1: %\documentclass{emulateapj}
2: \documentclass[12pt,preprint]{aastex}
3:
4: \shorttitle{ISAAC observation of IRAS 19254-7245}
5: \shortauthors{G. Risaliti et al.}
6:
7: %\received{2003 May 20}
8: \begin{document}
9:
10: \title{Revealing the AGN in the {\it Superantennae} through L-band spectroscopy\altaffilmark{1}}
11:
12: \altaffiltext{1}{Based on observations collected at the European Southern Observatory, Chile (proposal ESO 69.A-0643)}
13:
14: \author{G. Risaliti\altaffilmark{2,3}, R. Maiolino\altaffilmark{2},
15: A. Marconi\altaffilmark{2}, L. Bassani\altaffilmark{4},
16: S. Berta\altaffilmark{5}, V. Braito\altaffilmark{5,6}, R. Della Ceca\altaffilmark{6},
17: A. Franceschini\altaffilmark{5}, M. Salvati\altaffilmark{2}}
18: %R. Maiolino\altaffilmark{1},
19: %A. Marconi\altaffilmark{1} }
20:
21: \email{risaliti@arcetri.astro.it}
22:
23: \altaffiltext{2}{INAF - Osservatorio di Arcetri, L.go E. Fermi 5,
24: Firenze, Italy}
25: \altaffiltext{3}{Harvard-Smithsonian CfA, 60 Garden st.
26: Cambridge, MA}
27: \altaffiltext{4}{TESRE/CNR, Bologna, Italy}
28: \altaffiltext{5}{Dipartimento di Astronomia, Universit\`a di Padova, Italy}
29: %\altaffiltext{6}{INAF - Osservatorio Astrofisico di Padova, Italy}
30: \altaffiltext{6}{INAF - Osservatorio Astronomico di Brera, Milano, Italy}
31:
32:
33: \begin{abstract}
34: We present an L-band spectrum of the Ultraluminous Infrared
35: Galaxy IRAS 19254-7245 (the {\em Superantennae}), obtained with VLT-ISAAC.
36: The high signal to noise ratio allows a study of the main spectral features with
37: unprecedented detail for an extragalactic source. We argue that the main energy source
38: in the IR is an obscured AGN. This is indicated by the low equivalent width of
39: the 3.3~$\mu$m PAH feature, the broad absorption feature at 3.4~$\mu$m, and
40: the steep continuum at $\lambda>3.7~\mu$m ($f_\lambda\propto\lambda^{2.7}$).
41: The substructure of the 3.4~$\mu$m absorption feature indicates that the absorption is
42: due to hydrocarbon chains of 6-7 carbon atoms.
43: \end{abstract}
44:
45: \keywords{Galaxies: active; Galaxies: IRAS 19254-7245; Infrared: Galaxies}
46:
47: \section{Introduction}
48:
49: Ultraluminous Infrared galaxies (L$_{IR}>10^{12}~L_\odot$) have been extensively
50: studied at all wavelengths from the radio to the hard X-rays, in order to
51: unveil the energy source - starburst or Active Galactic Nucleus (AGN) - responsible for the huge infrared luminosity.
52: Determining the AGN role in ULIRGs is fundamental to evaluate the
53: contribution of accretion to the infrared background, and to have a
54: complete view of the population of AGNs in the near Universe.
55:
56: Several diagnostics have been proposed for this purpose. In the optical, the
57: presence of high-ionization narrow emission lines is an indication of the presence of
58: an AGN. However, it is difficult to estimate the contribution of the AGN
59: to the total luminosity just from the optical emission lines.
60: Hard X-rays are in principle a powerful
61: tool. The ratio between X-ray and IR emission in starburst-dominated ULIRGs is expected to be
62: $<10^{-4}$, while in AGN-dominated ULIRGs the ratio is $10^{-3}-10^{-1}$ (Risaliti et al.
63: 2000). However,
64: if the AGN is covered by a column density N$_H$ higher than 10$^{25}$ ~cm$^{-2}$, the
65: presence of the AGN and its total luminosity are hard to determine.
