of UV to IR lines from the literature. The compilation initially
456: included a list of \( \sim \! 120 \) measured line fluxes, which were obtained
457: over the last \( \sim \! 30 \) years using various instruments with different
458: apertures, spectral resolutions and reduction techniques. The observed lines
459: were taken from the following references, listed by spectral band with the aperture
460: used (where given): UV lines from Kriss et al (\cite{KDBFL92}) (\( 18\arcsec \oslash \));
461: optical lines from Osterbrock \& Parker (\cite{OP65}) (trailing long slit),
462: Anderson (\cite{Anderson70}) (\( 8\arcsec \times 8\arcsec \)), Wampler (\cite{Wampler71})
463: (\( 10\arcsec \oslash \)), Koski (\cite{Koski78}) (\( 2.7\arcsec \times 3.4\arcsec \))
464: and Neugebauer et al (\cite{Neugebauer80}) (\( \sim \! 10\arcsec \times 20\arcsec \));
465: optical to near-IR lines from Shields \& Oke (\cite{SO75}) (\( 10\arcsec \times 10\arcsec \)),
466: near-IR lines from Oliva \& Moorwood (\cite{OM90}) (\( 6\arcsec \times 6\arcsec \)),
467: Marconi et al (\cite{MWMO96}) (long slit of width \( 4.4\arcsec \)), Osterbrock
468: \& Fulbright (\cite{OF96}) (\( 3\arcsec \times 18\arcsec \)) and Osterbrock,
469: Tran \& Veilleux (\cite{OTV92}) (long slit of width \( 1.2\arcsec \)); IR
470: lines from Thompson (\cite{Thompson96}) (\( 2\arcsec \times 10\arcsec \))
471: and the ISO-SWS (\( 14\arcsec \times 20\arcsec \) to \( 20\arcsec \times 33\arcsec \)).
472:
473: As is discussed in detail by Alexander et al. (\cite{Alexander99}), the result
474: of such a compilation is generally not self-consistent, and only a small subset
475: of the of the \( \sim \! 120 \) lines can be used. First, we excluded low and
476: medium excitation lines (\( \Eion \lesssim 100\, \)eV) observed through apertures
477: whose smaller dimension is less than \( 3\arcsec \), so as to avoid significant
478: loss of light due to incomplete coverage of the line emitting region. The observed
479: emission from the high excitation lines is centrally concentrated in the inner
480: \( <4\arcsec \) (Marconi et al \cite{MWMO96}), which are covered even by
481: the smallest apertures used. Second, we excluded lines with \( \Eion <30\, \)eV
482: to avoid using lines that may be photoionized primarily by young hot stars or
483: other non-AGN, lower-energy excitation processes. This also reduces the loss
484: of light bias. Third, we excluded narrow lines whose measured flux is uncertain,
485: either because the flux is very low (flux less than \( 5\times 10^{-13}\, \)erg
486: s\( ^{-1} \) cm\( ^{-2} \), \( \sim \! 2\% \) of the strongest line), or
487: because it has a significant broad component, such as the \( \NV \) and \( \CIV \)
488: lines. Fourth, we excluded the \( \bFeXb \) and \( \bFeXIb \) lines, whose
489: collision strengths are highly uncertain and therefore cannot be modeled reliably.
490: The final, much reduced line list includes 22 lines (Table~\ref{tbl:flux}),
491: which we use for obtaining the best-fit SED. Whenever more than one measurement
492: of the line exists, we quote the average flux and use the rms scatter as an
493: error estimate. Important IR lines that were not included in the final list
494: were nevertheless compared to the best fit model predictions to verify that
495: there are no gross inconsistencies.
496:
497:
498: \section{The photoionization models}
499:
500: \label{s:models}
501:
502: The method of constructing photoionization models for the NLR and, in particular,
503: the ``\( \lgs \) fit procedure'' for obtaining the best fitting SED and
504: gas parameters is described in detail in Alexander et al. (\cite{Alexander99}),
505: and is summarized here briefly. We parameterize the SED as a piece-wise broken
506: power-law (Fig.~\ref{fig:temp}), and enumerate on the different possibilities
507: of connecting the power-law segments. We test a large number of simplified NLR
508: gas models, which consist of optically thick (radiation bounded) clouds whose
509: ionized surfaces partially cover a spherical shell around the continuum source.
510: Each cloud extends in the radial direction as far as it takes to effectively
511: absorb all the ionizing UV photons (specifically, until the hydrogen ionization
512: fraction falls below 2\%). The gas clouds are parameterized by their chemical
513: composition, the hydrogen density \( n \), the ionization parameter \( U \)
514: at the irradiated face of the cloud, and the filling factor \( F \). In addition,
515: an asymmetry parameter \( A \) expresses the ratio between the ionizing flux
516: directed towards the NLR and that directed towards the observer. This describes
517: situations where the continuum source is not isotropic, or where only a fraction
518: \( f_{\mathrm{refl}} \) of the continuum is reflected towards the observer
519: (\( A=1/f_{\mathrm{refl}} \)).
