AN INFRARED COMPARISON OF TYPE-1 AND TYPE-2 QUASARS

We used the DAOPHOT package in IRAF to measure the 70 μm and 160 μm fluxes of our targets. For the 70 μm measurements, we examined each image individually and determined the best aperture and sky annulus that maximized the signal-to-noise ratio (S/N). We then applied aperture corrections from the Spitzer Space Telescope MIPS Data Handbook,⁷ and compared the measured flux to a 3σ limit obtained by measuring the noise in the image.

Each of our 12 targets was observed in the 160 μm MIPS band, though many of our targets were not detected at this wavelength. We chose an example image with a strong detection and determined the aperture and sky annulus sizes that maximized the S/N. We used those parameters while measuring the other images with bright detections. The same procedure was used for the images with targets that were not detected. We then compared the measured fluxes of each of the targets to the 3σ limit determined from the noise in the image. Five of the 12 targets have fluxes less than the 2σ noise limit, and we count them as non-detections. Five other targets have fluxes >2σ, and only one has 160 μm flux >3σ. The remaining target is discussed in the following.

The target SDSS 1711+5855 required special attention, because there was a bright object near the target. We fit a point-spread function (PSF) using several other bright objects in the image, then subtracted sources from the image leaving only our target. We then measured the flux of SDSS 1711+5855 from the subtracted image. This procedure was successful at 70 μm, but our attempts at removing the extra source at 160 μm resulted in an image with significant residuals. Therefore, we only report an upper limit measured from the combined flux of the objects at 160 μm. This upper limit was not used in our modeling procedure and analysis.

Table 1 shows the results of our MIPS flux measurements combined with data from the xFLS.

We present the Spitzer IRS spectra for our sample in Figure 1. The spectra show distinct differences. The spectra of the type-1 quasars tend to have fewer features and are more homogeneous than those of the type-2's. The type-1 objects show significant continuum emission near 3 μm, little PAH emission, and no significant silicate absorption. The type-2 spectra are more heterogeneous, showing a variety of silicate absorption depths and PAH strengths, consistent with previous observations of AGNs (e.g., Hao et al. 2007; Spoon et al. 2007; Zakamska et al. 2008) as well as starburst galaxies, ULIRGs and submillimeter galaxies (e.g., Brandl et al. 2006; Sajina et al. 2007; Lutz et al. 2008; Menendez-Delmestre et al. 2009). The type-2 quasars have on average double the PAH luminosity of the type-1 quasars (see Table 3).

Zoom In Zoom Out Reset image size

Spoon et al. (2007) classify spectra based on their 6.2 μm PAH equivalent width and 9.7 μm silicate absorption strength. Our type-2 spectra are similar to their 1B, 1C, and 2B classifications. These classifications are dominated by Seyferts, starbursts, and ULIRGs, respectively. Our type-1 quasars have spectra most similar to the 1A classification of galaxies by Spoon et al.

The variation of silicate absorption strengths in the type-2 spectra of our sample is also consistent with that seen by Zakamska et al. (2008), who study a sample of optically selected type-2 quasars from the SDSS. Zakamska et al. find silicate absorption similar to some of our sample, including two objects that have more significant absorption than our sample (see Lacy et al. 2007). Four of their objects appear to have no silicate absorption, including one that may show silicate emission, consistent with our mid-IR-selected type-1 quasars. They also address the correlation of silicate absorption with axial ratio found by Lacy et al. (2007). Lacy et al. interpreted this correlation to indicate that the silicate dust resides in the host galaxy, and not the nuclear region. The objects of Zakamska et al. (2008) show no correlation, and they propose this could be a difference in samples due to selection biases based on luminosity.

We modeled the SEDs of our sample from the optical out to the far-IR. We used available SDSS and 2MASS photometry, IRAC bands 1 and 2 (3.6 μm and 4.5 μm, respectively) photometry (Lacy et al. 2005) and our Spitzer IRS mid-IR spectra and MIPS photometry. This additional photometry used in the modeling procedure is presented in Table 2. The modeling code is described by Sajina et al. (2006). Modifications to the code to enable modeling of the type-2 objects in our sample were briefly described by Lacy et al. (2007). Here, we provide a fuller description of the modeling, and the additional modifications required to fit the type-1 objects. The modeling is phenomenological in nature due to the current lack of a good physical understanding of the nature and geometry of the dust emitting regions, particularly in the mid-IR, thus many of the parameters do not have a direct physical interpretation. Nevertheless, we considered the effort worthwhile as it allows an empirical breakdown of the contributions of starlight, hot dust emission from the AGN, and cooler dust emission.

