Searching for nuclear obscuration in the infrared spectra of nearby FR-I radio galaxies

Active galactic nuclei (AGN) show an array of different spectra which are usually grouped into broad categories based on their radio luminosities, bolometric luminosities, and the presence or absence of broad optical emission lines. AGN unification models attempt to explain many of these different classes with a single type of object which differs in only a few parameters. Its particular focus is to explain the presence or absence of optical broad lines in objects which are otherwise similar (e.g., similar radio and X-ray profiles). These models consist of a central engine and some anisotropic distribution of gas and dust which scatters or absorbs light from the central engine. The central engine is matter accreting onto a supermassive black hole. Our description of this model follows the one by Antonucci (1993). For further reading see Antonucci (2012); Tadhunter (2016); Padovani et al. (2017).

In these models, material accreting onto the central supermassive black hole forms an optically thick but geometrically thin accretion disk which is obscured from some lines of sight by an optically thick, warm ($\sim 200-1000\,\text{K}$) mass of dusty material with a roughly toroidal geometry. This torus would absorb optical emission from the central engine and re-radiate that energy in infrared. Pier & Krolik (1992, 1993) modelled the torus as a smooth continuous cloud of variable density due to computing constraints but they believed it to be composed of discrete clouds (“clumpy”). The alternative to a torus is that the central engine is intrinsically dim in the optical rather than obscured, resulting in a lower mid-infrared luminosity where warm obscuring dust would radiate. Smooth torus models require fine-tuning in order to match the suppressed low-equivalent-width $10\,\mu\text{m}$ silicate features observed in infrared spectra (Granato & Danese, 1994; Granato et al., 1997).

Rowan-Robinson (1995) claimed that clumpy torus models can adequately explain silicate feature suppression observed in Seyfert 1 galaxies but, just like Pier & Krolik (1992), lacked a fully 3D radiative transfer code to fully explore it. Nenkova et al. (2002, 2008a, 2008b) formalized the clumpy torus model we consider in our decompositions in Section 5. This model describes an obscuring torus as being composed of small, discrete, optically thick clouds. The inner radius of the torus is determined by the sublimation radius $R_{d}$ for the component dust grains i.e., the minimum distance from the central engine at which a grain can survive. This sharp boundary is an approximation since larger dust grains can survive significantly closer to the central engine than the inner edge of the main torus (Pier & Krolik, 1993; Schartmann et al., 2005). The remaining parameters that Nenkova et al. (2008a, b) include in their torus model are the optical depth of a single torus clump at 550 nm, $\tau_{V}$ (assumed to be the same for all clumps), the mean number of clouds along a radial equatorial line, $N_{0}$, the radial extent of the torus in units of the dust sublimation, $Y$, the inclination of the torus with respect to the line of sight, $i$ (defined such that $0^{\circ}$ is pole-on), the angular scale height of the torus (effectively the standard deviation of the normally-distributed clumps around the equator), $\sigma$, and the power law index of the radial distribution of clumps, $q$. From these parameters they show how to calculate the half-opening angle of the torus, the torus height-to-radius ratio $H/R=\tan{(\sigma)}$, and the photon escape probability.

There are alternatives to the unification model in which the central engines of low-luminosity sources intrinsically differ from those of high-luminosity sources. These are largely beyond the scope of this paper but they often involve an advection dominated accretion flow (ADAF), in which a supermassive black hole is accreting at below 1% of the Eddington rate and the infalling material does not radiate much of its energy away. See e.g., Narayan et al. (1998); Best & Heckman (2012).

Fanaroff & Riley class I (FR-I, Fanaroff & Riley 1974) radio galaxies are low-luminosity radio galaxies with sub-relativistic jets that are brightest near their cores. FR-I radio galaxies typically do not show significant broad optical line emission though there are exceptions such as 3C 120 (French & Miller, 1980; Leipski et al., 2009; Torrealba et al., 2012). The question of how FR-I radio galaxies fit into AGN models can be reduced to two separate but related parts. These are the nature of the accretion structure that powers the radio jet and the structure of any obscuring material. These are related because if FR-Is are powered by an accretion structure with a high intrinsic optical luminosity such as a Keplerian thin accretion disk then the central engine must be obscured in order to match observations.