520:
521: For each NLR gas model, the fit procedure uses the observed line fluxes to derive
522: the best-fit SED for that gas model, and from it to derive in a self-consistent
523: way the corresponding covering factor \( C, \) the inner NLR angular radius
524: \( \theta \), the width of the ionized region \( \tion \) (defined here
525: as the radial extent of the Balmer lines emitting gas) and the reddening coefficient
526: \( \EBV \). These parameters are constrained by the observations and so can
527: be used to limit the range of acceptable NLR gas models. The \( \lgs \) fit
528: procedure assigns a score \( S \) to the best-fit model, which means that the
529: model line fluxes fit the observed ones up to a factor \( S \), on average.
530: The worst-fitting line and the factor by which it deviates from the observed
531: value are also recorded. Monte-Carlo simulations are used to calculate confidence
532: limits on the best-fit SED. In addition, we calculate the correlation between
533: the model-to-data line ratios and the lines wavelengths, ionization potentials,
534: critical densities and ``deplitivity'' (the tendency of an element to be depleted
535: into dust). These residual correlations test whether the remaining inconsistencies
536: in the best-fit model are related to inaccurate modeling of the reddening, the
537: spectral hardness / softness of the continuum, the gas density or its dust content.
538: The correct model should not display any such correlations. The final, global
539: best-fit SED is the one with the best \( S \)-score among all the NLR gas models
540: that are consistent with the observed NLR geometry and reddening, whose worst-fitting
541: line is not too far from the observed value, and which display no significant
542: residual correlations.
543:
544: We assume in all models that the gas clouds have constant density and that \( f_{\mathrm{refl}}=0.005 \).
545: We investigate two classes of models. The first consists of models with a single
546: type of cloud. We enumerate on different values for the ionization parameter
547: (\( \log U=-1,-2 \) and \( -3 \)), gas density (\( n=2000 \) and \( 10^{4}\, \mathrm{cm}^{-3} \))
548: and filling factor (\( \log F=-2,-3 \) and \( -4 \) at the ionized surface).
549: We test two different sets of non-solar abundances, the low oxygen set and high
550: nitrogen set (\S\ref{s:n1068-gas}). We test two different possibilities for
551: the radial run of the filling factor. The first is that \( F \) is constant,
552: which corresponds to either a static distribution of clouds, a rotating distribution
553: of clouds, a linear constant velocity outflow, a strongly decelerating outflow
554: at a constant opening angle, or an outflow where clouds are continuously added
555: to the flow (e.g. from the molecular torus). The second is \( F\propto r^{-2} \),
556: which corresponds to a constant velocity and constant opening angle outflow
557: where the clouds are formed at the base of the flow. The second class of models
558: have two types of gas clouds, and are constructed by combining all the possible
559: pairs of one component models. We assume that the clouds do not obscure one
560: another.
561:
562: The photoionization calculations were carried out using the numerical photoionization
563: code \noun{ion9}9, the 1999 version of the code \noun{ion} described by
564: Netzer (\cite{Netzer96}).
565:
566:
567: \section{Results}
568:
569: \label{s:results}
570:
571:
572: \subsection{Single component models}
573:
574: We find that single component models generally fail to fit the observed line
575: ratios and the observed constraints on the geometry of the NLR. Low \( U \)
576: models (\( \log U=-3 \)) cannot reproduce the high excitation lines regardless
577: of the values of the other model parameters. For the \( F\propto r^{-2} \)
578: models, the low \( U \) models significantly over-estimate the size of the
579: NLR, while the high-\( U \) models require unphysical covering factors exceeding
580: unity. Single component models with a high \( U \) and low constant filling
581: factor do somewhat better, although such models are problematic because a constant
582: filling factor is inconsistent with simple scenarios of NLR outflow. The best
583: fit single component model has low oxygen abundance with \( \log U=-1 \), \( \log n=3.3 \)
584: and \( \log F=-3 \). This model can reproduce the observed lines up to a factor
585: of 2 and predicts \( \theta =0.8\arcsec \), \( \tion =4.8\arcsec \), \( C=0.35 \)
586: and \( \EBV =0.18. \) These values are roughly consistent with the observed
587: constraints. However, this model under-predicts the observed \( \OVI \) flux
588: by a factor of \( \sim \! 5 \) and shows a residual negative correlation with
589: \( \lambda _{0} \) at the 5\% confidence level. Like all the models investigated
590: here, the best fit SED displays a deep trough at 4 Ryd (\( \log f=-27.4,\, -29.0,\, -27.4,\, -28.2 \)
591: at 2, 4, 8 and 16 Ryd, respectively). A fit of similar quality is obtained with
592: the high nitrogen abundance set.
