Mid-infrared imaging- and spectro-polarimetric subarcsecond observations of NGC 1068
We present sub-arcsecond 7.513 m imaging- and spectro-polarimetric observations of NGC 1068 using CanariCam on the 10.4-m Gran Telescopio CANARIAS. At all wavelengths, we find: (1) A 90 60 pc extended polarized feature in the northern ionization cone, with a uniform 44° polarization angle. Its polarization arises from dust and gas emission in the ionization cone, heated by the active nucleus and jet, and further extinguished by aligned dust grains in the host galaxy. The polarization spectrum of the jet-molecular cloud interaction at 24 pc from the core is highly polarized, and does not show a silicate feature, suggesting that the dust grains are different from those in the interstellar medium. (2) A southern polarized feature at 9.6 pc from the core. Its polarization arises from a dust emission component extinguished by a large concentration of dust in the galaxy disc. We cannot distinguish between dust emission from magnetically aligned dust grains directly heated by the jet close to the core, and aligned dust grains in the dusty obscuring material surrounding the central engine. Silicate-like grains reproduce the polarized dust emission in this feature, suggesting different dust compositions in both ionization cones. (3) An upper limit of polarization degree of 0.3 per cent in the core. Based on our polarization model, the expected polarization of the obscuring dusty material is 0.1 per cent in the 813 m wavelength range. This low polarization may be arising from the passage of radiation through aligned dust grains in the shielded edges of the clumps.
keywords:techniques: polarization, techniques: high angular resolution, galaxies: active, galaxies: Seyferts, infrared: galaxies
NGC 1068 is the archetypal type 2 active galactic nucleus (AGN), whose proximity (D = 12.5 Mpc, and 1 arcsec = 60 pc, adopting H km s Mpc) and high brightness make it an ideal target for polarimetry. This object is probably the best-studied AGN regarding polarimetry, and this has allowed us to obtain a better understanding of the AGN structure. The most important polarimetric study of NGC 1068, for the sake of the entire field, was certainly the detection of polarized broad emission lines in the optical (0.350.70 m) wavelengths by Antonucci & Miller (1985). This detection is interpreted through scattering of the radiation from the central engine (black hole and accretion disc) into our line of sight (LOS) by matter in the ionization cones. This result allowed us to understand that the central engine of NGC 1068 is obscured by a dusty structure, giving a major boost to the unified model of AGN. The unified model (i.e. Lawrence, 1991; Antonucci, 1993; Urry & Padovani, 1995) of AGN posits that the observational differences between AGN arise from the different orientations they present us. In this scheme, the AGN classification solely depends on the anisotropic obscuration of the central engine by the optically and geometrically thick distribution of dust.
Pioneering infrared (IR) polarimetric observations (Knacke & Capps, 1974; Lebofsky et al., 1978) found that the nucleus of NGC 1068 was highly obscured and polarized. The observed degree of polarization rises from 1.6 0.4 per cent at 1.25 m to 2.7 0.2 per cent at 3.45 m, and the position angle (P.A.) of polarization rotates from 99° 7° at 1.25 m to 125° 2° at 3.45 m in a 6 arcsec (360 pc) diameter aperture. They suggested that the IR polarization arises from an obscured power-law source, which is not directly observable at wavelengths shorter than 1 m. Using 813 m observations a the 3.9-m Anglo-Australian Telescope (AAT), Aitken et al. (1984) found a polarization spectrum with a uniform degree, 1.39 0.09 per cent, and a P.A. of 54.8° 1.9° within apertures of 4.2 arcsec (252 pc) and 5.2 arcsec (312 pc) diameter. Although they were not able to rule out non-thermal processes, they suggested that dust emission from non-silicate and featureless grains is responsible for the mid-IR (MIR) polarization. Using these data and additional 0.364.8 m broad-band imaging polarimetric observations, Bailey et al. (1988) concluded that the most likely polarization mechanism at near-IR (NIR) arises from the passage of centrally produced radiation through aligned dust grains in the obscuring dusty structure; a mechanism known as dichroic absorption. This result was further supported by the flip in the P.A. of polarization of 70° from NIR to MIR. They interpreted this result as the change from dichroic absorption to dichroic emission from aligned dust grains in the dusty structure obscuring the central engine. At shorter wavelengths, a wavelength-independent intrinsic (after correction for starlight dilution) polarization of 16 per cent with a constant P.A. of 97° from the ultraviolet (UV, Antonucci et al., 1994) to optical wavelengths (Antonucci & Miller, 1985) was found. The fact that the P.A. of polarization at UV and optical is different to that in the NIR, suggests that different mechanisms of polarization dominate in these wavelength regimes. Further modeling (Young et al., 1995; Watanabe et al., 2003) and observations (Capetti et al., 1995; Packham et al., 1997; Lumsden et al., 1999; Simpson et al., 2002; Lopez-Rodriguez et al., 2015) have shown that electron scattering is the dominant polarization mechanism from UV to optical, whilst dust distributed within the 23 arcsec (120180 pc) central region accounts for the polarization features in the IR (Bailey et al., 1988). Specifically, dichroic absorption from aligned dust grains within the dusty circum-nuclear obscurer in NGC 1068 is the dominant polarization mechanism in the 15 m wavelength range. Based on these studies, and together with the lack of evidence of variations in the P.A. of polarization over a period of months or years, non-thermal polarization mechanisms (e.g. synchrotron emission from a jet) can be ruled out as a polarization mechanism in the core of NGC 1068.
At large scales, imaging polarimetric observations (Scarrott et al., 1991) at the V band using the 3.9-m AAT showed that the optical polarization follows the arm and inter-arm structures of the galaxy. The degree of polarization increases in the large-scale inner bar (32 arcsec, 1.92 kpc) at a P.A. of 48°, the so-called NIR bar (Scoville et al., 1988; Schinnerer et al., 2000; Emsellem et al., 2006). Further high-spatial resolution optical (0.50.6 m) polarimetric observations (Capetti et al., 1995) using the Hubble Space Telescope (HST) showed a centro-symmetric polarization pattern along the ionization cones in the 10 arcsec (600 pc) central region. This polarization pattern is the signature of a central point source whose radiation is ultimately polarized through scattering by dust and/or electrons (and which can be polarized from multiple scattering into the funnel of the obscuring dusty structure and/or broad line clouds). At UV and optical wavelengths, these studies showed that electron scattering is the dominant polarization mechanism in the ionization cones of NGC 1068, with the optical polarization in the off-nuclear regions following the magnetic field of the NIR bar.
Using high-spatial resolution imaging polarimetric observations at 9.7 m on the 8.1-m Gemini telescope, Packham et al. (2007) found complex polarization structures within the 22 arcsec (120 pc 120 pc) central region of NGC 1068. Specifically, (1) polarization arising from aligned dust grains in the narrow line emission regions North of the core, (2) dust being channelled toward the central engine South, East and West of the core, and (3) a nuclear low, 0.6 per cent, polarization interpreted as the polarization arising from the compact (22 pc) obscuring dusty structure. Although only one filter was used in this study, making it difficult to disentangle the several polarization mechanisms, it showed the potential of MIR polarimetric observations to study the structure within and around AGN.
In this paper we examine the interaction of the AGN with the surrounding dusty structures within the central 2 arcsec (120 pc) region of NGC 1068. We also put constraints on the polarization of the obscuring dusty structure. We performed sub-arcsecond resolution 813 m imaging- and spectro-polarimetric observations using CanariCam on the 10.4-m Gran Telescopio CANARIAS (GTC) at the Roque de los Muchachos Observatory. The paper is organized as follows: Section 2 describes the observations and data reduction, Section 3 presents our polarimetric results. Polarization models are developed in Section 4 to explain the several polarization features, and then analyzed and discussed in Section 5. In Section 6 we present the conclusions.
2 Observations and data reduction
NGC 1068 was observed on December 29 and 30, 2012, using the imaging polarimetric mode (Packham et al., 2005) of CanariCam (Telesco et al., 2003) on the 10.4-m GTC in Spain. CanariCam uses a 320 240 pixel Si:As Raytheon array, with a pixel scale of 0.0798 arcsec pixel. The imaging polarimetric mode uses a half-wave retarder (half-wave plate, HWP), a field mask and a Wollaston prism. The Wollaston prism and HWP are made with sulphur-free CdSe. The HWP is chromatic, resulting in a variable polarimetric efficiency across the 7.513 m wavelength range that has been well determined (Packham et al., 2008). This mode is usable across the entire wavelength range. In standard polarimetric observations, the HWP is set in four P.A. in the following sequence: 0°, 45°, 22.5°and 67.5°. The field mask consisted of a series of slots of 320 pixels 25 pixels each, corresponding to a field of view (FOV) of 25.6 arcsec 2.0 arcsec, where a total of three slots can be used, providing a non-contiguous total FOV of 25.6 arcsec 6.0 arcsec.