The type-2 SEDs were modeled using the following components. (1) A single stellar population from Bruzual & Charlot (2003) with an age of 5 Gyr to represent the evolved stellar population. (2) A power-law component for the mid-IR emission, with an exponential cutoff at short wavelengths to represent dust sublimation, and a Fermi function cutoff at long wavelengths designed to match the SED shape of the least star-forming object in the sample of type-2's, SST 1721+6012. The functional form used was

$$\begin{equation} L_{\rm AGN}=\frac{L^{0}_{\rm AGN}\nu ^{1-\alpha } e^{h \nu / k_B T_{\rm sub}}}{e^{(\nu - \nu _{\rm hcut})/w}+1} \end{equation} \tag{ 1 }$$

where the normalization, L⁰_AGN, power-law index α and T_sub (a proxy for the sublimation temperature) are allowed to vary, and ν_hcut and w were fixed at 0.11 × 10¹⁴ Hz and 0.017 × 10¹⁴ Hz, respectively (h and k_B are the Planck and Boltzman constants, respectively). The Galactic Center extinction curve of Chiar & Tielens (2006) was then applied to this continuum to fit any silicate absorption feature. (3) A warm dust component to represent the small grain emission from H ii regions, represented by a power law with fixed high and low frequency cutoffs:

$$\begin{equation} L_{\rm SG} = L^{0}_{\rm SG}\nu ^{1-\gamma } e^{-\nu /\nu _{\rm sgl}} e^{-\nu _{\rm sgh}/\nu }. \end{equation} \tag{ 2 }$$

This component is poorly constrained in most of the fits, so γwas fixed at a typical value of two (e.g. Sajina et al. 2006). The high and low frequency cutoffs, ν_sgl and ν_sgh were set to 0.062 × 10¹⁴ Hz and 0.17 × 10¹⁴ Hz, respectively. (4) a thermal modified blackbody model for the far-infrared emission:

$$\begin{equation} L_{\rm FIR} = \frac{L^0_{\rm FIR} \nu ^{3+\beta }}{e^{h\nu /k_{\rm B}T}-1}. \end{equation} \tag{ 3 }$$

For most objects, we fixed the temperature of this component, T, to 45 K, because of the uncertainty in our 160 μm photometry. β was fixed at 1.5. In addition, the PAH model described in Lacy et al. (2007) was added when evidence of PAH emission was seen, and SST 1721+6012 had a silicate emission profile added, whose shape was approximated by that of the extinction curve.

For the type-1 quasars, component (1) was replaced by an SDSS composite quasar spectrum. As the composite of Vanden Berk et al. (2001) suffers from noticeable host galaxy contamination at the long wavelength end, we constructed a new composite by subtracting the continuum from the composite of Vanden Berk et al., and adding the residual emission line composite to a new continuum constructed using line-free SED points from the composite of Richards et al. (2006).

Most type-1 objects also required an extra near-infrared component to match the SEDs. This "very hot" component was modeled as a 1000 K modified blackbody (Equation (3) with T = 1000). The physical origin of this component is unclear, but it is most likely due to hot dust close to the sublimation radius (e.g., Netzer et al. 2007). Gallagher et al. (2007) show that the strength of this component correlates with quasar luminosity. We do not find this to be true for our sample, but note that Gallagher et al. (2007) have a much larger sample size spread over a range of luminosities.

The modeling results for the type-1 objects are presented in Figure 2, and the results for the type-2 objects are presented in Figure 3. Several differences between the type-1 sample and the type-2 sample are immediately apparent from the fitting results. The type-1 objects have a strong optical quasar component that can extend significantly into the mid-IR. The 1000 K component is significant in all six type-1's and also contributes to the mid-IR. Three of the six type-1 object SEDs showed no evidence of the warm small grain dust power-law emission. The type-2 object SEDs are dominated in the mid-IR by hot dust with no contribution from the optical stellar light and no evidence for the 1000 K component. Also, present in the mid-IR portion of the type-2 objects is a variety of PAH emission and silicate absorption. Five of the six type-2 objects have significant contributions to the mid and far-IR from the warm small grain dust power-law. SST 1721+6012 was the only type-2 not fit well with this component.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We note that the results of the fitting procedure for the object SST 1724+5915 underestimate both the observed amount of PAH emission and the far-IR emission. This is likely due to the large amount of silicate absorption also present in the spectrum of the object. In particular, the 9.7 μm absorption feature can affect the shape of the 7.7 μm PAH feature, making it difficult to fit with an empirical template. We include the data in our subsequent analysis, but note that star formation rates derived from the far-IR emission and the PAH emission likely underestimate the true star formation of this object (see Section 4.2).

We averaged the modeled type-1 SEDs and type-2 SEDs to highlight the differences between the two samples. We first normalized the SED of each target at 24 μm to ensure each object had equal weight. By choosing to normalize at 24 μm we can compare the relative amounts of hot and cool dust in the averaged SEDs. Though at 24 μm, extinction is still present, it is significantly reduced compared to shorter wavelengths. It is clear from Figure 4 that on average type-2 objects show more PAH emission, more silicate absorption and more far-IR emission than type-1 objects. In addition, while the very hot 3 μm bump is clear in the type-1 spectrum, no bump is seen in the type-2 spectrum. We note the caveat that the type-1 and type-2 objects could have equivalent amounts of PAH emission if that emission were significantly diluted by a strong mid-IR continuum in type-1 objects.

Zoom In Zoom Out Reset image size

The important question is if these observed differences could arise solely due to orientation effects. In that scenario, the optical through mid-IR light is absorbed and re-radiated in the far-IR. To test this, we integrated the optical through far-IR fluxes of the normalized and averaged spectra and found them to be identical. These results are consistent with the orientation hypothesis, though the geometry and distribution of the obscuring dust cannot be proved from this simple test.