593:
594:
595: \subsection{Two component models}
596:
597: A better fit to the observations is provided by the best-fit two component model
598: (Table~\ref{tbl:fit}). This model fits the 22 observed line fluxes to within
599: a factor of 1.9 on average (experience shows this is as well as one can expect
600: for AGN photoionization models). The worst fitting line, \( \bArVIb \), is
601: under-predicted by a factor of 4. The model-to-data ratios of individual lines
602: are displayed in Fig.~\ref{fig:ratios}. The low excitation IR lines (\( \Eion <30\, \mathrm{eV} \))
603: such as \( \bNeIIb \), \( \bSIIIbA \) and \( \bSIIIbB \), which were not
604: used in the fit, are nevertheless all consistent with the observations to within
605: a factor of 2. However, the model-to-data line ratios for these lines are not
606: much smaller than 1, as would be expected if there is a significant contamination
607: from gas photoionized by starbursts. The high excitation IR lines such as \( [\mathrm{S}\, {\textsc {ix]}}\, \lambda 1.25\mu \mathrm{m} \),
608: \( [\mathrm{Si}\, {\textsc {x]}}\, \lambda 1.4\mu \mathrm{m} \) and \( [\mathrm{Si}\, {\textsc {ix]}}\, \lambda 2.6\mu \mathrm{m} \),
609: which were not used in the fit, are also consistent with the observations to
610: within a factor of 3. There are no statistically significant residual correlations
611: between these ratios and \( \lambda _{0} \), \( \Eion \), the deplitivity
612: or \( n_{c} \). The best fit values of the extinction \( (\EBV \sim 0.2 \))
613: is in agreement with other estimates (\S\ref{s:n1068-gas}). The covering factors
614: are somewhat larger than is indicated by the observed opening angle of the emission
615: cone (\S\ref{s:n1068-nucleus}). The dense compact component requires a covering
616: factor of \( C=0.45 \), which corresponds to a bi-cone with an opening angle
617: of \( \sim \! 115\arcdeg \). The less dense, more extended component requires
618: a covering factor \( C=0.26 \), which corresponds to a single cone with an
619: opening angle of \( \sim \! 130\arcdeg \) or a bi-cone with opening angle
620: of \( \sim \! 90\arcdeg \). Figure~\ref{fig:mix} shows the best fit SED for
621: this gas model together with the 99.9\% confidence limits on it (the lower confidence
622: limits at 2, 4 and 16 Ryd were not calculated as they extend beyond the SED
623: grid). This SED shows the generic trough that is seen in the best-fit SED of
624: all the models we investigated. This model has a less than solar oxygen abundance
625: (Netzer \cite{Netzer97}; Netzer \& Turner \cite{NT97}). We find only a few
626: two-component models with high nitrogen abundance that come close to a reasonable
627: fit to the observations. Of these, almost all display a negative correlation
628: with deplitivity at the 5\% to 10\% significance, which may indicate that the
629: line emitting gas is more depleted than is assumed by the high nitrogen abundance
630: set.
631:
632: We conclude that best-fit two component model provides a better, but not an
633: overwhelmingly better fit to the observed line ratios than the single component
634: model. Its marked advantage over the single component models lies in its consistency
635: with the observed NLR geometry and kinematics. This is further discussed in
636: the next section.
637:
638:
639: \section{Discussion}
640:
641: \label{s:discuss}
642:
643: As was discussed in detail by Alexander et al. (\cite{Alexander99}), there
644: are various degeneracies between the parameters that describe the gas model
645: (\( n, \) \( A \), \( F \), \( U \), \( C \), \( \theta \), and \( \tion \)).
646: These degeneracies allow the fit procedure to converge to a robust best-fit
647: SED even when the assumed values of \( n \), \( A \), \( F \) or \( U \)
648: significantly differ from their true values, since this can be compensated to
649: a large degree by a suitable modification of the gas geometry (\( C \), \( \theta \),
650: and \( \tion \)). This property of the fit procedure is especially important
651: in the analysis of \( \NGC \), where observations indicate that the actual
652: properties of the NLR gas are much more complex than can be modeled by our family
653: of simplified gas models. For this reason, we place more weight on the fact
654: that all the best-fit SED models, whether one or two-component, display a deep
655: trough at 4 Ryd than on the determination of the exact values of the gas parameters.
656:
657: The trough in the SED is required for reproducing the relative line fluxes of
658: the high and low ionization species. To check the robustness of this result,
659: we attempted to re-fit the observed line fluxes with all of our one and two-component
660: models using an approximate single power-law SED (\( \log f=-25.8,\, -26.6,\, -27.4,\, -28.2 \)
661: at 2, 4, 8 and 16 Ryd, respectively), which was held fixed in the fit procedure.
662: In all cases the low excitation lines were over-estimated with respect to the
663: high excitation lines, regardless of the values of \( U \), \( n \) or \( F \).