Haas et al. (2004) analyzed 5 to 200 $\mu$m photometry from the photometer on the Infrared Space Observatory, ISOPHOT, of 75 objects in the Revised Third Cambridge Catalog of radio galaxies and quasars (Laing et al., 1983), including 5 FR-I sources (of which three overlap with our sample). They clearly detect a thermal component in the FR-Is despite the lower sensitivity compared to Spitzer (Kessler et al., 1996) but it was too cold to be consistent with the unification models we are considering. Chiaberge et al. (1999) detected the central core component in 85% of Hubble Space Telescope observations of FR-I radio galaxies and so put an upper limit on the covering fraction of any obscuring torus with the caveat that the base of an optical jet may appear as an alternate source of optical continuum emission. Whysong & Antonucci (2004) found that the MIR core component of M 87 is dominated by synchrotron emission from the base of its relativistic jet and Perlman et al. (2007) put further limits on a a warm obscuring torus in M 87. These findings effectively rule out the possibility of a significant amount of obscuring warm dust. van der Wolk et al. (2010a) found no evidence for warm dust tori in any of the eight FR-I sources in their observations from the Very Large Telescope’s (VLT) Imager and Spectrometer for the Mid Infrared. Note that one of the FR-I sources in the van der Wolk et al. (2010a) sample is 3C 270, which is also in our sample. In Hubble Space Telescope (HST) images of the optical nuclei of a sample of 54 FR-I and 55 FR-II radio galaxies at $z<0.3$, Kharb & Shastri (2004) found that the optical emission in these sources is primarily non-thermal emission from a relativistic jet. Their fits also showed no significant improvement from including a thermal accretion disk in the FR-II population as a whole but found some improvement from including a disk component in broad-line emitting FR-II. They claim that this is supported by the presence of a “big blue bump” which they attribute to an accretion disk. Kharb & Shastri (2004) additionally found that the luminosity of the optical nuclei in FR-I radio galaxies show strong dependence on the radio core prominence $R_{c}$, which is a common statistical indicator of orientation, and from this they conclude that their results are consistent with the lack of a torus in FR-I radio galaxies. In a study conceptually similar to ours, González-Martín et al. (2015) searched for AGN obscuration in Spitzer/IRS (see Section 3) spectra and Chandra photometry of a sample of low-ionization nuclear emission region (LINER) AGN. The dusty torus present in brighter sources disappeared below bolometric luminosities $L_{\mathrm{bol}}\simeq 10^{42}\,\mathrm{erg}\,\mathrm{s}^{-1}$, consistent with Elitzur & Shlosman (2006), and that the MIR spectra of such sources is dominated by host, jet, and/or ADAF emission.

Leipski et al. (2009) found a Spitzer/IRS band thermal excess attributed to warm nuclear dust emission in four of the fifteen galaxies with detected optical compact cores in their sample of 25 FR-I radio galaxies. Six of our ten sources — 3C 31, 3C 66B, NGC 3862, 3C 270, M 84, and M 87 — overlap with their sample. Of those, they only found thermal MIR excess in the nuclei of 3C 270 and M 84, which they tentatively attribute to an obscuring torus and starburst activity respectively. However, Wu & Cao (2006) give a star formation rate estimate of $0.20\,\mathrm{M}_{\odot}\,\mathrm{yr}^{-1}$ for M 84 which is inconsistent with a starburst. They also found a MIR excess in 3C 66B, which they attribute to non-thermal emission. Leipski et al. (2009) also found a very low ratio between the $7.7\,\mu\mathrm{m}$ and $11.3\,\mu\mathrm{m}$ PAH lines, which is not typical for starburst galaxies.

In contrast, Bianchi et al. (2019) found red and blue wings on line emission in NGC 3147 consistent with the profile of a mildly relativistic thin Keplerian accretion disk. NGC 3147 is an unobscured AGN with an accretion rate far below the Eddington rate ($\sim 10^{-4}L/L_{\mathrm{Edd}}$) which lacks a broad line region. The detection of an accretion disk in such an object suggests the possibility that extremely low accretion rates can still support thin accretion disks.

