Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.eso.org/~rsiebenm/sb_models/SKS.pdf
Дата изменения: Tue Mar 28 14:49:06 2006
Дата индексирования: Tue Oct 2 02:41:02 2012
Кодировка:
Поисковые слова: m 87 jet

A&A 414, 123139 (2004) DOI: 10.1051/0004-6361:20031633
c ESO 2004

Astronomy & Astrophysics

Midinfrared emission of galactic nuclei
TIMMI2 versus ISO observations and models
R. Siebenmorgen1 ,E.Krugel2 , and H. W. W. Spoon3 Ё
1 2 3

European Southern Observatory, Karl-Schwarzschildstr. 2, 85748 Garching b. Munchen, Germany Ё Max-Planck-Institut for Radioastronomy, Auf dem Hugel 69, Postfach 2024, 53010 Bonn, Germany Ё Kapteyn Institute, PO Box 800, 9700 AV Groningen, The Netherlands

Received 24 June 2003 / Accepted 16 October 2003
Abstract. We investigate the midinfrared radiation of galaxies that are powered by a starburst or by an AGN. For this end, we compare the spectra obtained at different spatial scales in a sample of infrared bright galaxies. ISO observations which include emission of the nucleus as well as most of the host galaxy are compared with TIMMI2 spectra of the nuclear region. We find that ISO spectra are generally dominated by strong PAH bands. However, this is no longer true when inspecting the midinfrared emission of the pure nucleus. Here PAH emission is detected in starbursts whereas it is significantly reduced or completely absent in AGNs. A physical explanation of these new observational results is presented by examining the temperature fluctuation of a PAH after interaction with a photon. It turns out that the hardness of the radiation field is a key parameter for quantifying the photodestruction of small grains. Our theoretical study predicts PAH evaporation in soft Xray environments. Radiative transfer calculations of clumpy starbursts and AGN corroborate the observational fact that PAH emission is connected to starburst activity whereas PAHs are destroyed near an AGN. The radiative transfer models predict for starbursts a much larger midinfrared size than for AGN. This is confirmed by our TIMMI2 acquisition images: We find that the midinfrared emission of Seyferts is dominated by a compact core while most of the starbursts are spatially resolved. Key words. infrared: galaxies galaxies: ISM galaxies: nuclei galaxies: dust

1. Introduction
Luminous infrared galaxies in the local universe with infrared luminosities exceeding 1011 L were discovered in the IRAS all sky survey (Soifer et al. 1987). Responsible for their huge luminosity may be a combination of starburst and AGN activity. In both cases, the nuclear emission is fuelled by an enormous concentration of gas and dust in the central region (Sanders & Mirabel 1996). Luminous infrared galaxies are also found to be a major component in the early universe (z > 1). Deep midinfrared surveys with the infrared space observatory, ISO, could identify individual objects which account for most of the cosmic infrared background light. It is argued that the emission of most of the distant infrared galaxies comes mainly from starbursts (Elbaz et al. 2002). This hypothesis may also be tested by inspecting the spectra of local galaxies. ISO spectra of local galaxies (Rigopoulou et al. 1999; Siebenmorgen et al. 1999;
Send offprint requests to: R. Siebenmorgen, e-mail: rsiebenm@eso.org Based on ESO: 68.B-0066(A) and observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries: France, Germany, The Netherlands and the UK) with the participation of ISAS and NASA.

Clavel et al. 2000; Laureijs et al. 2000; Laurent et al. 2000) usually show a set of infrared emission bands which are attributed to the emission of polycyclic aromatic hydrocarbons (PAH). The question as to which source, on average, is responsible for the luminosity of the cosmological background is still a matter of intense debate and, at the moment, different answers are given for different wavelength regions. The Xray background could be partly resolved by ROSAT into individual components and Hasinger et al. (1998) showed that at Xrays AGNs make the greatest contribution. At infrared wavelengths, on the other hand, starbursts seem to dominate, as infered from numerous ISO observations (Genzel & Cesarsky 2000). We want to re-address the question whether infrared bright galaxies are predominantly powered by dust-enshrouded AGN or starbursts. As a new observational technique, we employ midinfrared subarcsec imaging and midinfrared spectroscopy at various spatial scales. (For previous ground based midinfrared spectroscopy see for example: Smith et al. 1989; Roche et al. 1991 and Soifer et al. 2002). The midinfrared wavelength region is of particular interest as the optical depth is usually sufficiently low to penetrate the dust enshrouded nucleus. Modern midinfrared cameras have adequate spatial resolution to estimate the size of the emitting regions which is an important parameter to constrain the nature of the underlying

124

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei Table 1. A few parameters of our IRG sample. Velocities (Col. 3) and distances (Col. 4) are from NED. Column 5 gives the linear size of 1 at the distance D, and Col. 6 the infrared luminosity. v km s- Centaurus A Circinus IC 4329a IRAS 05189-2524 IRAS 08007-6600 M 83 Mrk 509 Mrk 1093 Mrk 1466 NGC 1068 NGC 1365 NGC 1386 NGC 2966 NGC 3256 NGC 4388 NGC 5506 NGC 5643 NGC 6000 NGC 6240 NGC 7552 NGC 7582 NGC 7674 PKS 2048-57 Sy 2. Sy 2. Sy 1. Sy 2. HI I SB Sy 1. SB SB Sy 2. Sy 1. Sy 2. SB SB Sy 2. Sy 1. Sy 2. SB Sy 2. SB Sy 2. Sy 2. Sy 2. 0 0 2 0
1

source: at comparable distances of the sources a resolved mid infrared emission points towards a starburst galaxy while an unresolved nucleus indicates an AGN or an extreme compact star cluster. As we will demonstrate in model calculations, infrared emission bands are a diagnostic tool to derive the activity type. The ratio of AGN versus starburst luminosity of local galaxies may be quantified by mid infrared spectra observed at different spatial resolutions. The paper is structured as follows: in Sect. 2 we introduce the sample of infrared bright galaxies observed with ISO and TIMMI2. The observation are summarised in Sect. 3, the data analysis procedure is explained in Sect. 4. Observational results of individual galaxies are given in Sect. 5. Sections 6 and 7 presents radiative transfer calculations, which also illustrate the effect of the optical depth on the SED, and a discussion of the heating and evaporation of PAH in starburst and AGN environments.

Name

Type

D Mpc 3.5 4 65 176 170 4 141 60 18 14 22 18 27 37 34 25 16 29 100 21 21 118 46

1 pc 17 19 315 852 822 19 684 290 87 68 106 87 131 178 164 120 78 141 483 102 102 573 222

L 1010 L 0.68 1.4 7.6 120 34 0.65 14 9.3 0.7 19 8.3 0.71 1.1 35 1.2 2.3 1.8 9.0 60 9.8 6.3 31 6.1

2

2. The sample
The ISO archive contains more than 250 galaxies that have been spectroscopically observed in the midinfrared. We have selected 22 of these galaxies for observation with TIMMI2 to compare the large aperture ISO spectra to the narrow slit TIMMI2 spectra. The selected galaxies all have 10 µm ISO flux densities above 100 mJy and are visible from La Silla (-29 South). Basic properties of the sample are presented in Table 1. The flux criterion was prompted by the sensitivity of TIMMI2 in spectroscopic mode which is 100 mJy/10 in 1 hour. The sample is heterogenous with respect to the galaxy type: it contains six starbursts (SB), one HII, two Sy 1 and thirteen Sy 2. The distances range from 3.5176 Mpc and the luminosities from 0.7120 в 1010 L . Our TIMMI2 sample also contains sources lacking an ISO spectrum. These sources were selected from the Markarian catalogue (Markarian et al. 1989) to fill gaps in our primary observing schedule.

0 8 0

0 9 0 0 0 0 0

547 439 4813 12 760 12 324 516 10 312 4441 1327 1148 1636 868 2044 2738 2524 1853 1199 2193 7339 1585 1575 8671 3402

3. Observations
The N band imaging and spectroscopic observations were made with the TIMMI2 midinfrared camera (Reimann et al. 2000; Kaufl et al. 2003) at the ESO 3.6 m telescope. The camЁ era uses a 240 в 320 pixel Raytheon Si:As array. In imaging we selected the 0.2 pixel scale providing a total field of view of 48 в 64 . For longslit spectroscopy the resolving power is / 400 and we used the 1.2 or 3 slit and 0.45 pixel field of view. Table 2 reports for each galaxy: observing date, total onsource integration time, filter and calibration standard. The filter is designated by its central wavelength and the entry grating refers to 813 µm spectroscopy. Observations were performed in chopping and nodding mode with 12 amplitude each, which is a compromise between sensitivity and the ability to measure extended structures. To avoid saturation of the detector by the high ambient photon background the elementary integration times were typically 20 ms for imaging and 40 ms for spectroscopy. Elementary images were coadded in real time during the chopping cycles. Chopping was in NorthSouth direction at a

frequency of 3Hz. After an interval of typically less than a few minutes the telescope was nodded in order to cancel the telescope emission and residuals in the sky subtraction. After coaddition of all chopping and nodding pairs there are two negative and one positive beam (slit) on the image which we combine applying a shiftandadd procedure. We note that TIMMI2 is not capable of measuring low surface brightness emission that is extended by more than half the chopping or nodding amplitude. Consequently, part of the diffuse emission is cancelled by the observing technique. TIMMI2 followup observations of ISO galaxies were obtained during January 28February 3, August 1314, 2002 and 4 photometric imaging observations on January 25, 2003. The observations of the targets were interleaved with those of nearby midinfrared calibration standard stars, typically 45 of which were measured each night. Model spectra for the standards have an absolute photometric uncertainty of a few percent (Cohen et al. 1999). However, for stars of spectral type later than F there might be fundamental molecular lines present in the photospheric spectra, such as SiO in the 9 µm region (Heras et al. 1998). These bands are not considered by the present photospheric models. Consequently, one must be careful not to over-interpret faint structures at these wavelengths. Furthermore close to the blue and red cutoff frequencies of the grating the terrestrial background varies strongly with ambient conditions and atmospheric corrections are more difficult.