The Si2 ( = 8.7 m, 1.1 m, 50 per cent cut-on/off), Si4 ( = 10.3 m, 0.9 m, 50 per cent cut-on/off) and Si5 ( = 11.6 m, 0.9 m, 50 per cent cut-on/off) filters provide the largest wavelength coverage with the best combination of sensitivity and spatial resolution for the filter set of CanariCam in the 10 m atmospheric window. We thus used this combination of filters. Additionally, to study possible star formation regions in and around the core of NGC 1068, we performed observations using the PAH2 ( = 11.3 m, 0.6 m, 50 per cent cut-on/off) filter. Observations were made using a standard chop-nod technique to remove time-variable sky background and telescope thermal emission, and to reduce the effect of 1/f noise from the array/electronics. In all observations, the chop-throw was 8 arcsec, the chop-angle was 0° E of N, and the chop-frequency was 1.93 Hz. The angle of the short axis of the array with respect to North on the sky (i.e. instrumental position angle, IPA) was 90° E of N, and the telescope was nodded every 52s along the chopping direction. Thus, only one slot with a FOV of 25.6 arcsec 2.0 arcsec was used in the observations, with the negative images (produced by the chop-nod technique) within the same slot. For each filter, two observational sets were observed, except for the PAH2 filter where three observational sets (one observational set the first night and two observational sets the second night) were observed. A summary of the observations is shown in Table 1. To improve the signal-to-noise ratio (S/N) of the observations, the negative images on the array were also used, providing the total useful on-source time shown in Table 1.
For the imaging-polarimetric observations, the total on-source time was estimated accounting for the positive and negative images, produced by the chop-nod technique, on the array. FWHM of the PSF.
Data were reduced using custom idl routines. The difference for each chopped pair was calculated and the nod frames, then differenced and combined to create a single image per HWP P.A. During this process, all nods were examined for high or variable background that could indicate the presence of clouds or variable precipitable water vapor. Fortunately, no data needed to be removed for these reasons. As NGC 1068 was observed in different sets, each HWP P.A. frame was registered and shifted to a common position, then images with the same HWP P.A. were co-added. Next, the ordinary (o-ray) and extraordinary (e-ray) rays, produced by a Wollaston prism, were extracted and the Stokes parameters , and were estimated according to the ratio method (e.g. Tinbergen, 2005). Then, the degree P and P.A. , of polarization were estimated.
The instrumental polarization was corrected based on data provided by the GTC website111Information at http://www.gtc.iac.es/instruments/canaricam/canaricam.php#Instrumental_Polarization. Specifically, the instrumental polarization is P = 0.6 0.2 per cent in all filters with a dependence on the P.A. of polarization given by P.A. = 29.6°, where is the Nasmyth rotator mechanical angle; and is the telescope elevation. The instrumental polarization was corrected as follows. The normalized Stokes parameters, q and u, of the instrumental polarization were estimated using the degree P and position angle P.A. of the instrumental polarization. Then, q and u were subtracted from the normalized Stokes parameters of NGC 1068. Finally, the polarization efficiency was corrected based on data provided by the GTC website222Information at: http://www.gtc.iac.es/instruments/canaricam/canaricam.php#Polarization_Measurement_Efficiency. Specifically, the polarization efficiency is 90.3 per cent, 99.5 per cent, 95.9 per cent, and 97.0 per cent at 8.7, 10.3, 11.6, and 11.3 m PAH, respectively. The measurements of the degree of polarization were corrected for polarization bias using the approach by Wardle & Kronberg (1974).
The polarized young stellar object, AFGL 2403, was observed at 8.7 m and 11.3 m immediately before NGC 1068, with an on-source time of 73s in both filters. In total flux, AFGL 2403 was used as point-spread function (PSF) calibrator. Table 1 shows the full width at half maximum (FWHM) of AFGL 2403, where a Moffat function with two parameters, FWHM and , best described the delivered PSF (e.g. Radomski et al., 2008). In polarimetry, AFGL 2403 allowed us to characterize the polarization observations because it is bright, 95.2 Jy, and polarized, 1.5 0.6 per cent and 1.2 0.3 per cent at 8.7 m and 11.3 m, respectively. The degree of polarization of AFGL 2403 corrected for instrumental polarization and polarization efficiency at 8.7 m, was estimated at 1.3 0.2 per cent. Our measurement is in excellent agreement with the degree of polarization of 1.4 0.8 per cent at 8.7 m measured by Smith et al. (2000, fig. 2). The zero-angle of the P.A. of polarization was calculated as the difference of the P.A. of polarization from our measurement, 100° 2°, and Smith et al. (2000), 39° 4°, i.e. = 61° 5° at 8.7 m. At other wavelengths, the zero-angle was estimated using the wavelength dependence of the P.A. of polarization observed by Smith et al. (2000, fig. 2).
Dedicated flux-standard stars were not observed. Flux calibration was performed using the N-band spectra observed with Michelle on the 8.1-m Gemini North (Mason et al., 2006). Specifically, flux calibration was performed using the spectral points at 8.7, 10.3, 11.3, and 11.6 m of the 0.4 arcsec wide slit oriented 20° E of N and centered at the peak of NGC 1068 (see Mason et al., 2006, fig. 2). The extracted fluxes were 7.5, 8.3, 11.0, and 11.9 Jy, respectively. Then, measured counts in a 0.4 0.4 arcsec simulated slit aperture from our images at each wavelength were equated to the flux densities from the MIR spectra by Mason et al. (2006). Finally, the factor Jy counts was estimated and used in the measurements of the flux densities in the several apertures shown in Section 3.1.1. These observations can be found at GTC Public Archive.
NGC 1068 was observed on July 22, 2013, during commissioning of the spectro-polarimetric mode (Packham et al., 2005) of CanariCam on the 10.4-m GTC. The spectro-polarimetric mode uses a HWP, a field mask, a slit and a Wollaston prism. A 0.41 arcsec (5 pixels) wide slit oriented at 0° E of N to cover the North-South polarized features of NGC 1068 was used. The low resolution N-band ( = 10.4 m, 5.2 m, 50 per cent cut-on/off) grating was used, resulting in a dispersion of 0.019 m pixel and R 175.
Observations were made using a standard chop-nod technique as described in Section 2.1 with a chop-throw of 8 arcsec, chop-angle of 90° E of N and a chop frequency of 1.93 Hz. The IPA was 0° E of N, and the telescope was nodded every 45.5s along the chopping direction. Thus, only one slot with a FOV of 25.6 arcsec 2.0 arcsec was used in these observations. We took two spectro-polarimetric observational sets with an on-source time of 947s each (Table 1). Before these observations, we obtained acquisition N-band imaging polarimetric observations with an on-source time of 79s. The slit was aligned to the peak of the total flux of NGC 1068 to better than 2 pixels (0.16 arcsec).
Data were reduced as described in Section 2.1. For each spectro-polarimetric observational set of NGC 1068, the o- and e-rays were extracted in two different locations (Section 3.2) of the slit: (1) PSF-extraction of the Southern feature, and (2) at 0.35 arcsec (4.4 pixels) North of the peak of the total flux density image using a fixed 0.4 arcsec (5 pixels) aperture. CanariCam shows a slight (1 pixel) curvature across the array in the spectral direction, which was measured using the observations of the polarized young stellar object, AFGL 2591. This curvature was taken into account during the extraction of the spectra. For each observational set, Stokes parameters, , and , were estimated according to the ratio method, and then co-added to obtain the final spectra. Finally, the degree and P.A. of polarization were estimated.
The instrumental polarization was corrected as described in Section 2.1. The polarization efficiency was corrected using observations of the unpolarized standard star, HD 186791, through a wire-grid in the optical path of CanariCam. These observations allow us to estimate the polarization efficiency spectra of the HWP across the 7.513 m wavelength range. The polarization efficiency was corrected by multiplying the spectrum of the degree of polarization of NGC 1068 by the inverse of the polarization efficiency spectrum. Although the variation of the polarization efficiency was accounted for, a variable S/N of the Stokes parameters as a function of the wavelength still remains. We corrected this effect by binning the and spectra across the wavelength axis in bins of 0.285 m (15 pixels) to obtain higher S/N across the spectrum. The measurements of the degree of polarization were corrected for polarization bias using the approach by Wardle & Kronberg (1974).