Overall, recent studies favor the two-mode models in which some, but not all, FR-I radio galaxies have some sort of obscuring material around their central engines. The remainder are believed to have central engines that are too dim in infrared (IR) and optical to distinguish from their host galaxies. These dim central engines would still produce jets which may appear in the IR as a synchrotron spectrum. This is likely due to environmental factors such as the source of the accreting material. E.g., cold source material may be more likely to form a disk while hot material may be more likely to form an ADAF. The structure of this obscuring material is, however, not necessarily a torus like that described by Nenkova et al. (2002, 2008a, 2008b).

To determine what powers the central engine in FR-I radio galaxies and how they fit into the broader picture of AGN we search for the signature of warm obscuring dust in the infrared spectra of ten nearby FR-I radio galaxies from a well-studied sample (see Section 2). Our primary focus is on continuum emission in the band covered by the Spitzer Space Telescope’s Infrared Spectrograph. We analyse archival spectra using a pair of specialized fitting codes based on Markov chain Monte Carlo techniques. We seek to determine whether there is evidence for obscuration consistent with the clumpy torus model by Nenkova et al. (2002) in a sample of ten nearby low luminosity radio galaxies. We used Markov-chain Monte Carlo fitting algorithms to fit spectral lines and and decompose the underlying spectra to IR observations.

We provide an overview of the concepts involved and previous evidence surrounding the possibility of nucleus obscuration in this section. We then describe our sample of FR-I radio galaxies in Section 2 followed by an overview of the archival data used in this research in Section 3 along with the processing of our broadband ancillary data. In Section 4 we provide our analysis of the narrow spectral features in the Spitzer/IRS spectra for each source, with particular focus on AGN and star formation tracers and the 10 and 18 $\mu$m silicate features. In Section 5 we show the results of fitting various spectral energy distribution models to our entire broadband data set for each source and provide upper limits on any possible torus component (or equivalent mid-infrared thermal component) in those sources which do not require one. In Section 6 we discuss the implications of the models which best agree with the observations and compare our results to those of previous authors.

We use $L_{\text{[Ne\,II]}}$ and $L_{\text{[Ne\,III]}}$, the luminosities [Ne II] and [Ne III] lines respectively, to estimate the star formation rate (SFR) in each sample galaxy. This estimate by Ho & Keto (2007), $\mathrm{SFR}_{\text{Ne}}$, is given as

$$\mathrm{SFR}_{\text{Ne}}=4.73\times{10}^{-41}\,M_{\astrosun}\,{\text{yr}}^{-1}\text{s}\leavevmode\nobreak\ {\text{erg}}^{-1}\leavevmode\nobreak\ \big{[}L_{\text{[Ne\,II]}}+L_{\text{[Ne\,III]}}\big{]}.$$

(1)

Similarly, we use the combined luminosity of the 6.2 and 11.3 $\mu$m PAH lines, $L_{\text{PAH}}$ to get a second independent estimate of the SFR in each source. This estimate by Farrah et al. (2007), $\text{SFR}_{\text{PAH}}$, is given by

$$\text{SFR}_{\text{PAH}}=1.18\times{10}^{-41}\,M_{\astrosun}\,{\text{yr}}^{-1}\text{s}\leavevmode\nobreak\ {\text{erg}}^{-1}\leavevmode\nobreak\ \big{[}L_{\text{PAH}}\big{]}.$$

(2)

Finally, we also show SFR estimates using scalings by Roussel et al. (2001) using $7\,\mu\mathrm{m}$ ($L_{7\,\mu\text{m}}$) and $15\,\mu\mathrm{m}$ ($L_{15\,\mu\text{m}}$) “photometry” measurements created by integrating the Spitzer/IRS spectra over 16.18 THz and 6.75 THz bandpasses respectively to approximate the bandpasses of the relevant Infrared Space Observatory filters. These scalings are given by

$$\mathrm{SFR}_{7\,\mu\text{m}}=6.3\times{10}^{-43}\,M_{\astrosun}\,{\text{yr}}^{-1}\text{s}\leavevmode\nobreak\ {\text{erg}}^{-1}\leavevmode\nobreak\ \big{[}L_{7\,\mu\text{m}}\big{]},$$