664: For example, when the power-law SED is applied to the best-fit two-component
665: gas model (Table~\ref{tbl:fit}), the low excitation lines (\( \Eion \lesssim 50\, \mathrm{eV} \))
666: are over-estimated by the model by up to a factor of 13, whereas the high excitation
667: lines (\( \Eion \gtrsim 100\, \mathrm{eV} \)) are under-estimated by up to
668: a factor of 23. The overall mismatch of this SED with the observed line fluxes
669: is reflected in both the poor fit score of \( S=4.1 \) and in the very strong
670: residual anti-correlation between the line ratios and \( \Eion \), whose random
671: probability is \( 10^{-4} \). We caution against generalizing this result to
672: mean that every AGN that exhibits a hard emission line spectrum has an absorbed
673: ionizing SED. The emission line spectrum reflects the gas parameters, such as
674: \( U \), \( n \) and \( F \), no less than it does the ionizing SED. It is
675: necessary to have some knowledge of the likely range of values for these parameters
676: in order to interpret the hardness of the line spectrum. Alexander et al. (\cite{Alexander99})
677: provide a counter-example where subtle cancellations between the SED and the
678: gas properties lead to a situation where an AGN (NGC4151) with a hard absorbed
679: SED has a \emph{softer} emission line spectrum than another AGN (Circinus) with
680: a soft unabsorbed Big Blue Bump.
681:
682: Although we do not claim to fix the gas parameters with certainty, the best
683: fit model (Table~\ref{tbl:fit}) is reassuringly consistent with the observations,
684: which broadly indicate that the integrated NLR emission originates in two components.
685: Component A of the model can be interpreted as a system of dense (\( n=10^{4} \)~\( \mathrm{cm}^{-3} \)
686: ), centrally concentrated (\( 0.3\arcsec <\theta <1.4\arcsec \)) outflowing
687: gas clouds (\( F\propto r^{-2} \)) with a relatively high filling factor (\( F\sim 0.01 \))
688: and high ionization parameter \( (\log U=-1 \)). Component B of the model can
689: be interpreted as a more extended distribution (\( 1.9\arcsec <\theta <4.5\arcsec \))
690: of lower density gas (\( n=2\times 10^{3} \)~\( \mathrm{cm}^{-3} \)) with
691: no net outflow (\( F=\mathrm{const}. \)), with a lower filling factor (\( F=0.001) \)
692: and a lower ionization parameter (\( \log U=-2 \)). Component A contributes
693: 58\% of the total line flux in 22 lines listed in Table~\ref{tbl:flux}, with
694: the contribution to individual lines ranging from 45\% of the low excitation
695: \( \bSIVb \) (\( \Eion =34.8\, \mathrm{eV} \)) to more than 99.9\% of the
696: very high excitation line \( \bSiIXbB \) (\( \Eion =303.2\, \mathrm{eV} \)).
697: Its large covering factor indicates that there is probably a significant contribution
698: of flux from the inner \( \sim \! 1\arcsec \) of the diffuse SW emission cone
699: as well as from the bright NE cone. Component B contributes the remaining 42\%
700: of the total line flux, mainly in the lower excitation lines. Its covering factor
701: is small enough for it to be concentrated mostly in the NE bright emission cone,
702: as is observed.
703:
704: The best-fit procedure indicates that models with the low oxygen abundance set
705: fit the observed line fluxes somewhat better than models with the high nitrogen
706: abundance set. In particular, the \( \mathrm{O}\, {\textsc {iii]}}\, \lambda 1663 \),
707: whose unusual relative weakness was an important argument for assuming non-solar
708: abundances (Netzer \cite{Netzer97}; Kraemer et al. \cite{KRC98}), is well
709: reproduced by the best-fit low oxygen model with a model-to-data line flux ratio
710: of 1.4 (Fig.~\ref{fig:ratios}), even though it was not included in our fit
711: since it didn't pass the minimal flux criterion.
712:
713: The trough in the best fit SED (Fig.~\ref{fig:mix}) can be interpreted as an
714: absorption trough due to an absorber between the continuum source and the NLR.
715: The energy resolution of the SED template, which is limited by the computational
716: cost of enumerating over all the SED combinations, is too low to allow detailed
717: modeling of the absorber. Figure~\ref{fig:abs_sed} shows an example of how
718: a quasi-thermal big blue bump that is absorbed by neutral hydrogen would appear
719: in our low resolution SED reconstruction. We find that the trough is consistent,
720: for example, with an absorber that shadows the entire NLR (\( C_{\mathrm{abs}}=1) \)
721: and has a column density of \( N_{H^{0}}=6\times 10^{19}\, \mathrm{cm}^{-2} \)
722: in neutral hydrogen, or with an absorber that allows a small leakage of unfiltered
723: radiation (\( C_{\mathrm{abs}}=0.999 \)) and a column density of \( N_{H^{0}}=10^{20}\, \mathrm{cm}^{-2} \).
724: A similar trough, consistent with an absorber of \( C_{\mathrm{abs}}>0.99 \)
725: and \( N_{H^{0}}=5\times 10^{19}\, \mathrm{cm}^{-2} \), was discovered in the
726: reconstructed SED of Seyfert 1 galaxy NGC~4151 (Alexander et al. \cite{Alexander99}).