The polarized young stellar object, AFGL 2591, was observed using the spectro-polarimetric mode immediately before NGC 1068. In total flux, AFGL 2591 was used as a PSF calibrator, giving us the FWHM across the wavelength range for the PSF-extracted spectra. A Moffat function with two parameters, FWHM and , best described the delivered PSF. In polarimetry, AFGL 2591 allowed us to estimate the zero-angle calibration of the observations because it is bright, 281.5 Jy, and highly polarized, 5.6 1 per cent (Smith et al., 2000, fig. 2). The zero-angle calibration, , was estimated as the difference of the measured P.A. of polarization of our observations, 58.5° 0.3°, and the P.A. of polarization by Smith et al. (2000), 170° 1°. We measured a constant P.A. of polarization across the 7.513 m wavelength range, in excellent agreement with Smith et al. (2000). Thus, the zero-angle of polarization was estimated to be 111.5° 1.0°.
The total flux spectrum of NGC 1068 was wavelength-calibrated to an accuracy of 0.019 m pixel using 10 m window sky lines present in the spectrum. Then, the total flux spectrum was binned to a spectral resolution of 0.076 m (4 pixels) to improve the S/N. Next, the spectrum of NGC 1068 was divided by the spectrum of the standard star, HD 186791 (K3III). Cohen & Davies (1995) show that late K and M stars have fundamental vibration-vibration bands of SiO in their spectra, which significantly depress the spectrum in the 7.510 m wavelength range, and therefore affect the ratio with the science object. To remove this band and obtain the original spectrum of NGC 1068, the total flux spectrum of NGC 1068 was multiplied by the template spectrum of the standard star HD 186791, provided by the Gemini website333Template at http://www.gemini.edu/sciops/instruments/mid-ir-resources/spectroscopic-calibrations#waveca. Flux calibration was achieved by using the N-band spectra of NGC 1068 (in Jy) observed with Michelle on the 8.1-m Gemini North (Mason et al., 2006). The measured nuclear spectrum (in counts) in a 0.4 arcsec aperture was extracted and equated to the flux density from the MIR spectrum by Mason et al. (2006). The factor Jy counts was estimated and used in several extractions of the spectra from our observations.
Fig. 1 shows the total flux images at 8.7, 10.3, 11.6, and 11.3 m (PAH). At all wavelengths NGC 1068 appears as an unresolved core in the E–W direction, with extended emission along N–S. The N–S extended emission slightly bends to the N–W direction up to 0.5 arcsec (30 pc), and then to the N–E direction up to 1.4 arcsec (84 pc); this feature was termed as ‘tongue’ (see Bock et al., 2000, fig. 2). At 1.4 arcsec N–E from the core, the NE knot is located. These structures are in excellent agreement with previously published high-spatial resolution MIR observations (e.g. Bock et al., 2000; Tomono et al., 2001; Galliano et al., 2005; Poncelet et al., 2007).
We made measurements of the nuclear total flux density images centred at the peak of the total flux density at several diameter apertures (hereafter, aperture refers to diameter) to compare with previously published values (Table 2). In all cases, photometric errors were estimated by the variation of the counts in subsets of the data. Our photometric measurements are consistent with published flux density measurements. We also made photometric measurements of the Southern feature and North knot (Section 3.1.2, Fig. 2 and 3) in a 0.4 arcsec (24 pc) aperture centered on the peaks of polarized flux, for all features, and at each wavelength (Fig. 2 and 3).
Fig. 2 shows the polarized flux images with the overlaid polarization vectors at 8.7, 10.3, 11.6 and 11.3 m (PAH). In these figures, the overlaid polarization vectors are proportional in length to the degree of polarization, and their orientation shows the P.A. of polarization. Only polarization vectors with are shown. is the uncertainty of the degree of polarization per pixel, estimated as , where and are the uncertainties of the normalized Stokes parameters (Wardle & Kronberg, 1974). Fig. 3 shows the contours of the degree and P.A. of polarization for those polarization vectors with . At all wavelengths, several polarized features are found:
The most prominent feature is the uniform P.A. of polarization with an extension of 1.5 arcsec 1 arcsec (90 pc 60 pc) North of the peak of the total flux density (white crosses in Fig. 2). The P.A. of polarization is roughly constant, 44°, in an area of 1.4 arcsec 0.8 arcsec (84 pc 48 pc), with variations of 10° (Fig. 3, bottom). In this region, we found a resolved polarized knot at 0.4 arcsec (24 pc) North from the peak of the total flux images, termed ‘North knot’. For the North knot, the degree of polarization shows a steep increase from South to North, with a peak in the degree of polarization, which then decreases smoothly to the NW and SW directions (Fig. 3-up). The degree of polarization decreases at longer wavelengths (Table 4, Fig. 4) with a constant P.A. of polarization across the wavelength range.
The polarized flux images (Fig. 2) show a peak slightly shifted, 0.16 arcsec (9.6 pc), SE from the peak of the total flux images. The degree of polarization increases at longer wavelengths with a P.A. of polarization increasing from 12° 10° to 37° 5° at 8.7 m and 11.6 m, respectively.
At 0.5 arcsec (30 pc) South from the peak of the total flux images, an extended polarized feature is found. This structure is spatially coincident with the ‘SW Lobe’ observed in polarized flux at 2.0 m using HST/NICMOS by Simpson et al. (2002) and in the K band using MMT/MMT-Pol by Lopez-Rodriguez et al. (2015).
We made measurements of the nuclear polarization to compare with the literature (Table 3). In all cases, polarimetric errors were estimated by the variation of the counts in subsets of the data. Our measured nuclear polarization in a 2 arcsec (120 pc) aperture at all wavelengths is higher than the degree of polarization of 1.30 0.05 per cent in a 2 arcsec aperture in the N-band using the 3.9-m AAT (Lumsden et al., 1999). These authors measured a P.A. of polarization of , which is in marginal agreement with our measurements within the uncertainties. We interpret these results as (1) an increase in the measured degree of polarization by an improvement in the spatial resolution, (2) the mix of different polarization structures within the aperture, and (3) the use of a wide bandwidth within the 10 m atmospheric window, which implies that their measurements were more strongly affected by the wavelength dependence of the different mechanisms of polarization. We note that the P.A. of polarization by Lumsden et al. (1999) is in better agreement with our measured P.A. of polarization in the Northern ionization cones (Table 4). This result indicates that their nuclear polarization measurement was dominated by the polarization arising from the Northern ionization cone (Section 5.1), as a larger amount of polarized flux is coming from the North regions within their 2 arcsec (120 pc) aperture. Within the nuclear 1.7 1.2 arcsec (102 72 pc) elliptical aperture, our measured degree of polarization of 2.2 0.7 per cent at 8.7 m is in excellent agreement with the measured degree of polarization of 2.48 0.57 per cent at 9.7 m using the 8.1-m Gemini (Packham et al., 2007). We note a difference in the measured P.A. of polarization of 26.7° 15.3° at 9.7 m by Packham et al. (2007), and our measured 46° 9° and 53° 8° at 8.7 m and 10.3 m, respectively. Both measurements are in marginal agreement within the uncertainties, however we note that the P.A. of polarization by Packham et al. (2007) is in better agreement with our measurements of the southern polarized feature (Table 4). This difference can be understood as the polarimetric observations of Packham et al. (2007) were more sensitive to the central 1 arcsec due to their low S/N. Our degree and P.A. of polarization are calibrated to standard stars for polarization measurements, but we note that the total and polarized flux morphologies of the Gemini/MICHELLE polarimetric data by Packham et al. (2007) seem to be rotated by 180° from our dataset. We confirmed the CanariCam orientation through comparing our morphology with other total flux measurements, including de-convolved images, and find a high degree of spatial coincidence. We speculate that perhaps the world coordinate system (WCS) used to indicate North in the Gemini/MICHELLE polarimetric data by Packham et al. (2007) was incorrect, perhaps due to an additional reflection or translation in the polarimetry mode compared to direct imaging.
Based on the observed polarized features and their S/N, we made polarimetric measurements of (1) the North knot in a 0.4 arcsec aperture, and (2) the Southern feature at 0.16 arcsec (9.6 pc) to the peak of the total flux density in a 0.4 arcsec (24 pc) aperture. The total flux density, degree of polarization, P.A. of polarization and polarized flux density for each feature and filter are shown in Table 4. Figure 4 shows the measurements (blue squares) for each feature, as well as the used apertures (blue circles in the central panel). We found an aperture dependence on the polarization in the Northern ionization cone. Specifically, a constant degree, 5 per cent, and P.A., , of polarization is measured using a 1.4 0.8 arcsec (84 48 pc) elliptical aperture with P.A. .