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei Table 2. Log of TIMMI2 observations. Table 2. continued. Target NGC 7552 Date (2002) A A A A A A A A A ug. ug. ug. ug. ug. ug. ug. ug. ug. 15 15 15 14 14 14 15 14 14 Time (s) 929 2580 3096 619 2580 2709 619 619 2460 Filter (µm) 11.9 grati grati 11.9 grati grati 11.9 11.9 grati ng ng ng ng Calibrator HD HD HD HD HD HD HD HD HD 1522 169916 1522 178345 178345 1522 178345 178345 178345

125

Target Centaurus A Circinus IC4329A IRAS 05189-2524

Date (2002) Jan. 30 Jan. 30 Jan. 30 Jan. 30 Jan. 29 Jan. 29 Feb. 01 Feb. 01 25.01.03 25.01.03 Feb. 01 Feb. 01 Feb. 01 25.01.03 Jan. 30 Jan. 30 Jan. 30 Jan. 30 Aug. 14 Aug. 14 Aug. 14 Feb. 02 Feb. 02 Feb. 04 Feb. 02 Feb. 02 Jan. 29 Jan. 29 Jan. 31 Jan. 31 Jan. 30 Jan. 30 Jan. 31 Jan. 31 Aug. 15 Aug. 15 Feb. 03 Jan. 31 Jan. 29 Jan. 29 Jan. 31 Feb. 03 Feb. 03 Feb. 02 Feb. 02 Feb. 02 Feb. 02 Aug. 14 Aug. 14 Jan. 29 Feb. 01 Feb. 01 Aug. 13 Aug. 15 Aug. 13 Aug. 15

Time (s) 645 3225 268 967 403 2419 361 2257 600 600 361 469 1935 360 322 301 2419 322 619 619 2580 602 2903 903 903 2419 268 1209 150 604 645 3225 150 1935 619 2580 602 180 537 3628 3064 602 2419 602 1612 602 1612 602 1612 215 301 2096 619 619 3096 2580

Filter (µm) 10.4 grating 10.4 grating 10.4 grating 8.6 grating 10.4 11.9 8.6 11.9 grating 11.9 10.4 11.9 grating 10.4 11.9 11.9 grating 11.9 grating 8.6 11.9 grating 10.4 grating 8.6 grating 10.4 grating 8.6 grating 11.9 grating 11.9 8.6 10.4 grating grating 11.9 grating 11.9 grating 11.9 grating 11.9 grating 10.4 8.6 grating 11.9 11.9 grating grating

Calibrator HD81420 HD 81420 HD 81420 HD 81420 HD 81420 HD 81420 HD 32887 HD 32887 HD 123139 HD 123139 HD 32887 HD 32887 HD 32887 HD 123139 HD 81420 HD 81420 HD 81420 HD 81420 HD 196171 HD 196171 HD 196171 HD 32887 HD 32887 HD 123139 HD 123139 HD 123139 HD 32887 HD 32887 HD 32887 HD 32887 HD 32887 HD 32887 HD 32887 HD 32887 HD 1522 HD 1522 HD 123139 HD 123139 HD 81420 HD 81420 HD 123139 HD 123139 HD 123139 HD 123139 HD 123139 HD 123139 HD 123139 HD 123139 HD 123139 HD 123139 HD 123139 HD 123139 HD 123139 HD 133774 HD 123139 HD 133774

NGC 7582

NGC 7674 PKS 2048-57

ng

IRAS 08007-6600

M83

Mrk 509

The same holds for wavelength regions of strong terrestrial lines near 9.58 µm, 11.73 µm and 12.55 µm. The N band imaging of the stars was performed in the same filter as the target and used for photometric flux conversion from photon count rates into astronomical units (mJy) and to establish the point spread function (PSF). The spectra of the stars were used for atmospheric corrections and for flux calibration of the target spectrum.

Mrk 1093 Mrk 1466

4. Data analysis

4.1. Midinfrared imaging
We apply multiaperture photometry on the shiftandadd processed image, which we refer to as raw or original image. For each aperture centred on the brightest pixel of the source we estimate the background as the mean flux derived in a 6 pixel wide annulus which is put 6 pixels away from the outer source aperture. The thus estimated aperture fluxes, F (r), were plotted against the aperture radii, r, to provide the growth curve. The flux where the growth curve of the aperture photometry flattens was identified interactively and is taken as the total flux density. The absolute photometric uncertainty is 10%. This error estimate is based on internal consistency of the calibration observations and monitoring of photometric calibration standard stars. We note that part of this error is due to large scale (10 ) sensitivity gradients on the detector array. Up to now no suitable flat fielding procedure has been established to further reduce the absolute photometric error. The growth curve is used to extract spatial information. For example, if one calculates the growth curve of an Airy function up to its first dark ring then the aperture radius, r50% ,where the flux is at 50% of its maximum value is half of the full width at half maximum (FWHM). Still the diffraction pattern of the telescope and the seeing has to be subtracted. If for an extended e target one derives a 50% flux radius r50% and for a PSF a rap dius r50% , a size estimate of the source extension may be app o e proximated by (r50% )2 = (r50% )2 - (r50% )2 . The size estimate by the growth curve depends on the total flux or a normalisation at large radii. Another diagnostic of the source extension is given by the differential flux profile, which is a function F (ri+i ) - F (ri ), and as well difficult to normalise, this time at o small radii. We take the sum of the aperture fluxes at r r50% .

NGC 1068

NGC 1365

NGC 1386 NGC 2966 NGC 3256

NGC 4388 NGC 5506 NGC 5643

NGC 6000

NGC 6240

126

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei

Table 3. Midinfrared flux densities of TIMMI2 observations. Photometric uncertainty 10%. Name Centaurus A Circinus IC 4329A IRAS 05189-2524 IRAS 08007-6600 Mrk 509 Mrk 1093 Mrk 1466 M 83 NGC 1068 NGC 1365 NGC 1386 NGC 2966 NGC 3256 NGC 4388 NGC 5506 NGC 5643 NGC 6000 NGC 6240 NGC 7552 NGC 7582 NGC 7674 PKS 2048-57 F (8.6) (Jy) 0.42 0.21 0.09 14.8 0.40 0.45 0.21 F (10.4) (Jy) 0.65 4.10 0.64 0.42 0.11 17.6 0.46 0.39 0.20 F (11.9) (Jy) 0.57 0.23 0.22 0.16 0.11 0.23 0.40 <0.03 0.29 1.06 0.31 0.28 1.87 0.69 0.26 0.92

Fig. 1. NGC 3256 at 10.4 µm. On the raw (topleft), PSF (topmiddle) and deconvolved image (bottom-left) contours are in black at 10%, 30% and 50% and in white at 75%, 90% and 95% levels of the peak flux; dotted contours are at -1 and -3 and dashed contours at +1 and +3, respectively. On the residual map (bottom-middle) black contours are at -98%, -90%, -50%, white contours at 50%, 90%, 98% of the peak. The peak fluxes and values are specified in Table 4. For all gray scale images North is up and East is left. Growth curves (topright) and differential flux distributions (bottomright) are shown for raw (dashed), PSF (dotted) and deconvolved image (full line). Error bars are 1 rms.