(3)

and

$$\text{SFR}_{15\,\mu\text{m}}=1.7\times{10}^{-42}\,M_{\astrosun}\,{\text{yr}}^{-1}\text{s}\leavevmode\nobreak\ {\text{erg}}^{-1}\leavevmode\nobreak\ \big{[}L_{15\,\mu\text{m}}\big{]},$$

(4)

with approximately $\sim 50\%$ calibration uncertainties. Farrah et al. (2007) showed that Equations 1 and 2 give equivalent results in their Figure 15; this is also supported by Willett et al. (2010). However, Jensen et al. (2017) found that AGN activity can dominate PAH emission in Seyfert galaxies and it is currently unknown whether this finding will generalize to include low-luminosity radio galaxies. Despite the different estimates, they are consistently below a few solar masses per year so we do not expect star formation to contribute a large thermal component in the MIR which could be mistaken for warm nuclear obscuring material.

The high $\text{SFR}_{\text{PAH}}$ estimates in 3C 31 and NGC 3862 are consistent with the high PAH emission Leipski et al. (2009) found in these objects. Our relatively low $\text{SFR}_{\text{PAH}}$ estimates in 3C 66B and 3C 270 are also consistent with their findings of no PAH features in 3C 66B and weak PAH features in 3C 270. However, Leipski et al. (2009) found significant 11.3$\,\mu$m PAH emission in M 84 which is inconsistent with our results.

Significant emission lines from ionization states with a high ionization potential require a large source of high energy photons. These photons can come from massive stars but the central engine itself is the obvious photon source candidate. Indeed, Spinoglio & Malkan (1992) showed that you can distinguish star formation from accretion based on which lines are excited; AGN produce photons that can ionize atoms with a higher ionization potential. We list the luminosities of 4 high-ionization lines which are accretion tracers in Table 12: $[\mathrm{NeVI}]$ 8 $\mu$m, $[\mathrm{NeV}]$ 14 $\mu$m, $[\mathrm{NeV}]$ 24 $\mu$m, and $[\mathrm{OIV}]$ 26 $\mu$m. High-energy ionizing photons from an accretion disk should be unable to penetrate a dusty obscuring torus, so the detection of infrared spectral lines from highly-ionized species suggests that we are seeing unobscured or minimally-obscured (in infrared) emission from clouds near the central engine. Note that the presence of these lines does not completely prohibit an obscuring torus, since these lines are infrared radiation which can pass through dust clouds better than UV or optical photons from the central engine. Still, they suggest some other explanation may be preferable.

We compare our 10 and 18 $\mu$m silicate feature observations to model sets by Sirocky et al. (2008) to test whether our results are consistent with clumpy torus models, as this is an important diagnostic of the clumpiness of an obscuring torus. We show this comparison in Figure 18, where we compare our best-fit silicate line strengths $S_{\text{sil}}$, defined as

$$S_{\text{sil}}=\log{\bigg{(}\frac{f_{\text{obs}}}{f_{\text{cont}}}\bigg{)}},$$

(5)

in terms of the observed flux of the feature $f_{\text{obs}}$ and the interpolated continuum flux $f_{\text{cont}}$ at the peak wavelength) that we found in our line fits to the various Sirocky et al. (2008) clumpy torus and smooth shell models.

Nenkova et al. (2008b) model the radial distribution of the clumpy torus $N_{r}$ as the power law distribution

$$N_{r}\propto r^{-q}.$$

(6)

Figures 12-14 show the comparison to model sets, each with various numbers $N$ of clouds along the line of sight, by Sirocky et al. (2008) with power law index $q$ of 0-2 respectively.

Similarly, the mass distribution of the single continuous cloud in the smooth shell model also follows a power law,

$$\rho_{r}\propto r^{-q},$$

(7)

still described by the power law index, $q$, and dependent on the radial extent of the cloud, $Y$ (given in units of the dust sublimation radius).

We find that the silicate line strengths are inconsistent with smooth shell models of AGN obscuration and with clumpy models with power law indices $q>1$ and are consistent with clumpy models with $q\leq 1$ and $N\leq 5$ although a few are also consistent with the complete absence of any $18\,\mu\mathrm{m}$ silicate feature. Note that in Section 5 we find that only four of our ten sources show a significant thermal component in their continua, which we denote with blue and yellow markers in Figures 12-14.