Flux densities estimated in a 0.4 arcsec (24 pc) aperture centred in the Southern feature and North knot, respectively.
Fig. 4 shows the total flux density (black solid line), the degree and P.A. of polarization and the polarized flux density (black filled-dots) for the PSF-extracted spectra of the 0.4 arcsec (24 pc) fixed aperture spectra of the North knot (left), and the Southern feature (right). The 0.41 arcsec (5 pixels) slit (black solid vertical lines) is shown. The total flux spectra were binned to a 0.076 m (4 pixels) resolution, whilst a 0.285 m (15 pixels) bin was used for the degree, P.A. of polarization and polarized flux. The uncertainties were estimated as a variation of the counts within the binned data.
For the Southern feature, the total flux spectrum shows strong dust continuum emission with a broad silicate absorption feature. This spectrum is similar and consistent with the 0.4 0.4 arcsec (24 24 pc) slit aperture of the 0.4 arcsec SSW spectrum using the 8.1-m Gemini by Mason et al. (2006). The degree of polarization slightly increases with increasing wavelength. The P.A. of polarization slightly increases from 12° 10° to 37° 5° across the 812 m wavelength range. The polarized flux increases with wavelength.
For the North knot, the total flux spectrum is similar to and consistent with the 0.4 arcsec NNE spectrum of Mason et al. (2006), also showing strong dust continuum emission with a broad silicate absorption feature and narrow fine-structure lines. In this spectrum, a [\ionAriii/\ionMgvii] blend at 8.99 m, and a [\ionSiv] fine-structure at 10.51 m are detected. The degree of polarization decreases with increasing wavelength and the P.A. of polarization is roughly uniform through the spectrum. The polarized flux increases with increasing wavelength.
Aitken et al. (1984) measured a uniform degree, 1.29 0.09 per cent, and P.A., 54.8° 1.9°, of polarization across the 8 m wavelength range using a 4.2 arcsec (252 pc) beam. We note that their measured P.A. of polarization is consistent with the P.A. of polarization of the northern ionization cone measured in our observations. We thus gather that the measurements of Aitken et al. (1984) were dominated by the northern ionization cone within their 4.2 arcsec (252 pc) beam, as a larger amount of polarized flux is coming from the North regions within their beam. This result makes it difficult to obtain information about the core of NGC 1068, and shows the potential of subarcsecond resolution polarimetric MIR observations to obtain more sensitive observations with the aim to study the cores of AGNs.
4 Polarization model
We aim to reproduce the observed polarization in the 7.5-13 m wavelength range of the several observed features discussed in Section 3. Specifically, we wish to reproduce the observed degree and position angle of polarization for 1) the northern and southern ionization cones, and 2) the obscuring dusty structure.
4.1 Components of the polarization model
MIR spectro-polarimetric observations of 55 objects (e.g. young stellar objects, star formation regions, and active galactic nuclei) using 4-m class telescopes found that 90 per cent of the objects can be explained by dichroism (Smith et al., 2000). Dichroic extinction and emission can compete at a given wavelength. Both mechanisms need to be disentangled through a multi-wavelength study. Fortunately, some dust grain features, such as silicates, are present in the MIR wavelength range, and can be used to distinguish between the two mechanisms. In the case of silicates, the absorptive polarization shows a peak at 10 m, while the emissive polarization is less structured. In general, if a rotation of the polarization angle with wavelength is observed, then more than one mechanism of polarization may be present. Unless the several polarization mechanisms have intrinsically the same polarization angle, the net polarization angle will be a function of wavelength. Both the degree and angle of polarization profiles are crucial to investigate the emissive and absorptive polarization components at MIR wavelengths.
The dichroic components can be disentangled following the procedure put forward by Aitken et al. (2004). These authors presented a procedure to separate and identify the absorptive, p, and emissive, p, polarization components within the 7.513 m wavelength range. The emissive polarization component is given by p p, where is the wavelength-dependent extinction curve. This condition can be used as long as the difference in orthogonal optical depths of the dust grains is less than unity for p, which makes p independent of the optical depth. This situation will hold if a few tens. For the extinction curve, the standard silicate-graphite interstellar dust grain of R is assumed444Note that recent studies (Gao, Li & Jiang, 2013) suggested that a trimodal grain size distribution with a combination of R, and is required to achieve a good fit to the extinction curve of the Galactic center from the UV to the MIR., which appears to be in agreement with the MIR extinction at different sightlines through the Galactic center (Weingartner & Draine, 2001). Variations in our model of less than 2% are found when extinction curves of R, and were used. For the absorptive component, p, we took the Becklin-Neugebauer (BN) object in Orion (Aitken et al., 1989) because it is the best-defined absorptive component with the highest S/N, and commonly used in these studies (Smith et al., 2000).
The polarization model to separate the absorptive and emissive polarization components is constructed as follows. It assumes an emission source that can be either polarized or unpolarized, and which is viewed through a cold dichroic sheet of dust grains. The Stokes parameters from both components simply add up, and the observed Stokes parameters, and , can be fitted as a linear combination of both components:
where , , and are the constants used for the fitting procedure. We followed the fitting procedure described by Aitken et al. (2004). Then, the final degree of polarization, P, and the position angle, P.A, are estimated.
The total flux spectrum is defined as an extinguished blackbody component with a characteristic temperature T. We have assumed an extinction curve with R, and the characteristic temperature was sampled in steps of 10 K. The polarized flux was estimated as the modeled total flux, times the total degree of polarization, P.
4.2 Fitting of the ionization cones
In the case of the northern ionization cone, a constant P.A. of polarization (Table 4, Fig. 4-left) suggests a unique polarization mechanism. In objects in our Galaxy where MIR polarization arises through dichroic absorption by aligned dust grains, a prominent silicate peak in the degree of polarization is present just longwards of 10 m (i.e. Aitken et al., 1986; Smith et al., 2000). No such feature is present in the polarization spectrum of the North knot in NGC 1068, which instead shows a slow decrease in the degree of polarization with increasing wavelength. Thus, non-silicate features, i.e. non 9.7 m feature, are present in the MIR polarization. We give a further interpretation of these features in Section 5.1.
In the southern ionization cone, the rotation of the P.A. of polarization (Table 4, Fig. 4-right) suggests that more than one mechanism contributes to the polarization. Following the procedure described in Section 4.1, we simultaneously fit the measured total flux, degree and P.A. of polarization of the Southern feature (Figure 5). The polarization model has four free model parameters: 1) the visual extinction A towards the core, 2) the blackbody characteristic temperature T, 3) the degree of polarization, and 4) the P.A. of polarization. The fit was considered acceptable when the reduced was minimized. We obtain a reduced = 0.21, 2.22, 15.8 and 14.2 for the flux density, degree and P.A. of polarization, and polarized flux density, respectively. Although we are aware of the limitations of the fitting procedure, we are more interested on explaining the overall behaviour of the polarization spectrum rather than optimize the goodness of the fitting, as we are limited by the low S/N of the observations. Further development of a Bayesian fitting procedure is out of the scope of this paper, but we think it will benefit the statistical analysis presented here as shown by Lopez-Rodriguez (2016). For the best fit, we found a blackbody component with a characteristic temperature of T K, extinguished with A 9 1 mag. The emissive polarization component is dominant and describes the observed wavelength dependence of the degree of polarization with a P.A. = 35° 1°. The absorptive component contributes 20 per cent of the polarization with a P.A. = 127° 1°. We give a further interpretation of this feature in Section 5.2.
4.3 Fitting of the dusty central structure
We now investigate the expected MIR polarization of the obscuring dusty structure in terms of dichroism. Dichroic absorption is the dominant mechanism of polarization at NIR wavelengths in the core of NGC 1068. This allows us to constrain the absorptive and emissive polarization components in the 213 m wavelength range. For the absorptive component, we assume a Serkowski (Serkowski et al., 1975) curve up to 8 m, followed by the absorptive profile of the BN object in Orion (Section 5.2.1) in the 8-13 m wavelength range. This composition takes into account the silicate feature at MIR wavelengths, that, if not used, would lead to the expected MIR polarization being underestimated by a factor 100. Using NIR polarimetric adaptive optics observations, Lopez-Rodriguez et al. (2015) estimated an intrinsic (starlight-corrected) polarization at K of 7.0 2.2 per cent in a 0.5 arcsec (30 pc) aperture. Using this value to constrain the absorptive polarization allows us to estimate the intrinsic polarization of the obscuring dusty structure in NGC 1068. To constrain the extinction curve (with R 5.5 as in Sec. 5.2.1), lower and upper limits of the visual extinction towards the core of NGC 1068 must be assumed. For the upper limit we used A 39 mag (Packham et al., 1997; Watanabe et al., 2003; Lopez-Rodriguez et al., 2015), interpreted as the extinction to the NIR emitting regions. For the lower limit, A is estimated using the 9.7 m silicate feature, 0.41, of the nuclear spectrum in a 0.4 0.4 arcsec (24 24 pc) slit aperture by Mason et al. (2006). This result was then converted555The conversion of the 9.7 m silicate feature to visual extinction is A (Roche & Aitken, 1984) to A 8 mag. The difference of both visual extinctions can be interpreted as a temperature gradient and inhomogeneities (clumpiness) in the obscuring dusty material (i.e. Pier & Krolik, 1992; Imanishi & Ueno, 2000; Levenson et al., 2007).