With such normalisation the differential flux distribution of an extended target should be less steep than for the PSF. The spatial resolution of the images are at best limited by the diffraction pattern of the 3.6 m telescope, which provides a FWHM of 0.7 at 10 µm. In oder to increase the spatial resolution we applied a deconvolution procedure. We use the maximum entropy multiresolution wavelet decomposition by Starck & Murtagh (2002). The gradient algorithm has been chosen with 4 wavelet scales assuming Gaussian distribution of the noise and we applied a 5 threshold, where is the noise as estimated on the object free part of the image. The method provides a residual map, R, which is a convolution of the deconvolved image, D, with the PSF, P, and subtracted from the original image, O; symbolically: R = D P - O. The iteration of the deconvolution is stopped so that the noise properties of the residual map is at similar level as the noise of the original image. However, for bright pointlike sources with little extension we note that the residual map is polluted by the very bright central peak. Residuals are typically less than 5% of the peak flux of the deconvolved image. They are caused by finite sampling of a type function. From the growth curve of the d deconvolved image we calculate its 50% flux radius, r50% . The residual map is used for error propagation To study the stability of the PSF and its influence on the deconvolution procedure, we calculate for each PSF the deconvolution and growth curves with any other PSF available of the same filter. For the deconvolved PSF images we find a d mean size of r50% = 0.098 ± 0.046 . Size estimates from the growth curves as measured on the raw images give a mean of

o r50% = 0.224 ± 0.074 , which already closely matches the static aberration of the telescope. Altogether we verified that seeing in the mid IR gives only a small contribution to the PSF at the 3.6 m. Nevertheless for deconvolution we apply the measured PSF matched closely in time to the target observation. A source is regarded to be unresolved if its radius measured on the raw and deconvolved image is below r50% + 3(r50% ). d o Therefore, resolved sources have r50% > 0.24 and r50% > 0.44 , respectively. For galactic nuclei fulfilling these criteria, the differential source profile also shows signatures of extended emission. Observed midinfrared flux densities for the galactic nuclei are given in Table 3. For 10 out of 22 galaxies the nucleus is resolved. For these we show in Figs. 111 raw, PSF, deconvolved and residual maps and present growth curves as well as differential flux distributions. Absolute levels of the contours, shown on top of the gray scale images, can be derived together with Table 4, where we specify the peak flux densities of the raw and deconvolved images and the residual map. For the residual map we specify the 1 rms noise level. The 1 rms noise of the raw image is given in Col. 3 of Table 5. It is an indicator of the sensitivity limit which we reached with TIMMI2. Table 5 lists the central wavelength of the TIMMI2 filter (Col. 2) together with the 50% flux radius, r50% , as derived from the raw (Col. 4) and deconvolved (Col. 5) image. The latter radius is our estimate of the source size (Col. 6).

4.2. TIMMI2 spectroscopy
The coaddition of all chopping and nodding pairs of raw frames in TIMMI2 spectroscopic observing mode gives two negative

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei

127

Fig. 2. NGC 7552 at 11.9 µm, see Fig. 1. On the raw image (topleft), the PSF and the deconvolved image contours are set in black at 8%, 20% and 30% and in white at 50%, 75%, and 95% levels of the peak flux

Fig. 5. Centaurus A at 10.4 µm, see Fig. 1.

Fig. 6. NGC 1068 at 10.4 µm, see Fig. 1. Fig. 3. Mrk 1093 at 11.9 µm, see Fig. 1.

Fig. 7. NGC 1386 at 11.9 µm, see Fig. 1. Fig. 4. Circinus at 10.4 µm, see Fig. 1.

and one positive longslit spectrum on the image. We developed an optimal extraction procedure for TIMMI2 spectra similar to Horne (1986): a source profile is extracted by collapsing the image along the dispersion direction and applying a median

filter. Extension of the sources are considered to a 2 cutoff, where is calculated for each column in crossdispersion direction and on the source-free part of the array. Preliminary wavelength calibration is performed using tables provided by the instrument team. This calibration is fine-tuned to sub-pixel accuracy by applying a linear wavelength shift. The shift is measured as an offset to atmospheric lines. The procedure is

128

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei

Fig. 8. NGC 7582 at 11.9 µm, see Fig. 1. On the raw image, the PSF and the deconvolved image contours are set in black at 14%, 20% and 50% and in white at 75%, 90%, and 95% levels of the peak flux, respectively.

Fig. 11. NGC 7674 at 11.9 µm, see Fig. 1.

Table 4. Peak surface brightness and rms of nuclei with extended mid infrared emission. Name Raw Peak
mJy arcsec2

Deconvolved Peak
mJy arcsec2

Peak
mJy arcsec2

Residual rms
mJy arcsec2

Fig. 9. Mrk 509 at 11.9 µm, see Fig. 1.

Centaurus A Circinus Mrk 509 Mrk 1093 NGC 1068 NGC 1386 NGC 3256 NGC 6240 NGC 7552 NGC 7582 NGC 7674

881 3902 637 52 10 078 512 296 280 997 275 287

2301 14 806 10 10 31 196 136 211 245 277 91 123

100 167 2.6 7.5 371 33 35 56 110 12 16

17 31 0.4 1.8 80 4.7 7.3 11 10 2.3 4.0

astronomical (Jy) units. The calibrated spectra are finally high frequency noise filtered by the method described by Starck et al. (1997). Statistical errors are computed from unfiltered data. Because not all detector channels were always fully operational, the spectra of M 83, CenA, Circinus, IC 4329A, and NGC 1068 could only be extracted longward of 9 µm.

4.3. ISO spectroscopy
Whenever possible, we compare midinfrared spectra at different spatial scales: highresolution longslit TIMMI2 spectroscopy (1 -3 ) and ISO spectra of much lower resolution (24 в 24 with ISOPHT, Lemke et al. 1996; 14 в 20 with ISOSWS in the N band, de Graauw et al. 1996; and ISOCAM spectra which refer, unless otherwise stated, to the total emission of the galaxy, Cesarsky et al. 1996). The diffraction limits of the telescopes at 10 µmare: 0.7 at the ESO 3.6 m and 8 with ISO. For the brightest objects of our sample ISOSWS grating or line scans are reported by Thornley et al. (2000), Sturm et al. (2002) and Verma et al. (2003); ISOCAM spectra

Fig. 10. NGC 6240 at 11.9 µm, see Fig. 1.

applied to the target and the calibration standard star. A division of both spectra removes the telluric lines. To minimise residuals of the sky line cancellation, we observed standard and target at similar airmasses. From the standard star spectrum we calculate the conversion factor from technical (ADU/s) into

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei Table 5. Midinfrared sizes of galactic nuclei. 1 Name 2 µm Centaurus A Circinus IC4329a IRAS 05189-2524 IRAS 08007-6600 Mrk 509 Mrk 1093 NGC 1068 NGC 1068 NGC 1365 NGC 1386 NGC 3256 NGC 4388 NGC 5506 NGC 6240 NGC 7552 NGC 7582 NGC 7674 PKS 2048-57
b o

129

Table 6. Important features and lines between 8 to 13 µm. 4 5
d r50%

3 b
mJy arcsec2

6
d r50%

Name

o r50%

a µm 8.6 8.991 9.009 9.537 9.660 9.7 10.511 11.3 12.814 12.7

Ib eV - 27.6 186.5 99.1 - - 34.8 - 21.6 -

Mode

pc <0.37 0.46 <0.28 <0.24 <0.23 1.2 1.9 0.53 0.58 <0.28 0.50 1.14 <0.30 <0.30 1.11 2.79 1.18 0.6 <0.2 0.20 0.20 0.21 0.20 0.27 2.4 1.69 0.27 0.35 <0.35 0.53 0.88 <0.36 <0.20 0.47 2.65 0.87 0.46 <0.23 < < < < < <3 <4 <66 <170 <222 1642 490 18 24 <37 46 157 < 59 <24 227 270 89 264 <51

10. 10. 10. 8. 8. 11. 11. 8. 10. 8. 11. 10. 11. 11. 11. 11. 11. 11. 11.

4 4 4 6 6 9 9 6 4 6 9 4 9 9 9 9 9 9 9

14 22 19 18 66 10 13 48 18 64 42 28 33 22 31 82 17 22 15

PAH Ar I I I Mg VII F e VI I H2 0-0 S(3) Silicate S IV PAH Ne I I PAH
a b

C- H - - - - Si-O - C- H - C- H

Rest wavelength. Ionisation energy with respect to the next lower ionisation level.

d

1 rms noise measured on the objectfree part of the raw image. Midinfrared radius where the flux is at 50% of the total as derived from the growth curves of the object (raw) and PSF images. Midinfrared radius where the flux is at 50% of the total as derived from the growth curve of the deconvolved image.

hard radiation only produced by AGNs (>200 eV). At the resolving power of TIMMI2 (/ 400) the lines are generally unresolved. The [Ar III] and [Mg VII] line occur at the same wavelength but their excitation requirements are very different. The [Fe VII] line may be blended by the ozone band, [S IV] is in a clean atmospheric region. The PAH band emission at 12.7 µm is close to the [Ne II] at 12.8 µm. However, as we will demonstrate by model calculations in Sect. 7, detection or nondetection of PAH features can serve as a diagnostic tool to distinguish the dominating excitation within the nucleus: starbursts or AGN; although the dividing line is not sharp. The mid-IR spectra of spirals show dust continuum and strong PAH bands. Close to massive stars there is additional hot dust continuum emission and some photodestruction of PAH (Siebenmorgen 1993). So starburst galaxies still show PAH bands but at lower PAH bandtocontinuum ratios than observed in spirals. In the harsh environment close to an AGN one expects much lower PAH bandtocontinuum ratios than in starbursts and most often PAH bands are absent. In the immediate environment of the hard X-ray and intense UV radiation of an AGN the PAH are destroyed. In addition significant emission by hot dust from the inner torus wall of the unified model is expected which further decreases the PAH bandtocontinuum ratios. The silicate band if present in the galactic nuclei is seen so far only in absorption. From its depth one can estimate the visual extinction. We apply V 18 · 9.7 µm . However, 9.7 µm is difficult to estimate without detailed model calculations of the PAH contribution to the midinfrared continuum (Siebenmorgen et al. 2001). Whenever possible we interpolate a line to the spectra from 8to 11 µm and so derive a crude estimate of 9.7 µm and the total visual extinction. This procedure maybe uncertain within a factor 2.

by Laurent et al. (2000), and ISOPHT spectra by Clavel et al. (2000), Laureijs et al. (2000), Rigopoulou et al. (1999), Siebenmorgen et al. (1999) and Spoon et al. (2002). For the remaining galaxies (Cen A, Circinus, IC 4329A, IRAS 08007-6600, Mrk 509, NGC 1068, NGC 1365, NGC 1386, NGC 3256, NGC 4388, NGC 5506, NGC 5643, NGC 6000, NGC 6240, NGC 7552, NGC 7582) we extracted ISOPHT raw data from the ISO archive and reduced the spectra with standard routines of the ISOPHT interactive analysis system (PIA). Most of the ISOPHT spectra appear for the first time in the refereed literature.

