Dust and PAH emission in the star-forming active nucleus of NGC 1097 R. E. Mason Gemini Observatory, Northern Operations Center, 670, N. A’ohoku Place, Hilo, HI 96720 USA 7 0 [email protected] 0 2 N. A. Levenson n a J Department of Physics and Astronomy, University of Kentucky, 177 Chemistry/Physics 5 Building, Lexington, KY 40506, USA 1 v [email protected] 4 2 C. Packham 1 1 0 Department of Astronomy, University of Florida, PO Box 112055, 211 Bryant Space 7 Center, Gainesville, Fl 32611, USA 0 / h [email protected] p - o M. Elitzur r t s a Department of Physics and Astronomy, University of Kentucky, 177 Chemistry/Physics : v Building, Lexington, KY 40506, USA i X r [email protected] a J. Radomski Gemini Observatory, Southern Operations Center, Casilla 603, La Serena, Chile [email protected] A. O. Petric Astrophysics Department, Columbia University, 520 W. 120th Street, New York, NY10025, USA. [email protected] G. S. Wright – 2 – UK Astronomy Technology Center, Royal Observatory Edinburgh, Blackford Hill, Edinburgh, EH9 3HJ, UK [email protected] ABSTRACT The nucleus of the nearby galaxy, NGC 1097, is known to host a young, compact (r < 9 parsec) nuclear star cluster as well as a low-luminosity active galacticnucleus(AGN).Ithasbeensuggestedboththatthenuclearstellarcluster is associated with a dusty torus, and that low-luminosity AGN like NGC 1097 do not have the torus predicted by the unified model of AGN. To investigate these contradictory possibilities we have acquired Gemini/T-ReCS 11.7 µm and 18.3 µm images of the central few hundred parsecs of this galaxy at <45 parsec angular resolution, in which the nucleus and spectacular, kiloparsec-scale star- forming ring are detected in both bands. The small-scale mid-infrared (mid- IR) luminosity implies thermal emission from warm dust close to the central engine of this galaxy. Fitting of torus models shows that the observed mid-IR emission cannot be accounted for by dust heated by the central engine. Rather, the principal source heating the dust in this object is the nuclear star cluster itself, suggesting that the dust that we detect is not the torus of AGN unified schemes (although it is also possible that the dusty starburst itself could provide the obscuration invoked by the unified model). Comparison of Spitzer/IRS and Gemini/GNIRSspectrashowsthat,althoughpolyaromatichydrocarbonemission (PAH) bands are strong in the immediate circumnuclear region of the galaxy, PAH emission is weak or absent in the central 19 parsecs. The lack of PAH emission can probably be explained largely by destruction/ionization of PAH molecules by hard photons from the nuclear star cluster. If NGC 1097 is typical, PAH emission bands may not be a useful tool with which to find very compact nuclear starbursts even in low-luminosity AGN. Subject headings: dust, extinction — galaxies: active — galaxies: nuclei — infrared: galaxies — galaxies: individual (NGC 1097) 1. Introduction The dusty molecular torus, cornerstone of the unified model of active galactic nuclei (AGN), has been the subject of much observational and theoretical scrutiny in recent years. – 3 – The detection of broad emission lines in polarised light in a number of type 2 objects (Antonucci & Miller 1985; Miller & Goodrich 1990; Young et al. 1996; Moran et al. 2000; Tran 2001) demonstrates the overall validity of the unified model, and structures thought to correspond to the torus itself have now been resolved through mid-IR interferometry of one Seyfert galaxy (Jaffe et al. 2004). However, questions remain as to the origin and precise nature of the obscuration and the extent to which it can account for the differences between the many classes of AGN. In particular, the possibility that the torus and star formation are intimately connected has been raised by several authors on theoretical grounds. For instance, the material in the nucleusneedstohaveacertainscaleheighttoobscurethenucleusandaccountfortherelative numbers of types 1 and 2 AGN, and it has been suggested that the energy necessary to main- tain this thickness might be supplied by supernovae in nuclear