Furthermore, under the orientation hypothesis, the dust tori of the type-1 and type-2 quasars should have the same intrinsic luminosities. Our modeling results show that the tori of the type-2 quasars are slightly more luminous than the tori of the type-1 quasars, though the difference is not statistically significant. This may be a result of our mid-IR selection technique, as those wavelengths are still somewhat affected by extinction. In the next section, we examine the PAH emission and the star formation contribution to the far-IR emission to gain more insight into whether the far-IR emission is different between the two objects.

We characterized and compared the star formation rates of the quasar host galaxies by comparing the PAH emission with the far-IR luminosity of our sample (Figure 5). We measured the far-IR luminosity from only the modeled warm (Equation (2)) and cold dust (Equation (3)) contributions to the SEDs (hereafter L'_FIR), excluding dust heated by the central AGN and emitting in the mid-IR. This definition of L'_FIR specifically excludes dust heated by the AGN while still measuring dust heated in star forming regions, even if the two emit at similar or overlapping wavelengths. This is in contrast to using a defined wavelength cutoff, which would measure dust heated by either source. However, it is still important to note that the far-IR emitting dust could also be heated by the AGN if that dust were at large enough distances from the nucleus. It is important to consider this potential AGN contribution to the far-IR emission when calculating SFRs.

Zoom In Zoom Out Reset image size

The PAH emission is an independent indicator of the star formation activity in the host galaxy. UV and X-ray emission from the AGNs destroy PAHs (Voit 1992), but soft UV emission from star formation excites PAHs into emission. In Figure 5, we find a correlation between PAH emission and far-IR emission similar to previous works (e.g., Schweitzer et al. 2006, Menendez-Delmestre et al. 2009). We also normalized the PAH emission to the far-IR emission, so we can compare the relative contribution of star formation to the far-IR emission. We measured the PAH luminosities by integrating our scaled templates, including the full profiles of each individual PAH feature in the template. This is different from many previous works (e.g., Schweitzer et al. 2006, Menendez-Delmestre et al. 2009), where only individual PAH features are measured, and definite wavelength cutoffs are applied. We find that the quasars as a whole have a fairly constant ratio across a range of far-IR luminosities. The median L_PAH/L'_FIR ratio for the type-1 objects of our sample is 0.012, and the median for the type-2 objects is 0.021. We note that the difference between the type-1 objects and type-2 objects is not significant considering the scatter in the sample. The median of our overall sample is 0.017. In comparison, Marshall et al. (2007) measure PAH luminosity using a similar method to ours and find that the starburst galaxies NGC 7714 and NGC 2623 have L_PAH/L_FIR ratios of 0.052 and 0.021, respectively. We cannot draw strong conclusions based on measurements of two objects, and note that many objects of our sample have ratios consistent with these objects. However, the average L_PAH/L_FIR from Marshall et al. is 0.037, higher than the median value of our sample. If taken at face value, these numbers indicate that some of the cool far-IR emitting dust is heated by AGN emission instead of stellar emission.

In Table 3, we present the integrated mid- to far-IR emission (8–1000 μm, hereafter L_IR). We find that nearly all of our targets have log (L_IR/L_☉) ⩾ 11.75. SST 1721+6012 is the only object with log (L_IR/L_☉) < 11.75. We also list the SFRs of our sample in Table 3. We calculated SFRs in two different ways. First, we converted the integrated warm and cold dust components of our model into SFRs using the Kennicutt (1998) calibration for starburst galaxies:

$$\begin{equation} {\rm SFR}\; (M_\odot \ {\rm yr}^{-1}) = 4.5 \times 10^{-44}L^{\prime }_{{\rm FIR}}\ ({\rm erg\ s}^{-1}). \end{equation} \tag{ 4 }$$

Although our objects have IR luminosities similar to those of ULIRGs, their SFRs are lower than those of typical ULIRGs. This is because some of the IR luminosity of our objects is due to AGN-heated dust, and we explicitly exclude this component when calculating SFRs. The SFRs of our sample are instead typical of LIRGs, which can range from 20–200 M_☉ yr⁻¹ (e.g., Sanders et al. 2003). We find that the type-2 objects show higher SFRs on average (92 M_☉ yr⁻¹) than the type-1 objects (67 M_☉ yr⁻¹). However, the scatter is large, and the object with the lowest SFR is a type-2, SST 1721+6012. We also independently calculated the SFR from the PAH emission. This can be done by predicting a far-IR luminosity based on the PAH luminosity. We integrated our scaled PAH template to measure the PAH luminosity of each object in our sample. We adopt an average ratio of L_PAH/L_FIR ∼ 0.037 based on the results of Marshall et al. (2007). Subsequently, for all objects in our sample except two, the SFR predicted from the PAH luminosity is lower than the SFR predicted by integrating the warm and cold dust components of the far-IR (L'_FIR). Again we found that the average type-2 quasar has a higher SFR (87 M_☉ yr⁻¹) than the average type-1 quasar (37 M_☉ yr⁻¹).