5. Results
Here we present our high resolution midinfrared imaging and spectroscopy data using TIMMI2 together with the low resolution ISO spectra. The data of the individual objects are briefly discussed and whenever necessary compared to observations available at other wavelengths. We will first describe the starbursts and then the Seyferts. A variety of broad-band dust features and fine-structure lines falls within the N band; they are summarised in Table 6. Gas emission lines imply either the presence of ionising shocks (Contini & Contini 2003) or a hard photon flux. For ionisation potential below 50 eV, photoionisation by hot stars is the most likely explanation for the excitation mechanism. In the 813 µm spectral range there are no lines which require very

130

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei

Fig. 12. Midinfrared flux densities (in Jy) of starburst galaxies: Data presented are: ISOPHT (24 aperture, histogram), TIMMI2 spectroscopy (3 , full line) with 3 errors (shadow) and TIMMI2 photometry with 1 error bars (Table 3).

5.1. Starburst galaxies: Imaging and spectroscopy
5.1.1. NGC 6000
ISO observations and radiative transfer models for the dust in NGC 6000 are presented by Siebenmorgen et al. (1999). At 12 µm, the flux from IRAS is stronger than from ISOPHT which suggests that the mid IR source is extended beyond 24 . TIMMI2 images at 8.6 and 10.4 µm are fuzzy, without a point source. Both spectra in Fig. 12 are dominated by PAH emission. In the TIMMI2 spectrum, the 11.3 µm PAH band tocontinuum ratio is about two times smaller (Table 8). The [Ne II] line is detected, but the 12.7 µm PAH band is seen only marginally. We estimate from the depth of the silicate absorption in the ISOPHT (TIMMI2) spectrum a visual extinction of 25 mag (20 mag); both values are consistent with AV = 29 mag inferred from the radiative transfer models.

bandtocontinuum ratio is lower in the smaller aperture spectrum by about a factor 2 (Table 8). There is strong [Ne II] emission (Table 7). Due to the 8.6 µm PAH band and because the atmospheric ozone band at 9.5 µm has not been fully eliminated, it is difficult to say whether there is silicate absorption at all.

5.1.4. NGC 7552
The deconvolved 11.9 µm TIMMI2 image at 0.45 angular resolution (Fig. 2) shows four hot spots, equidistant from the center, that form a ringlike circumnuclear starburst. Two of the hot spots have already been detected by Schinnerer et al. (1997) in their 10.5 µm map with 0.8 resolution. These authors also present near infrared images and spectral synthesis models consistent with the idea of a starburst ring. The slit during the TIMMI2 grating observations was positioned at the brightest knot, 3 North of the centre; the [Ne II] line is visible as well as the 11.3 µm PAH band (Table 7). The ISOPHT spectrum shows the [Ar II] line and strong PAH emission. The 11.3 µm PAH bandtocontinuum ratio is a factor 3 lower in the smaller aperture spectrum (Table 8). Silicate absorption is weak in both spectra, it corresponds to AV 3 mag (Fig. 12).

5.1.2. Mrk 1466
The signal-to-noise in the TIMMI2 image is low, hence we cannot determine the source size. The only feature in the TIMMI2 spectrum detected at high confidence is the 11.3 µm PAH band (Fig. 12 and Table 8). ISO spectra are not available for comparison.

5.1.3. NGC 3256
The brightness in th center. The emission in the core (Fig. 1). dominated by PAH e TIMMI2 image increases towards the is resolved and there is no point source Both ISOPHT and TIMMI2 spectra are emission (Fig. 12). The 11.3 µm PAH

5.1.5. Mrk 1093
The source at 11.9 µm is extended in the EastWest direction (Fig. 3). TIMMI2 spectroscopy (Fig. 12) reveals the [Ne II] line and PAH bands at 8.6 and 11.3 µm. ISO spectra are not available for comparison.

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei Table 7. Integrated atomic line fluxes and 3 upper limits (10-20 W/cm2 ) of TIMMI2 and ISOSWS spectra. Name Aperture Centaurus A Circinus IRAS 05189+2524 IRAS 08007-6600 IC 4329a M83 Mrk 1093 Mrk 1466 Mrk 509 NGC 1068 NGC 1365 NGC 1386 NGC 3256 NGC 4388 NGC 5506 NGC 5643 NGC 6000 NGC 6240 NGC 7552 NGC 7582 (1.2 ) NGC 7582 (3.0 ) PKS 2048-57
a

131

[S IV] 3 <2.4 <3.6 <1.7 <1.2 1.8 ± 0.5 <0.3 <2.2 <1.8 <1.3 30.1 ± 6.2 <1.7 <2.8 <1.6 3.4 ± 0.9 3.5 ± 0.9 <1.3 <2.0 <1.4 <8.3 <3.7 <1.6 <2.7

[S IV] 14 в 20 1.4b 12.7b

[Ne II] 3 22.6 ± 1.4 6.6 ± 0.9 1.6 ± 0.3 1.7 ± 0.2 3.0 ± 0.5 <2.3 <1.3 <14.2 <1.8 <2.8 28.5 ± 1.0 5.4 ± 0.9 2.9 ± 0.9 1.6 ± 0.4 15.1 ± 1.1 5.0 ± 1.6 49.0 ± 5.6 <8.8 8.9 ± 1.2 <3.6

[Ne II] 14 в 27 22.1b 90b

58b 2.6b 0.9c 5.4b

1.3b 70b 40.9b 89.2c 5.9b

0.3 1.8b 1.8b

c

17a 68c 14.8b 14.8b 2.1b

Thornley et al. (2000), b Sturm et al. (2002), c Verma et al. (2003).

10.0

M 83

1.0

for the presence of 8.6 µm PAH emission. This feature is, however, present in the larger aperture ISOPHT spectrum (Laureijs et al. 2000). The absence may be due to low signal-to-noise of our TIMMI2 spectrum. The silicate absorption is weak and indicates AV 6 mag from the central 200 pc (Table 5).

GC

5.1.7. M 83
M 83 is the nearest starburst galaxy in our sample. TIMMI2 images at 10.4 and 11.9 µm show a fuzzy structure without striking point sources. As typical for starburst galaxies, the large aperture ISOPHT spectrum is dominated by PAH emission bands. In contrast, the TIMMI2 spectrum of the central 60 pc does not show any sign of PAH emission features and is dominated instead by a smooth continuum with a weak [Ne II] emission line (Fig. 13). The shape of the continuum is consistent with the profile of the silicate absorption feature as found towards the ISOSWS position Sgr A in the Galactic center (Lutz 1999). Given their similarity, we speculate that the conditions in the nucleus of M 83 may be similar to those in the Galactic center. In the latter, PAHs are thought to be destroyed by the local intense radiation field (106.5в ISRF) contributed by massive stars found in the Galactic center (Lutz 1999). Indeed far infrared spectroscopic images of fine structures lines in M 83 indicate a strong central starburst activity headed by O9 stars (Stacey et al. 1998). So that this is our best example where PAH destruction occurs by the strength of the radiation field rather

0.1

6

7

8

9 10 11 wavelength (µm)

12

13

Fig. 13. Midinfrared flux densities (in Jy) of M 83: data presented are: ISOPHT (24 aperture, histogram), TIMMI2 spectroscopy (3 , full line) with 3 errors (shadow) and TIMMI2 photometry with 1 error bars (Table 3). For comparison we show the spectrum of the Galactic Center (dashed).

5.1.6. IRAS 08007-6600
This is one of the most distant galaxies in our sample (170 Mpc). The TIMMI2 images at 8.6 µm and 11.9 µm are unresolved. The energy distribution peaks near 60 µm (Vader et al. 1993). The TIMMI2 spectrum does not show evidence

132

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei

than by its hardness. The latter process is discussed in Sect. 6. From the depth of the silicate feature we estimate a visual extinction 20 mag.