This source favors a torus component and that torus component is a significant part of the best fit, contributing $\sim 70\%$ of the flux at 15 and 30 $\mu$m and $\sim 50\%$ at 60 $\mu$m. There is possible degeneracy between the torus component and NLR component but we doubt that this is a significant concern in this case due to the fact that the BIC will penalize the torus model more harshly due to its higher number of parameters. We therefore see no particular reason to doubt the presence of an obscuring warm torus-like structure in this AGN.

The presence of a torus component in the spectral energy distribution of NGC 315 is consistent with the discovery of broad polarized H$\alpha$ emission in this source by Barth et al. (1999). Which may indicate the presence of a Thomson-scattered hidden broad line region.

Similarly, Gu et al. (2007) detected a low-luminosity compact AGN in IRAC and MIPS images of NGC 315 after removal of stellar IR emission and accounting for emission from a central dusty disk. They make no claims about the nature of this AGN however its presence is consistent with our result that the IR spectrum of NGC 315 is not entirely dominated by emission from the host galaxy.

This source appears to be dominated by emission from the host galaxy in the IRS band. This source contains an optical jet, first noticed by Butcher et al. (1980), and a radio jet which Croston et al. (2003) identifies as the same structure as in the optical. The optical jet does not contribute substantially to the IR core flux. It is worth noting, however, that Lanz et al. (2011) detect extended NIR emission from the jet at kpc scales in Spitzer/IRAC images. As shown in Figure 21 and in Table 16, when a torus component is forced it contributes a significant amount of flux at $15\,\mu\mathrm{m}$ and $30\,\mu\mathrm{m}$ with best fit fractional contributions of 34.2% and 18.7% respectively. However, as shown in Table 40, the BIC difference between galaxy models with a forced torus and galaxy-only models is non-negligable (the BIC of the forced-torus model is $22.2$, or $\sim 6\%$, higher) and we notice no obviously unphysical best fit model parameters, therefore we conclude that we do not find evidence for a warm obscuring torus in the infrared spectra of 3C 31. We note, however, that Constantin et al. (2015) detected broad H$\alpha$ emission with a FWHM of $2710\,\mathrm{km}\,\mathrm{s}^{-1}$, contributing 86% of the combined narrow+broad H$\alpha$ flux. This broad H$\alpha$ is consistent with the presence of a broad line region and therefore possibly an accretion disk.

Not only does the best fit to the continuum emission in this source not require a torus component but the best fit contribution from a forced torus component is not significant, contributing only $\sim 20\%$ at its brightest. The largest difference between the best fit model and the data is the 70$\,\mu$m but that point is an upper limit and the best fit model is within the uncertainty of the data point. We therefore conclude that we can explain all the emission with only contributions from the host galaxy and none from the nucleus and so rule out any significant MIR-bright obscuration.

The best fit for this source requires a prominent hot dust component and a significant NLR model component. The best fit to the diffuse ISM component is unusually dim but the uncertainties are wide.

The upper limit provided by the fit with a forced torus component is a significant fraction of the total flux but the forced torus is unusually cold, suppresses the diffuse ISM component, and those two fits have a large BIC difference (best: 472.3, forced torus: 506.8) of 34.5. We note that the BIC difference may be primarily due to the removal of the hot dust component since we expect the most degeneracy between the NLR and torus components as they are both MIR thermal components. Overall, we cannot rule out the possibility of a torus, especially given the presence of another MIR thermal component, but we also do not favor the presence of a torus so we consider our results inconclusive with regard to the presence of a clumpy torus but our results quite strongly indicate the presence of some MIR thermal component.

The best fit for NGC 3801 requires a NLR component, however it is very dim so we suspect it is likely either a fitting artifact or possibly associated with a large-scale structure in the host galaxy. A likely candidate for this component from the host is the large dust lane visible near the AGN in HST images by Verdoes Kleijn et al. (1999). The forced-torus fit is significantly worse, with a BIC of 453.7 compared to the best-fit BIC of 423.5. We also show an extincted galaxy-only fit with a BIC of 437.3 in Figure 27 as a comparison since the dust lane which is a candidate thermal source would also provide foreground extinction on the core. Overall, we do not find evidence for nuclear obscuration in the form of a clumpy torus and our results are ambiguous for any other warm thermal component in the nucleus itself due to likely emission from the host galaxy.