727: That absorber was also detected in the \emph{HUT} absorption line spectra of
728: the UV continuum of NGC~4151 (Kriss et al. \cite{Kriss92}, \cite{Kriss95})
729: . The \emph{HUT} spectra of \( \NGC \) (Kriss et al \cite{KDBFL92}) do not
730: have a high enough S/N to allow the detection of absorption lines in the scattered
731: UV continuum of this AGN or against the stellar background (G. Kriss, private
732: comm.). We predict that future sensitive absorption line studies should reveal
733: the presence of such an absorber.
734:
735: The bias in our results due to the fact that we neglected the line emission
736: from the absorbing gas is likely to be small if the absorber is similar to the
737: dense, high velocity UV absorber that was detected in NGC~4151. Such an absorber
738: will not emit forbidden lines, and its permitted lines will be broader than
739: typical NLR lines. Only 3 of the 22 lines we used in our fit are permitted lines,
740: and we did not use lines that are contaminated by broad components. A highly
741: ionized and optically thin absorber will produce strong \( \OVI \) line emission
742: in excess of the typical NLR emission. It is therefore interesting that unlike
743: the two forbidden \( \bOIIIbB \) and \( \bOIVb \) lines and the semi-forbidden
744: \( \OIIIb \) line, which are well reproduced by the best fit model, the observed
745: \( \OVI \) line is 3.2 stronger than predicted (Fig.~\ref{fig:ratios}).
746:
747: We have, up to now, applied our SED reconstruction method to \emph{ISO-SWS}
748: observations of IR coronal lines of three AGN: the Seyfert 2 Circinus galaxy
749: (Moorwood et al. \cite{Moorwood96}; Alexander et al. \cite{Alexander99}),
750: the Seyfert 1 galaxy NGC~4151 (Alexander et al. \cite{Alexander99}), and the
751: Seyfert 2 galaxy \( \NGC \) (this work). In one of these (Circinus), we detect
752: a Big Blue Bump that peaks at \( \gtrsim 50\, \mathrm{eV} \). In the other
753: two we detect deep troughs, which are consistent (but not exclusively so) with
754: a Big Blue Bump that is absorbed by neutral gas interior to the NLR. Our findings
755: thus far are consistent with the picture that luminous Seyfert galaxies are
756: powered by thin accretion disks that produce a quasi-thermal Big Blue Bump,
757: and that in a large fraction of them the NLR sees a partially absorbed ionizing
758: continuum, as suggested by Kraemer et al. (\cite{KRC98}).
759:
760: \acknowledgments
761:
762: This work was supported by DARA under grants 50-QI-8610-8 and 50-QI-9492-3,
763: and by the German-Israeli Foundation under grant I-0551-186.07/97.
764:
765: \begin{thebibliography}{1992a}
766: \bibitem[1999]{Alexander99}Alexander T., Sturm E., Lutz D., Sternberg A., Netzer H., \& Genzel R., 1999,
767: \apj, 512, 204
768: \bibitem[1983]{Alloin83}Alloin D., Pelat D., Boksenberg A., \& Sargent W. L. W., 1983, \apj, 275, 493
769: \bibitem[1970]{Anderson70}Anderson, K. S., 1970, \apj, 162, 743
770: \bibitem[1993]{Antonucci93}Antonucci R., 1993, \araa, 31, 473
771: \bibitem[1985]{AM85}Antonucci R. R. J., Miller J. S., 1985, \apj, 297, 621
772: \bibitem[1998]{Axon98}Axon D. J., Marconi A., Capetti A., Macchetto F. D., Schreier E., \& Robinson
773: A., 1998, \apj, 496, L75
774: \bibitem[1991]{BSC91}Bland-Hawthorn J., Sokoloski J., \& Cecil G., 1991, \apj, 375, 78
775: \bibitem[1993]{BH93}Bland-Hawthorn J., \& Voit G. M., 1993, Rev, Mexicana Astron. Af., 27, 73
776: \bibitem[1991]{BTA91}Bruhweiler F. C., Truong K. Q. \& Altner B., 1991, \apj, 379, 596