starbursts (Wada & Norman 2002). Indeed, some or all of the nuclear obscuration has been proposed to arise in dusty clouds ejected from such starbursts (Cid Fernandes & Terlevich 1995; Fabian et al. 1998; Watabe & Umemura 2005) or in optically thick stellar winds (Zier & Biermann 2002). Ob- servationally, young starsarefoundquitecommonlyintheinner fewhundredpcofAGN(e.g. Gonz´alez Delgado et al.1998;Storchi-Bergmann et al.2000;Imanishi2002,2003;Rodr´ıguez-Ardila & Viegas 2003). However, while few-hundred-parsec-scale star formation is often observed in AGN, few young stellar clusters have to date been detected at smaller scales. The outer diameter of the torus remains rather unclear, but its warm, mid-IR-emitting regions have not yet been resolved in single-dish 10-20 µm observations of even the nearest objects (e.g. Bock et al. 2000; Radomski et al. 2003; Soifer et al. 2003; Packham et al. 2005; Mason et al. 2006), and interferometric measurements of NGC 1068 suggest that its mid-IR-emitting regions are only ∼3 pc (∼0.04′′) in diameter (Jaffe et al. 2004). Near-IR adaptive optics imaging of the Circinus galaxy (Prieto et al. 2004) resolves a source of FWHM∼2 pc, consistent with a parsec-scale torus at those wavelengths. The discovery by Storchi-Bergmann et al. (2005, hereafter SB05) of UV absorption sig- natures of a 106M , 106 year-old stellar cluster within only 9 pc of the nucleus of the nearby ⊙ LINER/Seyfert 1 galaxy, NGC10971, is therefore of obvious interest. The star cluster — but not the broad emission lines from the nucleus itself — suffers about 3 magnitudes of visual extinction, consistent with the possibility of starburst activity physically associated with a 1AlthoughNGC1097hostsaclassicLINERnucleus(Phillips et al.1984),broad,double-peakedemission lines in its spectrum are characteristic of rotating gas in an AGN accretion disk (Storchi-Bergmann et al. 1993). – 4 – dusty AGN torus. Young superstellar clusters or recent star formation are also known to exist in the central few parsecs of a handful of other AGN (Colina et al. 2002; Davies et al. 2006b; Mueller S´anchez et al. 2006). While the results of SB05 are suggestive of a starburst genuinely associated with an obscuring torus, the observation that many LINERs host nuclear UV point sources raises some doubt as to whether LINER-type nuclei in fact host obscuring tori (Maoz et al. 2005). This is expected on theoretical grounds in the context of the disk-wind scenario for the torus: below a certain luminosity the accretion onto the black hole can no longer sustain the mass outflow necessary to obscure the nucleus (Elitzur & Shlosman 2006). Moreover, how the properties of LINER tori, if they exist, might relate to those observed in Seyfert galaxies is completely open to question. Further observations of LINER nuclei to probe their detailed nuclear structure are therefore desirable. In addition, the UV evidence of star formation detected by SB05 is very susceptible to extinction, so UV detections of starbursts in AGN tori are likely to be limited to cases of nuclei in which the stellar clusters are not too deeply buried within the nuclear dust. Inter- preting the energetics of UV-detected starbursts will be complicated by uncertain extinction corrections. A more useful tracer of the nuclear starburst phenomenon would be an infrared star-formation indicator that is both unaffected by dust extinction and accessible from the ground, to take advantage of the spatial resolution achievable from ground-based telescopes. The 3.3 µm polycyclic aromatic hydrocarbon (PAH) emission band fits both of these crite- ria. Usually attributed to vibrational relaxation of aromatic hydrocarbon molecules excited by far-UV photons, the PAH bands have been used extensively as tracers of star formation in extragalactic environments (e.g. Moorwood 1986; Roche et al. 1991; Genzel et al. 1998; Lutz et al. 1998; Rigopoulou et al. 1999; Laurent et al. 2000; Tran 2001; Imanishi & Wada 2004; Peeters et al. 2004; Risaliti et al. 2006). The suite of PAH emission features is commonly detected in large-aperture ISO spec- tra