5.2. Seyfer t galaxies: Imaging and spectroscopy
5.2.1. Circinus
Circinus is the closest galaxy of our sample that contains an obscured Seyfert nucleus. In the TIMMI2 10.4 µm image (Fig. 4), half the flux comes from an unresolved core, the other half from a 1 sized blob elongated NorthSouth. At a comparable scale, resolved emission at 8.5 µm is reported by Siebenmorgen et al. (1997). The photometry of ISOPHT (this work) and ISOSWS (Moorwood et al. 1996) give similar fluxes for > 9 µm, but ISOSWS is systematically weaker 20%30% between 6-9 µm. From the silicate absorption feature we estimate a visual extinction AV > 15 mag. In the large aperture ISOSWS spectrum, one sees the silicate absorption feature, strong line emission ([S IV], [Ne II] ) and PAH bands. In the TIMMI2 spectrum, [S IV] and [Ne II] are not detected so they cannot be of nuclear origin. The TIMMI2 absolute fluxes agree well with Roche et al. (1991) made with a 5 aperture. Moorwood & Oliver (1994) find that the central 6 pc region is dominated by an AGN whereas star formation, as traced by H and Br recombination lines, occurs in a (partial) ring of 200 pc radius. This ring lies outside our map of Fig. 3. Model calculations of the dust re-processing of the AGN are presented by Siebenmorgen et al. (1997) and Ruiz et al. (2001). Generally speaking, when one observes a Sy1 galaxy with a large aperture, one has a free view to the inner wall of the torus and detects strong midinfrared emission from hot (large) grains. The PAH bandtocontinuum ratio is then relatively low. In a Sy2, like Circinus, the inner torus is obscured and hence, the hot grains are invisible and the PAH bandto continuum ratio is higher. This picture is supported in the case of Circinus, but also for Sy1 and other Sy2 galaxies (see below).

Bock et al. (2000) with MERLIN at Keck. However, we are not sensitive enough to confirm the individual hot cores found in those maps. PAHs are detected in the integrated ISOCAM (Laurent et al. 2000) and in our ISOPHT spectrum, but in a smaller aperture they disappear (Fig. 14, see also the ISOSWS spectrum of Lutz et al. 2000). The continuum is of similar strength for ISOSWS and TIMMI2 indicating that the AGN is indeed the major source in the central 1 kpc region. The ISOCAM spectrum of the central 9 by Laurent et al. (2000) is, like ISOSWS and TIMMI2, free of PAHs. The [Ne II] line is absent in TIMMI2, but strong in the ISOSWS spectrum, [S IV] is seen in both. The strength of the silicate absorption band increases with spatial resolution; for the central 20 pc, we estimate AV 5mag.

5.2.4. NGC 5643
This Sy2 galaxy shows high-excitation optical lines (Kinney et al. 1993). The emission in [O III] and H is extended up to 20 (Schmitt et al. 1994). The 11.9 µm TIMMI2 image is fuzzy, without a point source. The [Ne II] line is detected by TIMMI2 and for the [S IV] line we get an upper limit (Table 7). PAHs are strong in the ISOPHT aperture, whether they are seen with TIMMI2 is unclear (Fig. 14).

5.2.5. NGC 1386
The TIMMI2 image of this Sy2 galaxy (Ferruit et al. 2000) has an unresolved core (Table 5) and an extended component (Fig. 7). The ISOPHT spectrum shows PAH bands and tentative, weak silicate absorption. In the TIMMI2 spectrum, (Fig. 14) one sees the silicate feature, corresponding to AV 30 mag, but no PAHs.

5.2.6. NGC 1365
A faceon Sy1 galaxy with an unresolved core in the TIMMI2 image. Its midinfrared emission is dominated by PAHs everywhere except in the very center. The bandtocontinuum ratio decreases with decreasing aperture size (Fig. 14). PAH bands are not present in the TIMMI2 spectrum (Table 8). [S IV] and [Ne II] emission is detected only with ISOSWS, but not with TIMMI2. The TIMMI2 continuum is flat without evidence of silicate absorption.

5.2.2. Centaurus A
This is the closest example of an AGN in our sample. In the TIMMI2 image it shows an unresolved core of less than 0.5 (Fig. 5). The raw image and the growth curve (Fig. 5) indicate weak extended emission over 2 with a surface brightness ten times lower than at the peak. The ISOCAM (Laurent et al. 2000) and ISOPHT (this work) spectra are PAH dominated whereas the TIMMI2 spectrum is featureless except for the presence of a strong [Ne II] line (Fig. 14). The visual extinction estimated from the silicate absorption detected in the ISOPHT and TIMMI2 spectra is around 14 mag.

5.2.7. NGC 4388
A galaxy with a Sy2 nucleus where Ho & Ulvestad (2001) see at radio wavelengths two components, 1.9 (150 pc) apart, none of them coincides with the position of the optical peak. A hard X-ray peak is found at the position of the active nucleus (Iwasawa et al. 2003). As in the previous examples, the TIMMI2 image has a single unresolved core (Table 5), and there are strong PAH bands in the ISOPHT, but none in the TIMMI2 spectrum (Fig. 14). The [S IV] and [Ne II] lines are detected with TIMMI2 (Table 7).

5.2.3. NGC 1068
TIMMI2 resolves the nucleus of this prototype Sy2 galaxy (Fig. 6) and our image of the core agrees with the long integrations by Alloin et al. (2000) using CAMIRAS and

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei

133

Fig. 14. Midinfrared flux densities (in Jy) of Seyfert galaxies: Data presented are: ISOPHT (24 aperture, histogram), ISOCAM (total field or nuclear spectrum, dashed), ISOSWS (14 20 , dotted), UCL (5 , diamonds), TIMMI2 spectroscopy (3 , full line) with 3 errors (shadow) and TIMMI2 photometry with 1 error bars (Table 3).

5.2.8. IRAS 05189-2524
In this most distant object of our sample, the central source in the TIMMI2 image is unresolved (Table 5). The same conclusion was reached by Soifer et al. (2000) using MERLIN at Keck. ISOPHT sees weak PAH emission (Laureijs et al. 2000) and an absorption feature at 6 µm attributed to water ice (Spoon et al. 2002). PAH bands (Fig. 14) as well as gas emission lines are absent in the TIMMI2 spectrum. From the silicate absorption, we estimate AV 12 mag.

similar at both resolutions (Fig. 15): [Ne II] is present, the PAH band is detected at 11.3 µm but not at 8.6 µm. From the silicate absorption we estimate AV 20 mag. The 11.3 µm PAH band tocontinuum ratio is in TIMMI2 lower than in ISOPHT by about a factor 6 (Table 8).

5.2.10. NGC 5506
The optical spectrum of this Sy2 galaxy is comparable to what one finds in young starburst galaxies, however, the high excitation lines ([Ne V], [Fe VII], [He II]) mandate the presence of a nonthermal continuum (Oliva et al. 1999). The Xray spectrum is highly absorbed (Pfefferkorn et al. 2001). In the 8 13 µm wavelength range, [Ar III]/[Mg VII], [S IV] and [Ne II] emission lines are detected with ISOSWS, [S IV] and [Ne II] lines at similar strengths with TIMMI2 (Table 7). Evidence of PAH emission is found in the ISOPHT spectrum at 6.2 µm (Rigopoulou et al. 1999), whereas TIMMI2 sees a featureless spectrum. TIMMI2 and ISOPHT agree very well photometrically (Fig. 15). We estimate an optical depth of over 15 mag towards the core.

5.2.9. NGC 7582
The spectrum is a superposition of a Sy2 nucleus and a starburst (StorchiBergmann et al. 2000). The AGN is responsible for Xray emission over a few kpc (Levenson et al. 2000). Optical and near IR images (Regan & Mulchaey 1999) indicate a complex dust morphology. There is a lane that runs from the southeast to an absorbing ring north of the chain of blue clumps; the TIMMI2 image reflects this structure (Fig. 8). PAH bands are detected with ISOPHT, but not [S IV]. The excellent seeing conditions during the TIMMI2 observations allowed us to use besides the 3 also the 1.2 slit. The spectra look quite

134

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei

2.0 1.0
3.0" 1.2"

1.0

0.1 1.2 0.8 0.4 0.5

NGC 7582 PKS 2048-57

NGC 5506
0.0

IC 4329A

0.9 0.7 0.5

Mrk 509
0.3 0.1 6 8 10 wavelength (µm) 12 6

NGC 6240

1.0 0.6 0.2

8

10 wavelength (µm)

12

Fig. 15. Midinfrared flux densities (in Jy) of Seyfert galaxies: data presented are: ISOPHT (24 aperture, histogram), TIMMI2 spectroscopy (3 , full line and 1.2 dotted) with 3 errors (shadow) and TIMMI2 photometry with 1 error bars (Table 3).

5.2.11. PKS 2048-57
In the nearinfrared, the nuclear source of this object is very red (K -L = 2.1 mag) and heavily polarised (17%) which is explained by a non-thermal contribution (Hough et al. 1987). Its spectral energy distributions peaks at 60 µm (Hessler & Vader 1995) and the radio power is 2 orders of magnitude greater than that of typical Seyferts. The mid IR image is dominated by an unresolved core. The ISOPHT spectrum shows weak PAH emission. The TIMMI2 spectrum is badly affected by telluric sampling noise and no PAH features are visible. [Ne II] and [Ar III]/[Mg VII] is detected by ISOSWS but not with TIMMI2.

observations (Winge et al. 2000). The ISOPHT spectrum is typical for a Sy1 (Clavel et al. 2000), it shows weak PAH emission over a strong continuum. In the TIMMI2 spectrum PAH bands are not detected (Table 8). Both spectra do not show evidence of silicate absorption. [Ne II] is detected by ISOSWS but not with TIMMI2.