777: \bibitem[1995]{Capetti95}Capetti A., Axon D. J., \& Macchetto F. D. Sparks W. B., Boksenberg A., 1995,
778: \apj, 446, 155
779: \bibitem[1996]{Capetti96}Capetti A., Axon D. J., \& Macchetto F. D. Sparks W. B., Boksenberg A., 1996,
780: \apj, 469, 554
781: \bibitem[1997]{CAM97}Capetti A., Axon D. J., \& Macchetto F. D., 1997, \apj, 487, 560
782: \bibitem[1978]{CBT90}Cecil G., Bland J., Tully R. B., 1990, \apj, 355, 70
783: \bibitem[1998]{DW98}Dietrich M., Wagner S. J., 1998, \aap, 338, 405
784: \bibitem[1991]{Evans91}Evans I. N., Ford H. C., Kinney A. L., Antonucci R. R. J., Armus L., Caganoff
785: S., 1991, \apj, 369, L27
786: \bibitem[1998]{George98}George I. M.,Turner, T. J., Netzer H., Nandra K., Mushotzky R. F., \& Yaqoob
787: T., 1998, \apjs, 114, 73
788: \bibitem[1999]{Kishimoto99}Kishimoto M., 1999, \apj, 518, 676
789: \bibitem[1978]{Koski78}Koski A. T., 1978, \apj, 223, 56
790: \bibitem[1998]{KRC98}Kraemer S. B., Ruiz J. R., \& Crenshaw D. M., 1998, \apj, 508, 232
791: \bibitem[1999]{KTCG99}Kraemer S. B., Turner T. J., Crenshaw D. M., \& George I. M., 1999, \apj, in
792: press
793: \bibitem[1992a]{Kriss92}Kriss G. A., et al., 1992a, \apj, 392, 485
794: \bibitem[1992b]{KDBFL92}Kriss G. A., Davidsen A. F., Blair W. P., Ferguson H. C., \& Long K. S., 1992b,
795: \apj, 394, L37
796: \bibitem[1995]{Kriss95}Kriss G. A., Davidsen A. F., Zheng W., Kruk J. W., \& Espey B. R., 1995, \apj,
797: 454, L7
798: \bibitem[1988]{LNN88}Lipovetsky V. A., Neizvestny S. I., \& Neizvestnaya O. M., 1988, Soobshch. SAO.,
799: 55, 5
800: \bibitem[2000]{Lutz00}Lutz D., et al., 2000, \apj, in preparation
801: \bibitem[1994]{Macchetto94}Macchetto F., Capetti A., Sparks W. B., Axon D. J., \& Boksenberg A., 1994,
802: \apj, 435, L15
803: \bibitem[1996]{MWMO96}Marconi A., van der Werf P. P., Moorwood A. F. M., Oliva E., 1996, \aap, 315,
804: 335
805: \bibitem[1986]{MP1986}Meaburn J., Pedlar A., 1986, \aap, 159, 336
806: \bibitem[1993]{Marshal93}Marshall et al., 1993, \apj, 405, 168
807: \bibitem[1990]{MGM91}Miller J. S., Goodrich R. W., \& Mathews W. G., 1991, \apj, 378, 47
808: \bibitem[1996]{Moorwood96}Moorwood A. F. M., et al. 1996, \aap, 315, L109
809: \bibitem[1996]{Muxlow96}Muxlow T. W. B., Pedlar A., Holloway A. J., Gallimore J. F., \& Antonucci R.
810: R. J., 1996, \mnras, 278, 854
811: \bibitem[1996]{Netzer96}Netzer, H., 1996, \apj, 473, 781
812: \bibitem[1997]{Netzer97}Netzer, H., 1997, \apss, 248, 127
813: \bibitem[1997]{NT97}Netzer H., \& Turner T. J., 1997, \apj, 488, 694
814: \bibitem[1980]{Neugebauer80}Neugebauer G., et al, 1980, \apj, 283, 502
815: \bibitem[1990]{OM90}Oliva E., Moorwood A. F. M., \apj, 1990, 348, L5
816: \bibitem[1996]{OF96}Osterbrock D. E., Fulbright J. P. , 1996, \pasp, 108, 183
817: \bibitem[1965]{OP65}Osterbrock D. E., Parker R., 1965, \apj, 141, 892
818: \bibitem[1992]{OTV92}Osterbrock D. E., Tran H. D., Veilleux S., \apj, 389, 196
819: \bibitem[1994]{Pier94}Pier E. A., Antonucci R., Hurt T., Kriss G., \& Krolik J., 1994, \apj, 428,
820: 124
821: \bibitem[1991]{Planesas91}Planesas P., Scoville N. Z. \& Myers S. T., 1991, \apj, 369, 364
822: \bibitem[1975]{SO75}Shields G. A., \& Oke J. B., 1975, \apj, 197, 5
823: \bibitem[1973]{SS73}Shakura N. I., \& Sunyaev R. A., 1973, \aap, 24, 337
824: \bibitem[1986]{Snijders86}Snijders M. A. J., Netzer H., \& Boksenberg A., 1986, \mnras, 222, 549
825: \bibitem[1994]{SGT94}Sternberg A., Genzel R., \& Tacconi L., 1994, \apj, 436, L131
826: \bibitem[1994]{Tacconi94}Tacconi L. J., Genzel R., Blietz M., Cameron M., Harris A. I., \& Madden S.,
827: 1994, \apj, 426, 77