66: In general, all these methods are useful when they can directly measure the
67: AGN luminosity, but they fail in determining whether an AGN is present or not in case
68: of a nondetection.
69:
70: Mid-infrared spectroscopy provides a powerful way to disentangle the starburst
71: and AGN contribution. Genzel et al. (1998) used the Polycyclic Aromatic Hydrocarbon (PAH)
72: molecules emission lines as an indicator of starburst activity in ISO spectra
73: of the brightest ULIRGs. Since PAH molecules are
74: destroyed by the X-ray radiation emitted by AGNs, a
75: high equivalent width (EW) of PAH lines is an indicator of the absence of a strong AGN.
76: Genzel et al. (1998) used the EW of the
77: PAH emission feature at $\sim 7.7~\mu$m as main diagnostics, and concluded
78: that most ULIRGs are starburst-dominated.
79: Recently Imanishi \& Dudley (2000) showed that ground-based telescopes can
80: provide similarly good -or probably better- data for this kind of diagnostics.
81: In particular, L-band spectroscopy of low redshift ULIRGs allows the direct
82: measurement of the $3.3~\mu$m PAH feature (indicator of starburst activity, similarly
83: to the 7.7~$\mu$m feature) and of the
84: carbonaceous dust absorption dip at $\sim 3.4~\mu$m (indication of an absorbed point
85: source like an AGN). Most importantly, the L-band is wide enough for a correct
86: estimate of the continuum level, needed to measure both the
87: equivalent width of the PAH feature and the detection of the carbon dip.
88: In this way Imanishi et al. (2001) and Imanishi \& Maloney (2003)
89: clearly discovered signatures of an AGN dominating
90: the energy output in UGC 5101, which was previously classified as
91: starburst-dominated using ISO spectra only.
92:
93: The ULIRG IRAS 19254-7245 (the {\it Superantennae}, Mirabel, Lutz \& Maza 1991)
94: is optically classified
95: as a Seyfert 2. ISO mid-infrared spectroscopy indicates the presence of an AGN in IRAS 19254-7245,
96: but it does not provide a clear indication
97: on which is the dominant energy source, the EW of the 7.7~$\mu$m PAH feature being
98: intermediate between typical AGN and typical starburst values (Genzel et al. 1998).
99: Recently Charmandaris et al. (2002) classified the source as AGN-dominated, based
100: on the same ISO data.
101:
102: XMM-Newton data in the 2-10 keV band suggests that IRAS 19254-7245
103: is AGN-dominated (Braito et al. 2002). The X-ray spectrum suggests that the source
104: is Compton-thick, i.e. the absorbing column density is higher than 10$^{24}$~cm$^{-2}$.
105: The total luminosity of the AGN is estimated to be of the order or higher than
106: 10$^{44}$~ergs~s$^{-1}$.
107: %, but is starburst-dominated according to ISO mid-infrared
108: %spectroscopy (Genzel et al. 1998). However, an XMM-Newton observation in the
109: %2-10 keV band suggests that it is AGN-dominated (Braito et al. 2003). The same
110: %conclusion has been reached by Vanzi et al. (2001) based on optical and near-IR
111: %spectroscopy.
112: Berta et al. (2003) fitted the SED of IRAS 19254-7245 from the U to the mm-band,
113: with a starburst+AGN model, estimating a contribution of the AGN to the
114: bolometric luminosity of 40-50\%.
115:
116: Here we present an L-band spectrum of IRAS 19254-7245, obtained with the
117: instrument ISAAC on the VLT as part of a mini-survey of
118: bright ULIRGs. Thanks to the superb quality of both the
119: telescope and the instrument, this is probably the best L-band spectrum of
120: a ULIRG ever published, and shows the potentiality of ground-based L-band
121: spectroscopy in the study of active and star forming galaxies.