With the constraints on each component given above, we estimated the absorptive and emissive intrinsic polarization of the obscuring dusty structure of NGC 1068 in the 213 m wavelength range (Fig. 6). At wavelengths 4 m, the absorptive component is dominant for all values of A. The emissive component is dominant for low values of the visual extinction in the wavelength ranges of 48 m and 11 m, while both absorptive and emissive components compete in the 810 m wavelength range. Based on this model, the expected polarization from extrapolating the NIR absorptive component is 0.1 per cent in the 813 m wavelength range. We note that while there could be a component of polarized emission from those dust grains, if they are sufficiently warm to emit in the MIR, the uncertainties in estimating any contribution are very large. We give a further interpretation of this feature in Section 5.3.
5 Analysis and discussion
We outline the general physical structure of the central 100 pc of NGC 1068, which will aid us in the analysis of the several observed polarization features introduced in Section 3. The polarized structures are discussed in the following order: 1) the northern ionization cone, 2) the southern ionization cone, and 3) the obscuring dusty structure.
We present the structure of the inner 100 parsecs of NGC 1068. It shows a complex geometry in the inner few parsecs. Through 5 and 8.5 GHz radio observations using the very long baseline array (VLBA), Gallimore et al. (2004) detected the location of the hidden active nucleus (labeled as S1 in their paper) as a parsec-sized (0.8 pc diameter) structure with a major axis at P.A. 105°. The emission of this component most likely indicates strong free-free emission from hot (1010 K) ionized gas. This structure is oriented nearly perpendicular to the inner radio jet axis, P.A. 12° (Gallimore et al., 1996). The kinematics of the maser clouds indicate that the masers are located in a non-Keplerian disk with inner radius 0.65 pc and outer radius 1.1 pc at a P.A. of 40° (Gallimore et al., 2001). Through MIR interferometric observations, Jaffe et al. (2004) estimated a warm (T 320 K) dust structure with a 2.1 pc thick and 3.4 pc diameter, surrounding a hot (T 800 K) compact (1 pc) structure. Additional MIR interferometric observation by Raban et al. (2009) suggested a hot (800 K) 1.35 pc long and 0.45 pc thick structure with a larger 3 4 pc and warm (T 300 K) dust structure at a P.A. = 42°. These authors suggested that the dust and the maser disk are co-spatial, where the dusty torus begins at the outer edge of the maser disk. These results are consistent with the established relationship between the water maser excitation and warm (600 K) molecular dust in AGNs (Neufeld et al., 1994). Where the maser disk ends and the dusty torus structure begins may be more a question of semantics rather than a true physical boundary. Another implication from their results is that as the masers are seen edge-on, so is the torus, consistent with the classification of NGC 1068 as a Type 2 AGN. A sketch of the inner few parsecs of NGC 1068 is shown in Figure 7-left. Note that this figure is a detailed sketch of the central parsec of NGC 1068 adapted from fig. 6 in Lawrence & Elvis (2010). Based on the geometry of the inner few parsecs, misalignments between these structures are found. Although misaligned structures on similar spatial scales have been observed in other AGNs (i.e. Miyoshi et al., 1995; Wilson et al., 1995; Greenhill et al., 1997; Hagiwara et al., 1997; Trotter et al., 1998), it is not the scope of this paper to investigate these, and we refer to the reader to the discussion in Lawrence & Elvis (2010).
At larger scales (10 pc), NGC 1068 displays a variety of structures. Most of the radiation emerges from the interaction of the radio jet with material in the ionization cones partially extinguished by the disk of the galaxy and/or the NIR bar. The inner few parsecs of the ionization cone are roughly aligned with the radio jet in the N-S direction. At 24 pc in the northern ionization cone, the radio jet changes direction when it interacts with the molecular cloud (i.e. North Knot) (e.g. Gallimore et al., 2001). The radio-jet emission between the central few parsecs and the molecular cloud is dominated by synchrotron emission, while the jet-molecular cloud interaction radiates at all wavelengths. After this interaction, the radio jet expands and is roughly aligned with the extended northern ionization cone on a 50100 pc scale. The radio jet in the southern ionization cone does not change direction and moves away from our LOS. The radio jet and molecular cloud are extinguished by the NIR bar with an extension of 1.92 kpc at a PA 48° (Scoville et al., 1988; Schinnerer et al., 2000; Emsellem et al., 2006). Figure 7-right outlines the large scale structures.
5.1 The northern ionization cone
We analyze here the polarization mechanism from the northern extended polarization feature, and the North knot. As we noted in the introduction, NGC 1068 has a NIR bar with an extension of 32 arcsec (1.92 kpc) at a P.A. = 48° (Scoville et al., 1988; Schinnerer et al., 2000; Emsellem et al., 2006). In this region, the optical (Scarrott et al., 1991) and, J and H (Packham et al., 1997) polarization shows a P.A. of polarization consistent with the orientation of the NIR bar. This result is most easily interpreted as polarization arising from extinguished starlight passing through aligned dust grains in the NIR bar. Dichroic absorption dominates the polarization in the NIR bar, suggesting that the dust grains are aligned by the galactic magnetic field at P.A. = 48°. The measured P.A. of polarization of 44° in the northern ionization cone is consistent with the measured P.A. of polarization in the optical, J, and H bands. This result suggests that the dichroic absorption mechanism may dominate from the optical to the MIR in the NIR bar at the northern ionization cone.
To understand the polarization of the North knot and its relation with the extended polarization feature, we further compare the degree of polarization map at 8.7 m, and maps of the emission lines Paschen- (Pa) at 1.875 m (Fig. 8-middle) and [\ionOiii] at 0.5007 m (Fig. 8-right). We obtained archival images of NGC 1068 at Pa (ID: 7215, observed on 1998 Feb. 21) and [\ionOiii] (ID: 5754, observed on 1995 Jan. 17) emission lines using HST. We found that the MIR polarization is spatially coincident with the Pa (Fig. 8-middle) and [\ionOiii] (Fig. 8-right) emission lines. Pa line emission can be interpreted as a tracer of young (30 Myr) star formation or gas emission. Storchi-Bergmann et al. (2012) found an episode of young (30 Myr) stellar population in a ring-like structure at 100 pc from the nucleus, coincident with the ring of molecular gas detected in the 2.12 m warm molecular hydrogen line and in cold molecular gas detected by ALMA (García-Burillo et al., 2014). We found that the morphology of this young stellar population and cold molecular gas resembles neither the Pa line emission nor the MIR features in the northern ionization cone. Additionally, we found no evidence of 11.3 m PAH feature in the northern ionization cone in our spectropolarimetric observations (Fig. 4-left), consistent with MIR spectroscopic observations by Mason et al. (2006). This comparison between Pa line emission with the degree of polarization suggests that the Pa line emission is tracing gas photoionized by the AGN and/or gas excited by shocks in the ionization cones. This result is further supported by the spatial correspondence of Pa and [\ionOiii] emission lines, as the latter is a good tracer of the gas emission in the narrow line regions (e.g. Evans et al., 1991; Emsellem et al., 2006; Kraemer et al., 2015). Based on these results, we interpret the MIR polarization of the northern ionization cone as arising from extinguished radiation from dust and gas heated by the AGN (e.g. Crenshaw & Kraemer, 2000; Kraemer & Crenshaw, 2000a; Das et al., 2007) and the interaction of the jet (e.g. Gallimore et al., 1996) with dust and gas in the northern ionization cone passing through the aligned dust grains in the NIR bar producing the uniform P.A. of polarization in the northern ionization cone. The fact that the P.A. of polarization is not perfectly aligned along the P.A. of the NIR bar can be explained as the radiation from the jet-molecular cloud interaction being polarized with a slightly different P.A. of polarization. Variations of the S/N in the northern regions can also explain the 10° fluctuations of the P.A. of polarization.