828: \bibitem[1996]{Thompson96}Thompson R. I., 1996, \apj, 459, L61
829: \bibitem[1992]{ULPA92}Unger S. W., Lewis J. R., Pedlar A., \& Axon D. J., 1992, \mnras, 258, 371
830: \bibitem[1971]{Wampler71}Wampler E. J., 1971, \apj, 164, 1
831: \bibitem[1987]{Ward87}Ward et al M. J., 1987, \apj, 316, 138
832: \bibitem[1988]{Wilson88}Wilson A. S., 1988, Advances in Space Research, 8, 27
833: \bibitem[1983]{WU83}Wilson A. S., Ulvestad J, S., 1983, \apj, 275, 8
834: \end{thebibliography}
835:
836: \clearpage
837:
838: \figcaption[sed_template.eps]{The template used for enumerating on the SED of
839: \( \NGC \). The most and least luminous SEDs are indicated by the lines, together with
840: the relative sense of their spectral hardness between 1 and 150 Ryd. \label{fig:temp}}
841:
842: \figcaption[best_mix.eps]{The best fit SED of the two-component model
843: (Table~\ref{tbl:fit}). \label{fig:mix}}
844:
845: \figcaption[lratios.eps]{The model to data line ratios for the best fit
846: two-component model (Table~\ref{tbl:fit}). The $\OIIIb$ line ratio (white
847: circle), which was not used in the fitting procedure is also
848: displayed. \label{fig:ratios}}
849:
850: \figcaption[abs_sed.eps]{An example of how
851: a quasi-thermal Big Blue Bump that is absorbed by neutral hydrogen
852: would appear in the low resolution reconstructed SED. The unabsorbed
853: bump (thin line) and the bump after absorption by a $C_{\rm abs}=1$,
854: $N_{H^0}=6\times10^{19}$\,cm$^{-2}$ absorber (long dashed line) and a
855: $C_{\rm abs}=0.999$, $N_{H^0}=10^{20}$\,cm$^{-2}$ absorber (short
856: dashed line) are superimposed on the best fit SED
857: model.\label{fig:abs_sed}}
858:
859: \clearpage
860:
861: \begin{deluxetable}{lr@{.}lr@{.}lr@{.}lr@{.}l} \small
862: \tablecaption{The compiled emission line flux list. \label{tbl:flux}}
863: \tablewidth{0pt}
864: \tablehead{
865: \colhead{Line} &
866: \multicolumn{2}{c}{\(\lambda_{0}\)}&
867: \multicolumn{2}{c}{\(E_{\textrm{ion}}\)\tablenotemark{a}}&
868: \multicolumn{2}{c}{\(f_{\ell}\)\tablenotemark{b}}&
869: \multicolumn{2}{c}{\(\Delta f_{\ell}\)\tablenotemark{c}}\\
870: \colhead{}&
871: \multicolumn{2}{c}{\(\mu\)m}&
872: \multicolumn{2}{c}{eV}&
873: \multicolumn{4}{r}{\(10^{-13}\,{\textrm{erg}}\,{\textrm{s}}^{-1}{\textrm{cm}}^{-2}\)}
874: }
875: \tablecolumns{9}
876: \startdata
877: \cutinhead{UV, optical and NIR lines}
878: O\,{\textsc{vi}}& 0&1032+1038 & 113&9 & 37&4& 3&1\nl
879: N\,{\textsc{iv}}]& 0&1487 & 47&4 & 5&1 & 1&1\nl
880: He\,{\textsc{ii}}& 0&1640 & 54&4 & 17&7 & 5&2\nl
881: [Ne\,{\textsc{v}}]& 0&3426 & 97&1 & 15&7 & 6&7\nl
882: [Ne\,{\textsc{iii}}]& 0&3868+3969 & 41&0 & 19&2 & 2&6\nl
883: %[O\,{\textsc{iii}}]& 0&4363 & 35&1 & 2&54 & 0&96\nl
884: He\,{\textsc{ii}}& 0&4686 & 54&4 & 6&15 & 1&50\nl
885: [O\,{\textsc{iii}}]& 0&4959+5007 & 35&1 & 256& & 27&\nl
886: %[Fe\,{\textsc{vii}}]& 0&6086 & 99&1 & 3&23 & 0&46\nl
887: [Si\,{\textsc{vi}}]& 1&96 & 166&8 & 7&5 & 0&5\nl
888: [Si\,{\textsc{vii}}]& 2&48 & 205&1 & 8&3 & 0&9\nl
889: \cutinhead{ISO-SWS IR lines}
890: %[Si\,{\textsc{ix}}]& 2&585 & 303&2 & 3&0 & 0&6\nl
891: [Mg\,{\textsc{viii}}]& 3&028 & 224&9 & 11& & 1&1\nl
892: [Si\,{\textsc{ix}}]& 3&936 & 303&2 & 5&0 & 0&6\nl
893: [Mg\,{\textsc{iv}}]& 4&49 & 80&1 & 7&6 & 1&5\nl
894: [Ar\,{\textsc{vi}}]& 4&528 & 75&0 & 15& & 3&\nl
895: [Mg\,{\textsc{vii}}]& 5&50 & 186&5 & 13& & \multicolumn{2}{l}{?}\nl
896: [Mg\,{\textsc{v}}]& 5&61 & 109&2 & 18& & 2&\nl
897: [Ne\,{\textsc{vi}}]& 7&64 & 126&2 & 110& & 11&\nl
898: %[Fe\,{\textsc{vii}}]& 7&813 & 99&1 & 3&0 & 0&6\nl
899: [S\,{\textsc{iv}}]& 10&51 & 34&8 & 58& & 6&\nl