5.2.14. NGC 6240
A famous ultraluminous merger with a deeply hidden AGN. The AGN reveals itself by the strong iron line at 6.4 keV and the powerful continuum at 100 keV (Matt et al. 2000). Two hard X-ray nuclei have been recently detected by Komossa et al. (2003). Near infrared spectra indicate very low metalicity (1/10 solar, Oliva & Origlia 1998). NICMOS images (Scoville et al. 2000) show two nuclei separated by 1.6 (0.8 kpc) NorthSouth. There are also two nuclei at radio wavelengths 1.4 apart (Condon et al. 1996), the CO (21) peak lies in between (Tacconi et al. 1999). The TIMMI2 image at 11.9 µm (Fig. 8) reveals the nuclear region to be dominated by a strong core, which we identify with the Southern nucleus (Tecza et al. 2000). A fainter structure, at 10% of the peak flux of the core, extends 2 to the North and matches the structure observed at lower resolution at near-infrared wavelengths (Tecza et al. 2000). The ISOPHT spectrum is dominated by strong PAH emission and shows in addition emission lines of [Ar II], H2 00 S(5) (blended with [Ar II]) and H2 0-0 S(3). Interestingly, the latter line is absent in our TIMMI2 spectrum. Most of the [Ne II] emission (2/3) originates from outside the 3 slit (see Table 7). Both spectra (Fig. 15) show prominent PAH

5.2.12. IC 4329A
This Sy1 galaxy is pointlike in our TIMMI2 images, but shows extended [O III] and H emission (Mulchaey et al. 1996). The TIMMI2 and ISOPHT spectra appear to trace the same components, except for the [S IV] line which is stronger in the large aperture (Fig. 15). [Ne II] is seen with TIMMI2; the feature near 9.4 µm is probably telluric.

5.2.13. Mrk 509
The emission of this Sy1 galaxy is at 11.9 µm dominated by an unresolved core and marginal resolved emission below the 10% level of the peak flux (Table 4) is found toward the South East (Fig. 9). This may be supported by the slight increase of the midinfrared emission in the larger ISOPHT aperture as compared to the TIMMI2 spectrum (Fig. 15). Source extension towards the south is also reported by near infrared

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei Table 8. Continuum subtracted 11.3 µm PAH band fluxes, F , (in 10-20 W/cm2 ) and PAH bandtocontinuum ratios, R, of the TIMMI2 (3 slit) and ISOPHT (24 aperture) spectra. Name Circinus Centaurus A IC4329a IRAS 05189+2524 IRAS 08007-6600 M83 Mrk 509 Mrk 1093 Mrk 1466 NGC 1068 NGC 1365 NGC 1386 NGC 3256 NGC 4388 NGC 5506 NGC 5643 NGC 6000 NGC 6240 NGC 7552 NGC 7582 PKS 2048-57 F
TIMMI2

135

R

TIMMI2

F

ISO

R

ISO

<50 <11 <6 <10 <6 <2 <7 10 4 <95 <11 <16 30 <11 <16 <10 16 12 33 19 <22

< < < < < < < 1. 1. < 0. < 0. < < 0. 0. 0. 0. 0. <

0. 0. 0. 0. 0. 0. 0. 0 1 0. 2 0. 4 0. 0. 2 6 6 3 2 0.

1 1 1 1 1 1 1

1 1 2 1

1

530 100 <10 <3 <10 252 17 - - <750 125 165 107 17 <15 20 67 21 162 110 <50

0.4 0.9 <0. <0. <0. 1.2 0.1 - - <0. 0.6 0.6 1.0 0.6 <0. 0.5 1.1 1.0 0.9 1.3 <0.

1 1 1

1 Fig. 16. The absorption efficiency of a graphite sphere of 10 е radius. Mie theory is still assumed to be valid at Xrays. Optical constants from Laor & Draine (1993). 1

1

emission. The 8.6 µm/7.7 µm band ratio is small in ISOPHT and the PAH band at 8.6 µm feature is not even detected with TIMMI2. Between 9 and 10.5 µm the TIMMI2 spectrum is noise dominated. The silicate absorption, as of ISOPHT suggests AV > 25 mag.

above 2500 K, evaporation is more effective (see Fig. 8 of Voit 1991). The evaporation rate is proportional to the number of surface atoms, Nsurf , for which approximately Nsurf N 2/3 if N is the total number of atoms in the grain. Let P1 and P2 be the probabilities that a surface atom is at an energy level below or above the binding energy Eb , respectively. Usually, P2 1, so P1 1, and obviously P1 + P2 = 1. For a Boltzmann distribution, P2 /P1 exp(-Eb /kT ). Due to phonon collisions, an atom changes its level at a rate given by the Debye frequency D . In one second, the atom is P2 D times above Eb , the evaporation rate is therefore R
ev

5.2.15. NGC 7674
This Sy2 is slightly extended in the radio and has within 0.5 a double source (Kukula et al. 1995). The deconvolved 11.9 µm TIMMI2 image is also extended and at 0.5 (Fig. 11, Table 5) dominated by a single core. Previous mid IR maps by Miles et al. (1996) have insufficient resolution to resolve the galaxy but their absolute photometry is consistent with ours. The 5 resolution UCL spectrum (Roche et al. 1991) is rather featureless, whereas the ISOPHT spectrum contains several PAH bands (Schulz, priv. communication). We could not yet perform TIMMI2 spectroscopy.

= Nsurf D e

-Eb /kT

.

(1)

Evaporation rises exponentially with temperature. To counterbalance it, atoms must be built into the grain. The rate at which this occurs is proportional to the thermal velocity of atoms, or to T 1/2 , and to the density of the interstellar medium. The exponential process of evaporation wins over accretion by orders of magnitude in a fairly narrow range of T , the density is not relevant. Let us assume that the grains evaporate if a certain fraction of them, fev , is above some critical temperature T cr . The latter should be somewhat greater than T ev . Relegating all complicated or unknown physics to the parameters fev and T cr , we write the condition for evaporation as I
ev

6. Heating and evaporation of PAHs
We discuss the survival of PAHs in starburst and AGN environments with energetic photons of, at least, a few hundred eV. Such photons can only be emitted by an Xray source.

T

cr

P(T )dT > fev .

(2)

Reasonable numbers for T cr in Eq. (2) are 2500 K and fev 10-8 , but such suggestions do, of course, not constitute a theory.

6.1. The evaporation condition
Let P(T ) be the temperature distribution of an ensemble of very small grains, for example, PAHs, so that P(T )dT gives the probability of finding a particular grain in the temperature interval dT around T . Whereas grain cooling usually proceeds through emission of infrared photons, in a very hot particle,

6.2. Xray cross sections for dust absor ption
An AGN differs from a starburst nucleus of the same bolometric luminosity by the hardness of its radiation field. We are here interested in the question how the presence of very energetic photons in AGNs affects the chance of survival of small grains.

136

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei

First, we remark briefly on the Xray absorption cross section of dust. When a grain absorbs an optical or UV photon of frequency , its internal energy rises by the full amount h. Furthermore, when the grain is not smaller than the wavelength, the absorption efficiency is approximately one. The situation is different at Xrays. The interaction of hard photons with interstellar dust and the limits of Mie theory have been studied by Voit (1991), Laor & Draine (1993) and Dwek & Smith (1996). Below 10 keV, the main absorption process is photoionization of inner atomic shells. The excited electron loses energy in inelastic scattering as it travels through the grain, but may also leave the grain and carry away kinetic energy as a photo electron. When an Xray photon has created a gap in the innermost shell, the gap will be filled by a downward transition of an electron from an upper shell and the energy may escape either as a photon, or nonradiatively through ejection of an Auger electron. The computations of Dwek & Smith (1996) indicate that for graphite particles of 50 е radius, soft Xray photons with h < 100 eV deposit practically all their energy in the grain and Mie theory is still valid. For photon energies above h = 100 eV and particle radii a < 50 е, only part of the photon energy is deposited in the grain and we therefore reduce the absorption efficiency by a factor 1/. Figure 16 illustrates the absorption efficiency, Qabs , from Mie theory of graphite grains with radii from 10 to 1000 е (see Fig. 2 of Laor & Draine 1993 for a wider range of parameters). The jump at 43 е (290 eV) is due to the ionization threshold of the Kshell of the carbon atoms. At still higher frequencies, one has the typical -3 cross section for ionizing solitary atoms. The drop in Qabs at wavelengths short of the first ionization level at 10 eV is less steep (Qabs -2 for a 100 е); for big grains (a > 1000 е), Qabs is essentially flat at these wavelengths. When, as for Xrays, absorption is due to the excitation of inner atomic shells, the cross section of a grain, K abs , depends only weakly on the way the atoms are held together in the solid. One can therefore roughly set K abs proportional to the number of atoms in the grain. The dust particle in Fig. 16 with 10 е radius consists of NC = 520 carbon atoms and has Qabs 0.2 in the interval 10eV h 20 eV, and Qabs -2.2 for h > 20 eV. For the cross section per carbon atom, C abs , 10-17 cm2 for 10 eV h 20 eV, and this implies C abs abs -17 C 10 в (20/h)2.2 cm2 for h > 20 eV. We will use these crude numbers to calculate the cross section of PAHs.

Fig. 17. The distribution function of the temperature, P(T ), for graphite grains of 5 е radius. Monochromatic flux F = 104 erg cm-2 s-1 , photon energies from 10 to 80 eV.