of nearby active galaxies with circumnuclear star-forming regions (Rigopoulou et al. 1999), but on moving to apertures of 3-5′′ (∼ few hundred pc) or less, the bands are often weak or absent (Roche et al. 1991; Le Floc’h et al. 2001; Siebenmorgen et al. 2004; Roche et al. 2006a,b). Modelling suggests that the carriers of the bands are easily de- stroyed by the harsh radiation field close to the active nucleus itself (Roche & Aitken 1985; Siebenmorgen et al. 2004; Voit 1992). However, this appears not always to be the case. For instance, Imanishi & Wada (2004) find that the surface brightness of the nuclear PAH emission in Seyfert 1 galaxies is comparable to that in starburst galaxies, and the 3.3 µm PAH band has been shown to be present in several Seyfert galaxies of both types 1 and 2 on fairly small spatial scales, including, in the case of NGC 3227, within only 60 pc of – 5 – the central engine (Rodr´ıguez-Ardila & Viegas 2003). To exist at such small distances from these AGN, it seems likely that the fragile PAH band carriers are shielded from the nuclear X-rays. This intriguing result could perhaps be explained if the carriers exist within the dense, dusty clouds of the torus itself. To further investigate this possibility, we have performed L band (∼3–4 µm) spec- troscopy of the nucleus of NGC 1097 (D=17 Mpc for H = 75 km sec−1 Mpc−1; 1′′= 85 pc) 0 to search for the 3.3 µm PAH emission band. We have also obtained images at 11.7 and 18.5 µm, to search for the compact mid-IR source expected to be a signature of the torus and to constrain its properties. The data were taken using the Thermal Region Camera and Spectrograph (T-ReCS) and the Gemini Near-Infrared Spectrograph (GNIRS) on the Gemini South 8.1 m telescope, yielding ≤0.52′′ spatial resolution and therefore probing dis- tances ≤ 22 pc from the nucleus. We also discuss complementary Spitzer Space Telescope Infrared Spectrograph (IRS) observations of the central 3.6 × 3.6′′ (approx. 300 × 300 pc) of this galaxy. §2 describes all of these observations and the reduction of the data. In §3 we present the nuclear data and describe the characteristics of the compact source detected in the mid-IR, and compare our subarcsecond-resolution data and that of Prieto et al. (2005) with the clumpy torus models of Nenkova et al. (2002). The conclusions that can be drawn from this work are discussed in §4. 2. Observations and data reduction 2.1. Mid-IR imaging Narrow-band N (11.7 µm) and Q (18.3 µm) band images of NGC 1097 were taken on 2005September 16and17usingtheGeminiSouthmid-IRinstrument T-ReCS(Telesco et al. 1998). T-ReCS uses a Raytheon 320 x 240 pixel Si:As IBC array, providing a plate scale of 0.089′′ per pixel. The detector was used in correlated quadruple sampling mode (Sako et al. 2003). Images were obtained in the 11.7µm (δλ = 1.13µm, 50% cut-on/off) and 18.3µm (δλ = 1.51µm, 50% cut-on/off) filters using the standard chop-nod technique to remove time- variable sky background, telescope thermal emission and ”1/f” noise. The chop throw was 15′′ and the telescope was nodded 15′′ in the direction of the chop every 30 s. The chop angle was chosen based on the 5′′-aperture N band data of Telesco & Gatley (1981) to minimise the extended emission in the off-source beam and was fixed at 90◦ east of north. The total on-source time for the 11.7 µm and 18.3 µm observations was 456 s and 912 s respectively Thedatawerereducedusing theGeminiIRAFpackage. Thedifferenceforeachchopped pair for each nod-set was calculated, and the nod-sets then differenced and combined until a – 6 – singleimagewascreated. Choppedpairsobviously compromisedbyvariableskybackground, highelectronicnoise, orotherproblems weresearched forbutnodataneeded tobediscarded. The Cohen standards SAO 216405 and SAO 193679, and PPM 278407 were observed for flux and PSF calibration. Observations of the standards taken approximately 1 hr apart showed a variation in counts of around 10%, typical of mid-IR photometry. The PSF ob- servations were made immediately before or after the NGC 1097 observations and with the same instrumental configuration. The FWHM of the PSF star was 0.41′′ (standard de- viation ∼0.03′′) at 11.7 µm and 0.52′′ in the 18.3 µm filter. Short PSF or flux standard observations are comparable to longer source observations as the closed-loop active optics of Gemini provides a similar PSF when taken at a similar telescope pointing and time (see e.g. Radomski et al. 2003; Packham et al. 2005, although seeing variations may of course affect the final observed PSF). As judged by the FWHM of the standard stars, the PSF was stable over the course of these observations. 