Das et al. (2005) detected a flat-spectrum non-thermal core in NGC 3801 in $3\,\mathrm{mm}$ images from the Berkeley-Illinois-Maryland Association millimeter-wave array. This does not appear to contribute significantly to the IRS band and we have too little far-infrared photometry to properly compare. They also detected carbon monoxide CO(1-0) emission from the dust disk seen in HST images (Verdoes Kleijn et al., 1999) with a velocity gradient indicating a $\sim 2\,\mathrm{kpc}$ rotating disk of molecular gas and dust with an inferred $3\times 10^{8}\,\mathrm{M}_{\odot}$ of molecular hydrogen. They also detected a $\sim 10^{8}\,\mathrm{M}_{\odot}$ infalling molecular gas clump which they attribute to a recent merger.

Hota et al. (2012) found evidence in GALEX images of a kinematically decoupled core in NGC 3801 in which star formation has recently declined following a post-merger burst but the jet-driven shock has not yet triggered a burst of star formation in the outer regions to the galaxy. In comparison, our SFR estimates based on PAH fit results in Section 4 are higher than average for our sample and our SFR estimates based on narrow Neon lines are typical for the sample. Our continuum models have a bright diffuse ISM which may be associated with the large-scale gas Hota et al. (2012) predict will begin to form stars within the next 10 Myr. Our stellar population models show a high stellar age of 9.4 Gyr but as mentioned earlier our model only accounts for one burst of star formation followed by passive evolution so our results only suggest that the population of low-mass stars is still dominated by the older sub-population.

The best fit model for this source requires only components from the host galaxy suggesting that the AGN does not contribute significantly to the IR flux. The largest difference between the best fit model and the data is the 70$\,\mu$m but that point is an upper limit and within the uncertainty on the best fit model spectrum.

The forced torus model in this source is quite reasonable albeit rather warm with a peak around $10\,\mu\mathrm{m}$, providing a firm upper limit on any possible torus component of $\sim 17\%$ of the flux at $5\,\mu\mathrm{m}$, $\sim 50\%$ of the flux at $15\,\mu\mathrm{m}$, and $\sim 13\%$ of the flux at $30\,\mu\mathrm{m}$. However, it is also unnecessary to explain the infrared spectrum as the best fit has a BIC of $391.9$ and the forced-torus fit has a BIC of $420.3$. Therefore, we see no evidence for a warm IR-bright obscuring structure in NGC 3862.

We show our best fit for 3C 270 in Figure 31, the best fit with a forced torus component in Figure 32, and the fractional contributions to the flux in each of the two models in Tables 25 and 26 respectively. We suspect that the preference for a narrow-line region model in this source is due to a bias in our analysis since BIC favors models with fewer parameters and the NLR model has fewer parameters than a clumpy torus model. Therefore we cannot rule out a torus in this source. The presence of a torus in 3C 270 would be consistent with results by Antonucci (private communication) who detected broad polarized H$\alpha$ emission. This is potentially indicative of a Thomson-scattered hidden broad line region such as that expected in a high-excitation AGN, however Antonucci (private communication) caution that it could instead be due to narrow lines in the complex. Additionally, although van der Wolk et al. (2010a) ultimately do not conclude the existence of a warm MIR thermal component in any of the eight FR-I radio galaxies in their sample they list 3C 270 as a possible exception due to a weak MIR consistent with 200 K dust. Jaffe et al. (1993) note a $\sim 60\,\mathrm{pc}$ disk of cold dust in HST images of 3C 270 surrounding its unresolved nucleus, the inner regions of which may be heated by the AGN (or star formation but our SFR estimates are too low for that) to produce the MIR excess detected.

The best fit model for this source requires two AGN components: a hot dust component and an NLR. Unfortunately, we have no MIPS photometry and only one IRAC channel for M 84.