Fig. 18. The fraction of graphite grains of 5 е (solid lines) and 10 е (dashed lines) radius above a critical temperature T cr = 2500 K (see Eq. (2)). The strength of the monochromatic flux F is indicated.

6.3. The temperature distribution of PAHs in a hard radiation field
To investigate how the distribution function P(T ) and, in particular, the evaporation condition (2) depend on the photon energy, we consider a source with a monochromatic flux of strength F , all photons have then the same energy h. While keeping the value of the flux fixed, we change h, so if Nph is the number of photons, Nph h = F stays constant. Low energy photons impinge more frequently, but raise the enthalpy H of the grain only by a small amount. Xray absorption,

on the other hand, is a relatively rare event, but accompanied by a big jump in H . Temperature distributions P(T )are shown in Fig. 17 for a flux F = 104 erg cm-2 s-1 corresponding to a luminosity L = 2.4 в 1011 L at a distance of 100 pc. The photon energies vary between 10 and 80 eV. Grains much above the evaporation temperature of the solid (2000 K) are, of course, unrealistic. The distributions P(T ) in Fig. 17 display a kink at the temperature where the thermal energy of the grain equals h. Still higher grain (momentary) temperatures are achieved when a second photon is absorbed before the grain has cooled off from the excitation by the first. Figure 18 displays the integral Iev in equation (2) as a function of photon energy h for various fluxes and two grain sizes. The important point is here that for the smallest graphite grains with 5 е radius (65 carbon atoms), Iev grows very rapidly in the range from 20 to 40 eV and then declines only gradually. The energy of photons, h , emitted by the hottest O stars (T eff = 5 в 104 K) is only 16 eV when averaged over the

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei

137

1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 6
Fig. 19. The fraction of PAHs with NC = 100 carbon atoms above a critical temperature T cr = 2500 K (see Eq. (2)). The strength of the monochromatic flux F is indicated.

SB 4pc

SB 10pc

SB 100pc

AGN 4pc AGN 10pc AGN 100pc 8 10 12 14

8 10 12 14 6 8 10 12 14 6 wavelength (µm)

Fig. 20. Optically thin dust emission (full line) between 5 and 15 µmin a starburst (top) and an AGN (bottom) at various distances to the central heating source. The total luminosity is L = 1011 L . The emission of large grains, representing the continuum, is shown dashed.

total spectrum, and equals 22 eV when averaged over the Lyman continuum ( < 912 е). These numbers are thus upper limits for the radiation field in a starburst. Furthermore, the stellar flux of an O star plummets at wavelengths below 228 е, or above h = 54.1 eV, when helium becomes doubly ionized. AGNs, on the other hand, have a spectrum typically declining with -0.7 and emit copious amounts of extreme UV and soft Xray photons as witnessed by the observation of fine structure lines from ions with high (>100 eV) ionization potentials. We conclude from Fig. 18 that small graphite grains (5 е radius) survive in a starburst as long as fluxes are below 104 erg cm-2 , but they will be destroyed near an AGN of the same flux F . The fate of PAHs of equivalent mass is similar (see Fig. 19). Bigger grains (see curves for 10 е radius in Fig. 18) are much more resistive. They need higher fluxes to get vaporized and survive in a starburst environment.

carbon grains. The spectra are shown in Fig. 20. They are in arbitrary units and normalised at the maximum. One notices that large grains become hotter as the radiation field hardens. At a distance of 4 pc to the heating source, PAHs invariably evaporate, at 10 pc evaporation occurs only near an AGN, and at 100 pc they survive in both environments, starburst and AGN.

7.2. Optically thick dust emission
Nuclei of luminous infrared galaxies are generally dust enshrouded and not transparent. Radiative transfer calculations of optically thick dusty nuclei have been carried out in various approximations (Krugel & Tutokov 1978; Rowan-Robinson & Ё Crawford 1989; Pier & Krolik 1993; Loar & Draine 1993; Granato & Danese 1994; Krugel & Siebenmorgen 1984; Ё Siebenmorgen et al. 1997; Silva et al. 1998; Efstathiou et al. 2000; Ruiz et al. 2001; Siebenmorgen et al. 2001; Nenkova et al. 2002). In the subsequent spherical models, for starbursts and AGNs, the total luminosity is again 1011 L . The dust cloud has a constant density, a visual extinction of AV 25 mag (from the surface to the centre), and an outer cloud radius of 500 pc. To be a bit more realistic the dust consists here, besides large carbon grains, also of large silicate grains of the same size distribution and with a 12% larger total volume. Absorption and scattering efficiencies are calculated using Mie theory and optical constants by Loar & Draine (1993). In addition, there are small graphites and small silicates of 10 е radius, with an abundance of 0.5% that of the large carbon grains. PAH parameters are as described in the previous section. The radiative transfer in a dust cloud around a starburst is computed following Siebenmorgen et al. (2001), around an AGN after Siebenmorgen et al. (1997) and the inclusion of soft X-rays as in this work. The major difference between both radiative transfer models is that AGN are heated by a central source whereas starburst galaxies are heated by stars which are

7. Radiative transfer models
Here we present comprehensive model calculations for the midinfrared emission of starburst and AGN radiation environments in the optical thin and optical thick regime.

7.1. Optically thin dust emission
In a first attempt to understand and to illustrate the influence of the hardness of the radiation field, we compute the optically thin dust emission at various distances from the source. A pointlike heating source of total luminosity L = 1011 L has either the spectrum of an OB star (T = 25 000 K) to mimic a starburst, or of an AGN with L -0.7 in the range from 10 е to 2 µm (the exact upper limit is unimportant as the dust has little extinction there). The distance to the source, and thus the flux which the dust receives, is varied from 4 to 100 pc. For simplicity we consider in this section only two dust components: Large carbon grains with a n(a) a-3.5 size distribution and radii between 300 and 2400 е, and PAHs consisting of 100 C and 28 H atoms and an abundance of 10% of the large

138

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei

1.0

2.0 flux (Jy) 1.5 1.0 0.5 6 8

SB

0.8 0.6 0.4 AGN SB

AGN

0.2

12µm

25µm

10 12 14 wavelength (µm)

16

1

10 100 radius (pc)

Fig. 21. Comparison of radiative transfer model spectra of dusty starbursts (solid) and AGN (dashed). In both models the visual extinction towards the cloud centre is AV = 25 mag and the luminosity is L = 1011 L . Spectra are calculated for a distance of 50 Mpc.

Fig. 22. Normalised intensities at 12 (solid lines) and 25 µm (dashed lines) for starbursts and AGN models as presented in Fig. 21. At both wavelengths the predicted FWHM of the starburst galaxy is much larger than for the AGN.

distributed over a large volume. In case of a starburst, each of the OB stars have a uniform luminosity of 20 000 L and a stellar temperature T = 25 000 K. The stellar density changes with galactic radius r up to 500 pc from the center like 1/r (there must be a cutoff at r = 0, but its exact position is unimportant). Each OB star is surrounded by a hot spot of constant dust density, corresponding to n(H) 104 cm-3 . The size of the hot spot is determined by the condition of equal heating of the dust from the star and from the radiation field in the galactic nucleus. As for the AGN models, the inner radius of the hot spots is given by the photodestruction or evaporation radius of the grains. Our model spectrum for the starburst and the AGN is displayed in Fig. 21. The starburst shows strong PAH emission and a continuum which rises steeply beyond 10 µm, whereas the AGN is almost free of PAH emission. Cross cuts over the model nuclei, at various wavelengths, are shown in Fig. 22. At 12 µm and 25 µm, the AGN is much more compact than the starburst. Therefore the midinfrared size may be used to distinguish between the two kinds of energy sources. This statement holds unless the stars are distributed by more than a few pc.

8. Conclusion
We presented mid infrared imaging and spectroscopy for a sample of 23 galactic nuclei. The observations were obtained with the ISO (14 -24 resolution) and with TIMMI2 at the 3.6 m ESO telescope (3 and 0.5 resolution for spectroscopy and imaging, respectively). The galaxies are mostly starbursts and AGNs, with apparent midinfrared fluxes of 20 mJy or more. Our main results are: a) Seyfert nuclei are dominated by a compact core while starbursts do not display such a central concentration; they are usually spatially resolved by TIMMI2. b) Almost all our ISO spectra (18 out of 19) are marked by bands at 6.2, 7.7, 8.6 and 11.3 µm, generally attributed to

PAHs. In the TIMMI2 spectra, which refer to the innermost nuclear regions, these bands are present in starbursts, but not detected or significantly reduced in AGNs. This corroborates previous suggestions that PAHs evaporate much easier in the hard photon environment of an AGN, but are often able to survive in a starburst. c) Applying a simple criterion for dust evaporation, we quantify the conditions under which small grains are photo destructed in a monoenergetic hard radiation field. Analyzing their temperature distributions, we find that PAHs or very small graphite particles with less than 100 carbon atoms cannot survive if the photons have energies above 20 eV and their flux is greater than 103 erg cm-2 s-1 . The detection or nondetection of PAH bands can thus serve as a diagnosis whether a galaxy is predominantly powered by an AGN or a starburst. Of course a very dense stellar cluster provides a high radiative density and destroys PAH (Siebenmorgen 1993). Therefore the dividing line between AGN and starbursts is not sharp, as demonstrated by the nuclear spectrum of M 83 and Circinus. d) In radiative transfer models that include soft Xrays and incorporate the above dust evaporation criterion, the starburst galaxies show PAH bands everywhere, and in the midinfrared they are stronger and more extended than AGNs of the same luminosity. e) The [S IV] and [Ne II] fine structure lines at 10.5 and 12.8 µm, which can still be excited in an O star environment, are in both types of galaxies generally stronger in the large (ISO) aperture spectra indicating that their emission comes from outside the nucleus.
Acknowledgements. PIA is a joint development by the ESA Astrophysics Division and the ISOPHT Consortium. This research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.