900: [Ne\,{\textsc{v}}]& 14&32 & 97&1 & 97&0 & 9&7\nl
901: [Ne\,{\textsc{iii}}]& 15&56 & 41&0 & 160& & 32&\nl
902: [Ne\,{\textsc{v}}]& 24&32 & 97&1 & 70& & 7&\nl
903: [O\,{\textsc{iv}}]& 25&89 & 54&9 & 190& & 20&\nl
904: [Ne\,{\textsc{iii}}]& 36&01 & 41&0 & 17& & 3&\nl
905: \enddata
906: \tablenotetext{a}{
907: The ionization energy required to produce the emitting ion from the
908: preceding ionization stage.}
909: \tablenotetext{b}{
910: The observed flux. For permitted lines, the flux of the decomposed narrow
911: component is quoted.}
912: \tablenotetext{c}{
913: The error estimate on the observed flux. The ISO-SWS errors are estimated by
914: the scatter in various methods for defining the underlying continuum and
915: measuring the line. The errors do not include systematic calibration
916: errors, which are generally smaller than 30\%. A question mark means that
917: error estimates are unavailable.}
918: \end{deluxetable}
919:
920: \clearpage
921: \begin{deluxetable}{llcc}
922: \small
923: \tablecaption{The best-fit two-component model. \label{tbl:fit}}
924: % Mixture of models lowO F0 m12 and lowO F2 m09
925: \tablewidth{5in}
926: \tablehead{
927: \colhead{} & \colhead{} & \colhead{Component A} & \colhead{Component B}
928: }
929: \tablecolumns{4}
930: \startdata
931: {Input parameters:}&&&\nl
932: &$U$ & 0.1 & 0.01 \nl
933: &$n$ &$10^4$ cm$^{-3}$ &$2\times10^3$ cm$^{-3}$\nl
934: &Composition & low oxygen & low oxygen \nl
935: &$A$ & $200$ & $200$ \nl
936: &$F$\tablenotemark{a}&
937: $10^{-2}h$ & $10^{-3}h$ \nl
938: &$\case{{d\log F}}{{d\log r}}$
939: & $-2$ & $0$ \nl
940: {Best-fit results:\tablenotemark{b}}&&&\nl
941: &$S$ & \multicolumn{2}{c}{$1.9$} \nl
942: &$\max S_\ell$\tablenotemark{c}
943: & \multicolumn{2}{c}{4.0$^{-1}$} \nl
944: &worst line
945: & \multicolumn{2}{c}{$\bArVIb$} \nl
946: &$\log f_2$\tablenotemark{d}
947: & \multicolumn{2}{c}{-29.0 (-29.0, -26.6)} \nl
948: &$\log f_4$ & \multicolumn{2}{c}{-29.0 (-29.0, -26.6)} \nl
949: &$\log f_8$ & \multicolumn{2}{c}{-27.4 (-27.4, -26.6)} \nl
950: &$\log f_{16}$& \multicolumn{2}{c}{-29.0 (-29.0, -28.2)} \nl
951: &$C$ & 0.45 & 0.29 \nl
952: &$\theta$ & $0.26\arcsec$ & $1.9\arcsec$ \nl
953: &$\tion$ & $1.1\arcsec$ & $2.6\arcsec$ \nl
954: &$\EBV$ & \multicolumn{2}{c}{0.23 (0.18, 0.29)} \nl
955: &$\qion$ & \multicolumn{2}{c}{0.24 s$^{-1}$\,cm$^{-2}$}\nl
956: &$\log Q_{\rm ion}$/s$^{-1}$ \tablenotemark{e}&
957: \multicolumn{2}{c}{54.2} \nl
958: &$\langle h\nu \rangle$\tablenotemark{f}&
959: \multicolumn{2}{c}{5.1 Ryd} \nl
960: {Residual correlations:}&& Correlation & Random prob. \nl
961: &$\lambda_0$ & $-0.19$ & $0.23$\nl
962: &$\Eion$ & $ 0.00$ & $1.00$\nl
963: &depletion & $+0.11$ & $0.70$\nl
964: &$n_c$ & $+0.03$ & $0.86$\nl
965: \enddata
966: \tablenotetext{a}{Assuming $h=0.65$.}
967: \tablenotetext{b}{Values in parentheses are the 99.9\% confidence intervals.}
968: \tablenotetext{c}{Model to data ratio, given as the reciprocal when $<1$.}
969: \tablenotetext{d}{Flux in erg\,s$^{-1}$\,cm$^{-2}$\,Hz$^{-1}$.}
970: \tablenotetext{e}{Ionizing photon luminosity assuming isotopic emission.}
971: \tablenotetext{f}{Mean ionizing photon energy between 1 and 16 Ryd.}
972: \end{deluxetable}
973:
974: %
975: % Figures attached at end
976: %
977:
978: %\setcounter{figure}{0}
979:
980: %\clearpage
981: %\plotone{\texdir/sed_template.eps}
982: %\clearpage
983: %\plotone{\texdir/best_mix.eps}
984: %\clearpage
985: %\plotone{\texdir/lratios.eps}
986: %\clearpage
987: %\plotone{\texdir/abs_sed.eps}
988:
989: \end{document}
990: