Anomalous Silicate Dust Emission in the Type 1 LINER Nucleus of M81 Howard A. Smith Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138 [email protected] Aigen Li, M.P. Li, and M. K`hler Department of Physics and Astronomy University of Missouri, Columbia, MO 65211 M. L. N. Ashby, G. Fazio, J-S Huang, M. Marengo, Z. Wang, S. Willner, Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138 A. Zezas 1 Physics Department, University of Crete, 71003 Heraklion, Greece L. Spinoglio Istituto di Fisica dello Spazio Interplanetario, CNR, via Fosso del Cavaliere 100, I-00133 Rome, Italy and Y.L. Wu Spitzer Science Center, California Institute of Technology, Pasadena CA 91101 1Also IESL, Foundation for Research and Technology-Helas, 7110 Heraklion, Greece; Harvard- Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138 1 ABSTRACT We report the detection and successful modeling of the unusual 9.7μm Si--O stretching silicate emission feature in the type 1 (i.e. face-on) LINER nucleus of M81. Using the Infrared Spectrograph (IRS) instrument on Spitzer, we determine the feature in the central 230 pc of M81 to be in strong emission, with a peak at ~10.5μm. This feature is strikingly different in character from the absorption feature of the galactic interstellar medium, and from the silicate absorption or weak emission features typical of galaxies with active star formation. We successfully model the high signal-to-noise ratio IRS spectra with porous silicate dust using laboratory-acquired mineral spectra. We find that the most probable fit uses micron-sized, porous grains of amorphous silicate and graphite. In addition to silicate dust, there is weak PAH emission present (particularly at 11.3μm, arising from the C--H out-of-plane bending vibration of relatively large PAHs of ~500--1000 C atoms) whose character reflects the low-excitation AGN environment, with some evidence that small PAHs of ~100--200 C atoms (responsible for the 7.7μm C--C stretching band) in the immediate vicinity of the nucleus have been preferentially destroyed. Analysis of the infrared fine structure lines confirms the LINER character of the M81 nucleus. Four of the infrared H rotational lines are detected and fit to an excitation temperature of 2 T~800K. Spectral maps of the central 230pc in the [NeII] 12.8μm line, the H 17μm line, and 2 the 11.3μm PAH C--H bending feature reveal arc- or spiral-like structures extending from the core. We also report on epochal photometric and spectroscopic observations of M81, whose nuclear intensity varies in time across the spectrum due to what is thought to be inefficient, sub- Eddington accretion onto its central black hole. We find that, contrary to the implications of earlier photometry, the nucleus has not varied over a period of two years at these infrared wavelengths to a precision of about 1%. 2 Subject headings: Galaxies: individual (M81) -- galaxies: nuclei -- galaxies: Seyfert-- infrared: galaxies – ISM: dust Online material: color figures Submitted to The Astrophysical Journal August 31, 2009; Revised: February 15, 2010; Accepted March 11, 2010. 3 1. INTRODUCTION M81 is one of the nearest large spirals and a member of an interacting group of about 25 galaxies that includes M82. Its estimated distance is 3.63 " 0.34 Mpc (Freedman, 1994), a value we use in all subsequent calculations. The nucleus of M81 has been classified as a LINER (Heckman 1980; Ho et al. 1997) based on a wide range of observations, with properties that mark it as being closely related to Seyferts. Ho et al. (1996) used HST and ground-based observations to probe the inner few parsecs, and concluded that the main source of excitation is photoionization from a non-stellar continuum, consonant with the conclusion that it is a genuine AGN with a likely accretion rate of ~2x10-5 M yr-1 . Its estimated black hole mass is 7x107M (Devereux et À À al. 2003; 2007), it has a nuclear X-ray (0.5-2.4keV) luminosity of 5x106 L (Devereux and À Shearer 2007), and a circumnuclear (64” radius) total IR luminosity of -4x108 L (Perez- À Gonzalez et al. 2006; Devereux et al. 1995). HST WFPC2 images of the nucleus in V-band find a single object less than 0.04” (0.7pc) in size, with no other nuclei or stellar components within about 4” (Devereux, Ford and Jacoby 1997); ground-based 10 μm data at ~arcsecond resolution are consistent with a point-like nucleus. Devereux, Ford and Jacoby (1997) and Davidge and Courteau (1999) studied the inner 1 kiloparsec (including the region whose Spitzer spectrum we discuss shortly) and its near-IR spectral energy distributions (SED) in J, H, K, and 2.26μm filters, concluding that thermal emission from hot dust can (as in other galaxies) contribute about 20% of the light at K-band within 0.5” of the nucleus. Davidge and Courteau interpret the hot dust as a sign of nuclear activity, and also conclude there is no population of extremely red stars 4 in the central arcsecond. Although the M81 nucleus is relatively inactive, as is characteristic of LINERS, temporal variations have been reported across the spectrum due to what is thought to be inefficient, sub-Eddington accretion onto its central black hole. The nuclear emission varies by a factor of about 2 in visible light (Devereux et al. 2003) and near 10μm (Grossan et al. 2001) on timescales estimated at about 25 years. Willner et al. (2004; hereafter Paper I) published the first IRAC observations of M81, including the nuclear region. The IRAC flux densities in a 3.5" square aperture centered on the nucleus (including starlight) were 109, 75, 65, and 56 mJy at 3.6, 4.5, 5.8, and 8 μm, respectively, with uncertainties of "10%. After subtraction of the stellar component, they found a residual point- like source evident at 4.5μm and longer wavelengths, and perhaps at 3.6μm as well. Flux densities of the point-like source at the four IRAC wavelengths were 6, 21, 22, and 39 mJy, respectively, with the first being uncertain by more than a factor of 2. Willner et al. compared the 8 μm flux density with the reported, much stronger 10 μm flux density of 159 mJy observed in 1999 by Grossan et al., (2001), and suggested based on this comparison that the nuclear source may have varied over the previous 4 years. They noted, however, that the spectral shape of the nuclear emission between 8 and 10 μm was not well known, although an ISO spectrum of the source in a 24" beam did show a steep rise (Rigopoulou et al. 1999). Since it would seem difficult to account for the differences in measured flux densities by spectral shape alone -- it would require a factor of two increase in the continuum level between 8 and 10 μm -- the authors suggested that the difference could instead be due to variability, thus supporting the idea that IRAC was seeing AGN activity. The IRAC images also revealed a filament or bar of material leading from a ringlike structure at ~1 kpc (1') to an inner structure and to the AGN. The 5 prominence of the ring at 8μm suggests that much of its emission is from dust, probably mostly aromatic features. These inner features were suspected of being related to the structures seen in HST images and in the light of Hα (Devereux et al. 1997). The possibility of time variations in the infrared flux from the M81 nucleus convinced us to begin a series of epochal observations of the source with IRAC Guaranteed Time; the possibility that an unusual spectral energy distribution (SED) might be responsible for the apparent discrepancy prompted us to examine the IRS spectra in the Spitzer archive. Photometric redshift estimates are powerful and increasingly common ways to estimate redshifts of faint, distant galaxies, and for galaxies out to z ~ 2 the 8-10μm SED can be sampled from IRS spectra (e.g., Huang & Faber, 2009). A steep and unusual SED in M81 (and, as we discuss below, in some other LINERs and quasars) will strongly influence the estimated photometric redshifts of any cosmological objects of a similar nature. In Section 2 we review all the observations; section 3 presents an analysis of the peculiar SED in M81 and its spectral features, and compares it to the SEDs seen in other galaxies. Section 4 addresses the spectral lines in the nucleus. Section 5 models the SED as arising from silicate dust, and discusses the nature of that dust and possible explanations for its presence. Section 6 presents our spectral imaging data of the nuclear region; the conclusions are in Section 7. 6 2. OBSERVATIONS AND ANALYSIS 2.1 IRAC, MIPS and IRS Observations After the original Willner et al. (2004; PID 1035) conclusions were published, we began a four- epoch campaign to look for possible variability of the M81 nucleus (Fazio, PID 121). This was motivated in part by Grossan et al. (2001), who reported from their ground-based N-band (10.79μm) observations compared to earlier observations from Rieke and Lebofsky (1978) that there was a factor of two flux variability, and that therefore the M81 nucleus had a substantial nonstellar component to its flux at 10μm. The Willner observations were taken on 2003 November 6, 2004 December 21, 2005 May 6, and 2005 October 24. These, and in addition the observations from SINGS on 2004 June 4, were reanalyzed using the IRACproc software package (Schuster, Marengo, and Patten 2006; e.g., Ashby et al. 2009). The combined, reanalyzed three-color IRAC image (3.6, 4.5, and 8.0 μm) spanning 1.7 years is shown in Figure 1a after subtraction of a stellar contribution as per Paper I; Figure 1b is a zoom into the nuclear region (inner 30”) with its ring and filamentary arcs. Figure 1c, for comparison, is of the same nuclear region as seen with HST in Hα  (Devereux, Ford and Jacoby, 1997), and shows the same arc-like structures. All of the features noted in Paper I are even more clearly seen in the deeper IRAC mosaic, in particular the central point-source. We employed a simple, iterative scale-and-subtract procedure to determine whether any changes in the apparent brightness of M81's nucleus could be detected in epochs subsequent to 2003 November. Specifically, we added a constant offset to the epochal mosaics to match the background in the 2003 November mosaic. We then determined the scale factors needed to produce residuals with a mean average of zero when the subsequent epoch mosaics were subtracted from the 2003 November mosaic. 7 This procedure was repeated for all epochs in all four IRAC channels. In some cases, small sub- pixel offsets were necessary to achieve proper spatial alignment. We find that in all epochs, for all channels, we obtain scale factors that are consistent with unity to within 2σ or less. At 3.6, 4.5, 5.8, and 8.0μm, respectively, the scale factors derived are 0.991" 0.005, 0.991"0.006, 0.977"0.016, and 0.985"0.015. Given the errors attending this procedure, our results suggest that the variability of the M81 nucleus is less than 1% at 3.6 and 4.5μm, and less than 2% at 5.8 and 8.0μm. Given that these results are comparable to the 3% error in the IRAC absolute gain calibration and the 1% errors attending repeated point-source photometry according to v3.0 of the IRAC Data Handbook, we cannot rule out unknown systematic effects that would dominate our photometry at the level probed by our difference- image technique. Thus, we see no evidence for flux variability in the M81 nucleus in the IRAC bands. The MIPS 24μm data were taken as part of the SINGS Legacy Survey program (Kennicutt et al. 2004). We downloaded these data from the Spitzer archive and reduced them independently. We used MIPS scan map observations from the PID 717 and PID 159 programs as generated by version S16.1 of the MIPS pipeline software. After first subtracting object-masked median stacked images on a by-AOR basis, the individual exposures were coadded in the standard way into three separate single-AOR mosaics with 2.5 arcsecond pixels. The flux within a 6x5 pixel region centered on the galaxy nucleus was measured in each of the three mosaics produced (two from PID 159, and one from PID 717). The result, aperture corrected per the MIPS guidelines, is a 24μm flux of 0.50 +- 0.05 Jy. We also used the post-BCD 70μm mosaic from PID 159, which 8 we downloaded from the Spitzer archive. As before, pipeline version S16.1 was used. Within a square region 16” wide centered on the nucleus of M81 we measured a 70μm flux of 0.8 " 0.2Jy. 1 The IRS data were obtained as part of the SINGS Legacy Survey program (Kennicutt, PID 159, 193), on 2004 April 15, and 2005 April 11 and 22, and retrieved from the Spitzer archive. The data were processed with pipeline S17, and then manually reduced using CUBISM (Smith et al. 2007), and with the SMART (Higdon et al. 2004) software packages used for the spectral analyses using Gaussian line fit routines. (SMART routines were also used to manually extract the spectra as a check, using a tapered column fit, with care taken to include the nuclear region and to exclude as much of the off-nuclear regions as possible. The results were consistent with CUBISM reductions.) The CUBISM extractions gave spectral orders matched in flux at the overlap wavelengths to better than about 8%, both for low-res and for high-res extractions; matching the continuum levels between low-res and hi-res spectra was a bit worse, with the high-res continuum systematically about 15% higher than low-res and as much as 20% higher at wavelengths longer than about 35μm. We refined the results by cross-referencing the flux values to the IRAC and MIPS24 photometry of the region. Figure 2 shows the full, composite-module spectrum of the nuclear region extending over 230pc; Table 2 lists the observed lines and their strengths. We used the SINGS MIPS images of M81 to extract the 24μm flux of the nuclear region and compare it with the value measured with IRS. Both give F = 0.30 " .03 Jy. The 24 agreement gave us added confidence in assembling the full spectrum for the region from the 1 We note that this figure is about 1% of the total 70μm flux of M81. At 24μm, according to Dale et al. 2007, our 0.5 Jy measurement is about 10% of the total 24μm emission from the galaxy of 5.09 +/- 0.20 Jy 9 different IRS modules. 3. THE MID-IR DUST EMISSION 3.1 The Unusual Silicate Emission Profile of the M81 Nucleus The continuum SED of the M81 nucleus (Figure 2) is striking: it shows a strong silicate emission bump with its flux increasing by a factor of two between 8 and 10μm. In fact, the continuum shape in M81 changes by exactly the amount needed to explain the apparent discrepancy between the 1999 ground-based photometry and the IRAC photometry, and thus obviates the need for time-variability as an explanation. Thus we find no evidence for time variability, either within the internal set of epochal IRAC infrared data reported in Section 2, or in comparison with the filter-corrected, ground-based observations. The silicate emission profile of the M81 nucleus warrants careful attention. The spectral shape is not only steeply rising, but it peaks at wavelengths of ~10.5μm, longer than the 9.7μm peak usually seen in most of the galactic sources (see Figure 3a). Its width is also much broader than that of the galactic sources: while the silicate emission feature of the M81 nucleus has a FWHM of ~4.2μm, the 9.7μm silicate absorption feature of the local interstellar medium (ISM) along the line of sight toward Cyg OB2 #12 has a FWHM of only ~2.6μm (Whittet et al.1997); the ISM sightline toward the Galactic Center object Sgr A*, with a FWHM of ~1.8μm (Kemper, Vriend, & Tielens 2004), has an even narrower 9.7μm silicate absorption feature. Also shown in Figure 3a is the silicate absorption feature of IRAS08572+3915, an ultraluminous IR galaxy (ULIRG). Similar to that of the galactic sources, it peaks at 9.8μm and has a FWHM of ~2.2μm 10