122:
123: \section{Data reduction and analysis}
124:
125: IRAS 19254-7245 is a system consisting of two colliding spiral
126: galaxies, at z=0.062. The separation of
127: the two nuclei is 8.5'' ($\sim10$~kpc\footnote{We assume H$_0$=75 km s$^{-1}$
128: Mpc$^{-1}$.}). The southern galaxy is classified as a Seyfert 2 (Mirabel et al.
129: 1991), while the northern galaxy shows an optical spectrum typical of
130: starbursts (Colina, Lipari \& Macchetto 1991). At 2~$\mu$m, the southern galaxy is
131: brighter by $\sim$1 magnitude than the northern galaxy (Duc, Mirabel \& Maza 1997).
132: Images at 10~$\mu$m show that the mid-IR emission is concentrated in the
133: nuclei, and the southern one is more than $\sim$5 times brighter than the northern one.
134: Finally, ISOCAM observations (Charmandaris et al. 2002)
135: show that more than 90\% of the 5-20~$\mu$m
136: emission is due to the southern source.
137:
138: We performed 2 observations of IRAS 19254-7245 with ISAAC-VLT, on June 02, 2002, and on June
139: 3, 2002. Each observation was 1 hour long. The two nights were photometric, and
140: the seeing was around 1''.
141: In the image acquired before the spectroscopic observation, we detected the
142: two nuclei and we oriented the 1'' slit in order to obtain spectra of both.
143: However, the northern nucleus turned out to be too faint to obtain a useful
144: spectrum. Therefore, we will not discuss it further.
145:
146: The spectroscopic observations have been performed in ``chopping'' mode,
147: with
148: single exposures of 0.56 sec. The data were merged,
149: flat-fielded and sky-subtracted using standard procedures in the IRAF
150: package. A spectrum of the spectrophotometric standard star HIP 183 (B4III, L=5.5,
151: T$_{eff}$=15,800 k) was acquired in the
152: same way for both observations, and used to obtain the instrumental
153: response, in order to correct the source spectrum.
154:
155: Since the calibration lamp lines were too faint to be useful,
156: the wavelength calibration was performed using the nominal instrumental range
157: and the wavelength of the carbon absorption features (as described in the next
158: Section).
159: Throughout this paper we always refer to rest frame wavelengths, unless
160: otherwise stated.
161:
162: The absolute flux
163: calibration has been obtained analyzing the profile of the star along the
164: slit, and estimating the fraction of flux inside the slit, assuming a perfect
165: centering. We estimate that this procedure has an error of the order of
166: 10\%.
167: We checked that the two final spectra were consistent, and we finally merged
168: the results of the two observations.
169: The final spectrum, rebinned by a factor of 5, is shown in Fig. 1.
170: The error bars are estimated from the Poissonian noise in the sky counts (which are
171: by far the dominant source of noise).
172:
173: %\clearpage
174:
175: \begin{figure}
176: \plotone{f1.eps}
177: \figcaption{\footnotesize{
178: Flux-calibrated spectrum of IRAS 19254-7245. In the observed wavelength
179: range 3.2-3.35~$\mu$m (shaded region) the atmospheric transmission is poor.
180: The solid line is an adopted continuum.}}
181: \end{figure}
182: %\clearpage
183:
184: \section{Discussion}
185:
186: Previous optical and infrared observations indicated the
187: presence of an AGN in the nucleus of IRAS 19254-7245, but the
188: relative importance of the AGN and starburst components was not
189: clear: the 7.7~$\mu$m PAH feature detected by ISO (Genzel et al. 1998)
190: has an intermediate strength (S=0.8, where S is the ratio between
191: the peak 7.7~$\mu$m flux and the continuum at the same wavelength) with
192: respect to pure starbursts and pure AGN (S$\sim$3.6 and S$\sim$0.04
193: respectively, Genzel et al. 1998).
194: However, the determination of the continuum level is highly uncertain in
195: the ISOPHOT spectra, due to the presence of absorption features and
196: the poor signal-to-noise ratio.
197: %Based on the same data, Charmandaris et al. (2001) concludes that the
198: %source is probably dominated by the AGN emission.