R. Siebenmorgen et al.: Midinfrared emission of galactic nuclei

139

References
Antonucci, R. R. J., & Miller J. S. 1985, ApJ, 297, 621 Alloin, D., Pantin, E., Lagage, P. O., & Granato, G. L. 2000, A&A, 363 Bock, J. J., Neugebauer G., Matthews K., et al. 2000, ApJ, 120, 2919 Cesarsky, C., Abergel, A., Agnese, P., et al. 1996, A&A, 315, L32 Cohen, M., Walker, R. G., Carter, B., et al. 1999, AJ, 117, 1864 Condon, J. J., Helou, G., Sanders, D. B., & Soifer, B. T. 1996, ApJS, 103, 81 Contini, M., & Contini, T. 2003, MNRAS, 342, 299 Clavel, J., Schulz, B., Alteri, B., et al. 2000, A&A, 357, 839 Genzel, R., & Cesarsky, C. 2000, ARA&A, 38, 761 de Graauw, T., Haser, L. N., Beintema, D. A., et al. 1996, A&A, 315, L49 Dwek, E., & Smith, R. K. 1996, ApJ, 459, 686 Efstathiou, A., Rowan-Robinson, M., & Siebenmorgen, R. 2000, MNRAS, 313, 734 Elbaz, D., Cesarsky, C. J., Chanial, P., et al. 2002, A&A, 384, 848 Ferruit, P., Wilson, A. S., & Mulchaey, J. 2000, ApJS, 128, 139 Granato, G. L., & Danese, L. 1994, MNRAS, 268, 235 Hasinger, G., Burg, R., Giacconi, R., et al. 1998, A&A, 329, 482 Hessler, C. A., & Vader, J. P. 1995, AJ, 110, 87 Heras, A. M., Shipman, R. F., Price, S. D., et al. 1998, in The Universe as seen by ISO, ed. P. Cox, & M. F. Kessler, Paris, ESA-SP, 427 Ho, L. C., & Ulvestad, J. S. 2001, ApJS, 133, 77 Horne, K. 1986, PASP, 98, 609 Hough, J. H., Brindle, C., Axon, D. J., Bailey, J., & Sparks, W. B. 1987, MNRAS, 224, 1013 Iwasawa, K., Wilson, A. S., Fabian, A. C., & Young, A. J. 2003, MNRAS, 345, 369 KЁ , H.-U., Sterzik, M., & Siebenmorgen, R. 2003, SPIE, 4841, 117 aufl Kinney, A. L., Bohlin, R. C., Calzetti, D., Panagia, N., & Wyse, R. F. G. 1993, ApJS, 86, 5 Komossa, S., Burwitz V., Hasinger G., et al., 2003, ApJ 582, L15 Krugel, E., & Tutokov, A. V. 1978, A&A, 63, 375 Ё Krugel, E., & Siebenmorgen, R. 1994, A&A, 282, 407 Ё Kukula, M. J., Pedlar, A., Baum, S. A., & O'Dea, C. P. 1995, MNRAS, 276, 126 Laurent, O., Mirabel, F. A., Charmandaris, V., et al., 2000, A&A, 359, 887 Laureijs, R. J., Watson, D., Metcalfe, L., McBreen, B., & O'Halloran, B., 2000, AA, 359, 900 Lemke, D., Klaas, U., Abolins, J., et al. 1996, A&A, 315, L64 Levenson, N. A., Weaver, K. A., & Heckman, T. M. 2001, ApJS, 133, 269 Loar, A., & Draine, B. T. 1993, ApJ, 402, 441 Lutz, D. 1999, in The Universe as seen by ISO, ed. P. Cox, & M. F. Kessler, Paris, ESA-SP 427, 623 Lutz, D., Sturm, E., Genzel, R. et al. 2000, ApJ, 536, 697 Markarian, B. E., Lipovetskii, V. A., Stepanian, J. A., et al. 1989, Soobsch. Spets. Astrof. Obs., 62, 5 Matt, G., Fabian, A. C., Guainazzi, M., et al. 2000, MNRAS, 318, 173 Miles, J. W., Houck, J. R., Hayward, T. L., & Ashby, M. L. N. 1996, ApJ, 465, 191 Moorwood, A. F. M., & Oliva, E. 1994, Infrared Phys. Tech., 35, 349 Moorwood, A. F. M., Lutz, D., Oliva, E. et al. 1996, A&A, 315, L125

Moran, E. C., Barth, A. J., Kay, L. E., & Filippenko, A. V. 2000, ApJ, 391, L73 Mulchaey, J. S., Wilson, A. S., & Tsvetanov, Z. 1996, ApJS, 102, 309 Nenkova, M., Ivezic, Z., & Elitzur, M. 2002, ApJ, 355, 456 Oliva, E., & Origlia, L. 1998, A&A, 332, 460 Oliva, E., Origlia, L., Maiolino, R., & Moorwood, A. F. M. 1999, A&A, 350, 90 Pfefferkorn, F., Boller, T., & Rafanelli, P. 2001, A&A, 368, 797 Pier, E. A., & Krolik, J. H. 1993, ApJ, 418, 673 Regan, M. W., & Mulchaey, J. S. 1999, AJ, 117, 2676 Reimann, H.-G., Linz, H., Wagner, R., et al. 2000, SPIE, 4008, 1132 Rigopoulou, D., Spoon, H. W. W., Genzel, R., et al. 1999, AJ 118, 2625 Roche, P. F., Aitken, D. K., Smith, C. H., & Ward, M. J., 1991, MNRAS, 248, 606 Rowan-Robinson, M., & Crawford, J. 1989, MNRAS, 238, 523 Ruiz, M., Efstathiou, A., Alexander, D. M., et al. 2001, MNRAS, 316, 49 Sanders, D. B., & Mirabel, I. F. 1996, ARA&A, 34, 749 Schinnerer, E., Eckart, A., Quirrenbach, A., et al. 1997, ApJ, 488, 174 Schmitt, H. R., Storchi-Bergmann, T., & Baldwin, J. A. 1994, ApJ, 423, 237 Schutte, W. A., Tielens, A. G. G. M., & Allamandola, L. J. 1993, ApJ, 415, 397 Scoville, N. Z., Evans, A. S., Thompson, R., et al. 2000, AJ, 119, 991 Siebenmorgen, R. 1993, ApJ, 408, 218 Siebenmorgen, R., Moorwood, A., Freudling, W., & Kaufl, H. U. Ё 1997, A&A, 325, 450 Siebenmorgen, R., Krugel, E., & Zota, V. 1999, A&A, 351, 140 Ё Siebenmorgen, R., Krugel, E., & Laureijs, R. 2001, A&A, 377, 735 Ё Silva, L., Granato, G. L., & Bressan, A. 1998, ApJ, 506, 600 Smith, C. H., Aitken, D. K., & Roche, P. 1989, MNRAS, 241, 425 Soifer, B. T., Sanders, D. B., Madore, B. F., et al. 1987, ApJ, 320, 238 Soifer, B. T., Neugebauer, G., Matthews, K., et al. 2000, ApJ, 119, 509 Soifer, B. T., Neugebauer, G., Matthews, K., et al. 2002, AJ, 124, 2980 Spoon, H. W. W., Keane, J. V. Tielens, A. G. G. M., et al. 2002, A&A, 385, 1022 Starck, J., Siebenmorgen,, R., & Gredel, R. 1997, ApJ, 482, 1011 Starck, J.-L., & Murtagh, F. 2002, Astronomical Data and Image Analysis (Springer, ISBN 3-540-42885-2) Stacey, G. J., Swain M. R., Bradford C. M., et al. 1998, in The Universe as seen by ISO, ed. P. Cox, & M. F. Kessler, Paris, ESASP 427, 973 Storchi-Bergmann, T., Raimann, D., Bica, E. L. D., & Fraquelli, H. A. 2000, ApJ, 544, 747 Sturm, E., Lutz, D., Verma, A., et al. 2002, A&A, 393, 821 Tacconi, L. J., Genzel, R., & Tecza, M. 1999, ApJ, 524, 732 Tecza, M., Genzel, R., Tacconi, L. J., et al. 2000, ApJ, 537, 178 Thornley, M. D., ForsterSchreiber, N. M., Lutz, D., et al. 2000, ApJ, Ё 539, 641 Vader, J., Frogel, J. A., Terndrup, D. M., & Heisler, C. A. 1993, AJ, 106, 1743 Verma, A., Lutz, D., Sturm, E., et al. 2003, A&A, 403, 829 Voit, G. M. 1991, ApJ, 379, 122 Winge, C., Storchi-ergmann, T., Ward, M., & Wilson, A. S. 2000, MNRAS, 316, 100