Based on the model presented in Section 4, the dust grains in the northern ionization cone are different from those in the interstellar medium (ISM) in the Milky Way. This confirms the spectropolarimetric observations of Aitken et al. (1984), who also found a relatively featureless polarization spectrum in a 4.2 arcsec (252 pc) beam. They concluded that it was not produced by standard silicate grains. The North knot region produces the bulk of the nuclear polarization in NGC 1068, and our observations confirm that it mostly arises from non-silicate grains, despite the clear signature of silicate absorption in the total flux density spectrum (Fig. 4, left).
We considered other polarization mechanisms to explain the MIR polarimetric observations of the northern ionization cone. Scattering off electrons and/or dust grains can be ruled out due to the uniform P.A. of polarization in this region; there is no evidence of a centro-symmetric pattern that would be expected for scattering of radiation from a nuclear source. We considered also the possibility of the jet-molecular cloud interaction as a dust grain alignment mechanism in the northern ionization cone. For this study, we obtained archival images of NGC 1068 at 8 GHz (Observer: AC467, observed on 1999 Sep. 08) using the Very Large Telescope (VLA). We plot the degree of polarization map at 8.7 m and radio emission at 8 GHz in Figure 6, left. If the jet-molecular cloud interaction dominated in the northern ionization cone, we would expect the magnetic field to be parallel to the interface of the jet-molecular cloud interaction (e.g. Drury, 1983). The magnetic field is compressed in the interface and increases in strength, resulting in better dust grain alignment in the northern ionization cone than in the ISM. As noted in Section 5, the jet bends at the position of the North knot, which would imply variations of P.A. of polarization of 30° as a consequence of the different interface between jet and surrounding molecular dust in the northern ionization cone. Despite the registration uncertainties of 0.1-0.3 arcsec in Fig. 6-left, the uniform P.A. extends further (1.5 1.0 arcsec, i.e. 90 60 pc) than the radio jet emission. These observations suggest that the jet-molecular cloud interaction cannot be the dominant polarization mechanism in the whole northern ionization cone.
5.2 The southern ionization cone
We present here an analysis of the southern ionization cone and its Southern feature. The optical and NIR spectral lines and continuum emission are much fainter to the South of the core of NGC 1068 (Macchetto et al., 1994; Bruhweiler et al., 2001; Thompson et al., 2001, see also Fig. 6), with a deeper silicate feature, , within 1 arcsec SSW of the core (Mason et al., 2006). To show these results, we produced (Fig. 9) a NIR colormap of NGC 1068 using archival HST/NICMOS images, specifically F110W and F160W (ID: 7215, observed on 1998 Feb. 21). Next, we modeled the PSF at each filter using TinyTim (Krist et al., 2011), and then the PSFs were subtracted from the HST/NICMOS images. Finally, the F110W / F160W colormap (Fig. 9) was produced. Fig. 9 shows larger amount of host galaxy dust to the South of the core of NGC 1068, suggesting that the southern ionization cone suffers from higher extinction than the northern ionization cone. This result is in agreement with the increase of the NIR polarization to the SW (Young et al., 1995; Packham et al., 1997; Simpson et al., 2002; Lopez-Rodriguez et al., 2015), interpreted as an additional scattering medium in this region. In addition, the dusty disc of the galaxy has an inclination of 29° (García-Gómez, Athanassoula & Barberà, 2002), screening the southern but not the northern ionization cone (e.g. Packham et al., 1997; Bock et al., 1998; Kraemer & Crenshaw, 2000b). Thus, the southern ionization cone is more extinguished than the northern one, as is also suggested by the deeper silicate feature (Mason et al., 2006).
We found that the MIR polarization of the southern ionization cone is spatially coincident with the Pa (Fig. 8-middle) and [\ionOiii] (Fig. 8-right) emission lines. The P.A. of polarization (Fig. 3-bottom) is 30° along the Pa emission (Fig. 8-middle), whereas a P.A. of polarization of 10° is observed where not spatial coincident with this emission line is found. For those regions where MIR polarization and Pa emission are spatially coincident, we interpret the observed polarization in the southern ionization cone as dust directly heated by the jet; this dust emission passes through both the NIR bar and the disc of the galaxy. For those regions where MIR polarization and Pa emission are not spatially coincident, the polarization arises from dichroic absorption in the galaxy. As the southern ionization cone is affected by several extinction components, its P.A. of polarization differs from that observed in the northern regions.
As in the northern regions, scattering off electrons and/or dust grains can be ruled out, as there is no evidence of the centro-symmetric pattern that would be expected for scattering of radiation from a nuclear source. If the radio jet were responsible for the dust grain alignment, then a constant P.A. of polarization would be expected. Such a uniform polarization pattern is however not observed, and thus polarization arising from dust emission of aligned dust grains by the radio jet can be ruled out.
5.2.1 The Southern feature
Based on the model presented in Section 4.2, we interpret the polarization of the Southern feature as a polarized dust emission component passing through a large concentration of silicates as well as the dusty disc of the galaxy along our LOS. The polarized dust emission component in the Southern feature can arise from 1) magnetically aligned dust grains directly heated by the jet in the southern ionization cone close to the AGN, or 2) aligned dust grains in the obscuring dusty structure surrounding the central engine (Section 5.3). As the core is included within the aperture used to perform the measurements of the Southern feature, we cannot distinguish between dust grains in the southern ionization cone and those grains associated with the circum-nuclear dusty obscuring structure. Unfortunately, to obtain polarimetric measurements in an aperture further south of the nucleus to avoid the core of NGC 1068 produced low S/N spectra making it difficult to interpret the polarization.
As noted in Section 4, a prominent peak at wavelengths 10 m would be present in the degree of polarization if silicate grains were present. It is interesting that the polarization of the Southern feature can be reproduced with silicate emission/absorption, while the North knot does not show such silicate feature in the polarization spectrum. This suggests different dust composition within the central 120 pc of NGC 1068. If the interaction of the jet with the dust in the northern ionization cone is changing the dust properties, then the observed silicate feature of the Southern feature may favour the observed polarization arising from dust grains in the obscuring dusty structure surrounding the central engine of NGC 1068. We study the contribution of the obscuring dusty structure in Section 5.3.
5.3 The obscuring dusty structure
Although the MIR total flux density peaks at the position of the AGN (i.e. Braatz et al., 1993; Cameron et al., 1993; Lumsden et al., 1999; Bock et al., 1998; Alloin et al., 2000; Bock et al., 2000; Tomono et al., 2001; Jaffe et al., 2004; Galliano et al., 2005; Mason et al., 2006; Tomono et al., 2006; Raban et al., 2009; López-Gonzaga et al., 2014), a very low polarization with an upper-limit of 0.3 per cent at this location is detected within the 813 m wavelength range (Fig. 2 and 3). We present here an interpretation of the MIR polarization of the obscuring dusty structure of NGC 1068 in the context of the Clumpy666Clumpy models can be found at: www.clumpy.org torus models (Nenkova et al., 2002, 2008a, 2008b). Clumpy models the obscuring dusty structure as an optically and geometrically thick torus surrounding the central engine of an AGN, and comprised of dusty clouds.
Lopez-Rodriguez et al. (2015) interpreted the NIR polarization of NGC 1068 in terms of the level of magnetization through the estimation of the thermal-to-magnetic pressure ratio in the clumps. Specifically, the NIR polarization they measured is determined in the outermost material of lower optical depth in the clumps. Based on these results, and the fact that the absorptive/emissive polarization components should arise from the same aligned dust grains, but with orthogonal polarization angles, we are able to give an interpretation of the IR polarization. The NIR polarization arises from the passage of NIR radiation from directly illuminated clumps through the low-density dusty regions of the clumps with a visual extinction of A. The same aligned dust grains emit at MIR wavelengths, and this radiation passes through lower extinction, A, than the NIR photons. The lower the extinction, the higher the emissive intrinsic polarization component, i.e. P P. Since our observations do not show a MIR polarization at the core, and from our polarization model, the minimum visual extinction that the MIR radiation passes through is A 8 mag. For lower A, a degree of polarization higher than 0.1 per cent is expected (Fig. 6), which would be detectable in our observations. Thus, the model of emissive polarization arising from aligned dust grains in the clumps is consistent with the non-detection of MIR polarization at the core of NGC 1068. Based on these results, Figure 10 shows the possible configuration for a clump compatible with the observations.