199: In the optical and near-IR, the presence of an AGN is clearly revealed from the
200: high [OIII]/H$\beta$ ratio and from the presence of strong coronal lines
201: (Vanzi et al. 2002).
202:
203: The ISAAC spectrum presented here is probably the highest
204: S/N L-band spectrum of a ULIRG so far. Thanks to the high signal
205: and to the moderately high spectral resolution, the starburst and
206: AGN indicators can be studied with unprecedented detail.
207:
208: The signatures of the AGN activity in the L-band spectrum of IRAS 19254-7245 are
209: quite clear: 1) the deep absorption at $\sim3.4~\mu$m (rest frame) strongly
210: suggesting the presence of a point source behind a screen of dusty gas;
211: 2) the PAH feature at 3.3~$\mu$m with a much lower equivalent width
212: than typical starburst-dominated sources; 3) the rather steep continuum slope above
213: $\sim3~\mu$m, suggesting the presence of warm, AGN-heated
214: dust. We discuss each of these points in the following.
215:
216: \subsection{The absorption feature at 3.4~$\mu$m}
217:
218: The broad absorption feature at $\sim3.4~\mu$m (rest frame) is present
219: in the spectra of many AGNs and ULIRGs (Imanishi \& Dudley 2000, Imanishi
220: 2000). We clearly detected such feature in the spectrum of
221: IRAS~19254-7245, with an optical depth $\tau\sim0.8$.
222: The absorption is believed to be due to C-H stretching vibration
223: in hydrocarbon dust grains (Sandford et al. 1991 and references therein).
224: %A strong such feature is considered a reliable indicator of the presence of an active
225: %nucleus.
226: Imanishi \& Maloney (2003) showed that an optical depth of this feature
227: $\tau_{3.4}$ higher than $\sim0.2$ requires a centrally-concentrated source,
228: i.e. an AGN, unless dust absorption in the host galaxy is significant.
229: %, both for theoretical and observational reasons:\\
230: %- A deep continuum absorption requires a dense ({\rm quanto denso? possiamo
231: %mettere qualche numero?}) dusty medium fully obscuring the
232: %energy source. This is what is expected in obscured AGN. On the other side, a
233: %starburst emitting $\sim10^{45}$~erg~s$^{-1}$ or more would have a linear
234: %size of at least 1 kpc ({\rm ???? ragionevole??? ????} ) and it is unlikely
235: %that such a large region is completely covered by this dusty absorber.\\
236: Indeed, absorption features at 3.4~$\mu$m are commonly observed in AGNs
237: (Imanishi 2000) while they are never found in galaxies known to be dominated by
238: starburst.
239:
240: Since the same dust grains are present in the Galactic interstellar medium,
241: the most detailed studies of the 3.4~$\mu$m absorption have been done on Galactic center
242: sources (Sandford et al. 1991, Pendelton et al. 1994).
243: In these cases the high statistics available allows to resolve the
244: substructure of the absorption feature in at least two major components,
245: due to -CH$_2$ and -CH$_3$ groups, respectively.
246: %The ISAAC spectrum of IRAS 19254-7245 is the first of an extragalactic source
247: %where this sub-structure is clearly identified.
248: In Fig. 2b we plot the ratio between the observed spectrum and the continuum
249: shown in Fig. 1, obtained with a polynomial fit of the regions with
250: observed wavelengths in the ranges 2.9-3.1~$\mu$m, 3.3-3.5~$\mu$m, 3.8-4.1~$\mu$m.
251: A similar spectrum of the Galactic center source IRS7 from Pendelton et al.
252: 1994 is plotted in Fig. 2a. The similarity between the two profiles is
253: remarkable, and suggests a similar composition of the carbonaceous dust
254: grains. A deblending of the single absorption components is quite uncertain:
255: at least three components are present at rest frame wavelengths of 3.38,
256: 3.42 and 3.485~$\mu$m (Fig. 2). Moreover, the absorbed continuum
257: is blended with the long wavelength tail of the PAH emission feature at 3.3~$\mu$m.