The forced torus component in the forced-torus fit to M 84 is quite reasonable albeit unusually dim with a best-fit AGN luminosity of $1.2\times 10^{41}\,\mathrm{erg}\,\mathrm{s}^{-1}$ (see Table 38 and discussion in Section 6). However, it has the largest difference in goodness-of-fit out of any of our pairs of models with the best fit having a BIC of $386.9$ compared to $444.4$ for the forced-torus fit. At least part of that is likely due to forced suppression of the hot dust component. Recall that the purpose of the forced-torus fits was to provide an upper limit on any potential torus contribution. Overall, we conclude that there is likely an optically thin thermal hot dust component and that the bumps in the spectrum toward the MIR are probably a thermal component although likely not in the form of a Nenkova et al. (2002) clumpy torus.

Although every source in our sample has a radio jet and a couple have large scale optical jets, M 87 is the only source in which the red power-law synchrotron spectrum of the inner jet provides a significant contribution to the IR flux. This power law has a best-fit index of $\alpha=1.2$ (recall that we use the $L_{\nu}\propto\nu^{-\alpha}$ convention) which is consistent with the optical spectral indices of $\alpha_{o}\sim 1.0-1.2$ Perlman et al. (2001) found for inter-knot regions of the inner jet. The model with a “forced” torus component, shown in Figure 36, actually has a marginally lower BIC (392.0) than the otherwise “best” fit (398.7), shown in Figure 35, but this must be a fitting artifact as it contributes no flux. This lack of a torus and strong power law spectrum is consistent with findings by Whysong & Antonucci (2004), previously mentioned in Section 1.2, that M 87 is weak in IR and that its nuclear IR emission is consistent with a synchrotron spectrum from the base of its jet. Similarly, the absence of a torus is consistent with the upper limits on MIR thermal emission established by Perlman et al. (2007).

The model of NGC 7052 with the lowest BIC and the model with a forced torus have very similar goodness-of-fit with the best fit having a BIC of 425.7 and the forced-torus fit having a BIC of 429.8. However, inspection of the forced torus fit shows that it is unusually cold for a circumnuclear torus. Indeed, HST images of NGC 7052 by Verdoes Kleijn et al. (1999) show a promising candidate for such a cold thermal source in the form of a kpc-scale dust disk which crosses the core region just off to the side of our line-of-sight into the nucleus. Overall, we conclude that a warm thermal component, such as a torus, is unnecessary to explain the IRS flux.

Some of the stellar age parameters for the best fits, shown in Table 33, are unusually low for the various SFR estimates. This could be explained by those sources having recently quieted after a starburst, indeed Das et al. (2005) claim something similar for NGC 3801, but it seems improbable that so many of such a small sample would be within 2 Gyr of the end of a starburst and NGC 3801 is not one of the sources with these low stellar ages. The fits with a forced torus, shown in Table 34, are better in this regard with only 3C 66B and M 87 having this unusual combination. We do not put much stock in this result because stellar ages are generally poorly constrained by infrared data since young stars are brightest at shorter wavelengths and, as mentioned earlier, the GRASIL stellar population model assumes a single Gyr-long starburst followed by passive evolution which may not be a good approximation in these galaxies. This would likely be clearer with more NIR spectrometry and optical spectroscopy (especially in blue) would help characterize the population of young stars.

The diffuse ISM components of the two model sets, the best fit models shown in Table 35 and the forced-torus models shown in Table 36, show lower limits on their interstellar radiation fields, PAH fractions, and ISM luminosities which are generally consistent between the two sets with the exception of M 87 which shows an improbably weak radiation field and significantly higher ISM luminosity in the forced-torus model.

The fit parameters for the torus, when present (forced-torus models and NGC 315) are shown in Tables 37 and 38. Of particular interest is the low inclination of NGC 3801, the large half-opening angle and high escape probability of NGC 3801, NGC 3862, and M 84. These are unusual for sources in which we can’t see broad lines from the central engine. The masses of obscuring material in the forced-torus fits to NGC 3801 and M 84 are remarkably small compared to the others with both being $\sim 10^{-3}\,\mathrm{M}_{\odot}$. And finally the power law indices of the cloud distributions in the forced-torus models are inconsistent with the silicate line ratios in several sources, especially three previously mentioned in this paragraph: NGC 3801, NGC 3862, and M 84. When considered together these issues suggest that the forced-torus fits to NGC 3801, NGC 3862, and M 84 are inconsistent with independent observations.