Although the Clumpy models assume only discrete dust clumps with no inter-clump medium, an inter-clump medium should be considered as a more physical realistic model. We here put constraints on the extinction of the inter-clump medium to be used in two-phase clumpy torus models. Assuming a magnetohydrodynamical framework in the obscuring dusty structure of NGC 1068 (Lopez-Rodriguez et al., 2015), a global magnetic field at the same location of the obscuring dusty material aligns the dust grains in both the inter-clump medium and in the clumps. In this scheme, the IR polarization can arise from the passage of radiation through aligned dust grains in the clumps, as described above, and/or in the the inter-clump medium. To obtain an upper-limit on the extinction of the inter-clump medium, we assume that the clumps are optically thick with uniform density, then the aligned dust grains in the inter-clump medium would be responsible for the observed dichroic polarization at IR. Specifically, the IR radiation from the clumps passes through the aligned dust grains in the inter-clump medium. We now define the ratio of the optical depth of the clump, , and inter-clump medium, , with the contrast parameter C = /. C is a positive quantity, with C = 1, describing the obscuring dusty material in a smooth distribution of dust, where large values of C will enhance the clumpy distribution of dust grains. An optical depth of the clumps of 49 was took from Alonso-Herrero et al. (2011) using Clumpy models to fit the nuclear IR spectral energy distribution (SED) of NGC 1068. Based on our polarization model, the minimum visual extinction to be compatible with the observations is A 8 mag, or 8.7. Taking this value as the extinction of the inter-clump medium, we obtain a contrast of C = 5.6 at V-band. This result favors the clumpy distribution of dust versus a smooth-density obscuring material. The clumpy models (Stalevski et al., 2012; Siebenmorgen et al., 2015) including a two-phase medium are crucial to generate SED based on the MIR polarimetric observational constraints. However, the development of these models is beyond the scope of this paper.
We presented sub-arcsecond resolution MIR (7.513 m) imaging- and spectro-polarimetric observations of NGC 1068 obtained with CanariCam on the 10.4-m GTC. Using the wide MIR wavelength coverage, we were able to disentangle several polarized structures in the complex 3 2 arcsec (180 120 pc) central region through their polarization signatures. At all wavelengths, we found:
Northern ionization cone: the most prominent feature is the uniform P.A. of polarization, 44°, with an extension of 1.5 1.0 arcsec (90 60 pc). Based on our polarization model, we found that the polarization arises from dust and gas emission in the ionization cone, heated by the AGN and jet, and further extinguished by aligned dust grains in the NIR bar. In this region, we found that the North knot, located at 0.4 arcsec (24 pc) from the core, is spatially coincident with the interaction of the jet and molecular cloud. The North knot shows a uniform P.A. of polarization and a slow decrease in the degree of polarization with increasing wavelength. No silicate feature is found in the polarization spectrum of the North knot, suggesting that the dust grains are different from those in the ISM.
Southern ionization cone: we found a polarized feature at 0.16 arcsec (9.6 pc) South of the core. Based on our polarization model, the Southern feature polarization arises from an emissive polarization component passing though a larger concentration of dust than in the northern ionization cone. For this emissive component, we cannot distinguish between 1) magnetically aligned dust grains directly heated by the jet in the southern ionization cone close to the AGN, and 2) magnetically aligned dust grains in the obscuring dusty structure surrounding the central engine. The polarization spectrum shows a prominent silicate feature, suggesting that different dust grains composition or sizes may be present in the northern and southern ionization cones.
Obscuring dusty structure: An upper limit of 0.3 per cent in the degree of polarization from the core is detected in our observations. Based on our polarization model, the expected polarization from extrapolating the NIR absorptive component is 0.1 per cent in the 813 m wavelength range. We note that while there could be a component of polarized emission from those dust grains, if they are sufficiently warm to emit in the MIR, the uncertainties in estimating any contribution are very large. We suggested that the MIR polarization arises from the passage of MIR radiation through aligned dust grains in the shielded edges of the clumps located at the innermost regions of the obscuring dusty structure (Fig. 10). Alternatively, an obscuring clumpy structure with an inter-clump medium, and with an optical depth contrast between the clumps and the inter-clump medium of C = 5.6 at V-band, can also reproduce the upper-limit of the degree of polarization in our observations.
The sub-arcsecond resolution polarization observations of NGC 1068 demonstrate that there are a number of different regions and mechanism operating within the central 10s of parsecs in this AGN. Similar complexity may occur in other AGNs, where the current observations cannot resolve these structures. Further sub-arcsecond MIR imaging- and spectro-polarimetric observations to several AGNs are crucial to find ordinary and/or extraordinary properties in AGN allowing us to refine and/or modify the AGN unified model.
It is a pleasure to acknowledge discussion with R. Mason. We would like to thank the anonymous referee for their useful comments, which improved the paper significantly. Based on observations made with the Gran Telescopio CANARIAS (GTC), installed in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofísica de Canarias, in the island of La Palma. E.L.R and C.P. acknowledge support from the University of Texas at San Antonio. C.P. acknowledges support from NSF-0904421 grant. A.A.-H. acknowledges financial support from the Spanish Ministry of Economy and Competitiveness (MINECO) under the 2011 Severo Ochoa Program MINECO SEV-2011- 0187. A.A.-H. acknowledges financial support from the Spanish Ministry of Economy and Competitiveness through grant AYA2012-31447, which is party funded by the FEDER program, P.E. from grant AYA2012-31277, and L.C. from grant AYA2012-32295. R.N. acknowledges support by FONDECYT grant No. 3140436. C.R.A. is supported by a Marie Curie Intra European Fellowship within the 7th European Community Framework Programme (PIEF-GA-2012-327934). E.P. acknowledges support from NSF grant AST-0904896.
- Aitken et al. (1984) Aitken D. K., Briggs G., Bailey J. A., Roche P. F., Hough J. H., 1984, Nature, 310, 660
- Aitken et al. (1986) Aitken D. K., Briggs G. P., Roche P. F., Bailey J. A., Hough J. H., 1986, MNRAS, 218, 363
- Aitken et al. (1989) Aitken D. K., Smith C. H., Roche P. F., 1989, MNRAS, 236, 919
- Aitken et al. (2004) Aitken D. K., Hough J. H., Roche P. F., Smith C. H., Wright C. M., 2004, MNRAS, 348, 279
- Alloin et al. (2000) Alloin D., Pantin E., Lagage P. O., Granato G. L., 2000, A&A, 363, 926
- Alonso-Herrero et al. (2011) Alonso-Herrero A., et al., 2011, ApJ, 736, 82
- Antonucci (1993) Antonucci R., 1993, ARA&A, 31, 473
- Antonucci & Miller (1985) Antonucci R. R. J., Miller J. S., 1985, ApJ, 297, 621
- Antonucci et al. (1994) Antonucci R., Hurt T., Miller J., 1994, ApJ, 430, 210