258:
259: A qualitative analysis is possible comparing our data with those
260: of the source IRS 7. Following Sandford et al. (1991) the
261: absorption features at 3.38~$\mu$m and at 3.42~$\mu$m are due to
262: -CH$_3$ and -CH$_2$ groups, respectively, in saturated
263: aliphatic hydrocarbons. The ratio between the two absorption peaks is
264: an indication of the average length of hydrocarbon chains\footnote{The
265: structure of a saturated hydrocarbon chain is CH$_3$-(CH$_2)_n$-CH$_3$,
266: therefore the ratio between the strength of -CH$_3$ and -CH$_2$ absorption features
267: is a direct measure of $n$.}.
268: Since the relative strength of the two features
269: in IRAS 19254-7245 is compatible within the errors with
270: those in IRS~7, the average ratio between -CH$_2$ and -CH$_3$ groups
271: is the same, i.e. $\sim 2-2.5$. This
272: corresponds to relatively complex molecules made of 6-7 carbon atoms.
273: %like hexane.
274: The third absorption feature is due to the same C-H stretching vibrations,
275: when the -CH$_2$ and CH$_3$ groups are perturbed by an electronegative group,
276: such as -C$\equiv$N, -OH, or aromatic chains. The higher depth of this
277: feature in IRAS 19254-7245 with respect to the Galactic source IRS~7
278: indicates that the absorbing medium in IRAS 19254-7245 is rich of
279: organic molecules with such electronegative groups (see Fig. 7 in Sandford et
280: al. 1991).
281:
282: %\clearpage
283:
284: \begin{figure}
285: \epsscale{0.8}
286: \plotone{f2.eps}
287: \figcaption{\footnotesize{(a): Continuum-subtracted L-band spectrum of the Galactic
288: center source IRS~7, from Pendelton et al. 1994. (b): The same for IRAS
289: 19254-7245. The similarity in the structure of the absorption feature
290: indicates a common origin, i.e.vibrational transitions of -CH$_3$ and -CH$_2$
291: groups in hydrocarbon grains.
292: }}
293: \end{figure}
294:
295: %\clearpage
296:
297: \subsection{The emission feature at 3.3~$\mu$m}
298: The broad emission feature at 3.3~$\mu$m is due to Polycyclic Aromathic Hydrocarbon
299: molecules.
300: It has been empirically shown that this emission feature is weak or
301: absent in AGNs, while is strong in starbursts (Roche et al.
302: 1991).
303: This is probably due to the destruction of PAH molecules by the strong X-ray
304: and EUV emission
305: of AGNs, and to the dilution of the continuum stellar component . As a consequence, a low
306: equivalent width (EW) of the 3.3~$\mu$m emission feature is an indicator of
307: AGN emission. Typical EW for starburst galaxies are in the range
308: 1000-1500~\AA~(Moorwood 1986, Imanishi \& Dudley 2000), while in AGN-dominated sources
309: EW $\leq$100-300~\AA~(Imanishi 2002).
310: The estimated value for IRAS 19254-7245 is EW=120$\pm10$~\AA, suggesting that
311: the emission is dominated by the active nucleus.
312: %The error on this estimate due to aperture effects
313: %is estimated to be smaller than 10\%.
314:
315: We emphasize that the broad spectral range is crucial for a correct estimate of
316: the EW. In particular, if the continuum at rest-frame wavelengths $\lambda>3.6~\mu$m is
317: not available, the line flux is easily over-estimated,
318: while the continuum at 3.4~$\mu$m is under-estimated because of the
319: broad 3.4~$\mu$m absorption feature. To illustrate this fact,
320: we tried to fit the rest-frame 3.0-3.7~$\mu$m
321: spectrum with a simple power law and an emission line, and we obtained an
322: equivalent width of 700~\AA, in the range typical of starburst-dominated
323: sources.