2.2. L-band spectroscopy Lbandspectroscopy ofthenucleus ofNGC1097wasobtainedusingGNIRS,thenear-IR spectrograph on the Gemini South telescope, on 2005 November 13. The short red camera and 3-pixel (0.45′′) slit were used together with the 32 l/mm grating, resulting in R∼1100. The total integration time was 30 minutes, and the telescope was nodded by 20′′ every 30 seconds in an ABBA pattern so that the sky background emission could be subtracted out. Sky conditions were photometric during these observations. The nucleus was acquired in the H band to take advantage of the contrast between object and sky at those wavelengths. A point source contributes 52% of the H band flux in a 0.4′′ diameter aperture, that fraction increases with wavelength (Prieto et al. 2005), and we observe an unresolved source at 11.7 µm (§ 3.1), suggesting that the location at which the slit was centered at 1.6 µm also corresponds to the flux peaks at 3.3 and 11.7 µm. The slit was orientated at 9◦ east of north in order to encompass one of the circumnuclear emission knots visible in the T-ReCS image, but that knot proved too faint to yield a useful spectrum and will not be considered further. The data were reduced using the Gemini IRAF package in combination with standard IRAF and Starlink FIGARO routines. After bad pixel masking, flatfielding and subtracting in pairs, the individual sky-subtracted spectra were averaged together, with the positive and negative beams cross-correlated and shifted in both the spatial and dispersion directions before combining. The central three pixels (0.45′′) of the galaxy spectrum were extracted – 7 – using the standard star spectrum to define the curvature of the extraction aperture across the array. Wavelength calibration was accomplished using the numerous telluric absorption lines in the spectrum. The galaxy spectrum was then cross-correlated with the spectrum of the standard star, BS 466 (F7V), and divided by that spectrum. In order to improve the cancellation of telluric lines, the resulting spectrum was multiplied by a logarithmically- scaled model atmospheric spectrum2, binned to the appropriate spectral resolution. This had the effect of slightly reducing the amplitude of some of the residuals in the 3.3 µm region. The spectrum was then multiplied by a 6240 K blackbody curve. Flux calibration was with reference to BS466, using a Vband magnitude of 6.24 and V-L = 1.36. No attempt was made to correct for uncertain differential slit losses between the galaxy and standard star (due to e.g. seeing variations and different intrinsic spatial flux distributions) and, as is usual for flux calibration of narrow-slit spectral data, the uncertainty of the derived flux density is considerable. 2.3. IRS spectroscopy Publically-available spectra of the nucleus of NGC 1097 taken with the short-low (first order) module of the Spitzer Space Telescope’s Infrared Spectrograph (Houck et al. 2004, IRS) were retrieved from the Spitzer data archive (PID 14; PI Houck). These R∼100 ob- servations represent a total integration time of 56 s. Files output from the IRS pipeline at the Spitzer Science Center (which includes ramp fitting, dark-sky subtraction, droop cor- rection, linearity correction, flatfielding and wavelength calibration) had background light subtracted by subtracting the first-order slit exposed during second-order observations of NGC 1097 taken a few minutes previously. Spectra were extracted from the background- subtracted, coadded frames using Spice, the SPItzer Custom Extractor. The extraction window scales