- Bailey et al. (1988) Bailey J., Axon D. J., Hough J. H., Ward M. J., McLean I., Heathcote S. R., 1988, MNRAS, 234, 899
- Bock et al. (1998) Bock J. J., Marsh K. A., Ressler M. E., Werner M. W., 1998, ApJ, 504, L5
- Bock et al. (2000) Bock J. J., et al., 2000, AJ, 120, 2904
- Braatz et al. (1993) Braatz J. A., Wilson A. S., Gezari D. Y., Varosi F., Beichman C. A., 1993, ApJ, 409, L5
- Bruhweiler et al. (2001) Bruhweiler F. C., Miskey C. L., Smith A. M., Landsman W., Malumuth E., 2001, ApJ, 546, 866
- Cameron et al. (1993) Cameron M., Storey J. W. V., Rotaciuc V., Genzel R., Verstraete L., Drapatz S., Siebenmorgen R., Lee T. J., 1993, ApJ, 419, 136
- Capetti et al. (1995) Capetti A., Axon D. J., Macchetto F., Sparks W. B., Boksenberg A., 1995, ApJ, 446, 155
- Cohen & Davies (1995) Cohen M., Davies J. K., 1995, MNRAS, 276, 715
- Crenshaw & Kraemer (2000) Crenshaw D. M., Kraemer S. B., 2000, ApJ, 532, 247
- Das et al. (2007) Das V., Crenshaw D. M., Kraemer S. B., 2007, ApJ, 656, 699
- Drury (1983) Drury L. O., 1983, Reports on Progress in Physics, 46, 973
- Emsellem et al. (2006) Emsellem E., Fathi K., Wozniak H., Ferruit P., Mundell C. G., Schinnerer E., 2006, MNRAS, 365, 367
- Evans et al. (1991) Evans I. N., Ford H. C., Kinney A. L., Antonucci R. R. J., Armus L., Caganoff S., 1991, ApJ, 369, L27
- Galliano et al. (2005) Galliano E., Pantin E., Alloin D., Lagage P. O., 2005, MNRAS, 363, L1
- Gallimore et al. (1996) Gallimore J. F., Baum S. A., O’Dea C. P., 1996, ApJ, 464, 198
- Gallimore et al. (2001) Gallimore J. F., Henkel C., Baum S. A., Glass I. S., Claussen M. J., Prieto M. A., Von Kap-herr A., 2001, ApJ, 556, 694
- Gallimore et al. (2004) Gallimore J. F., Baum S. A., O’Dea C. P., 2004, ApJ, 613, 794
- Gao et al. (2013) Gao J., Li A., Jiang B. W., 2013, Earth, Planets, and Space, 65, 1127
- García-Burillo et al. (2014) García-Burillo S., et al., 2014, A&A, 567, A125
- García-Gómez et al. (2002) García-Gómez C., Athanassoula E., Barberà C., 2002, A&A, 389, 68
- Greenhill et al. (1997) Greenhill L. J., Ellingsen S. P., Norris R. P., Gough R. G., Sinclair M. W., Moran J. M., Mushotzky R., 1997, ApJ, 474, L103
- Hagiwara et al. (1997) Hagiwara Y., Kohno K., Kawabe R., Nakai N., 1997, PASJ, 49, 171
- Imanishi & Ueno (2000) Imanishi M., Ueno S., 2000, ApJ, 535, 626
- Jaffe et al. (2004) Jaffe W., et al., 2004, Nature, 429, 47
- Knacke & Capps (1974) Knacke R. F., Capps R. W., 1974, ApJ, 192, L19
- Kraemer & Crenshaw (2000a) Kraemer S. B., Crenshaw D. M., 2000a, ApJ, 532, 256
- Kraemer & Crenshaw (2000b) Kraemer S. B., Crenshaw D. M., 2000b, ApJ, 544, 763
- Kraemer et al. (2015) Kraemer S. B., Sharma N., Turner T. J., George I. M., Crenshaw D. M., 2015, ApJ, 798, 53
- Krist et al. (2011) Krist J. E., Hook R. N., Stoehr F., 2011, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series. p. 0
- Lawrence (1991) Lawrence A., 1991, MNRAS, 252, 586
- Lawrence & Elvis (2010) Lawrence A., Elvis M., 2010, ApJ, 714, 561
- Lebofsky et al. (1978) Lebofsky M. J., Kemp J. C., Rieke G. H., 1978, ApJ, 222, 95
- Levenson et al. (2007) Levenson N. A., Sirocky M. M., Hao L., Spoon H. W. W., Marshall J. A., Elitzur M., Houck J. R., 2007, ApJ, 654, L45
- López-Gonzaga et al. (2014) López-Gonzaga N., Jaffe W., Burtscher L., Tristram K. R. W., Meisenheimer K., 2014, A&A, 565, A71
- Lopez-Rodriguez (2016) Lopez-Rodriguez E., 2016, MNRAS, 455, 2656
- Lopez-Rodriguez et al. (2015) Lopez-Rodriguez E., et al., 2015, MNRAS, 452, 1902
- Lumsden et al. (1999) Lumsden S. L., Moore T. J. T., Smith C., Fujiyoshi T., Bland-Hawthorn J., Ward M. J., 1999, MNRAS, 303, 209
- Macchetto et al. (1994) Macchetto F., Capetti A., Sparks W. B., Axon D. J., Boksenberg A., 1994, ApJ, 435, L15
- Mason et al. (2006) Mason R. E., Geballe T. R., Packham C., Levenson N. A., Elitzur M., Fisher R. S., Perlman E., 2006, ApJ, 640, 612
- Miyoshi et al. (1995) Miyoshi M., Moran J., Herrnstein J., Greenhill L., Nakai N., Diamond P., Inoue M., 1995, Nature, 373, 127
- Nenkova et al. (2002) Nenkova M., Ivezić Ž., Elitzur M., 2002, ApJ, 570, L9
- Nenkova et al. (2008a) Nenkova M., Sirocky M. M., Ivezić Ž., Elitzur M., 2008a, ApJ, 685, 147
- Nenkova et al. (2008b) Nenkova M., Sirocky M. M., Nikutta R., Ivezić Ž., Elitzur M., 2008b, ApJ, 685, 160
- Neufeld et al. (1994) Neufeld D. A., Maloney P. R., Conger S., 1994, ApJ, 436, L127
- Packham et al. (1997) Packham C., Young S., Hough J. H., Axon D. J., Bailey J. A., 1997, MNRAS, 288, 375
- Packham et al. (2005) Packham C., Hough J. H., Telesco C. M., 2005, in Adamson A., Aspin C., Davis C., Fujiyoshi T., eds, Astronomical Society of the Pacific Conference Series Vol. 343, Astronomical Polarimetry: Current Status and Future Directions. p. 38
- Packham et al. (2007) Packham C., et al., 2007, ApJ, 661, L29
- Packham et al. (2008) Packham C., Mason R. E., Boreman G. D., 2008, Optical Engineering, 47, 126401
- Pier & Krolik (1992) Pier E. A., Krolik J. H., 1992, ApJ, 401, 99
- Poncelet et al. (2007) Poncelet A., Doucet C., Perrin G., Sol H., Lagage P. O., 2007, A&A, 472, 823
- Raban et al. (2009) Raban D., Jaffe W., Röttgering H., Meisenheimer K., Tristram K. R. W., 2009, MNRAS, 394, 1325
- Radomski et al. (2008) Radomski J. T., et al., 2008, ApJ, 681, 141
- Roche & Aitken (1984) Roche P. F., Aitken D. K., 1984, MNRAS, 208, 481
- Scarrott et al. (1991) Scarrott S. M., Rolph C. D., Wolstencroft R. W., Tadhunter C. N., 1991, MNRAS, 249, 16P
- Schinnerer et al. (2000) Schinnerer E., Eckart A., Tacconi L. J., Genzel R., Downes D., 2000, ApJ, 533, 850
- Scoville et al. (1988) Scoville N. Z., Matthews K., Carico D. P., Sanders D. B., 1988, ApJ, 327, L61
- Serkowski et al. (1975) Serkowski K., Mathewson D. S., Ford V. L., 1975, ApJ, 196, 261
- Siebenmorgen et al. (2015) Siebenmorgen R., Heymann F., Efstathiou A., 2015, preprint
- Simpson et al. (2002) Simpson J. P., Colgan S. W. J., Erickson E. F., Hines D. C., Schultz A. S. B., Trammell S. R., 2002, ApJ, 574, 95
- Smith et al. (2000) Smith C. H., Wright C. M., Aitken D. K., Roche P. F., Hough J. H., 2000, MNRAS, 312, 327
- Stalevski et al. (2012) Stalevski M., Fritz J., Baes M., Nakos T., Popović L. Č., 2012, MNRAS, 420, 2756
- Storchi-Bergmann et al. (2012) Storchi-Bergmann T., Riffel R. A., Riffel R., Diniz M. R., Borges Vale T., McGregor P. J., 2012, ApJ, 755, 87
- Telesco et al. (2003) Telesco C. M., et al., 2003, in Iye M., Moorwood A. F. M., eds, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series Vol. 4841, Instrument Design and Performance for Optical/Infrared Ground-based Telescopes. pp 913–922
- Thompson et al. (2001) Thompson R. I., Chary R.-R., Corbin M. R., Epps H., 2001, ApJ, 558, L97
- Tinbergen (2005) Tinbergen J., 2005, Astronomical Polarimetry. Cambridge University Press
- Tomono et al. (2001) Tomono D., Doi Y., Usuda T., Nishimura T., 2001, ApJ, 557, 637
- Tomono et al. (2006) Tomono D., Terada H., Kobayashi N., 2006, ApJ, 646, 774
- Trotter et al. (1998) Trotter A. S., Greenhill L. J., Moran J. M., Reid M. J., Irwin J. A., Lo K.-Y., 1998, ApJ, 495, 740
- Urry & Padovani (1995) Urry C. M., Padovani P., 1995, PASP, 107, 803
- Wardle & Kronberg (1974) Wardle J. F. C., Kronberg P. P., 1974, ApJ, 194, 249
- Watanabe et al. (2003) Watanabe M., Nagata T., Sato S., Nakaya H., Hough J. H., 2003, ApJ, 591, 714
- Weingartner & Draine (2001) Weingartner J. C., Draine B. T., 2001, ApJ, 548, 296
- Wilson et al. (1995) Wilson A. S., Braatz J. A., Henkel C., 1995, ApJ, 455, L127
- Young et al. (1995) Young S., Hough J. H., Axon D. J., Bailey J. A., Ward M. J., 1995, MNRAS, 272, 513