324: Another possible source of uncertainty in the measurement of the 3.3~$\mu$m feature
325: EW is the overlap with the 3.4~$\mu$m absorption feature.
326: However, the 3.3~$\mu$m emission line appears to be symmetric with respect its
327: peak wavelength (which is exactly at 3.3~$\mu$m), and the absorption feature in the
328: comparison source in Fig. 2a does not extend below $\sim3.34~\mu$m.
329: We conclude that the 3.3~$\mu$m PAH feature is not significantly affected
330: by the absorption at 3.4~$\mu$m.
331: %As a result, the EW of the
332: %PAH emission can be over-estimated by a factor 7-10, up to values as high as
333: %those typical of starburst-dominated sources.
334:
335: It is possible to use the measurement of the 3.3 PAH line to estimate
336: the AGN/starburst contribution to the infrared emission.
337: An interesting indicator
338: %of the starburst/AGN contribution to the
339: %infrared emission
340: is the ratio between the flux of the 3.3~$\mu$m feature,
341: $F_{3.3}$, and the total far infrared flux,F$_{IR}$,
342: as estimated from the four IRAS filters
343: (Sanders \& Mirabel 1996). The typical value for starburst galaxies is
344: $R\sim10^{-3}$, while in AGN-dominated sources $R$ is more than 1-2 orders of magnitude
345: lower (Imanishi 2002). For IRAS 19254-7245 we have F$_{IR}=5.3\times10^{-10}$
346: erg~cm$^{-2}$~s$^{-1}$ and F$_{3.3}=2.1\times10^{-14}$ erg~cm$^{-2}$~s$^{-1}$.
347: As a consequence, $R=3.9\times10^{-5}$.
348: This indicator suggests that the detected starburst can account for only a small
349: fraction of the infrared luminosity of IRAS~19254-7245.
350: %Again, this indicator confirms that
351: %the dominant energy source in IRAS 19254-7245 is an active nucleus.
352:
353: A direct way to obtain a rough quantitative estimate of the AGN contribution to the
354: infrared luminosity is to subtract a starburst template from the L-band spectrum,
355: and then compare the L-band flux $(\lambda f_\lambda)_{3\mu m}$
356: with the total Far Infrared flux $F_{FIR}$ as measured
357: by IRAS. We expect that in case of AGN dominance these two fluxes are $\sim$ equal.
358: We assumed that the PAH emission is entirely due to the starburst, and that the starburst
359: continuum is reproduced by a flat ($f_\lambda=$const) spectrum normalized in order
360: to have EW$_{3.3\mu m}$=1000~\AA (in agreement with starburst spectra in Imanishi \& Dudley 2000).
361: The resulting ``pure AGN'' spectrum is steeper ($\Gamma\sim3)$, with
362: %a 3.4~$\mu$m optical depth
363: $\tau_{3.4 \mu m}=0.9$ and
364: %a 3~$\mu$m flux
365: $(\lambda f_\lambda)_{3\mu m}=6\times10^{-12}$~erg~s$^{-1}$~cm$^{-2}$.
366: Assuming the Galactic extinction curve of Pendelton et al. 1994 ($A_V\sim140-250\tau_{3.4 \mu m}$,
367: $A_L\sim0.06~A_V$, where $A_L$ is the extinction in the L-band), we obtain $A_L\sim8-14$.
368: Correcting the observed 3~$\mu$m flux for this extinction we end up with
369: $(\lambda f_\lambda)_{3\mu m}=2-4500\times F_{FIR}$. This result is an indication that the
370: AGN is dominant in the IR, and that the extinction curve in the {\em Superantennae}
371: is probably quite different from the Galactic
372: one.
373: %: in order to have $(\lambda f_\lambda)_{3\mu m} = F_{FIR}$ the L-band extinction should be
374: %$A_V=6$.
375: %Further details on this issue will be discussed in a forthcoming paper (Risaliti
376: %et al. 2003, in preparation).
377: %\clearpage
378:
379: \begin{figure}
380: \plotone{f3.eps}
381: \figcaption{\footnotesize{1-6~$\mu$m spectrum of IRAS 19254-7245.