with wavelength but was defined to be 2 pixels (3.6′′) at 11.2 µm, equal to the 3.6′′ slit width and comparable to Spitzer’s diffraction limit at 11.2 µm (3.3′′). The spatial profile of the IRS spectrum shows flux between the nucleus and the circum- nuclear ring ∼10′′ distant, presumably from the wings of the PSFs of these bright sources combined with real diffuse emission from that region. Neither the point source nor the ex- tended source flux calibration provided by Spice is optimised to deal with this situation. However, the conclusions drawn from these observations do not depend critically on a highly accurate calibration, so we simply normalise the mean flux in the IRS spectrum to that given 2 Lord (1992), obtained from www.gemini.edu/sciops/ObsProcess/obsConstraints/ocTransSpectra.html – 8 – by Telesco & Gatley (1981) over a similar bandpass in a slightly larger aperture (5′′ diameter vs. 3.6 × 3.6′′). 3. Results 3.1. The mid-IR images and model fits The T-ReCS mid-IR images of NGC 1097 are shown in Figures 1 and 2. A compact nuclear source and the well-known circumnuclear star-forming ring are clearly detected at both 11.7 and 18.3 µm. The starburst ring emits strongly in the mid-IR, with each of the three knots in the northern part of the ring contributing a similar level of flux to that of the nucleus itself. The 15′′ maximum chop throw of the Gemini telescopes means that the positive and negative images partially overlap in the region of the circumnuclear ring. Inside the ring, faint extended emission is very probably present but is difficult to detect without a clear sample of the sky on which to measure the zero level of the background. The flux of the nuclear point source is approximately 48 mJy at 11.7 µm and 64 mJy at 18.3 µm (Table 1), roughly 75% and 27% of the broad-band, 5′′-aperture measurements of Telesco & Gatley (1981) at N and Q respectively, and consistent with the 60 mJy, 1.5′′-aperture N-band upper limit of Gorjian et al. (2004). Faint emission between the nucleus and ring likely accounts for the difference between our data and that of Telesco & Gatley (1981), most clearly affecting the Q band photometry. Table 1 gives the flux densities of the three regions of emission in the circumnuclear ring that are clearly detected at 18.3 µm and least likely to be affected by the negative chop beams. The aperture sizes used were determined partly by the proximity of neighboring emission regions and/or the edge of the detector. The N/Q flux density ratios are in the region of 0.4 for the ring, but rather higher, about 0.75, for the nucleus. The FWHM of the central compact source was measured to be 0.43′′ at 11.7 µm, com- parable to the mean FWHM of the comparison star observations (0.41±0.03′′, § 2). This translates to a diameter of < 37 pc, consistent with the NIR limit of 10 pc diameter of Prieto et al. (2005). Thus, like all AGN observed non-interferometrically to date, the nu- clearsourceofNGC1097isnotresolved atitsFWHMintheNband. LimitedS/Nprevented a meaningful measurement of the source size at 18.3 µm. – 9 – Table 1. Photometry of the nucleus and circumnuclear ring of NGC 1097 Location1 Aperturediameter 11.7µmfluxdensity 18.3µmfluxdensity (arcsec) (mJy) (mJy) Nucleus 3.02 48 64 A 1.6 34 83 B 2.0 75 169 C 1.6 25 63 1SeeFig.2 2An aperture of 3” diameter was used for the photometry of the nucleus, chosen based on the profiles of the standard stars to ensure that all of the flux from the unresolvedsource was collected. A surrounding”sky” annulus was usedto subtract off any residual sky background and diffuse emission surrounding the nucleus. The (unknown) spatial variationofsurroundingdiffuseemissionwillthereforedetermine the degree to which the photometry represents the unresolved source alone, but as anysuchemissionmustbeclosetothesensitivitylimitoftheinstrument,weexpect ittomakeanegligiblecontributiontothemeasurednuclear flux. – 10 – N E 1” Fig. 1.— Central portion of the T-ReCS 11.7 image of NGC1097, smoothed with a 2-pixel Gaussian. The (unguided) negative off-sourceimagesareevident ateither side ofthepositive galaxy image.