382: }}
383: \end{figure}
384:
385: %\clearpage
386:
387: \subsection{Continuum slope}
388: The slope of the continuum at wavelengths $\lambda>3.7~\mu$m is rather steep
389: ($f_\lambda\propto\lambda^{2.7}$). This suggests a strong emission by
390: warm dust, typical of AGN spectra (Granato, Danese \& Francecshini 1997).
391: Interestingly, analyzing the 3-4~$\mu$m spectra of ULIRGs available in the
392: literature
393: %(in particular in Imanishi \& Dudley 2000)
394: we note that a positive
395: slope in the 3.5-4.0~$\mu$m continuum is only found in AGN-dominated sources.
396: The opposite is not true: a few sources exist (MKN 231, NGC 6240) with
397: negative 3.5-4.0~$\mu$m slope, which are known to host a powerful AGN from
398: observations at other wavelengths. This issue will be discussed in further
399: detail in a forthcoming paper
400: %discussing ISAAC observations of 6 more ULIRGs
401: (Risaliti et al. 2003, in preparation).
402: At $\lambda<3.1~\mu$m an inversion
403: of the slope is clearly seen. In Fig. 3 we plot the L-band spectrum
404: with the near-IR spectrum obtained at the NTT (Vanzi et al. 2002)
405: and the ISOCAM point at 6~$\mu$m (Laurent et al. 2000).
406: The extrapolation of the
407: continuum matches well both the K spectrum and the 6~$\mu$m point. This shows
408: that the three observations are well cross-calibrated, and our flux
409: calibration error is lower than $\sim$5\%.
410:
411: The minimum in the L-band spectrum, at $\sim$3.1~$\mu$m is the absolute
412: minimum in the optical-infrared spectrum of IRAS 19254-7245. At shorter
413: wavelengths the emission is dominated by radiation from
414: the host galaxy, since the AGN is highly obscured-, while at longer
415: wavelengths the radiation is mainly due to the reprocessing of the AGN
416: direct emission. Imanishi \& Maloney (2003) detected a broad absorption feature
417: at 3.1~$\mu$m in several ULIRGs, due to ice-covered dust grains. In principle we
418: cannot exclude that the minimum observed at 3.1~$\mu$m is partly due to this effect.
419: However, we note that if a significant fraction of carbonaceous dust grains are covered
420: by ice, they do not contribute to the 3.4~$\mu$m absorption. This would imply that
421: the optical extinction $A_V$ for a given $\tau_{3.4 \mu m}$ is even higher than
422: that assumed in Sect. 3.2. If this is the case, the correction to the
423: observed 3~$\mu$m continuum would also be higher, worsening the problem
424: of the too high 3~$\mu$m flux compared with the FIR emission.
425:
426: \section{Conclusions}
427:
428: We presented a high signal-to-noise L-band spectrum of
429: the ultra-luminous infrared galaxy IRAS 19254-7245.
430: The signatures of a powerful AGN dominating the energy output in the infrared
431: are clear:\\
432: $\bullet$ The broad absorption feature at $\sim3.4~\mu$m, suggests
433: the presence of a powerful point source. The substructure of this feature shows
434: that the absorption is due to hydrocarbon molecules consisting of chains of 6-7 carbon atoms,
435: rich in electronegative groups, as -CN, -OH.\\
436: $\bullet$ The EW of the PAH emission feature at 3.3~$\mu$m, when the continuum is
437: correctly estimated taking into account the broad absorption at 3.4~$\mu$m, is
438: typical of AGN-dominated sources.\\
439: $\bullet$ The continuum at $\lambda>3.7~\mu$m is rather steep, indicating the
440: presence of hot dust.\\
441: The above conclusions show the potential of high signal-to-noise L-band
442: spectroscopic observations with VLT-ISAAC. We are now completing
443: a work in which we extend the present study to a sample of 6 more bright
444: ULIRGs of the Genzel et al. (1998) sample (Risaliti et al. 2003, in
445: preparation).
446:
447: \acknowledgements
448: We gratefully acknowledge Dr. D. Lutz for fruitful discussions, and
449: the anonymous referee for suggestions which significantly
450: improved this Letter.
451: This work has been partially supported by the Italian Ministry of Research (MIUR)
452: and the Italian Space Agency (ASI).
453: \begin{thebibliography}{}
454: \bibitem[]{384} Berta, S., et al. 2003, A\&A, submitted
455: \bibitem[]{385} Braito, V., et al. 2003, A\&A, 398, 107
456: %\bibitem[]{385} Bridger,
457: %A., Wright, G.~S., \& Geballe, T.~R.\ 1994, ASSL Vol.~190: Astronomy with
458: %Arrays, The Next Generation, 537
459: \bibitem[Colina, Lipari, \& Macchetto(1991)]{1991ApJ...379..113C} Colina,
460: L., Lipari, S., \& Macchetto, F.\ 1991, \apj, 379, 113
461: \bibitem[]{386} Charmandaris, V., et al. 2002, A\&A, 391, 429
462: \bibitem[Duc, Mirabel, \& Maza(1997)]{1997A&AS..124..533D} Duc, P.-A.,
463: Mirabel, I.~F., \& Maza, J.\ 1997, \aaps, 124, 533
464: \bibitem[]{387} Genzel, R., et al. 1998, ApJ, 498, 579
465: \bibitem[]{388} Granato, G.L., Danese, L., \& Franceschini, A.
466: 1997, ApJ, 486, 147
467: %\bibitem[]{388} Imanishi, M., Terada,
468: %H., Sugiyama, K., Motohara, K., Goto, M., \& Maihara, T.\ 1997, \pasj, 49, 69
469: \bibitem[]{389} Imanishi, M. 2000, MNRAS, 319, 331
470: \bibitem[]{390} Imanishi, M., \& Dudley, C.C. 2000, ApJ, 545, 701
471: \bibitem[]{391} Imanishi,
472: M., Dudley, C.~C., \& Maloney, P.~R.\ 2001, \apjl, 558, L93
473: \bibitem[]{392} Imanishi, M. 2002, ApJ, 569, 44
474: \bibitem[]{393} Imanishi, M.~\&
475: Maloney, P.~R.\ 2003, \apj, 588, 165
476: \bibitem[]{394} Laurent, O., et al. 2000, A\&A, 359, 887
477: \bibitem[]{395} Mirabel, I.F., Lutz, D., Maza, J., 1991, A\&A, 243, 367
478: \bibitem[Moorwood(1986)]{1986A&A...166....4M} Moorwood, A.~F.~M.\ 1986,
479: \aap, 166, 4
480: \bibitem[]{396} Pendleton, Y.J., Sandford, S.A., Allamandola, L.J.,
481: Tielens, A.G.G.M., \& Sellgren, K. 1994, ApJ, 437, 683
482: \bibitem[]{397} Risaliti, G., Gilli, R., Maiolino, R., \& Salvati, M.
483: 2000, A\&A, 357, 13
484: \bibitem[]{398} Roche, P. F., Aitken, D. K., Smith, C. H., \& Ward, M.
485: 1991, MNRAS, 248, 606
486: \bibitem[Sanders \& Mirabel(1996)]{1996ARA&A..34..749S} Sanders, D.~B.~\&
487: Mirabel, I.~F.\ 1996, \araa, 34, 749
488: \bibitem[]{399} Sandford, S.A., Allamandola, L.J., Tielens, A.G.G.M.,
489: Sellgren, K., Tapia, M., \& Pendleton, Y.J. 1991, ApJ, 371, 607
490: \bibitem[]{401} Vanzi, L., Bagnulo, S., Le Floc'h, E., Maiolino, R.,
491: Pompei, E., \& Walsh, W. 2002, A\&A, 386, 464
492: \end{thebibliography}
493:
494: \end{document}
495:
496: